INTRODUCTIONTO PROCESSCONTROL II

Table 1-3. Meanings of Identification Letters

SUCCEEDINGLE1TERS

FIRST LE1TER
Measured or
Inilialing
Variable
A

Readoulo,
Passive
Funclion

Modijier

User'scooice

Voltage (EMF)

F
G

F1ow rate

User', choice

User',cboic.

Control

Conductivity
(electrical)
Density (mass)
or specitX:

Modi]ier

Alarrn

Analysis
Burner flame

Output
Function

Differential

gravity
Primary element
Ratio (fraction)

Glas.

Gaging
(dimensiooal)
Hand (manually
initiated)

Current

(electrical)
Power

Time or time

High

Indicat.
Controlstation

scheduIe
L

Level

Moisture or

humidity
User's cboice

User's cboic.

Oritice
(restriction)

Pressure or

Paint (test
connection)

Middle or inter-

mediate
User's cboice

vacuum
Q

Quantity or
event

Radioactiviy

Speed or

Low

Light (pilot)

Integrate

User'schoice

User's cboice

or

totalize
Recordor pnn!
Switch

SafelY

frequency
Transmit

Temperature

MuItivariable

Viscosity

w
x

Weight or force

Well

Unclassified

Unclassified

Multifunction

Multifunction

Multifunction

Valve, damper,
or louver
Unclassified

User's cboice

Relay or compute

Position

Drive, actuate
or unclassified
final control

element

Unclassified

FEEDBACK PROCESSCONTROL

Fig. }-8. Heatexchanger.

lated variable) and out of (the controlled variable) the processo The
closed-loop control system achieves this balance by measuring the demand and regulating the supply to maintain the desired balance over
time.
The basic idea of a feedback controlloop is most easily understood
by imagining what an operator would bave to do if automatic control
did Dot exist. Figure 1-8 shows a common application of automatic
control found in many industrial plants: a beat exchanger that uses
steam to beat cold water. In manual operation, the amount of steam
entering the beat exchanger depends on the air pressure to the valve,
which is set on the manual regulator. To control the temperature manually, the operator would watch the indicated temperature, and by comparing it with thedesired temperature, would open or close the valve to
admit more or less steam. When the temperature had reached the desired value, the operator would simply hold that output to the valve to
keep the temperature constant. Under automatic control, the temperatufe controller performs the same function. The measurement signat to
the controller from the temperature transmitter is continuously compared to the set-point signat entered into the controller. Based on a
comparlson of the signats, the automatic controller can tell whether
the measurement signat is above or below the set point and move the
valve accordingly until the measurement (temperature) comes to its
final value.
The simple feedback controlloop shown in Figure 1-9 illustrates
the four major elements of any feedback controlloop.

13

INTRODUCTION TO PROCESS CONTROL

+
~;~OL~:'-=- --.or
I MEASUREMENT

SUPPLV-*
FINAL ACTUATOR

CONTROLLED
VARIABLE

Fig. 1-9. Feedbackcontrolloop.

I. Measurementmust be made to indicate the current value of the

variable controlled by the loop. Commonmeasurementsused in
industry include ftow rate, pressure,level, temperature,analytical
measurementssuchas pH, ORPand conductivity; and manyotherg particular to specificindustries.
2. For every processthere mustbe afina! actuator that regulatesthe
supply of energyor materialto the processand changesthe measurementsignal. Most oftenthis is somekind of valve, but it might
also be a belt or motor speed,louver position, and so on.3.
The kinds of processesfound in industrial plants are as varied as
the materials they produce. They range from the commonplace,
suchas loops to control ftow rate, to the large and complex, such
as distillation columns in the petrochemicalindustry. Whether
simple or complex, they all consistof somecombinationof capacity resistanceand deadtime.4.
The last elementof the loop is the automaticcontroller. Its job is to
control the measurement.To "control" meansto keep the measurement at a constant, acceptablevalue. In this chapter, the
mechanismsinsidethe automaticcontroller will Dotbe considered.
Therefore, the principIesto be discussedmar be applied equally
well to both pneumaticand electronic controllers and to the controllers from any manufacturer.All automaticcontrollers use the
samegeneralresponses,althoughthe internalmechanismsand the
definitions given for these responsesmar differ slightly from one
another.

14 FEEDBACK PROCESSCONTROL

One basic concept is thai for automatic feedback control to exist,

the automatic controlloop musi be closed. This means thai information
musi be continuously passed around the loop. The controller musi be
able to move the val ve, the valve musi be able to atrect the measurement, and the measurement signal musi be reported to the controller. If
ibis path is broken at any point, the loop is said to be open. As soon as
the loop is opened-for example, when the automatic controller is
placed on manual-the automatic unit in the controller is no longer able
to move the valve. Thus, signats from the controller in response to
changing measurementconditions do noi atrect the valve and automatic
control does noi exist.
Feedback Control
Several principIes associated with feedback control can be observed by
considering a familiar control situation-adjusting the temperature of
water in a bathtub. This is obviously a manually controlled system.
One hand feels the water in the tub while the other manipulates the
inftow to reach the desired temperature. If a thermometer were used to
measure the temperature, greater accuracy would resulto Improved
measurement generally results in improved control.
The bathtub also illustrates the important effect of process capacity. Capacity (Figure 1-2)is a measure ofthe amount ofenergy it takes
to change a system a unit amount; thermal capacity is Btu/F, or the
amount of beat required to increase the temperature 1F. Since the
bathtub has a large capacity, it can be controlled in any of several
ways-by partially filling the tub with cold water, for example, and
then adding enough bot water to reach the desired temperature; or by
mixing the bot and cold to get the same resulto

Controlling

the Process

In performing the control function, the automatic controller uses the

difference between the set-point and the measurement signats to develop the output signat to the valve. The accuracy and responsiveness
of these signats is a basic limitation on the ability of the controller to
control the measurement correctly. If the transmitter does Dot send an
accurate signat, or ifthere is a lag in the measurementsignat, the ability
of the controller to manipulate the process will be degraded. At the
same time, the controller must receive an accurate set-point signat. In

INTRODUCTIONTO PROCESSCONTROL 15

controllers using pneumatic or electronic set-point signals generated

within the controller, miscalibration of the set-point transmitter will
develop the wrong value. The ability of the controller to position the
valve accurately is ret another limitation. If there is friction in the
valve, the controller mar Dot be able to move the valve to a specific
stem position to produce a specific flow, and this will appear as a
difference between measurement and set point. Repeated attempts to
position the valve exactly mar lead to hunting in the valve and in the
measurement. Or, if the controller is able only to move the valve very
slowly, the ability of the controller to control the process will be degraded. One way to improve the response of control valves is to use a
valve positioner, which acts as a feedback controller to position the
valve at the exact position corresponding to the controller output signal. However, positioners should be avoided in favor of volume boosters on fast-responding loops such as flow and liquid pressure.
For proper process control, the change in output Cromthe controller must be in such a direction as to oppose any change in the measurement value. Figure 1-10 shows a direct-connected valve to control

CONTROLLER
SPAN OF
MEASUREMENT

--rINLET FLOW

Lr

,-',
,,"

[[],
=

". ,'~~~""'-1

...,...,-

~}f~

-"..,"~ 2--=~.;:

-~-=-~_.=~.

f
OUTLET

_E/
I
100

I
50

I
O

PERCENT OPENING OF VALVE

FLOW

Fig. 1-10. In proportionalcontrol, the controllingvalve's positionis proportionalto the

controlled variable(level).

,:--=('i
~--=~
~I

16 FEEDBACK PROCESSCONTROL

level in a tank at midscale. As the level in the tank rises, the fioat acts to
reduce the fiow rate coming in. Thus, the higher the lquid level, the
more the fiow will be reduced. In the same way, as the level falls, the
fioat will open the valve to add more lquid to the tank. The response of
this system is shown graphically. As the level moves from O to 100
percent, the valve moves from fully open to fully closed. The function
of an automatic controller is to produce this kind of opposing response
over varying ranges. In addition, other responses are available to control the process more efficiently.
Selecting

Controller

Action

Depending on the action of the valve, increases in measurement mar

require either increasing or decreasing outputs for control. All controllers can be switched between direct and reverse action. Direct action
means that, when the controller sees an increasing signal from the
transmitter, its output will increase. For reverse action, increasing
measurement signals cause the controller output to decrease. To determine which of these responses is correct, an analysis of the loop is
required. The first step is to determine the action of the valve.
In Figure 1-1, for safety reasons the valve must shut if there is a
failure in the plant air supply. Therefore, this valve must be air-toopen, or fail-closed. Second, consider the effect of a change in measurement. For increasing temperature, the steam ftow to the beat exchanger should be reduced; therefore, the valve must close. To close
this valve, the signal from the automatic controller to the valve must

decrease. Therefore, this controller requires reverse, or increase/

decrease, action. Ifdirect action is selected, increasing signals from the
transmitter will result in a larger steam ftow, causing the temperature to
increase further. The result would be a runaway temperature. The
same thing will occur on any decrease in temperature, causing a falling
temperature. Incorrect selection of the action of the controller always
results in an unstable controlloop as soon as the controller is put into
automatic.
Assuming that the proper action is selected on the controller, how
does the controller know when the proper output has been reached? In
Figure 1-10, for example, to keep the level constant, a controller must
manipulate the ftow in to equal the ftow out. Any difference will cause
the level to change. In other words, the ftow in, or supply, must balance
the ftow out, or demandoThe controller performs its job by maintaining
this balance at a steady rate, and acting to restore this balance between
supply and demand whenever it is upset.

INTRODUCTION TO PROCESS CONTROL

17

U psets

There are three conditions that require different ftows to maintain the
level in the tank. First, if the position of the output hand valve is
opened slightly, more ftow leaves the tank, causing the level to fall.
This is a change in demand, and to restore balance, the inlet ftow valve
must be opened to supply a greater ftow rate. A second type of unbalanced condition is a change in the set point. Maintaining any other level
besides midscale in the tank causes a different ftow out. This change in
demand requires a different input valve position. The third type of
up set is a change in the supply. If the pressure output of the pump
increases, even though the inlet valve remains in the same position, the
increased pressure causes a greater ftow, which at first causes the level
to begin to rise. Sensingthe increased measurement, the level controller must close the val ve on the inlet to hold the level at a constant value.
In the same way, any controller applied to the beat exchanger shown in
Figure 1-1 must balance the supply ofheat added by the steam with the
beat removed by the water. The temperature remains constant if the
ftow of beat in equals the ftow of beat out.

Process

Characteristics

and Controllability

The automatic controller uses changes in the position of the final actuator to control the measurement signal, moving the actuator to oppose any change it sees in the measurement signal. The controllability
of any process depends on the efficiency of the measurement signal
response to these changes in the controller output. For proper control,
the measurement should begin to respond quickly, but then Dot change
too rapidly. Because of the tremendous number of applications of automatic control, characterizing a process by what it does, or by industry, is an almost hopeless task. However, all processes can be described by the relationship between their inputs and outputs. Figure
1-11 illustrates the temperature response of the beat exchanger when
the control valve is opened by manually increasing the controller output signal.
At first, there is no immediate response at the temperature indication. Then the temperature begins to change, steeply at first, then approaching a final, constant level. The process can be characterized by
the two elements of its response. The first element is the dead time, or
the time before the measurement begins to respond. In this example, a
delay arises because the beat in the steam must be conducted to the

18 FEEDBACK PROCESSCONTROL

STEAM FLOW

(MANUAL)
I

OUTLET

WATER

TEMPERATURE

L~,I"/"""'
TIME

Fig. l-Il. Responseof beatexchangerto stepupset.

water before it can affect the temperature, and then to the transmitter
before the change can be seen. Dead time is a function of the physical
dimensions of a process and such things as belt speeds and mixing
rates. Second, the capacity of a process is the material or energy that
must enter or leave the process to change the measurements-for example, the gallons necessary to change level, the Btu's necessary to
change temperature, or the standard cubic feet of gas necessary to
change pressure. The measure of a capacity is its response to a step
input. Specifically, the size of a capacity is measured by its time constant, which is defined as the time necessary to complete 63 percent of
its total response. The time constant is a function of the size of the
process and the rate of material or energy transfer. For this example,
the larger the tank and the smaller the ftow rate of the steam, the longer
the time constant. These numbers can be as short as a few seconds, or
as long as several hours. Combined with dead time, they define the time
it takes the measurement signal to respond to changes in the valve
position. A process will begin to respond quickly, but then not change
too rapidly, if its dead time is small and its capacity is large. In short,
the larger the time constant of capacity compared to the dead time, the
better the controllability of the processo

Controli er Responses

The first and most basic characteristico the controller responsehas

beenshownto be either direct or reverseaction. Oncethis distinction

-..TIME

INTRODUCTION TO PROCESS CONTROL

19

has been made, severa! types o responses are used to control a processoThese are (1) on/otI, two-position, control, (2) proportional action, (3) integral action (reset), and (4) derivative action.
On/off Control

On/off control is illustrated in Figure 1-12for a reverse-acting controller

and an air-to-close valve. An on/off controller has only two outputs,
either full maximum or full minimum. For this system, it has been
determined that, when the measurement falls below the set point, the
valve must be closed to cause it to increase. Thus, whenever the signal
to the automatic controller is below the set point, the controller output
will be 100 percent. As the measurement crosses the set point, the
controller output goes to O percent. This eventually causes the measurement to decrease, and as the measurement again crosses the set
point, the output goes to maximum. This cycle will continue indefinitely because the controller cannot balance the supply against the
load. This continuo us oscillation mar or mar Dot be acceptable, depending on the amplitude and length ofthe cycle. Rapid cycling causes
freqent upsets to the plant supply system and excessive valve wear.
The time of each cycle depends on the dead time in the process because
the dead time determines the time it takes for the measurement signal
to reverse its direction once it crosses 'the set point and the output of
the controller changes. The amplitude of the signal depends on how
rapidly the measurement signal changes during each cycle. On large
capacity processes, such as temperature vats, the large capacity causes

MEASUREMENT

100
% SIGNAL TO
VALVE

Fig. 1-12. On-off control for reverse-actingcontrollerand air-to-closevalve.

20 FEEDBACK PROCESSCONTROL

Fig. 1-13.Automatic controller with artificial signal.

a long time constant. Therefore, the measurement can change only

slowly. As a result, the cycle occurs within a very narrow band around
the set point, and this control mar be quite acceptable, if the cycle is
Dot too rapid. The on/off control is the one used most frequently in
commercial and domestic installations. However, if the process measurement is more responsive to changes in the supply, the amplitude
and frequency ofthe cycle begins to increase. At some point, this cycle
will become unacceptable and some Corm of proportional control will
be required.
In order to study the remaining three modes of automatic control,
open-loop responses will be used. Open-Ioop means that only the response ofthe controller will be considered. Figure 1-13shows an automatic controller with an artificial signal Croma manual regulator introduced as the measurement. The set point is introduced normally and
the output is recorded. With this arrangement, the specific controller
responses to any desired change in measurement can be observed.
Proportional

Action

Proportional response is the basis for the three-mode controller. If the

other two, integral and derivative are present, they are added to the
proportional response. "Proportional" means that the percent change
in the output of the controller is some multiple of the percent change in
the measurement.
This multiple is caUedthe "gain" ofthe controUer. For some controUers, proportional action is adjusted by such a gain adjustment,
while for others a "proportional band" adjustment is used. Both bave
the same purposes and effect. (See Appendix for table showing the
controUer adjustments from one manufacturer to another.) Figure 1-14
iUustrates the response of a proportional controUer from an inputl
output pointer pivoting on one ofthree positions. With the pivot in the
center between the input and the output graph, 100 percent change in

INTRODucnON TO PROCESSCONTROL 21

% OUTPUT

Fig. 1-14.Responseof proportionalcontroller, input to output, for three proportional

band values.

measurement is required to obtain 100percent change in output, or full

va1vetrave1. A controller adjusted to respond in this way is said to bave
a 100 percent proportiona1 band. When the pivot is moved to the righthand position, the measurement input wou1d need to change by 200
percent in order to obtain full output change from Oto 100percent. This
is c'alled a 200 percent proportiona1 band. Finally, if the pivot were in
the 1eft-hand position, and if the measurement moved over on1y 50
percent of the sca1e,the output wou1dchange over 100 percent of the
scale. This is called a 50 percent proportiona1 band. Thus, the smaller
the proportional band, the smaller amount the measurement must
change to cause full valve travel. In other words, the smaller the proportional band, the greater the output change for the same size measurement change. This relationship is illustrated in Figure 1-15. This
graph shows how the controller output will respond as a measurement

Fig. 1-15. ProportionaI diagram.

22 FEEDBACK PROCESSCONTROL

deviates from set paint. Each line on the graph represents a particular
adjustment of the proportional band. Two basic properties of proportional control can be observed from this graph:
1. For every value of proportional band, whenever the measurement
equals the set paint, the normal output is 50 percent.
2. Each value of the proportional band defines a unique relationship
between measurement and output. For every measurement value
there is a specific output value. For example, using the 100percent
proportional band line, whenever the measurement is 25 percent
above the set paint, the output from the controller must be 25
percent. The output from the controller can be 25 percent only if
the measurement is 25 percent above the set paint. In the same
way, whenever the output from the controller is 25 percent, the
measurement will be 25 percent above the set paint. In short, there
is one specific output value for every measurement value.
For any process controlloop, only one value of the proportional
band is the best. As the proportional band is reduced, the controller
response to any change in measurementbecomes increasingly greater.
At some paint, depending on the characteristic of each particular process, the response in the controller will be large enough to drive the
measurement back so far in the opposite direction as to cause constant
cycling of the measurement. This proportional band value, known as
the ultimate proportional band, is a lmit on the adjustment of the
controller in that loop. On the other hand, if too wide a proportional
band is used, the controller response to any change in measurement is
too small and the measurement is Dot controlled as tightly as possible.
The determination of the proper proportional band for any application
is part of the tuning procedure for that loop. Proper adjustment of the
proportional band can be observed by the response ofthe measurement
to an upset. Figure 1-16shows several examples ofvarying the proportional band for the beat exchanger.
Ideally, the proper proportional band will produce one-quarter amplitude damping, in which each hal cycle is one-half the amplitude of
the previous hal cycle. The proportional band that will cause onequarter wave damping will be smaller, thereby yielding tighter control
over the measured variable, as the dead time in the process decreases
and the capacity increases.
One consequence of the application of proportional control to the
basic controlloop is offset. Offset means that the controller will maintain the measurement at avalue different from the set paint. This is
most easily seen in Figure 1-10. Note that ifthe load valve is opened,

INTRODUCTIONTO PROCESSCONTROL 23

WATER FLOW
(LOAO)

i
I

1~
TEMPERATURE

-j/'\/,~"-",,,
I

,-~

PROPORTI ONA L BAND

TOOWIOE

CHANGE IN
UEMANO

B
C

TOO NARROW
CORRECT

TIME

Fig. 1-16. Examples of varying proportional band for beat exchanger.

ftow will increase through the valve and the val ve would bave to open.
But note that, because of the proportional action of the linkage, the
increased open position can be achieved only at a lowered level. Stated
another way, in order to restore balance between the ftow in and the
ftow out, the level must stabilize at avalue below the set point. This
difference, which will be maintained by the control loop, is called
offset, and is characteristic of the application of proportional-only control to feedback loops. The acceptability ofproportional-only control
depends on whether this offset can be tolerated. Since the error necessary to produce any output decreases with the proportional band, the
narrower the proportional band, the less the offset. For large capacity,
small dead time applications accepting a very narrow proportional
band, proportional-only control will probably be satisfactory, since the
measurement will remain within a small percentage band around the set
point.
If it is essential that there be no steady state difference between
measurement and set point under all load conditions, an additional
function must be added to the controller. This function is called integral
action (an older term is reset).
Integral

Action

(Reset)

The open-loop response of the integral mode is shown in Figure 1-17,

which indicates a step change in the artificial measurement away from
the set point at some instant in time. As long as the measurement
remains at the set point, there is no change in the output due to the
integral mode in the controller. However, when any error exists between measurement and set point, the integral action will cause the
output to begin to change and continue to change as long as the error

24 FEEDBACK PROCESSCONTROL

% MEASUREMENT

L_-

SET
paiNT

I
I
I

...riME
Fig. l-I?

Open-loop response oCintegral mode.

exists. This function, then, causes the output to change until the proper
output is achieved in order to hold the measurement at the set paint at
various loads. This response is added to the proportional response of
the controller as shown in Figure 1-17. The step change in the measurement first causes a proportional response, and then an integral
response, which is added to the proportional. The more integral action
there is in the controller, the more quickly the output changes due to
the integral response. The integral adjustment determines how rapidly
the output changesas a function oftime. Among the various controllers
manufactured, the amount of integral action is measured in one of two
ways-either in minutes per repeat, or the number of repeats per minute. For controllers measuring integral action in minutes per repeat,
the integral time is the amount of time necessary for the integral mode
to repeat the open-loop response caused by proportional mode, for a
step change in error. Thus, for these controllers, the smaller the integral number, the greater the action of the integral mode. On controllers
that measure integral action in repeats per minute, the adjustment indicates how many repeats of the proportional action are generated by the
integral mode in one minute. Table 12-1 (p. 300) relates the controller
adjustments from one manufacturer to another. Thus, for these controllers, the higher the integral number, the greater the integral action.
Integral time is shown in Figure l-IS. The proper amount of integral
action depends on how fast the measurement can respond to the additional valve trave 1 it causes. The controller must nat drive the valve
faster than the dead time in the process allows the measurement to
respond, or the valve will reach its lmits before the measurement can
be brought back to the set paint. The valve will then remain in its
extreme position until the measurement crosses the set paint, whereupon the controller will drive the valve to its opposite extreme, where it

INTRpDUCTION TO PROCESSCONTROL 25

% MEASUREMENT

~
,

:+-RT

% OUTPUT

(I/D)

SET paiNT

-+1

RT. RESETTIME MIN./RPT.

::::::""J--r-

b~I~

I
I

a. b

TIME

Fig. l-IS. Open-loopresponseoCproportional plus integralmodes.

will remain until the measurementcrosses the set point in the opposite
direction. The result will be an integral cycle in which the valve travels
from one extreme to another as the measurementoscillates around the
set point. When integral action is applied in controllers on batch processes, where the measurement is away from the set point for long
periods between batches, the integral may drive the output to its maximum, resulting in "integral wind-up." When the next batch is started,
the output will Dot corne off its maximum until the measurement
crosses the set point, causing large overshoots. This problem can be
prevented by including a "batch function" in the controller, a function
specifically designed to prevent "wind-up."

Derivative

Action

The third response found on controllers is the derivative mode.

Whereasthe proportional mode respondsto the size of the error and
the integral mode respondsto the size and time duration of the error,
the derivative moderespondsto how quickly the error is changing.ln
Figure 1-19,two derivative responsesare shown.The first is a response
to a stepchangeof the measurementaway from the set point. For a
step, the measurementis changinginfinitely fast, and the derivative
mode in the controller causesa considerablechangeor spike in the
output, which dies immediatelybecausethe measurement
hasstopped
changingarterthe step. The secondresponseshowsthe responseofthe
derivative mode to a measurementthat is changingat a constantrate.
The derivative output is proportionalto the rate of changeofthis error.
The greaterthe rate of change,the greaterthe output dueto the deriva-

...TIME

26 FEEDBACK PROCESSCONTROL

% MEASUREMENT

-_J==~

--~

"~!

-t__J-

% OUTPUT

(IlO)

+TIME J...

~
Fig. 1-19. Two derivative responses.

tive response. The derivative holds this output as long as the measurement is changing. As soon as the measurementstops changing, regardless owhether it is at the set point, above or below it, the response due
to derivative action will cease. Among all brands o controllers, derivative response is commonly measured in minutes, as shown in Figure
1-20. The derivative time in minutes is the time that the open-loop,
proportional-plus-derivative response, is ahead o the response due to
proportional action alone. Thus, the greater the derivative number, the
greater the derivative response. Changes in the error are the result o
changes in either the set point or the measurement, or both. To avoid a
large output spike caused by step changes in the set point, most modem controllers apply derivative action only to ehanges in the measurement. Derivative action in controllers helps to control processes with
especially large time constants. Derivative action is unnecessary on

""

% MEASUREMENT

DT' DERIVATIVE TIME, MINUTES

,,

DT""
I

:.:::~J
% OUTPUT

r-

::::::::::~'
.f"

PROPORTIONAL ONLY

PROPORTIONAL + DERIVATIVE

Fig. 1-20. Open-loop response oCproportional plus derivative modes.

INTRODUCTIONTO PROCESSCONTROL 27

processes thai respond fairly quickly to valve motion, and cannot be

used at all on processes with noise in the measurement signal, such as
flow, since the derivative in the controller will respond to the rapid
changes in measurement it sees in the noise. This will cause large and
rapid variations in the controller output, which will keep the valve
constantly moving up and down, wearing the valve and causing the
measurement to cycle.
As previously described, the response of an element is commonly
expressed in terms of a time constant, defined as the time thai will
elapse until an exponential curve reaches 63.2 percent ora step change
in input. Although the transmitter does nat bave an exact exponential
response, the time constant for the pneumatic temperature transmitter
and its associated thermal system is approximately 2 seconds.
The reaction curve shown in Figure 1-21 incorporates the response
of the beat exchanger and measuring system, and it represents the
signal thai will actually reach the controller. This curve indicates thai,
given a sudden change inside the beat exchanger, more than 30 seconds
will elapse before the controller receives a signal thai is a true representation of thai change. From the reaction curve, or characteristic, we
can determine the type of controller required for satisfactory control
under ibis difficult, but common, delayed response characteristic.
Selecting

the Controller

The beat exchanger acts as a small-capacity process; thai is, a small

change in steam can cause a large change in temperature. Accurate

Fig. 1-21.The processreactioncurve is obtainedby imposinga stepchangeat input.

28 FEEDBACK PROCESSCONTROL

regulation of processes such as this calls for proportional rather than

on/otI control.
Variations in water rate cause load changes that produce offset, as
described previously. Thus, the integral mode should also be used.
Whether or Dot to include the derivative mode requires additional
investigation of the process characteristic. Referring to the reaction
curve (Figure 1-19), notice that the straight line tangent to the curve at
the point of inftection is continued back to the 150For 66C (starting)
level. The time interval between the start of the upset and the intersection of the tangentialline is marked TA;the time interval from this point
to the point of inftection is TB.IfTB exceeds TA,some derivative action
will prove advantageous. If TBis less than TA,derivative action mar
lead to instability because of the lags involved.
The reaction curve of Figure 1-21 clearly indicates that some derivative will improve control action. Thus, a three-mode controller with
proportional, integral, and derivative modes satisfies the needs of the
beat exchanger processo
Figure 1-22shows the combined proportional, integral, and deriva-

Fig. 1-22. Open-loopresponseofthree-modecontroller.

INTRODUCTIONTO PROCESSCONTROL 29

tive responsesto a simulated' beat exchanger temperature measurement

thai deviates from the set point due to a load change. When the measurement begins to deviate from the set point, the tirst response from
the controller is a derivative response proportional to the rate of change
of measurement thai opposes the movement of the measurement away
from the set point. This derivative response is combined with the proportional response. In addition, as the integral mode in the controller
sees the error increase, it drives the valve further still. This action
continues until the measurement stops changing, at which point the
derivative response ceases. Since there is still an error, the measurement continues to change due to integral action, until the measurement
begins to move back toward the set point. As soon as the measurement
begins to move back toward the set point, there is a derivative response
proportional to the rate of change in the measurement opposing the
return of the measurement toward the set point. The integral response
continues, because there is still error, although its contribution decreases with the error. AIso, the output due to proportional is changing.
Thus, the measurement comes back toward the set point. As soon as
the measurement reaches the set point and stops changing, derivative
response again ceases and the proportional output returns to 50 percent. With the measurement back at the set point, there is no longer any
changing response due to integral action. However, the output is at a
new value. This new value is the reguli ofthe integral action during the
time thai the measurement was away from the set point, and compengates for the load change thai caused the original upset.

Conclusion

This chapter has describedthe responsesof a three-modecontroller

when it is used in the feedbackcontrol of industrial measurements.
The
readershouldbave a clear understandingof the following points:
1. In order to achieve automatic control, the control loop musi be
closed.
2. In order to maintainastable feedbackcontrolloop, the most important adjustmentto the controller is the selectionof the proper
action, either reverseor direct, on the controller. Properselection
of tbis action will causethe controller output to changein sucha
way thai the movementofthe valve will opposeany changein the
measurementseenby the controller.
3. The proper value of the settingsof the proportionalband, the integral mode, and derivative time dependson the characteristicsof

30 FEEDBACK PROCESSCONTROL

the processoThe proportional band is the basic tuning adjustment on

the controller. The more narrow the proportional band, the more
the controller reacts to changes in the measurement. If too narrow
a proportional band is used, the measurement cycles excessively.
If too wide a proportional band is used, the measurement will
wander and the offset will be too large.4.
The function of the integral mode is to eliminate offset. If too much
integral action is used, the res uIt will be an oscillation of the measurement as the controller drives the valve from one extreme to the
other. If too little integral action is used, the measurement will
retum to the set point too slowly.5.
The derivative mode opposes any change in the measurement. Too
little derivative action has no significant effect. Too much derivative action causes excessive response of the controller and cycling
in the measurement.

Questions

l-I. All control systemsthat fit into the usual patternare:

a. Open-Ioop
c. Closed-loop
b. Nonself-regulating
d. On/off
1-2. If operatingproperly,automaticcontrol will always:
a. Reducemanpower
b. Reducecosts
c. Make the processoperatemore uniformly
d. Decreasemaintenance
1-3. Automatic controllersoperateon the differencebetweenset point and
measurement,which is called:
a. Offset
c. Error
b. Bias
d. Feedback
1-4. A two-positioncontroller (on/oil) always:
a. Controls with a fixed offset
b. Controlsarounda point
c. Automaticallyadjustsits integraltime
d. Requiresprecisetuning
1-5. Gainand proportionalbandsare:
a. Reciprocallyrelated
b. Two differentcontrol modes
c. Adjusted independentlyof one another
d. Controllerfunctionscalibratedin time units

INTR9DUCTION TO PROCESSCONTROL 31

1-6. Whenwe adjustintegraltime in a controller:

a. We determineanRC time constantin the controller's internalfeedback
path
b. We adjustthe time it will take for integralto equalderivative
c. We setthe processtime constantso that it will alwaysequal1
d. What happensspecificallydependson the type of controller,pneumatic
or electronic
1-7. Match the following:
Controller, two-position-a.
Derivative -b.
Deviation-c.
Final-controllingelement-d.
Proportionalband
adjustment-f.
Regulatedby control valve -g.
Reset-h.
Set point-

Gain
Rate
Integral
Controller, on/otI
e. Valve
Desiredvalue
Manipulatedvariable
Error

1-8. A proportionalcontroller will bave an otIsetditIerencebetweenset

point and control point:
a. At all times
b. Equal to the proportionalbandsetting
c. That dependsuponprocessload
d. That will eventuallyvanish
1-9. If it were possiblefor a proportionalcontrollerto bavea true Opercent
proportionalband, the controller gainwould bave to be:
a. Unity
c. 100
b. O
d. Infinite
1-10. If the proportionalbandof the controlleris adjustedto minimum
possiblevalue,the control actionis likely to be:
a. On/otI
c. Excellent
b. With maximumotIset
d. Inoperative
I-Il. The following symbol @
representsa:
8. Flow rate controller
b. Fixed control point

appearsin an instrumentdiagram.It
c. Frequencyconverter
d. Final control element

1-12. With a proportional-onlycontroller if measurement

equalsset point, the
output will be:
8. O

c. 50 percent

b. 100percent

d. Impossibleto define

1-13. If in a proportional-plus-integral
controller measurement
is away from
the set point for a long period,the controller's output will be:

32 FEEDBACK PROCESSCONTROL

8. Oor 100percent,dependingon actionselected

b. Unknown
c. O

d. 100percent
1-14. In the modemcontroller,derivative actionis appliedonly to the:
8. Error
c. Setpoint
b. Measurement
d. Integral circuit
1-15. The functionof the integral(reset)modeis to:
8. Opposechangein measurement
b. Automaticallyadjustthe controller's gain
c. Eliminate offset
d. Stabilizethe controlloop

Process/Pressure Measuring
Process/Pressure
Instruments
Instruments

Pressure is a universal
universal processing
processing condition.
condition. It is also
also a condition
oflife
Pressure
conditionoflife
on this planet:
at the
an atmospheric
that
atmospheric ocean
ocean that
on
planet: we live at
the bottom
bottom of an
extends upward
upward for many
miles. This mass
mass of air has
has weight,
weight, and
and this
extends
manymiles.
weight
pressing downward
downward causes
causes atmospheric
atmospheric pressure.
pressure. Water,
Water, a funfunweight pressing
damental
necessityof life, is supplied
most of us
supplied to most
us under
under pressure.
pressure. In
damental necessity
the typical process
process plant,
plant, pressure
pressure influences
influences boiling
boiling point
point temperatemperathe
tures, condensing
temperatures, process
process efficiency,
efficiency, costs,
costs, and
and
tures,
condensingpoint
point temperatures,
other important
important factors.
measurementand
control of pressure,
other
factors. The
The measurement
and control
pressure, or
lack of it-vacuum-in
it-vacuum-in the
plantis critical. Instruments
lack
the typical process
process plant
Instruments
are
How these
are available
available to measure
measure a wide
wide range
range of pressures.
pressures. How
these instruinstruments function
chapter.
ments
function is the subject
subjectof this chapter.
What
Is Pressure?
Pressure?
What Is

Pressure
Pressure is force dividel
divided by the area over which it is applied. Pressure
Pressure
is often defined in terms of
of "head."
"head." For example, assume
assume that we bave
have
a water column 1 foot square
square and 23
23 feet tall. We
We want to find the
pressure in the bottom
boUom ofthe
mar be
ofthe column. The weight ofthe
ofthe column may
calculated by first finding the volume of water. This is the area
of the
area of
33

34

FEEDBACK PROCESSCONTROL

base multiplied by height, or 1 times 23 equals 23 cubic feet. Water

weighs 62.43 pounds per cubic foot. So the weight of 23 cubic feet will
be 23 times 62.43, or 1,435.89pounds. The area ofthe base is 1 square
foot, or 12 inches times 12 inches, or 144square inches. The pressure
equals 1,434.89 divided by 144 equals 9.9715, or approximately 10
pounds per square inch. In practice, we find that only the height ofthe
water confits. It mar be present in a small pipe or beneath the surface
of a pond. In any case, at a depth of 23 feet, the pressure will amo unt to
approximately 10 pounds per square inch. If in your home the water
pressure is 50 pounds per square inch and the system uses a gravity
feed, the water tank, or reservoir, holds the water at a height of 50
divided by 10, or 5 times 23 equals 115 feet above the point where the
pressure measurement is made. Head and pressure, then, mar mean
the same thing. We must be able to convert from one to the other. You
mar encounter reference to inches of mercury for pressure measurement. Mercury is 13.596times as heavy as an equal volume of water.
Therefore, ahead ofmercuryexerts apressure 13.596times greaterthan
an equivalent head of water. Because it is hazardous, mercury no
longer is used commonly in manometers.
The head or pressure terms cited thus far are called, collectively,
"gauge pressure." For example, if a tire gauge is used to check the
pressure in your automobile tires, it measures gauge pressure. Gauge
pressure makes no allowance for the fact that on earth we exist under a
head of air, or an atmosphere. The height of this head of air varies with
elevation, and also to some degree with weather conditions. If rou ride
an elevator from the bottom to the top floor of a tall building, rou will
likely feel your ears "pop." This is caused by the change in atmospheric pressure.
A simple method of measuring atmospheric pressure would be to
take a length of small diameter (0.25 inches) glass tubing about 35
inches long, sealed at one end. Fill the tube entirely with mercury and
temporarily seal the end. Invert this end into a deep dish of mercury
and remo ve the seal. The result will be a column of mercury as shown
in Figure 2-1 with some space remaining at the top. Atmospheric
pressure on the surface of the exposed mercury will balance the height
of mercury in the tube and prevent it from running out of the tube. The
height of the mercury above the level in the dish is, then, a measure of
atmospheric pressure. At sea level, this would amount to approximately 29.9 inches, or 14.7pounds per square inch. When the etfect of
the atmosphere is included in our measurement, we then must use
absolute pressure (gauge pressure plus atmospheric pressure).

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 35

Fig. 2-1. Mercurybarometer.

Units ot Measurement

Every major country has adopted its own favorite units of measurement. The United States has traditionally employed the English system. However, international trade has made it necessaryto standardize
units of measurement throughout the world. Fortunately, during this
standardization, there has be en rationalization ofthe measurementsystem. This has led to the adoption of the System lnternatonal d'Untes
(SI), a metric system of units. The force of common usage is so strong
that the familiar English system will undoubtedly persist for many
years, but the changeover is definitely underway. The time will soon
corne when process industries will deal exclusively with SI units.

Pressure

Measurement

Perhaps the area that has caused the most concern in the change to SI
units is pressure measurement. The new unit of pressure, the pascal, is
unfamiliar even to those who have worked in the older CGS (centimetre, gram, second) metric system. Once it is accepted and understood,
it willlead to a great simplification of pressure measurement Cromthe
extremes of full vacuum to ultrahigh pressure. It will reduce the multiplicity ofunits now common in industry to one standard that is compatible with other measurements and calculations. To understand the pascal and its relationship to other units of pressure measurement, we
must return to a basic understanding of pressure.
As noted previously, pressure is force per unit area.

36 FEEDBACK PROCESSCONTROL

From Newton's laws, force is equal to magstimes acceleration. In

the English system, the distinction between mags and force became
blurred with common usage of terms such as weight and mags. We live
in an environment in which every object is subject to gravity. Every
object is accelerated toward the center of the earth, unless it is restrained. The force acting on each object is proportional to its mags. In
everyd ay terms ibis force is called the weight of the objecto
W= m' G

where:
W = weight of the object
m = its mass
G = accelerationdue to gravity
Because gravity on the earth's surface is roughly constant, it has
been easy to talk abolli a weight of 1 pound and a mass of 1 pound
interchangeably. However, in faci, force and mass, as quantities, are as
different as apples and pears, as the astronauts bave observed. A numbec of schemes bave be en devised to overcome ibis problem. For example, a quantity called the pound-force was invented and made equal
to the force on a mass of one pound under a specified acceleration due
to gravity. The very similarity between these two units led to more
confusiott. The pascal, by its definition, removes all these problems.
The Pascal
The SI unit of pressure is defined as the pressure or stress that arises
when a force of one newton (N) is applied uniformly over an area of one
square metre (m2). This pressure has been designated one pascal (Pa).
Thus, Pa = N/m2. This is a small unit, but the kilopascal (KPa),
1,000 pascals, and the megapascal (MPa), one million pascals, permit
easy expression of common pressures. The definition is simple, because gravity has been eliminated. The pascal is exactly the same at
every point, even on the moon, despite changes in gravitational acceleration.
In SI units, the unit offorce is derived from the basic unit for mass,
the kilogram (kg), and the unit of acceleration (metres per second per
second, mls2). The product of mass times acceleration is force and is

Gauge,

PROCESS/PRESSURE
MEASURING INSTRUMENTS 37

designated in newtons. One newton would be the force of one kilogram

accelerating at one metre per second per second (N = kg' mls2).
Bar versus Pascal

After the introduction of SI units, the use of the "bar" (lOSPa) gained
favor, especially in European industry, where it closely resembles the
CGS unit of kg/cm2 (kilograms per square centimetre). At thai time, the
SI unit was called the "newton per square metre." As well as being
quite a mouthful, it was found to be inconveniently small (one N/m2
equals 0.000145psi). The use ofthe millibar in meteorology lent weight
to the acceptance of the bar. However, the use of a multiple like (lOS)in
such an important measurement and the resulting incompatibility of
stress and pressure units led to the adoption ofthe N/m2, giving it a new
name, the pascal (Pa), in October 1971. The kilopascal (kPa), 1,000
pascals, equals 0.145 psi and most common pressures are thus expressed in kPa. The megapascal(mPa) equals 145 psi and is convenient
for expressing high pressures.
Absolute,

and Differential

Pressure

The pascal can be used in exactly the same way as the English or CGS
metric units. The pascal may be regarded as a "measuring gauge," the
size ofwhich has been defined and is constant. This gauge can be used
to measure pressure quantities relative to absolute vacuum. Used in
this way, the results will be in pascal (absolute). The gauge may also be
used to measure pressures relati ve to the prevailing atmospheric
pressure, and the results will be pascal (gauge). Ifthe gauge is used to
measure the difference between pressures, it becomes pascal (differen-

tial).
The use of gauge pressure is extremely important in industry, since
it is a measure ofthe stress within a vesseland the tendency offtuids to
leak auto It is really a special case of differential pressure measurement,
inside versus outside pressure. Where there is any doubt about whether
a pressure is gauge, differential, or absolute, it should be specified in
full. However, it is common practice to shaw gauge pressure without
specifying, and to specify by saying "absolute" or "differential" only
for absolute or differential pressures. The use of "g" as in psig is
disappearing, and the use of "a" as in psia is frowned upon. Neither g
flor a is recognized in SI unit symbols. However, M is recognized for
differential pressure in all units.

38 FEEDBACK PROCESSCONTROL

Understanding

the Effects

of Gravity

Before discussing the effects of gravity on pressure measurement, it is

well to keep in mind the size of error thai can arise if gravity is regarded
as constant.
The "standard" gravitational acceleration is 9.806650rn/S2.This is
an arbitrary figure selected as a near average of the actual acceleration
due to gravity found all over the earth. The following are typical values

at different places:
Melbourne (Australia)
Foxboro (USA)
Soest (The Netherlands)

9.79966 rn/S2
9.80368 rn/S2
9.81276 rn/S2

Hence the difference around the world is approximately :t 0.1 percent from the average. This is oflittle practicat importance in industrial
applications. However, with some transmitters being sold with a rated
accuracy of :t0.25 percent, it is well to consider the effect of a gravityinduced difference of more than hal the tolerance thai can arise if the
transmitter was calibrated in Europe and tested in Australia.
Gravity-Dependent

Units

Units such as psi, kglcm2, inches of water, and inches of mercury (Hg)
are all gravity dependent. The English unit pounds per square inch (psi)
is the pressure generated when the force of gravity acts on a mass of
one pound distributed aveT one square inch. Consider a dead weight
tester and a standard mass of one pound which is transported around
the earth's surface: the pressure at each paint on the earth will vary as
the gravitational acceleration varies. The same applies to units such as
inches of water and inches of mercury. The force at the bottom of each
column is proportional to the height, density, and gravitational acceleration.
Dead weight testers are primary pressure standards. They generate
pressure by applying weight to a piston that is supported by a fluid,
generally oil or air. By selecting the weights and the cross-sectional
aTea of the piston, the pressure generated in any gravity field can be
calculated. Therefore, dead weight testers are gravity dependent. For
accurate laboratory work, the gravity under which the tester was calibrated and that at the place of use must be taken into account. Similarly,
the pressure obtained by a certain height of fluid in a manometer depends on density and gravity. These factors must be corrected for the

PROCESS/PRE:SSURE
MEASURINGINSTRUMENTS 39

existing conditions ifprecision regulis are to be obtained. Factors given

in the conversion tables in the Appendix, it should be noted, deal with
units of force, noi weight. Dead weight testers will be discussed in
more detaillater in ibis chapter.
Gravity-lndependent

Units

While gravity plays no part in the definition of the pascal, it has the
same value wherever it is measured. Units such as pounds-force per
square inch and kilogram-force per square centimetre are also independent of gravity because a specific value of gravitational acceleration
was selected in defining these units.
Under equal gravity conditions, the pound-mass and pound-force
are numerically equal (which is the cause of considerable confusion).
Under nonstandard gravity conditions (the usual case), correction factors are required to compensate for the departure from standard. It
should be noted that the standard value of actual gravity acceleration is
llot recognized as such in the SI unit system, where only the SI unit of
acceleration of one metre per second per second is used. In the future,
only the measured actual gravity at the location of measurement (G)
will be used when gravity plays a part in the system under investigation. The pascal is a truly gravity-independent unit and will be used to
avoid the presently confusing question of whether a stated quantity is
gravity dependent.

Pressure

Standards

Now let us consider the calibration standards that are employed with
pressure-measuringinstruments and the basic instruments that are used
to measure pressure. It mar help to look at the ways in which the
standards for pressure calibration are established. You will recalI that
head is the same as pressure. A measure of head, then, can be a dependable measure of pressure. Perhaps the oldest, simplest, and, in
many respects, one of the most accurate and reliable ways of measuring pressure is the liquid manometer. Figure 2-2 shows a differential
manometer. When only a visual indication is needed and static
pressures are in a range that does Dot constitute a safety hazard, a
transparent tube is satisfactory. When conditions for the visual manometer are unsuitable, a variety of ftoat-type liquid manometers are
often employed.

40 FEEDBACK PROCESSCONTROL

HIGH
PRESSURE

Fig. 2-2. Simple U-tube manometer.

The simplest differential gauge is the liquid-filled manometer: it is

basic in its calibration and free of frictional effects. It is often used to
calibrate other instruments. The most elementary type is the U-gauge,
which consists of a glass tube bent in the form of a U, or two straight
glass tubes with a pressure connection at the bottom. When a differential pressure is applied, the difference between the heights of the two
columns of lquid is read on a scale graduated in inches, millimetres, or
other units. In the more advanced designs, vertical displacement of one
side of the manometer is suppressed by using a chamber of large surface area on that side. Figure 2-3 shows such a manometer. If the area
ratio is in the vicinity of 1,600 to 1, the displacement in the large
chamber becomes quite small and the reading on the glass tube will
become extremely close to true inches or true millimetres. The large
side would have to be of infinite area for the reading in the glass tube to
be exact. This problem is sometimes overcome with a special calibration of the scale. However, if the glass tube becomes broken and must
be replaced, the scale must be recalibrated.
A more common and quite reliable design features a zeroing gauge
glass as shown in Figure 2-4. The scale mar be adjusted to zero for each
differential pressure change, and the reading mar be taken from a scale
graduated in actual units of measurement arter rezeroing. The filling
LOW
PRESSURE

Fi2. 2-3. Well or reservoirmanometer.

PROCESS/PR;:SSURE
MEASURINGINSTRUMENTS 41

Fig. 2-4. Well manometerwith zeroingadjustment.

lquid is usually water or mercury, or some other stable fluid especially

compounded by the manometer manufacturers for use with their product. Incline manometer tubes, such as those shown in Figure 2-5, will
give magnified readings, but must be made and mounted carefully to
avoid errors due to the irregularities ofthe tube. It is also essential that
the manometers be precisely positioned to avoid errors due to level.
Still other types of manometers for functions other than simple
indication, including those used with high pressure and hazardous
fluids, employ a float on one leg of the manometer. When reading a
manometer, there are several potential sources of error. One is the
effect of gravity, and another is the effect of temperature on the material contained within the manometer. Correction tables are available
which provide the necessary correction for the conditions under which
the manometer is to be read. Perhaps even more important is the
meniscus correction (Figure 2-6). A meniscus surface should always be
read at its center-the bottom, in the case of water, and the top, in the
case of mercury. To be practical, gravity and temperature corrections
are seldom made in everyday work, but the meniscus correction, or
proper reading, must always be taken into account.

HIGH
PRESSURE

Fig. 2-5. Inclined manometer.

Fig.

42

FEEDBACK PROCESSCONTROL

LIQUID
GLASS

DOES NOT WEl

(MERCURY)

2-6. Readinga manometer.

The dead weight tester is shown in Figure 2-7. The principIe of a

dead weight is similar to thai of a balance. Gravity acts on a calibrated
weight, which in tom exerts a force on a known area. A known
pressure then exists throughout the fluid contained in the system. This
fluid is generally a suitable oil. Good accuracy is possible, bot requires
thai several factors be well established: (1) the piston area; (2) the
weight precision; (3) gravity corrections for the weight if the measurement is to be made in an elevation quite different Crom the original
calibration elevation; (4) buoyancy, since as each weight displaces its
volume of air, the air weight displaced should also be taken into consideration; (5) the absence offriction; (6) head oftransmitting fluid and (7)
operation technique (the weight should be spun to eliminate frictional
effect); unless the instrument being calibrated and the tester are at
precisely the same level, the head of material can contribute an appreciable error. Perhaps the most important part of the procedure is to
keep the piston floating. This is accomplished generally by spinning the
weight platform.

--- WEIGHT
DEAD WEIGHT

TESTER

Fig. 2-7. Ranges30psiandup: increasepressurewith crank unti! pressuresupportsan

accuratelyknownweight. An accuratetestgaugemar beusedwith hydraulic pumpin a
similarsetup.

TO

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 43

A properly operated dead weight tester should bave a pressure

output accurate to a fraction of a percent (actually, 0.1 percent of its
calibration or calibrated reading).
Still another type of dead weight tester is the pneumatic dead
weight tester. This is a self-regulating primary pressure standard. An
accurate calibrating pressure is produced by establishing equilibrium
between the air pressure on the underside of the ball against weights of
known mass on the top. A diagram of an Ametek pneumatic tester is
shown in Figure 2-8. In this construction, a precision ceramic ball is
ftoated within a tapered stainless steel nozzle. A ftow regulator introduces pressure under the ball, lifting it toward the annulus between the
ball and the nozzle. Equilibrium is achieved as soon as the ball beginsto
lift. The ball ftoats when the vented ftow equals the fixed ftow from
the supply regulator. This pressure, which is also the output pressure,
is proportional to the weight load. During operation, the ball is centered
with a dynamic film of air, eliminating physical contact between the ball
and the nozzle. When weights are added or removed fro~ the weightcarner,
the ball rises or drops, affecting the air ftow. The regulator
sensesthe change in ftow and adjusts the pressure beneath the ball to
bring the system into equilibrium, changing the output pressure accordingly. Thus, regulation of output pressure is automatic with change of
weight mass on the spherical piston or ball. The pneumatic dead weight

INSTRUMENT
UNOER TEST

Fig. 2-8. Pneumatictester. Courtesyof AMETEK Mfg. CD.

44 FEEDBACK PROCESSCONTROL

tester has an accuracy o :t 0.1 o 1 percent o indicated reading. It is

commonly used up to a maximum pressure o 30 psi or 200 kPa gauge.

Plant Instruments

That Measure

Pressure

Directly

Thus far in tbis chapter we bave been concemed with the detinition of
pressure, and some of the standards used bave been described. In the
plant, manometers and dead weight testers are used as standards for
comparison and calibration.
The working instruments in the plant usually include simple mechanical pressure gauges, precision pressure recorders and indicators,
and pneumatic and electronic pressure transmitters. A pressure transmitter makes a pressure measurementand generates either a pneumatic
or electrical signal output thai is proportional to the pressure being
sensed. We will discuss transmitters in detaillater in ibis chapter. Now
we will deal with the basic mechanical instruments used for pressure
measurement, how they operate and how they are calibrated. When the
amount of pressure to be measured is very small, the following instruments might be used.
Bell Instrument

This instrument measures the pressure difIerence in the compartment

on each gide ora bell-shaped chamber. Ifthe pressure to be measured is
gauge pressure, the lower compartment is vented to atmosphere. Ifthe
lower compartment is evacuated, the pressure measured will be in
absolute units. If the difIerential pressure is to be measured, the higher
pressure is applied to the top of the chamber and the lower pressure to
the bottom.
The bell chamber is shown in Figure 2-9. Pressure ranges as low as
Oto 1 inch (Oto 250 Pa) of water can be measured with this instrument.
Calibration adjustments are zero and span. The difficulty in reading a
manometer accurately to fractions of an inch are obvious, ret the manometer is the usual standard to which the bell difIerential instrument is
calibrated. The bell instrument finds applications where very low
pressures must be measured and recorded with reasonable accuracy.
Slack or Limp-Diaphragm

The slack or limp-diaphragminstrument is used when very small

pressuresare to be sensed.The most commonapplicationofthis gauge

Fig.

PROCESS/PRE$SURE
MEASURINGINSTRUMENTS 45

2.9. Bell instrument.

is measurement of fumace draft. The range of this type instrument is

CromO to 0.5 inches (125 Pa) of water to O to 100 inches (25 kPa) of
water above atmospheric pressure.
To make this instrument responsive to very small pressures, a large
aTea diaphragm is employed. This diaphragm is made of very thin,
treated nonporous leather, plastic, or rubber, and requires an extremely small force to deftect it. A spring is always used in combination
with the diaphragm to produce a deftection proportional to pressure.
Let us assume the instrument shown in Figure 2-10 is being used to
measure pressure. The low-pressure chamber on the top is vented tothe
atmosphere. The pressure to be measuredis applied to the high-pressure
chamber on the bottom. This causesthe diaphragm to move upward. It

Fig. 2-10. Slack or limp-diaphragm instrument.

46 FEEDBACK PROCESSCONTROL

will move until the force developed by pressure on the diaphragm is

equal to that applied by the calibrated spring. ln the process of achieving balance, the lever attached to the diaphragm is tipped and this
motion transmitted by the sealed link to the pointer. This instrument is
calibrated by means of zero and span adjustments. The span adjustment allows the ridged connection to the spring to be varied, thus
changing the spring constant. The shorter the spring the greater the
spring constant; thus, a greater force is required to deftect it. As the
spring is shortened, a higher pressure is required for full deftection of

the pointer. The zero adjustmentcontrols the spring's pivot point,

thereby shifting the Creeend of the calibrated spring and its attached
linkages. Thrning the zero adjustment changes the pointer position on
the scale, and it is normally adjusted to read zero scale with both the
high and low pressure chambers vented to the atmosphere. lf pressure
, is applied to both the high- and low-pressure chambers, the resultant
reading will be in differential pressure. This instrument is extremely
sensitive to overrange, and care should be taken to avoid this problem.
The other long-term difficulty that can develop is damage to the diaphragm. The diaphragm mar become stiff, develop leaks, or become
defective, which results in error. These difficulties mar be observed by
periodically inspecting the diaphragm.
Pressure Gauges
In the process plant, we frequently find simple pressure ganges scattered throughout the process and used to measure and indicate existing
pressures. Therefore, it is appropriate to devote some attention to the
operation of a simple pressure gauge.
The most common of all the pressure ganges utilizes the Bourdon
tube. The Bourdon tube was originally patented in 1840. In bis patent,
Eugene Bourdon stated: "I bave discovered that if a thin metallic tube
be ftattened out then bent or distorted Croma straight line, it has the
property of changing its Cormconsiderably when exposed to variations
of internal or external pressure. An increase of interna! pressure tends
to bring the tube to a straight cylindrical Corm, and the degree of
pressure is indicated by the amount of alteration in the Corm of the

tube."
Figure 2-11A shows such a tube. As pressure is applied internally,
the tube straightens out and returns to a cylindrical Corm. The excursion of the tube tip moves linearly with internal pressure and is
converted to pointer position with the mechanism shown. Once the

PROCESS/PRE;SSURE
MEASURINGINSTRUMENTS 47

Fig. 2-11. (A) Bourdontube. (B) Typical pressuregauge.

internal pressure is removed, the spring characteristic of the material

returns the tube to its original shape.
Ifthe Bourdon tube is overranged, that is, pressure is applied to the
point where it can no longer return to its original shape, the gauge mar
take a new set and its calibration becomes distorted. Whether the gauge
can be recalibrated depends on the extent ofthe overrange. A severely
overranged gauge will be ruined, whereas one that has been only
slightly overranged mar be recalibrated and reused.
Most gauges are designed to handle approximately 35 percent of
the upper range value as overrange without damage. Typically, a gauge
will exhibit some amo unt of hysteresis, that is, a difference due to
pressure moving the tube in the upscale direction versus the spring
characteristic of the tube moving it downscale. A typical gauge mar
also exhibit some amount of drift, that is, a departure from the true
reading due to changes over a long period in the physical properties of
the materials involved. All of these sources of error are typically included in the manufacturer' s statement of accuracy for the gauge. A
typical Bourdon tube gauge, carefully made with a Bourdon tube that
has been temperature-cycled or stress-relieved, will bave an accuracy
of :t 1 percent of its upper range value. A carefully made test gauge will
bave an accuracy of :t0.25 to 0.50 percent ofits upper range value. The
range of the gauge is normally selected so that it operates in the upper
part of the middle third of the scale.

48 FEEDBACK PROCESSCONTROL

In addition to the gauges used for visual observations of pressure

throughout the plant, the typical instrument shop wi11bave a number of
test gauges which are used as calibration standards. The gauge must be
vertical to read correctly. A typical pressure gauge is shown in Figure

2-11 B.
Liquid

or Steam Pressure

Measurement

Whena liquid pressureis measured,the piping is arrangedto prevent

entrappedvaporswhich mar causemeasurement
error. If it is impossible to avoid entrappingvapors, vents should be provided at all high
points in the line.
When steampressureis measured,the steamshouldbe prevented
from enteringthe Bourdontube. Otherwise,the hightemperaturemar
damagethe instrument.Ifthe gaugeis belowthe paint of measurement,
a "siphon" (a singleloop or pigtail in a vertical plane)is provided in the
pressureline to the gauge.A cock is installed in the lme betweenthe
loop and the gauge.In operation,the looptraps condensate,preventing
the steamfrom enteringthe gauge.
Seals and Purges
If the instrument is measuring a viscous, volatile, corrosive, or extremely bot or cold fluid, the use of a pressure seal or a purge is
essential to keep the fluid out of the instrument. Liquid seals are used
with corrosive fluids. The sealing fluid should be nonmiscible with the
measured fluids. Liquid purges are used on bot, viscous, volatile, or
corrosive lquids, or lquids containing solids in suspension. The purging lquid must be of such a nature that the small quantities required (in
the order of I gph) (4 litres per hour) will not injure the product or
processoMechanical seals will be described in detaillater in this chap-

ter.
Pulsation Dampener
If the instrument is intended for use with a fluid under pressure and
subject to excessive fluctuations or pulsations, a deadener or damper
should be installed. This will provide a steady reading and prolong the
life of the gauge.
Two other elements that use the Bourdon principIe are the spiral
(Figure 2-12) and helical (Figure 2-13). The spiral and helical are, in
effect, multitube Bourdon tubes. Spirals are commonly used for

Fig.

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 49

2-12. Spiral.

Fig. 2-13. Helical.

pressure ranges up to o to 200 psi or 1.4 MPa, and helicals are made to
measure pressures as high as Oto 80,000 psi or 550 MPa. The higher the
pressure to be measured, the thicker the walls ofthe tubing Cromwhich
the spiral or helical is constructed. The material used in the construction mar be bronze, beryllium copper, stainless steel, or a special NiSpan C alloy. Spirals and helicals are designed to provide a lever motion of approximately 45 degrees with full pressure applied. If this
motion is to be translated into pen or pointer position, it is common
practice to utilize a four-bar linkage, and this necessitates a special
calibration technique. If, instead of measuring gauge pressure, it is
necessary to read absolute pressure, the reading must make allowances
for the pressure of the atmosphere. This mar be done by utilizing an
absolute double spiral element. In tros element, two spirals are used.
One is evacuated and sealed; the second has the measured pressure
applied.
The evacuated sealed element makes a correction for atmospheric
pressure as read on the second element. Thus, the reading can be in
terms of absolute pressure, which is gauge plus the pressure of the
atmosphere, rather than gauge.
An absolute double spiral element of this type mar be used to
measure pressures up to 100 psi or 700 Kpa absolute. This element is
shown in Figure 2-14.
Metallic

Bellows

A bellows is an expandableelementmade up of a series of folds or

corrugationscalled convolutions(Figure 2-15).Wheninternalpressure
is appliedto the bellows, it expands.Becausea sizablearcais involved.

SO FEEDBACK PROCESS CONTROL

Fig. 2-14. Absolutedoublespiral.

the applied pressure develops a sizable force to actuate an indicating or

recording mechanism. (In some instruments, the bellows is placed in a
sealed can and the pressure is applied externally to the bellows. The
bellows will then compress in a fashion similar to the expansion just
described. Such an arrangement is shown in Figure 2-16).
A variety of materials are used to fabricate bellows, including
brass, beryllium copper, copper nickel alloys, phospher bronze, Monel,
steel, and stainless steel. Brass and stainless steel are the most commonly used. Metallic bellows are used from pressure ranges of a few

Fig. 2-15. Bellows receiver unit.

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 51

Fig. 2-16. Bellows in sealedcan.

ounces to many pounds per square inch. A bellows will develop many
times the power available from a helical, spiral, or Bourdon tube.
A bellows is typically rated in terms of its equivalent square inch
area. To create a linear relationship between the excursion of the bellows and the applied pressure, it is common practice to bave the bellows work in conjunction with a spring, rather than with the spring
characteristic of the metal within the bellows itself. Each bellows and
spring combination has what is called a spring rate. The springs used
with the bellows are usually either helical or spiral. Typically, the
spring rate of the helical spring is ien times or more thai of the bellows
material itself.
Using a spring with a bellows has several advantages over relying
on the spring characteristics of the bellows alone. The calibration procedure is simplified, since adjustments are made only on the spring.
Initial tension becomes zero adjustment and the number of active turos
becomes span adjustment. A spring constructed of stable material will
exhibit long-term stability thai is essential in any component.
.When a measurement of absolute pressure is to be made, a special
mechanism employing two separate bellows mar be used. It consists of
a measuring bellows and a compensating bellows, a mounting support,
and an output lever assembly (Figure 2-17). The measuring and compensating bellows are fastened to opposite ends of the fixed mounting
support: the free ends of both bellows are attached to a movable plate
mounted between them. The motion of ibis movable plate is a measure
of the difference in pressure between the two bellows.

52 FEEDBACK PROCESSCONTROL

Fig. 2-17. Absolutepressurebellows.

Since the compensating bellows is completely evacuated and

sealed, ibis motion is a measure of pressure above vacuum, or absolute
pressure applied to the measuring bellows. This movable plate is attached to the output lever assembly which, in turo, is linked to the
instrument pen, or pointer.
In many applications, the bellows expands very little, and the force
it exerts becomes its significant output. This technique is frequently
employed in force-balance mechanisms which will be discussed in
some detail in Chapter 8.

Pressure

Transmitters

Signal Transmissions

In the process plant, it is impractical to locate the control instruments

out in the plant near the processoIt is also true thai mostmeasurements
are Dot easily transmitted from some remote location. Pressure measiIrement is an exception, but if a high pressure measurement of some
dangerous chemical is to be indicated or recorded several hundred feet
from the point of measurement, a hazard will be created. The hazard
mar be from the pressure and/or from the chemical carried in the line.
To eliminate ibis problem, a signal transmission system was developed. In process instrumentation, ibis system is usually either pneumatic (air pressure) or electrical. Becauseibis chapter deals with pressure,
the pneumatic, or air pressure system, will be discussed first. Later it
will become evident thai the electrical transmitters perform a similar
function.

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 53

U sing the transmission system, it will be possible to install most of

the indicating, recording, and control instruments in one location. This
makes it practical for a mnimum number of operators to rUll the plant
efficiently.
When a transmission system is employed, the measurementis converted to a pneumatic signal by the transmitter scaled from O to 100
percent ofthe measured value. This transmitter is mounted close to the
point of measurement in the processo The transmitter output-air
pressure for a pneumatic transmitter-is piped to the recording or control instrument. The standard output range for a pneumatic transmitter
is 3 to 15 psi, 20 to 100kPa; or 0.2 to 1.0 bar or kg/cm2. These are the
standard signals that are almost universally used.
Let us take a closer look at what this signal means. Suppose we
have a field-mounted pressure transmitter that has been calibrated to a
pressure range of 100 psi to 500 psi (689.5 to 3447.5 kPa).
When the pressure being sensedis 100psi (69 kPa), the transmitter
is designed to produce an output of 3 psi air pressure (or the approximate SI equivalent 20 kPa). When the pressure sensedris es to 300 psi
(2068.5 kPa), or midscale, the outputwill climb to 9psi, or 60kPa, and at
top scale, 500 psi (3447.5 kPa), the signal output will be 15 psi, or 100kPa.
This signal is carried by tubing, usually V4-inchcopper or plastic, to
the control room, where it is either indicated, recorded, or fed into a
controller.
The receiving instrument typically employs a bellows element to
convert this signal into pen or pointer position. Or, if the signal is fed to
a controller, a bellows is also used to convert the signal for the use of
the controller. The live zero makes it possible to distinguish between
true zero and a dead instrument.
The top scale signal is high enoughto be useful without the possibility of creating hazards.
Pneumatic Recorders and Indicators
Pressure recorders and indicators were described earlier in this chapter. Pneumatic recorders and indicators differ only in that they always
operate from a standard 3 to 15 psi or 20 to 100 kPa signal. The
indicator scale, or recorder chart, mar be 1abeledO to 1,000 psi. This
would represent the pressure sensed by the measuring transmitter and
converted into the standard signat that is transmitted to the receiver.

54 FEEDBACK PROCESSCONTROL

The receiver converts the signat into a suitable pen or pointer position.
Because the scale is labeled in proper units, it is possible to read the

measuredpressure.
A typical pneumatic indicator is shown in Figure 2-18 (top) and its
operation mar be visualized by studying Figure 2-18 (bottom, right).
The input signal passes through an adjustable needle valve to provide damping, then continues to the receiver bellows. This bellows,

Fig. 2-18. Pneumaticindicator.

PROCESS/PRESSURE MEASURING INSTRUMENTS

55

acting in expansion, moves a fQrce plate. A spring opposing the bellows

provides zero and span adjustments. Regulating the amo unt of spring
used (its effective length) provides a span adjustment. Setting the initial
tension on the spring provides a zero adjustment.
The force plate is connected to a link thai drives the pointer arbor
assembly. Changing the length ofthe link by turning the Dut on the link
provides an angularity adjustment. The arm connecting the pointer and
the arbor is a rugged crushed tube designed to reduce torsional effects.
A takeup spring is also provided to reduce mechanical hysteresis.
Overrange protection is provided to prevent damage to the bellows or
pointer movement assembly for pressures up to approximately 30 psi or
about 200 kPa.
A pneumatic recorder also uses a receiver bellows assembly as
shown in Figure 2-19 (top). This receiver provides an unusually high
torque due to an effective bellows area of 1.1 square inches.
The input signal tirst passesthrough an adjustable damping restrictor. The output of this signat is the input to the receiver bellows assembly. The receiver consists of a heavy duty impact extruded aluminum
can containing a large brass bellows working in compression. The large
effective area of the bellows assures an extremely linear pressure-topen position relationship. The motion, created by the input signal, is
picked up by a conventionallink and transferred to an arbor. Calibration is easily accomplished by tuming zero, span, and angularity adjustments on ibis simple linkage system. All calibration adjustments are
accessible through a door (Figure 2-19) on the gide of the instrument
and mar be made while instrument is in operation. The arm interconnecting the pen and the arbor incorporates a rugged tubular member to
reduce torsional effects and takeup spring to reduce mechanical hysteresis. Overrange protection is provided to prevent damage to the
bellows or pen movement assembly for pressures up to 30 psi or about
200 kPa.
Mechanical

Pressure

Seals

Application
A sealedpressuresystemis usedwith a pressuremeasuringinstrument
to isolate corrosive or viscousproducts, or products that tend to solidify, from the measuringelementand its connectivetubing.

Definition

A sealedpressuresystemconsistsof a conventionalpressuremeasuring elementor a force-balancepressuretransmittercapsuleassembly

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 57

connected, either directly orby capillary tubing, to a pressure seal as

seen in Figure 2-20. The system is solidly filled with a suitable lquid
transmission medium.
The seat itself mar take many forms, depending on process conditions, but consists of a pressure sensitive flexible member, the diaphragm, functioning as an isolating membrane, with a suitable method of
attachment to a process vessel or line.
Principie or Operation
Process pressure applied to the flexible member of the seal assembly
forces some of the filling fluid out of the seal cavity into the capillary
tubing and pressure measuring element, causing the element to expand
in proportion to the applied process pressure, thereby actuating a pen,
pointer, or transmitter mechanism. A sealed pressure system offers
high resolution and rapid response to pressure changes at the diaphragm. The spring rate ofthe flexible member musi be low when compared with the spring rate of the measuring element to ensure thai the
fill volume displacement will full stroke the measuring element for the
required pressure range. A low-diaphragm spring rate, coupled with
maximum fill volume displacement, is characteristic of the ideal system.

Fig. 2-20. SeaIconnected to 6-inch pressure gauge.

58 FEEDBACK PROCESSCONTROL

A sealed pressure system is somewhat similar to a liquid-filled

thermometer, the primary differences being a flexible rather than rgid
member at the process, and no initial filling pressure. The flexible
member ofthe seal should ideally accommodate any thermal expansion
of the filling medium without perceptible motion of the measuring element. A very stiff seal member, such as a Bourdon tube, combined with
a low-pressure (high-volume change) element, will produce marked
temperature effects Crom both varying ambient or elevated process

temperatures.
The Foxboro l3DMP Series pneumatic d/p Cell transmitters with
pressure seals (Figure 2-21) measure differential pressures in ranges of
O to 20 to O to 850 inches of water or O to 5 to O to 205 kPa at static
pressures Cromfull vacuum up to flange ralingo They transmit a proportional 3 to 15 psi or 20 to 100kPa or 4 to 20 mA dc signat to receivers
located up to several hundred yards or melers Cromthe point of measurement.

Fig. 2-21. Sealsconnectedto differentialpressuretransmitter.

PROCESS/PRESSURE MEASURING INSTRUMENTS

S9

FiUing Fluids
Ideally, the filling fluid used in a sealed pressure system should be
noncompressible, bave a high-boiling point, a low-freezing point, a low
coefficient of thermal expansion, and low viscosity. It should be noninjurious to the diaphragm and containing parts, and should Dot cause
spoilage in the event ofleakage. Silicone-based liquid is the most popular filling fluid.
The system is evacuated before the filling fluid is introduced. The
system musi be completely filled with fluid and free from any air pockets thai would contract or expand during operation, resulting in erroneous indications at the pen or pointer or in an output signat. The
degree of accuracy of any filled pressure system depends on the perfection of the filling operation.
Calibration

Techniques

The procedure for calibration of a pressure instrument consists of comparing the reading of the instrument being calibrated with a standard.
The instrument under calibration is then adjusted or manipulated to
make it agree with the standard. Success in calibration depends Dot
only on one' s ability to adjust the instrument, but on the quality of the
standard as well.
Field Standards
Field standards must be reasonably convenient to use and must satisfy
the accuracy requirements for the instrument under calibration. A 100inch water column, for example, is extremely accurate but Dot practical
to set up out in the plant. For practical reasons, we find that most field
standards are test ganges.The test gaugeis quite similar, in most cases,
to the regular Bourdon ganges. However, more care has gone into its
design, construction, and calibration, making it very accurate. A good
quality test gauge will be accurate to within ::\:0.25percent of its span.
This is adequate for most field use.
Under some conditions, a manometer mar be used in the field. This
usually occurs when a low-pressure range is to be calibrated and no
other suitable standard is readily available.
Portable Pneumatic Calibrator
All of the ingredients required to perform a calibration bave been combined into a single unit called a portable pneumatic calibrator. This unit

60

FEEDBACK PROCESSCONTROL

Fig. 2-22. Portablepneumaticcalibration.Courtesyo Wallace& Tieman.

contains an accurate pressure gauge or standard, along with a pressure

regulator and suitable manifold connections. The portable pneumatic
calibrator will accurately apply, hold, regulate, and measure gauge
pressure, differential pressure, or vacuum. The gauge case mar be
evacuated to make an absolute pressure gauge. The pointer in the
operation of the gauge in this calibrator differs in that it makes almost
two full revolutions in registering full scale, providing a scale length of
45 inches. This expanded scale makes the gauge very easy to read.
The calibrator is available in seven pressure ranges (SI unit
ranges). Of these, three are sized for checking 3 to 15 psi or 20 to 100
kPa pneumatic transmission instrumentation. The calibrator is shown
in both picture and schematic form in Figures 2-22 and 2-23. The air
switching arrangement makes it possible to use the calibrator as both a
source or signal and a precise gauge for readout. This portable pneumatic calibrator is manufactured by Wallace and Tieman, Inc., Belleville, New Jersey.
Force-Balance

Pneumatic

Pressure

Transmitter

A pneumaticpressuretransmitter sensesa pressureand converts that

pressureinto a pneumaticsignatthat mar be transmittedover a reason-

0/1"

PROCESS/PRE~SURE
MEASURINGINSTRUMENTS 61

~~

o~
~~
;c
~"
~m
~C

~~z
mO~
,,~~
cm~

z
~2
~~
~'"

~~~;~:~
;

l ,t'i'

" ",'
~, \c:
\\'i",:"

'

Fig. 2-23. Connectionsfor differentpressurereadouts(courtesyof Wallace& Tieman):

For GaugePressure:testpressureis appliedto the capsulethroughthe appropriateP
connection;the caseis opento atmospherethroughS.
For DifferentialPressure:high-testpressureis appliedto the capsulethroUghthe
appropriateP connection.Low-testpressureis appliedto the casethroughS.
For AbsolutePressure:testpressureis appliedto the capsulethroughthe appropriateP
connectionand the caseis continuouslysubjectedto full vacuumthroughS.
For Vacuum:the capsuleis opento atmospherethroughconnectionP; the caseis
connectedto test vacuumat S.
For Positiveand NegativePressures:testpressureis appliedto the capsulethroughthe
appropriateP connectionandthe caseis opento atmospherethroughS.

able distance. A receiving instrument is then employed to convert the

signat into a pen or pointer position or a measurement input signat to a
controller. The Foxboro Model llGL force-balance pneumatic
pressure transmitter is an example of a simple, straightforward device
that performs this service (Figure 2-24). Before its operation is discussed, the operation of two vital components must be described: the
ftapper and the nozzle, or detector; and the pneumatic amplifier, or
relay. These two mechanisms are found in nearly every pneumatic
instrument.
The ftapper and nozzle unit converts a small motion (position) or

62 FEEDBACK PROCESSCONTROL

FLE'~"

Fig. 2-24. ModelllGM pressure transmitter is a force-balance instrument that

measures pressure and transmits it as a proportional 3 to 15 psi or 20to 100kPa
pneumatic signa!.

force into an equivalent (proportional) pneumatic signal. Flapper

movement of only six ten-thousandths of an inch (0.0015 cm) will
change the nozzle pressure by 0.75 psi or 5.2 kPa. This small pressure
change applied to the pneumatic amplifier or relay becomes an
amplified change of 3 to 15 psi or 20 to 100kPa in the amplifier output.
Let us take a closer look at the operation of the alplifier or relay.
It is shown in a cross-sectional view in Figure 2-25.
Pneumatic

Relay

A relay is a pneumatic amplifier. Like its electronic counterpart, the

function of the relay is to convert a small change in the input signal (an
air pressure signal) to a large change in the output signal. Typically, a 1
psi or 7 kPa change in input will produce approximately a 12 psi or 80
kPa change in output.
The supply enters the relay through a port on the surface of the

PROCESS/PRE$SURE
MEASURINGINSTRUMENTS 63

INPUT
DIAPHRAGM
SUPPLY

STEM

VALVE

BALL VALVE
OUTPUT
EXHAUST

SPRING

Fig. 2-25. Cross-sectionalviewoCamplifier.

instrument on which the Telar is mounted. The input signal (nozzle

pressure) enters the Telar through another port and acts on the diaphragm. Snce the diaphragm is in contact with a stem valve, the two
move in unson.
As the input signal increases, the stem pushes against a ball valve
which in turo moves a flat spring, allowing the supply of air to enter the
Telar body. Further motion of the stem valve causes it to close off the
exhaust port. Thus, when the input pressure increases, the stem
(exhaust) valve closes and the supply valve opens; when the input
decreases, the stem valve opens and the supply valve closes. This
varies the pressure to the output.
Principie

ot Operation

The 11GM Pneumatic Transmitter (Figure 2-24) is a force-balance instrument that measures pressure and transmits it as a proportional 3 to
15 psi pneumatic signal (20 to 100kPa).
The pressure is applied to a bellows, causing the end of the bellows
to exert a force (through a connecting bracket) on the lower end of the
force bar. The metal diaphragm is a fulcrum for the force bar. The force
is transmitted through the ftexure connector to the range rod, which
pivots on the range adjustment wheel.
Any movement of the range rod causes a minute change in the
clearance between the ftapper and nozzle. This produces a change in
the optput pressure from the relay to the feedback bellows until the
force of the feedback bellows balances the pressure on the measure-

64 FEEDBACK PROCESSCONTROL

ment bellows. The output pressure which is established by this balance

is the transmitted signal and is proportional to the pressure IlPplied to
the measurement bellows.
Ifthe pressure
be measured
is high, such as 5,000 psi or 35.MPa,
a different
sensing to
element
is employed.
The pressure being measured is applied to a Bourdon tube. This
pressure tends to straighten the tube and causes a horizontal force to be
applied to the lower end of the force bar. The diaphragm seat serves as
both a fulcrum for the force bar and as a seal for the pressure chamber.
The force is transmitted through a ftexure connector to the range rod,
which pivots on the range adjustment wheel.
Any movement of the rangerod causes a minute change in the
clearance between the ftapper and nozzle. This produces a change in
the output pressure from the rely to the feedback bellows until the
force in the bellows balances the force created by the Bourdon tube.
The output pressure, which establishes the force-batancing, is the
transmitted pneumatic signal that is proportionat to the pressure being
measured. This signat is transmitted to a pneumatic receiver to record,
indicate, or control.
The calibration procedure for this transmitter is the same as for the
low pressure transmitter, but the clllibration pressure is developed with
a dead weight tester. For safety, oil or liquid, never air, should be used
for high pressure calibrations.
Pressure measurements found in some applications require that
absolute, rather than gauge pressure, be determined. To handle these
applications, the absolute pressure transmitter mar be used.
Absolute

Pressure Transmitter

The pressure being measured is applied to one side of a diaphragm in a

capsule. The space on the other side of the diaphragm is evacuated,
thus providing a zero absolute pressure reference (Figure 2-26).
The pressure exerts a force on the diaphragm thai is applied to the
lower end of the force bar. The diaphragm seal serves both as a fulcrum
for the force bar and as a seal for the pressure chamber. The force is
transmitted through the ftexure connector to the range bar, which
pivots on the range adjustment wheel.
Any movement of the range bar causes a minute change in the
clearance between the ftapper and nozzle. This produces a change in
the output pressure from the relay (Figure 2-26) to the feedback bellows

PROCESS/PRES:sURE
MEASURINGINSTRUMENTS 65

i'ig. 2-26. Pneumaticabsolutepressuretransmitter.

until the force in the bellows balances the force on the diaphragm

capsule.
The output pressure, which establishes the force-balane, is the
transmitted pneumatic signal, which is proportional to the 'absolute
pressure being measured. This signal is transmitted to a pneumatic
receiver to record, indicate, or control.

Ouestions
2-1. An ordinary commercial Bourdon gauge has a scale of O psi to 250 psi,
and an accuracy of:t 1.percent of span. If the gauge reads 1.75psi, within
what maximum and minimum values will the correct pressure fall?
8. 1.74to 1.76psi
c. 1.76.5to 1.78.5psi
b. 1.72.5to 1.77.5psi
d. 1.79to 1.80psi
2-2.
8.
b.
c.

A manometer, read carefully:

Always has zero error
Has an error caused by the liquid's impurity unless corrected
May bave an error caused by temperature and gravity effects on the
various components and so on.
d. Has an error that varies only with altitude

66 FEEDBACK PROCESSCONTROL

2-3.
a.
b.
c.
d.

Absolutepressuris:
Gaugepressureplus atmosphericpressure
Gaugepressurelessatmosphericpressure
Gaugepressureplus atmosphericpressuredivided by two
Always referencedto a point at the peakof Mt. Washington,NH

2-4. The pressureat the bottomof a pondwhere it is exactly46 feet deep

will be:
a. 100psi
c. 20 psi
b. 46 psi
d. 20 psi absolute
2-5. The advantagesof makingabsolutepressuremeasurements
rather than
gaugeare:
a. Greateraccuracy
b. Eliminateserrors introducedby barometricvariations
c. Is more indicativeof true processconditions
d. Is more relatedto safetythangaugepressure
2-6. A bellows elementis usedas a receiverfor a 3 to 15psi pneumatic
signaland by error a 30-psipressureis appliedto it. The result will be:
a. A damagedbellows
b. An instrumentin needof recalibration
c. No damage
d. A severezero shift
2-7. A pressureinstrumentis calibratedfrom 100to 600psi. The spanof this
instrumentis:
a.600
c. 500
b. 100
d.400
2-8. Instrumentsthat measurepressureare generallyclassifiedas:
a. nonlinear
c. free of hysteresis
b. linear
d. Doneof the above
2-9. Whenreadinga manometer,it is goodpracticeto:
a. Readat the bottomof the meniscus
b. Readat the top of the meniscus
c. Read at the centerof the meniscus
d. Carefullyestimatethe averageof the meniscus
2-10. If a pressure range of O to 1 inch of water is to be measured and
recorded, an instrument capable of doing this would be:
a. A bellows
c. A Bourdon tube
b. A bell with a lquid seal
d. A mercury manometer

2-11. Whenmeasuringa pressurethat fluctuatesseverely:

a. A large-capacitytank shouldbe installed
b. A pulsationdampenershouldbe employed

PROCESSlPRESSURE
MEASURINGINSTRUMENTS 67

c. No problemis created
d. Readthe peakand minimumand divide by two to obtainthe true
pressure
2-12. A measurement
of absolutepressureis to be madeusinga mechanism
employingtwo separatebellows. Measurementis appliedto one bellowsand
the other:
8. Is sealedat an atmosphericpressureof 14.7psi
b. Is completelyevacuatedand sealed
c. Containsalcoholfor temperaturecompensation
d. Has an activeareatwice that of the measuringbellowsand is sealedat a
pressuretwice atmospheric
2-13. The dangerof havinga high-pressureline carryinga dangerous
chemicalrupture in the control room is:
8. Eliminated by usingspecialduty piping
b. Ignored
c. Eliminated throughthe useof a transmissionsystem
d. Minimized by placingthe line within a protectivebarrier
2-14. The standardpneumatictransmissionsignalmost generallyused in the
United Statesis:
8. 3 to 27 psi
c. 3 to 15psi
b. 10to 50 psi
d. 2 to 12psi
2-15. A sealedpressuresystem:
8. Is similar in someways to a liquid-filled thermometer
b. Sealsthe processmaterialin the instrument
c. Must be usedat a fixed temperature
d. Is alwaysusedwith manualtemperaturecompensation
2-16. A pneumaticrelay:
8. Is a setof electricalcontactspneumaticallyactuated
b. Is a signalbooster
c. Is a pneumaticamplifier
d. Containsa regulatoractuatedby a bellows
2-17. Whenthe clearancebetweenflapper and nozzlechangesby 0.0006
inchesthe output of the transmitterwill changeby:
8. 3 psi
c. 12psi
b. 15psi
d. 6 psi
2-18. An instrumentis to be calibratedto measurea rangeof Oto 6,000psi.
For sucha calibration:
8. oil or lquid shouldbe used
b. Air mustbe used
c. An inert gas suchas nitrogenis required
d. Any sourceof highpressureis acceptable

68 FEEDBACK PROCESSCONTROL

2-19. The kilopascal(kPa),an SI Ilnit of pressure:

a. Is always equivalentto psi x 6.895
b. Isgravity dependent
c. Is a force of 1.000newtons (N) applied uniformly over an areaof one
squaremetre
d. Changessubstantiallyfrom placeto place
2-20. A pneumaticpressuretransmitteris calibratedto a pressurerangeof
100to 500psi. The signaloutput is 10.2psi. Whatis the measuredpressurein
psi?
8. 272psi
c. 267psi
b. 340psi
d. 333psi
2-21. A viscoseline is located 100feet from the control roomwherea record
of line pressuremustbe made.The maximumpressureis approximately85
psi. Selectthe instrumentationthat canhandlethis job:
8. A sealedsystemwith 100feet of capillaryconnectedto a special
elementin a recordinginstrument.
b. A sealedsystemconnectedto a nearbypneumatictransmittersendinga
signalto a pneumaticreceivingrecorder
c. A direct connectedspiral elementrecorder
d. A pneumatictransmitterdirectly connectedto the line
2-22. A pressurerecorderreads38 psi and the barometerreads30.12inches
of mercury. The absolutepressureis
8. 23.21 psi absolute
c. 52.79psi absolute
b. 68.12psi absolute
d. 38 psi absolute
2-23. The lowestrangesof pressuremar be measuredby
8. a watercolumn
c. a bell meter
b. a bellows
d. a slackor limp-diaphragm
2-24. A slack diaphragmindicatoris to be calibratedOto 1 inch of water. The
standardwould likely be:
8. a water column
b. a columnof kerosene
c. a mercury-filledinclined manometer
d. a water-filledinclined manometer

Level and Density Measurements

Level Measurement
Measurement Methods
Methods
Level

The
The typical process
process plant
plant contains
contains many
many tanks,
tanks, vessels,
vessels, and
and reservoirs.
reservoirs.
measurement
store or process
process materials.
materials. Accurate
Accurate measurement
Their function
function is to store
of thesecontainers
materialin the tanks
of the contents
contents ofthese
containers is vital. The
The material
tanks is
usually
occasionallyit may
mar consist
consistof solids.
usually liquid, but occasionally
solids.
Initially,
level measurement
measurementappears
presenta simple
Initially, level
appears to present
simple problm.
problem.
However,
closerlook soon
soonreveals
variety of problems
that must
mustbe
However, a closer
reveals a variety
problems that
be
resolved.
materialmay
mar be
very corrosive;
mar tend
resolved. The material
be very
corrosive; it may
tend to solidify; it
mar tend
mar contain
contain solids;
mar create
other
may
tend to vaporize;
vaporize; it may
solids; or it may
create other
difficulties.
difficulties.
The
commonmethods
The common
methods employed
employed for automatic
automatic continuous
continuous liquid
level measurements
level
measurements are
are as
as follows
follows (see
(see also
also Table
Table 3-1):
3-1):
1.
Float-and-eable
1. Float-and-cable
2.
2. Displacement
Displacement(buoyancy)
(buoyancy)
3. Head
Head (pressure)
(pressure)
4. Capacitance
Capacitance
5. Conductance
5.
Conductance
6. Radiation
Radiation(nucleonic)
(nucleonic)
7. Weight
7.
Weight
8. Ultrasonic
8.
Ultrasonic
9. Thermal
9.
Thermal
69

70 FEEDBACK PROCESSCONTROL

Table3-1. Liquid Level Measurement

Closedor
Method
Bubbletube
Diaphragm
box
Head Pressure
Diff. Pressure
Displacement
(buoyancy)
Float-andCable
Weight
Radiation
(nucleonic)
Capacitance
Ultrasonic
Conductance

Available Upper
Range Values

Open

10 in to 250 ft
0.25 to 75 m
4 in to 250 ft
0.lt075m
2 in to many ft.
50 mm to many metres
5 in to many ft.
0.25 to many metres
6 in to 12 ft
0.15 to 3.6 m
3 in to 50 ft.
75 mm to 15 m
lnches to Feet
mm to m
Depends on Tank Dim.

Open

Any type including

Open

corrosive or dirty
Clean

Tank

Open
Both
Both

Condition
of Liquid

Any type with proper

selection
Any type with proper
selection
Any type with proper
selection

Open

Any type

Both

Any type

Wide

Both

Any type

Wide
Wide
One or more Points

80th

Nonconductive

80th
80th

Any type
Conductive

Float-and-Cable
A ftoat-and-cable or ftoat-and-tape instrument (Figure 3-1) measures
liquid level by transmitting to a mechanism the rise and fall of a ftoat
thai rides on the surface of the liquido Mechanisms are available to
accommodate level variations ranging from a few inches to many feet.
Float-and-cable devices are used primarily in open tanks, whereas ftoat
level switches mar be designed to operate in a pressurized tank.
Float devices bave the advantage of simplicity and are insensitive
to density changes. Their major disadvantage is their limitation to reasonably clean liquids. Thrbulence mar also create measurement problems. The ftoat and cable technique does Dot lend itselfto the transmitter concept as well as do some of the following techniques.

LF;VEL AND DENSITY MEASUREMENTS 71

Fig. 3-1. Float-and-cablerecorder.

Displacement

(Buoyancy)

The displacement, or buoyancy, technique is a type of force-balance

transmitter (Figure 3-2). It mar be used to measure liquid level, interface, or density by sensingthe buoyant force exerted on a displacer by
the liquid in which it is immersed. The buoyant force is converted by a
force-balance pneumatic or electronic mechanism to a proportional 3 to
15 psi, 20 to 100kPa, 4 to 20 mAldc or 10 to 50 mAldc signal.

Fig. 3-2. Buoyancy type level measuring transmitter.

72

FEEDBACK PROCESSCONTROL

Archimedes' principIe states that a body immersed in a liquid will

be buoyed upward by a force equal to the weight of the liquid displaced. The displacer element is a cylinder of constant cross-sectional
area and heavier than the liquid displaced. It must be slightly longer
than the level change to be measured (Figure 3-3). Since the displacer
must accurately sense the buoyant force throughout the full range of
measurement, displacer lengths and diameters wiU vary according to
the particular process requirements.
The formula to determine the buoyant force span for liquid level
applications is as foUows:

where:
f
V
Lw
L
B

=
=
=
=
=

buoyant force span (lbf or N)

total displacer volume (cubic inches or cubic centimeters)
working length of displacer (inches or millimeters)
total displacer length (inches or millimeters)
a constant (weight of unit volume of water) (0.036 Ibf/in3
or 9.8 x 10-3N/cm3
SG = specific gravity of the lquid
The minimum buoyant force span for the buoyancy level transmitters is 1.47 pounds-force or 6.7 newtons. The maximum mass of the
displacer-plus-hanger cable must not exceed 12 pounds or 5.4 kilograms (or buoyant force span x 6, whichever is smaller).
For a cylindrical displacer, the formula for the volume is:

Fig. 3-3. Buoyancytype level transmitterinstallation.

LEVE.L AND DENSITY MEASUREMENTS

73

In the above formula, as well as the sizing of displacers, the working

volume considered is .1imitedto the cylindrical volume of the displacer
and does not include the volume of the dished end pieces or hanger
connection.
U nless certain precautions are taken, buoyancy transmitters are
not recommended for extremely turbulent process conditions. The turbulence mar cause the displacer to swing erratically, resulting in unpredictable measurement effects or physical damage to the displacer,
transmitter, or vessel. In such cases, some form of displacer containment should be used.
Buoyancy transmitters mar be applied readily to glass-lined vessels, vessels in which a lower connection is not permissible or possible,
density applications with ftuctuating pressures or levels, and high-

temperature service. Interface level measurementmar be accomplished by permitting the interface level to vary over the length of the
displacer.
f = V (B) (SG difi)
SG difI = SG lower liquid minus the SG upper liquid
Head or Pressure

Measurementof head, or pressure,to determine level is the most

common approach.The ways in which level mar be determinedby
measuringpressureare many and varied. Some of the more common
techniques are presented befe.
Hubble Thbe Method
In the air purge, or hubble tube system (Figure 3-4), liquid level is
determined by measuring the pressure required to force a gas into the
liquid at a point beneath the surface. In this way, the level mar be
obtained without the liquid entering the piping or instrument.
A source of clean air or gas is connected through a restriction to a
hubble tube immersed a fixed depth in the tank. The restriction reduces
the air flow to a minute amount, which builds up pressure in the hubble
tube until it just balances the fluid pressure at the end of the hubble
tube. Thereafter, pressure is kept at this value by air bubbles escaping
through the liquido Changes in the measured level cause the air pressure in the hubble tube to build up or dropo A pressure instrument
connected at this point can be made to register the level or volume of
liquido A small V-notch is filed in the bottom of the tube so that air

74 FEEDBACK PROCESSCONTROL

Fig. 3-4. Usingbubblepipe.

emerges in a steady stream of small bubbles rather than in intermittent

large bubbles (Figure 3-5).
An advantage ofthe hubble tube method is that corrosive or solids
bearing lquids can damage only an inexpensive, easily replaced pipe.
Diaphragm Box
The diaphragm box is shown in Figure 3-6. It is similar to the diaphragm seals used for pressure gauges (Figure 2-20, p. 57) except that
the fill is air and the diaphragm is very slack, thin, and flexible. The

DETAlL

OF NOTCH

IN BUBBLE PIPE

Fig. 3-5. Detail or notch in bubblepipe.

LEYEL AND DENSITY MEASUREMENTS 75

Fig. 3-6. Diaphragrn boxo

diaphragmbox is usuallysuspendedfrom a chain. The instrumentthat

sensesthe pressurechangesand relatesthemto level mar be mounted
above or below the vessel.The diaphragmbox is used primarily for
water level measurementin openvessels.
Pressure and Differential Pressure Methods Using Differential Pressure
Transmitters
These are the most popular methods of measuring liquid level and are
shown in Figure 3-7. With an open tank, the pressure at the high
pressure side of the cell is a measure of the liquid level. With a closed
tank, the etfect of tank pressure on the measurement is nullified by
piping this pressure to the opposite side of the cell.
When a sealing fluid is used, it musi possess a specific gravity
higher than that of the liquid in the vessel. The sealing fluid must also
be immiscible with the liquid in the vessel.

76 FEEDBACK PROCESSCONTROL

OPEN TANK

Fig. 3-7.
Span = xGL
Suppression= yGL+ zG,

Span = xGL
Suppression= yGL+ zG,

Span = xGL
Elevation =/dG. -yGL

Where GL = specific gravity oCliquid in tank

G. = specific gravity oCliquid in outside filled line or tines
IC transmitter is at level oClower tank tap, or iC air purge is used, z = O.
(Note: The density oCthe gas in the tank has been disregarded in these calculations.)

EXAMPLE: Assume an open tank withX = 80 inches,y = 5 inches, and

z = 10 inches. The specific gravity ofthe tank liquid is 0.8; the specific
gravity of the liquid in the connecting leg is 0.9.
Span = 80 (0.8) = 64 inches head of water
Suppression = 5 (0.8) + 10 (0.9)
= 4 + 9 = 13 inches head of water
Range = 13 to 77 inches head of water
EXAMPLE:Assume a closed tank with X = 70 inches, y = 20 inches,
and d = 100 inches. The specific gravity of the tank liquid is 0.8; a
sealing liquid with a specific gravity of 0.9 is Ilsed.
Span = 70 (0.8) = 56 inches head of water
Elevation = 100 (0.9) -20
= 90 -16

(0.8)

= 74 inches head of water

Range = -74 to -18 inches head of water

LEVEL AND DENSITY MEASUREMENTS 77

Fig. 3-8. The repeateris a flanged,force-balance,pneumaticinstrumentthat measures

pressureand delivers an outputequalto the measuredpressure.

(minus sign indicates thai the higher pressure is applied to the low side
of the cell.)
At times the liquid to be measured possessescharacteristics thai
create special problems. Assume, for example, thai the liquid being
measured will solidify if it is applied to a wet leg and it is virtually
impossible to keep the leg dry. This type of application could use
a pressure repeater. The repeater (Figure 3-8) is mounted above
the maximum level of the liquid, and the liquid level transmitter is
mounted near the bottom ofthe tank. The pressure in the vapor section
is duplicated by the repeater and transmitted to the instrument below,
(Figure 3-9). Thus, the complications of a wet leg are avoided, and a
varying pressure in the tank will Dot atrect the liquid level measurement.
The pressure range thai mar be repeated is O to 100 psi or Oto 700
kPa. The supply pressure musi exceed the pressure to be repeated by

78 FEEDBACK PROCESSCONTROL

Fig. 3-9. Liquid-level transmitter(electronicor pneumatic)used with a repeater.

20 psi or 140kPa. A somewhatsimilar configurationmar be arranged

by usinga sealedsystemas describedin Chapter2 and shownin Figure
2-21,p. 58.
Capacitance
If a probe is inserted into a tank and the capacitance measured between
it and the tank, a sizable change in capacitance will occur with liquid
level. This phenomenon is due primarily to the substantial difIerence
between the dielectric constant o air and that o the liquid in the tank.
This technique is best applied to nonconductive liquids, since it is best
to avoid the problems generated by conducting materials like acids

(Figure 3-10).

Fig. 3-10. Capacitance.

LEVEL AND DENSITY MEASUREMENTS 79

Conductance
Conductivity level sensors consist of two electrodes inserted into the
vessel or tank to be measured. When the level rises high enough to
provide a conductive path from one electrode to the other, a relay (solid
state or coil) is energized. The relay mar be used for either alarm or
control purposes. Conductivity then becomes either point control or an
alarm point. The lquid involved must be a conductor and must Dot be
hazardous if a spark is created. Level by conductivity finds occasional
applications in process plant s (Figure 3-11).
Radiation

A radiation level measurement generally consists of a radioactive

source on one gide of the tank and a suitable detector on the other. As
the radiation passes through the tank, its intensity varies with the
amount of material in the tank and can be related to level. One advantage is that nothing comes in contact with the liquido Among the
disadvantages are the high cost and the difficulties associated with
radioactive materials. Radioactive techniques do bave the ability to
solve difficult level-measuring problems (Figure 3-12).
Weight

Occasionally the measurement or the contents or a tank is so difficult

that Done or the usual schemes will work. When this occurs it may be
advantageous to consider a weighing system. Weight cells, either hy-

LlNE

Fig. 3-11. Conductivity.

80 FEEDBACK PROCESSCONTROL

DETECTOR

~~~Jr;~-~~'-==-L__~
\~::::-

--

\~"
\ ::;"""""
\ \'

\\"

\"

INDICATOR

\,

-+

~-g

i~CONVERTER

NUCLEONIC(RADIATION)
Fig. 3-12. Radiationtechnique.

draulic or strain gauge, are used to weigh the vessel and its contents.
The tare weight of the tank is zero adjusted out of the reading, which
will result in a signal proportional to tank contents (Figure 3-13).
One advantage of the weight system is that there is no direct contact with the contents of the tank and the sensor. However, the system
is not economical, and varying densities mar confuse the relationship
between signal and true level.
Ultrasonic

The ultrasonic level sensor(Figure 3-14) consists of an ultrasonic

generator or oscillator operating at a frequency of approximately

Fig. 3-13. Level measurement using weight method.

LEVEL AND DENSITY MEASUREMENTS 81

Fig. 3-14. Ultrasonictechnique.

20,000 Hz and a receiver. The time required for the sound waves to
travel to the liquid and back to the receiver is carefully measured. The
time is a measure of level. This technique has excellent reliability and
good accuracy. Furthermore, nothing comes in contact with liquid in
the tank, which minimizes corrosion and contamination. The only generallimitation is economic.
Thermal

One approach employing a thermal sensorto determine level relies on a

difference in temperature between the lquid and air above it. When the
lquid contacts the sensor, a point determination o level is made.
Another approach depends on a difference in thermal conductivity
between the lquid level sensedand the air or vapor above it. Thermal
techniques are inexpensive, but bave achieved only modest popularity
in process applications.
Density

Measurement

Methods

for Liquids

and Liquid

Slurries

Density is defined as mags per unit volume. Specific gravity is the

unitless ratio of the density of a substance to the density of water at
some standard temperature. In Europe, this is known as relative den-

sity.
The measurement and control of liquid density are critical to a great
number of industrial processes. Although density can be of interest, it
is usually more important as an inference of composition, of concentration of chemicals in solution, or of solids in suspension.

82 FEEDBACK PROCESSCONTROL

Many methods are available to determine concentration, including

measurement of changes in electrolytic conductivity and of rise in boiling paint. This section deals with common industrial methods of
measuring density continuously (Table 3-2).
Agitation in a process tank where density is being measured musi
be sufficient to ensure uniformity ofthe lquid. Velocity effects musi be
avoided; agitation should be sufficient to maintain a uniform mixture
without affecting head pressure at measurement points.
Nearly aU industrialliquid density instruments utilize the effect of
density changes on weight as a measurement. Such instrumentation
falls into three major categories, defined by the principIes employed.
These include two other categories, radiation and vibration methods.
For density measurement, a cylindrical displacer is located so as to
be fuUy submerged. The buoyant force on the displacer changes only as
a function of changing lquid density, and is independent of changing
lquid level in the vessel. Buoyancy transmitters based upon the Archimedes principIe may be readily adapted to the measurement of
lquid density or specific gravity.
Hydrostatic Head
Proved suitable for many industrial processes is the method that continuously measures the pressure variations produced by a fixed height
of liquid (Figure 3-15). Briefly, the principIe is as follows: The ditIerence in pressure between any two elevations below the surface (A and
B) is equal to the ditIerence in liquid head pressure between these
elevations. This is true regardless of variation in level above elevation

TABLE 3-2. Liquid Density Measurement

Minimum
Method

Span

Hydrometer

0.1
0.005
0.05
0.05
O.O~
0.05

Displacer
Hydrostatic head
Radiation
Weight oCfixed volume
Vibrating U-Thbe

Condition
oi Liquid

Accuracy

Clean
Clean

:!:1%

Any
Any

Clean
Clean

as % Span

:!:1%
~ to 1%
1%
1%
1-3%

Fig.

LEVEL AND DENSITY MEASUREMENTS 83

lz=ZZarZz=
3-15. Fixed heightofliquid for densitymeasurement.

A. This difference in elevation is represented by dimensionH. Dimension

(H) must be multiplied by the specific gravity (G) ofthe liquid to obtain
the difference in head in inches or metres of water, which are the
standard units for instrument calibration.
To measure the change in head resulting from a change in specific
gravity from GI to G2, it is necessary only to multiply H by the difference between G2 and GI. Expressed mathematically:
P = H (G2 -GJ

(3-3)

The change in head (P) is differential pressure in inches or metres of

water. GI is mnimum specific gravity and G2 is maximum specific
gravity.
It is common practice to measure only the span of actual density
changes. Therefore, the measurement zero is suppressed; thai is, the
instrument "zero," or lower range value, is elevated to the minimum head pressure to be encountered. This allows the entire instrument measurement span to be devoted to the differential caused by
density changes. For example, ifG1 is 1.0 andH is 100inches, the span
of the measuring device musi be elevated 100 inches of water
(H x G). For a second example, if GI is 0.6 and H is 3 metres, the
zero suppression should be 1.8 metres of water.
The two principal relationships thai musi be considered in selecting
a measuring device are:

Vibration

84 FEEDBACK PROCESSCONTROL

Span = H X (G2,- GJ
Zero suppression = H x GI,

(3-4)

For a given instrument span, a low gravity span requires a greater H

dimension (a deeper tank).
Radiation

Density measurements by this method are based on the principIe that

absorption of gamma radiation increases with the increasing specific
gravity of the material measured.
The principal instrumentation includes a constant gamma source
(usually radium), a detector, and an indicating or recording instrument.
Variations in radiation passing through a fixed volume of ftowing process liquids are converted iota a proportional electrical signal by the

detector.

Damping a vibrating object in contact with the process fluid increases

as the density ofthe process fluid increases. This principIe is applied to
industrial density measurement. The object that vibrates (from externally applied energy) is usually an immersed reed or plate. A tube or
cylinder that conducts, or is filled with, the process fluid can also be
used. Density can be inferred from one of two measurements: changes
in the natural frequency of vibration when energy is applied constantly,
or changes in the amplitude of vibration when the object is "rung"
periodically, inthe "bell" mode.
Temperature

Effects and Considerations

The density of a liquid varies with expansion due to rising temperature,

but Dot allliquids expand at the same rate. Although a specific gravity
measurement must be corrected for temperature effects to be completely accurate in terms of standard reference conditions for density
and concentration, in most cases this is Dot a practical necessity.
For applications in which specific gravity measurement is extremely critical, it m~y be necessary to control temperature at a constant value. The necessary correction to the desired base temperature
can then be included in the density instrument calibration.

LEVEL AND DENSITY MEASUREMENTS 85

Differential

Pressure Transmitter

There are a variety of system arrangements for hydrostatic head density measurements with d/p Cell transmitters. Although ftange-mounted
d/p Cell transmitters are often preferred, pipe-connected transmitters
can be used on lquids where crystallization or precipitation in stagnant
pockets will noi occur.
These d/p Cell transmitter methods are usually superior to those
using bubble tubes. They can be applied wherever the vessel is high
enough to satisfy the mnimum transmitter span. They are also well
suited for pressure and vacuum applications.
Constant level overftow tanks permit the simplest instrumentation
as shown in Figure 3-16. Only one d/p Cell transmitter is required. With
H as the height of lquid above the transmitter, the equations are still:
Span = H X (G2

G}. zero suppression = H X GI

Applications with level and/or static pressure variations require compensation. There are three basic arrangements for density measurement
under these conditions. FiTSt, when a seal fluid can be chosen that is
always heavier than the process fluid and will Dot mix with it, the
method shown in Figure 3-17 is adequate. This method is used extensively on hydrocarbons with water in the wet leg.

Fig. 3-16. Constant level, open overftow tanks require only one d/p Cell transmitter for
density measurement.

Fig.

86 FEEDBACK PROCESSCONTROL

y-:'--=(

\
H

DENSITY
~
SIGNA L

A I

=
UQUID

a':iJ-l

PURGE

(SP.GR.=GS)
3-17. In an open or closed tank with varying level and/or pressure. a wet leg can be
filled with seal fluid heavier than the process liquido

For a wet leg fluid o specific gravity Gs. an elevated zero transmitter
musi be used. The equations become:
Span = H x (G2 -GJ,

zero elevation = H x (GB -GJ

When no seal or purge lquid can be tolerated, there are ways to provide a "mechanical seal" for the low-pressure leg, or for both legs, if
needed. Figure 3-18 shows the use of a pressure repeater for the upper
connection.
The repeater transmits the total pressure at elevation B to the low
pressure side of the d/p Cell transmitter. In this way, the pressure at
elevation B is subtracted from the pressure at elevation A. Therefore,
the lower transmitter measures density (or H x G, where G is the
specific gravity of the lquid). The equations for the lower transmitter

are:
Span = H X (G2 -GJ
Zero suppression = H x G1

The equationfor the upper repeateris:

Output (maximum) = d 1 (max) x G2 + P (max),

Fig.
whered1

LEVEL AND DENSITY MEASUREMENTS 87

3-18. In an open or closed tank with varying level and/or pressure where seal fluid
or purge is Rot suitable, a pressure repeater can be used.

is the distance from elevation B to the lquid surface, andP is

the static pressure on the tank, if any. When there is no pressure on the
tank, the upper repeater output is a measurement of level.
Transmitter locations should be as high as possible above the bottom ofthe tank. The process lquid then can be drained below this level
for removal and maintenance of the transmitter.

Questions

3-1. An opentank containsa liquid of varying densityand the level within

the tank mustbe accuratelymeasured.The bestchoiceof measuringsystem
would be:
a. Hubble tube
b. Diaphragmbox
c. Float and cable
d. Head type with differentialpressuretransmitter
3-2. A chain-suspended
diaphragmbox and pressureinstrumentare usedto
measureliquid level in a tank which is 10feet in diameterand 15feetdeep.
The tank containsa liquid that hasa specificgravity of 0.9 at ambient
temperature.The spanof the pressureinstrumentrequiredis approximate1y:
a. 10 psi
c. 5 psi
b. 25 psi
d. 15psi

88 FEEDBACK PROCESSCONTROL

3-3. A pressurized tank contains a lquid, SG = 1.0, and the level measuring
pressure taps are 100inches apart. A pneumatic differential pressure
transmitter is used to measure level. The leg that connects the top of the tank
to the transmitter is filled with tank lquid. The top tap should be connected

to:
a. The low-pressuretap on the transmitter.
b. The high-pressuretap on the transmitter.
, Either the high- or the low-pressuretap.
d. The high-pressuretap but througha sealchamber.
3.4. In Qu~stion3-3the bottom connectionis madeto the low-pressuretap
on the transmitterand the top is madeto the high-pressure
tap. The signal
output, if the transmitteris calibratedOto 100inches,whenthe tank is full
will be:
a. Topscalevalue
c. O
b. Bottom scalevalue
d. A measureof density
3.5. A plant hasa water tower mountedon top of an 80-footplatform. The
tank is 30 feet high. Whatis the heightof water in the tank if a pressuregauge
on the secondfloor, height 15feet,reads 40 psi.
a. Full
c. 4.74feet
b. 12.42feet
d. 27.42feet
3-6. A displaceris 5 inchesin diameterand 30 incheslong. If it is submerged
to a depthof 20 inchesin lquid sa = 0.8, whatforce will it exert on the top
works?
a. 11.3pounds
c. 35.5pounds
b. 426.3pounds
d. 1.47pounds
3.7. A closed tank level (Sa = 0.83)is measuredwith a differentialpressure
transmitter.The level mar vary from 10to 100inches.The high-pressuretap
is 10inchesabovethe transmitterand a watersealfluid is used.A pressure
repeateris usedfor the top tank pressure.The differentialpressure
transmittershouldbe calibrated:
a. 10to 100inches
b. 10to 84.7inches
c. 8.3 to 74.7inches
d. 8.3 to 98.3inches
3.8. To resistthe corrosiveeffectsof a very unusual,highlyexplosive
chemical,a storagetank is lead-lined.Level measurement
is difficult because
the materialalso solidifiesonce it entersa measuringtap. To measurethe level
with accuracythe bestchoicewould be:
a. Weighthe tank and its contentsand zero out the tare weightof the tank.
b. Use a radioactivelevel measurement.
c. Install a conductivitylevel measurement.
d. Use a thermal conductivitylevel detector.

LEVEL AND DENSITY MEASUREMENTS 89

3-9. A displacer48 incheslong and 1%inchesin diameteris usedto measure

densityon a 4 to 20 mA dc transmitter.Its total submergedweightchanges
from 1.42poundsto 5.7pounds.If the densityis checkedand determinedto
be 0.4 whenthe output is 4 mA dc, the densityat an outputof 12mA dc is:
a. 0.6
c. 1.0
b. 0.8
d. 1.6
3-10. A densitymeasuringsystemis set up as shownin Figure3-18.The H
distanceis 100inchesand the lower transmitteris calibratedto a 120-inch
span.If the transmitteris pneumaticand deliveringa 12-psisignal,the
specificgravity of the tank's contentsis:

a. 1.0

c. 0.9

b. 0.83

d. 1.2

Flow Measurement

The process industries by their very nature deal constantly with flowing
fluids, and measurement of these flows is essential to the operation of
the plant. These measurements are indicated, recorded, totalized, and
used for control. Flow measurement is generally the most common
measurement found in the process plant.
Fluid flow measurement is accomplished by:
Ao Displacement
lo Positive displacement meters
20 Metering pumps
Bo Constriction Type, Differential Head
lo Closed conduit or pipe
ao Orifice plate

bo Venturi tube
Co Flow nozzle
do Pitot tube
eo Elbow
fo Target (drag force)
go Variable aTea (rotameter)
20 Open channel
ao Weir

bo Flume
90

FLOWMEASUREMENT 91

C. Velocity Flowmeters
1. Magnetic
2. Turbine
3. Vortex or swirl
4. Ultrasonic
5. Thermal
D. Mass Flowmeters
1. Weighttypes
2. Head and magnetictypes compensatedfor temperature,pressure,
and density
3. Gyroscopeprecisiontypes
4. Centrifugalforce (torque)types
Positive displacementmeters and meteringpumps measurediscrete quantities of flowing fluid. This flow is indicated in terms of an
integrated or totalized flow volume (gallons,cubic feet, litres, cubico
metres, and the like). A typical applicationof this type of flow measurementis custodytransfer, and familiar examplesaredomesticwater
metres and gasolinepumps. The other types of meters listed above
measureflow rate.
lt is flow rate-quantity per time, suchas gallons per minute or
liters per second-which is most generallyused for measurementand
related control applications in the processplant. The most common
rate meter is the constriction or headtype.
Constriction

or Differential

Head Type

In this type of flow measurement a primary device or restriction in the

flow line creates a change in fluid velocity that is sensedasa differential
pressure (head). The instrument u~ed to sensethis differential pressu~e
is called the secondary device, and this measurementis related to flow
rate. The flow-measuring system, orlen called a flowmeter, consists of
primary and secondary devices properly connected.
The operation of the head-type flow-measuring system depends on
a theorem first proposed by Daniel Bernoulli (1700-1782). According to
this theorem the total energy at a point in a pipeline is equal to the total
energy at a second point if friction between the points is neglected.
The energy balance between Points 1 and 2 in a pipeline can be
expressed as follows:

+ Z2 = ~p + ~
(VmJ2+ Zl

92 FEEDBACK PROCESSCONTROL

where:
P
Vm
Z
p
G

=
=
=
=
=

Static pressure, absolute

Fluid stream velocity
Elevation of center line of the pipe
Fluid density
Acceleration due to gravity

If the energy is divided into two forms, static head and dynamic or
velocity head, some of the inferential flow devices can be more easily
studied.
For example, with an orifice plate, the change in cross-sectional
aTeabetween the pipe and the orifice produces a change in flow velocity
(Figure 4-1). The flow increases to pass through the orifice. Since total
energy at the inlet to, and at the throat or, the orifice remains the same
(neglecting losses), the velocity head at the throat must in.crease,causing a corresponding decrease in static head. Therefore, there is a head
difference between a point immedlately ahead of the restriction and a
point within the restriction or downstream from it. The resulting differential head or pressure is a function of velocity that can then be related
to flow. Mechanical flow-measuring instruments use some device or
restriction in a flow line that results in such a differential head.
The following should clarify these relationships. Assume a tank as
shown in Figure 4-2. A flow line enters the tank and replaces the out-

Fig. 4-1. Orifice plate differential producer. The difference in head (pressure) atP, and
P 2is a function of velocity which can be related to ftow.

FLOW MEASUREMENT 93

-+
a=A.J2;

-..Q =AV2-;;h

Fig. 4-2. Free and tank-to-tankflow. usedto simplify conceptofflow formula(4-1).

ftow through the orifice located near the bottom of the tank. If the level
in the tank is H, the velocity of outftow will be:
y2 = 2GB

or

This is the lawof falling bodies that is developed in most physics

books. The volume of liquid discharged per time unit through the
orifice c3;nbe calculated by geometry.

=AV
Substituting the expression for velocity in this equation:

(4-1)
This expression may also be used to calculate the rate offtow past a
point in a pipe. The actual ftow rate will be less than this equation will

V=Y2GH
Q
Q=A\f2Gii

O)
Fig.

94

FEEDBACK PROCESS CONTROL;

calculate. The reduct"ionis caused by several factors, including friction

and contraction of the stream within the pipe. Fortunately all common
primary devices bave been extensively tested and their coefficients
determined. These data become part of the ftow calculations.
Primary Devices

The orifice plate is the most popular primary device found in most
process plants. Orifice plates are applicable to all clean fluids, but are
Dot generally applicable to fluids containing solids in suspension(dirty
fluids).
A conventional orifice plate consists of a thin circular plate containing a concentric hole (Figure 4-3). Th plate is usually made of stainless
steel but other materials, such as monel, nickel, hastelloy, steel, glass,
and plastics, are occasionally used. The most popular orifice plate is the
sharp-edged type. The upstream face of the plate usually is polished
and the downstream gide is often counterbored to prevent any interference with the flowing fluid. The bore in the plate is held to a tolerance
of a few ten-thousandths of an inch in small sizes and to within a few
thousandths in sizes above 5 inches in diameter.
ln addition to the conventional, sharp-edged, concentric plate,
there are others that bave been designed to handle special situations.
These special types constitute only a small percentage of the total, but
they do occasionally solve a difficult problem.
There are two plates (Figure 4-3) designed to accommodate limited
amounts of suspended solids. The eccentric plate has a bote that is
bored off-center, usually tangent to the bottom of the flow line (inside
periphery of the pipe). The segmentat orifice plate has a segment removed from the lower hal of the orifice plate. ln addition, there are

CONCENTRIC

4-3. Orificeplate types.

ECCENTRIC

SEGMENTAL

Reynolds

FLOW MEASUREMENT 95

quadrant-edged or special-purpose orifice plates with rounded edges.

This design minimizes the effect of low Reynolds number (see below).
Orifice plates along with other primary devices operate most satisfactorily at high ftow rates. The reason is the change in ftow coefficient
that occurs when the ftow rate changes from high to very low. When
this happens the ftow velocity distribution changes from uniform to
parabolic. This mar be related to a ratio of inertia forces to viscous
forces called Reynolds number.
Number = Average Fluid Velocity x Orifice Diameter
Absolute Viscosity
In most process plants the Reynolds numbers encountered will be
high-between 10,000 and 1,000,000. Under these conditions the ftow
coefficients stay fixed and head-type ftow measurementsare reasonably
accurate. To achieve the ultimate in accuracy, a Reynolds number
correction is made and this is described in the text Principies and Practices o/ Flowmeter Engineering by L. K. Spink. At low Reynolds number, the ftow changes from turbulent to parabolic or laminar and the
coefficients undergo a change. The changeover cannot be precisely
defined, which makes the application of head-type ftow measurements
difficult at low-ftow velocities. The quadrant-edged orifice has the ability to perform well at lower velocities than the square-edged plate.
However, it does Dot perform as well at high Reynolds numbers (velocities).
Orifice plates that are to handle lquids containing small amounts of
dissolved air should contain a small vent hole bored at the top to permit
the air to pass through the plate. Specifications bave been established
for the size of this vent hole in relation to the size of the orifice. Plates
for use with steam should bave a similar hole drilled at the bottom to
drain any condensed steam. In certain ftow measurements, the orifice
plate should also bave a bottom drain hole to permit the condensation

to drain.
Table 4-1 summarizes application factors to help in deciding
whether the orifice plate can do a particular job.
Another primary device that is frequently found in the process
plant is the Venturi tube shown in Figure 4-4. The Venturi tube produces a large differential with a mnimum permanent pressure loss. It
has the added advantage of being able to measure ftows containing
suspended solids. The most significant disadvantage is its cost which,
when compared with other primary devices, is high.

~~
G
VP
Fig.

96 FEEDBACK PROCESSCONTROL

Table4-1. Primary Flow MeasurementDevices

" t
:

o:
Accuracy (empirical
data)

None

Pressure loss #

, P

Use on dirty service

Differenlial produced
for given ftow and size

p
E

E
P
P

E
P
F

G
G
E

~
.9"

.s~

".."
.~
~~~~"to"~
~5

"='"
" E

E
EG

None
E
E

GG

F
None

Non.

VP

G
G
F

U
E

F
P

F
F

VP
E

P
F

For liquido containing

vapors

For vapors containing

condensate

For viscous ftows

First cost

EGG *
F
E

E
U
G

Ease of changing
capacity

E
G

Ease of instalIalion

Excellent

G
F

Good
FairP
Poor

VP
U
/I

Very poor
Unknown
E indicates lowest loss, P highest, etc.

EG

EG
p
E
G

None
E

u
uVP

E
F

G
F

.For measuring velocity at oRe point in conduit the

well-designed Pitot tube is reliable. For measuring
total ftow, accuracy depends on velocity traverse.
t Excellent in verticalline if ftow is upward.
4;Excellent in verticalline if ftow is downward.
Requires a velocity traverse.

Several other primary devices mar be classified as modifications of

the Venturi tube. The Lo-Loss and Dall tubes are similar in many ways
to the Venturi. The flow nozzle (Figure 4-5) has some similarity but
does not bave a diffuser cone, and this lmits its ability to minimize
permanent pressure loss. The Venturi family ofprimary devices is often

4.4. The HerschelVenturitube,developedin 1888,to producea largedifferentialpressure

with a smallheadloss.

Fig.

FLOW MEASUREMENT 97

4-5. Flownozzleassembly.

chosen to minimize permanent pressure loss or to cope with solids

suspended in the flowing material. Table 4-1 provides aid in making a
selection.
The Pitot tube is another primary device. It has the advantage of
practically no pressure dropo Its limitations are its inability to handle
solids carried by the flowing material and somewhat limited accuracy.
Solids tend to plug the openings in the tube, and the classical Pitot tube
senses impact pressure at only one point, thus decreasing accuracy. A
multi-impact opening type (annubar tube) is available which improves
the potential accuracy. In the process plant Pitot tubes are used more
for testing than for continuous use. It is possible to install a Pitot tube
into a flow line under pressure if the required equipment is available.
Occasionally a pipe elbow (Figure 4-6) mar be used as a primary
device. Elbow taps bave an advantage in thai most piping configurations contain elbows thai can be used. If an existing elbow is used, no
additional pressure drop is created and the expense involved is minimal. The disadvantages are thai accuracy will be Iacking and dirty
flows mar tend to plug the taps. Repeatability should be reasonably
good. Two taps locations are used at either 45 or 22Y2degrees from the

Fig.

98 FEEDBACK PROCESSCONTROL

rb

4-6. Elbow(usedas prirnarydevice).

inlet face ofthe elbow. At low-ftow velocities, the differential produced

is inadequate for good measurement, thus the elbow is a usable choice
only for high top-scale velocities.
The target or drag body ftow device (Figure 4-7) is in many ways a
variation of the orifice plate. Instead of aplate containing a bore
through which the ftow passes,this device contains a solid circular disc
and the ftow passes around il. An essential part of the measurement is
the force-balance transducer that converts this force into a signal. This
signal is related to ftow squared, just as it wouid be with any other
primary device connected to a differential pressure transmitter. The

4-7. Targetflowmeter.

FLOW MEASUREMENT 99

target meter mar be generally applied to the measurement of lquids,

vapors, and gases. Condensates pass freely along the bottom of the
pipe, white gases and vapors pass easily along at the top. The target
flow device has demonstrated the ability to handle many difficult assignments Dot possible with other primary elements. Both pneumatic
and electronic force-balance transducers can be used, making either
type signal, pneumatic or electrlc, available.
Secondary

Oevices

Secondaryinstrumentsmeasurethe differential produced at the primary devices and convert it into a signal for transmissionor into a
motion for indication, recording, or totalization.
Secondaryinstrumentsinclude the mercury manometer,the socalled mercurylessor diaphragm(beUows)meterand various types of
force-balanceand motion-balancepneumaticand electronictransmitters.
In the past the mercury manometer was an extremely popular secondary device, but the high price and danger of mercury bave all but
eliminated it from everyday use. However, a few mercury manometers
are still employed for ftow rate measurements in gas pipelines. The
bellows or diaphragm secondary devices are popular when a direct
indication or record is desired. The most popular secondary device is
the force-balance transmitter. The reasons for this popularity are virtually unlinited adjustability, no harm from extensive overrange, and
the a vailability of a standard signal that can be fed into a recorder, a
controller, or other instruments that may be combined to form a sys-

tem.
Still another detail is the mounting of the secondary device (locations are shown in Figure 4-8). With liquid ftow, the secondary device is

LIQUID FLOW

Fia. 4.8. Flow.

GAS FLOW

STEAM FLOW

100 FEEDBACK PROCESSCONTROL

below the pipeline being measured to ensure that the connecting lines
attached to the sicle of the pipe remain liquid filled. With steam, the
lines should always remain filled with condensate; but with gas, the
secondary device is mounted above the ftow line to drain away any
liquid that mar be present.
When any primary device is installed in a pipeline, the accuracy
will be improved if as much straight run as possible precedes il.
Straight run beyond the primary device is of far less concem.
Still another important consideration is the location of the pressure
taps. With nozzles, Venturi, Dall and Pitot tubes, and elbows, the
pressure tap locations are established. For orifice plates, however, a
variety of tap locations are used. These are ftange, corner, vena contracta, or D and D/2, along with full ftow or pipe taps. Figure 4-9
describes these tap locations, and Figure 4-10 shows the importance of
tap location. In general, ftange taps are preferred except when physical
limitations make pipe taps advantageous. Corner taps require a special
ftange and vena contracta tap locations relate to orifice openings.
Corner taps are advantageous for pipe sizes under 2 inches.
Figure 4-11 describes the permanent head loss caused by the primary device selected.
Relating

Flow Rate to Differential

The two basic formulas bave already been introduced. In the English
system, G = 32 feet per second squared, and in the metric system,
G = 980 centimetres per second squared. The height (h) would be expressed in feet in the English system and centimetres in the metric

Fig. 4-9. Taplocationsfor orifices.

FLOW MEASUREMENT 101

Fig. 4-10. Differentialpressureprofile with orifice plate.

t

system. In Equation 4-1, area (A) would be expressed in square feet

and in the metric system in square centimetres yielding cubic feet per
second or cubic centimetres per second. In the United States, the
English system is the most common. This system is in the process of
being changed to the SI metric, but it appears that the total conversion

Fig. 4-11. Pennanent head los s in a ditferential producer may be plotted as a percent of
measured ditferential for the several types of primary devices. These values must be
interpreted according to the acceptable head loss limit for any particular application.

102 FEEDBACK PROCESSCONTROL

will take a long time to complete. For tros reason the sample calculations that follow will use the English system.
In Equation 4-1, V is in fe et per second; the acceleration due to
gravity (G) is in feet per second squared; H is the height in feet of a
column of the fluid caused by the differential pressure across a primary
device. To express this equation in terms of an equivalent differential
(h), in inches of water, H is replaced by (h/12)G) where G is the
specific gravity of the fluid at flowing temperature.
Substituting V into Equation 4-1:

(4-2)
where:
Q = Volumetric flow (cubic feet per second)
A = Cross-sectional aTea of the orifice or throat of the primary
device (square feet)
h = Differential across the primary device (inches of water)
G, = Specific gravity of the fluid (dimensionless)
G = Acceleration due to gravity (a constant: 32.17feet per second2)
For lquids, it is more useful to express Q in gallons per minute.
AIso, it is convenient to express the area of the orifice or throat interms
of its diameter (d in inches). Substituting in Equation 4-2:

x~

(4-3)

Equation 4-3 most be modified to take into account several factors,

such as the contraction of the jet, frictionallosses, viscosity, and the
velocity of approach. This modification is accomplished by applying a
discharge coefficient (K) to the equation. K is defined as the actual
ftow rate divided by the theoretical ftow rate through the primary
device. Basic texts carry K factors for most primary devices. For
instance, K is listed according to vario us pressure tap locations, and
di D ratios. Where extremely accurate results are required, however,
the K factor must be determined in the laboratory by actual ftow
calibration of the primary device.

FLOW MEASUREMENT 103

Applying K to Equatiort 4-3 gives:

Since both K and d are unknown, another value is defined. The

quantity (S) is set equal to K (d/D)2. Since Kd2 then equals SD2,
substitution produces:

S values (sizingfactor) are tabulated againstbeta ratios (di D) for various

differential devices in Table 4-2.
j
Effect of Temperature on Flow Rate
The above equations give the flow through the primary device at conditions that exist at the time the measurement is made. In most industries, it is more desirable to know the equivalent volume flow at a stated
reference temperature, usually 600P(15.6C).
Por lquids, this correction can be applied by multiplying the equation by the ratio ofthe lquid specific gravity at the flowing temperature
(G,) divided by the lquid specific gravity at the reference temperature
(G). Thus Q gallons per minute at the reference temperature:
Q

= 5.667

SD2

fE::
V G;

Gi

(4-6)

which simplifiesto:
Q = 5.667 SIJ2 YG;l
GI

Similar equations can be developed for steam or vapor ftow in weight

units (such as pounds per ho ur) or for gas ftow in volume units such
as standard cubic feet per hour (scfh).

W (poundsper hour) = 359SD2v'h:y;

Q (scfh) = 218.4SD2~ ~

Pb" T;G

104 FEEDBACK PROCESSCONTROL

Table4-2. Sizing Factors

S'ta
or
d/D

Ratio
0.100
0.125
0.150
0.175
0.200
0.225
0.250
0.275
0.300
0.325
0.350
0.375
0.400
0.425
0.450
0.475
0.500
0.525
0.550
0.575

Square-Edged
Orifice, Flange
Corner or
Radius Taps

Ful/-Flow
Quadrant-

2\-2& 8D

Nozz/e
and

Lo-Loss

Dall

Taps

Venturi

Tub.

Tube

(PipJ

0.005990

0.006100

0.009364

0.009591

0.01349

0.01389

Edged
Orifice

0.01839

0.01902

0.02402

0.02499

0.0305

0.03044

0.03183

0.0390

0.03760

0.03957

0.0484

0.04558

0.04826

0.0587

0.05432

0.05796

0.08858

0.06390

0.06874

0.1041

0.07429

0.08068

0.1210

0.1048

0.08559

0.09390

0.1392

0.1198

0.09776

0.1085

0.1588

0.1356

0.1170

0.1267

0.1109

0.1247

0.1800

0.1527

0.1335

0.1443

0.1251

0.1426

0.2026

0.1705

0.1500

0.1635

0.1404

0.1625

0.2270

0.1900

0.1665

0.1844

0.1568

0.1845

0.2530

0.2098

0.1830

0.207

0.1745

0.2090

0.2810

0.2312

0.2044

0.232

0.1937

0.2362

0.3110

0.2539

0.2258

0.260

0.0700
0.0824
0.0959
0.1106

0.2144

0.2664

0.3433

0.2783

0.2472

0.292

O.(X)()

0.2369

0.3002

0.3781

0.3041

0.2685

0.326

0.625
0.650
0.675
0.700
0.725
0.750
0.775
0.800
0.820

0.2614

0.3377

0.4159

0.3318

0.2956

0.364

0.2879

0.3796

0.4568

0.3617

0.3228

0.3171

0.4262

0.5016

0.3939

0.3499

0.3488

0.4782

0.5509

0.4289

0.3770

0.3838

0.6054

0.4846

0.4100

0.4222

0.6667

0.5111

0.4430

O.4646

0.5598

0.4840

0.5113

0.6153

0.5250

0.6666

0.5635

FLOW MEASUREMENT 105

In Equations 4-8 and 4-9: .

'Yf= Specific weight of the steam or vapor at operating conditions
in pounds per cubic foot
T b = Reference temperature (absolute); (i. e., 460 plus the reference
temperature in F.)
Pb = Reference pressure (psi absolute)
Tf = Operating temperature at the primary device (absolute); (i.e.,
460 plus operating temperature in F.)
Pf = Operating pressure (psi absolute)
G = Specific gravity ofthe gas (molecular weight ofthe gas divided
by the molecular weight of air or the weight of a volume ofthe
gas at a given temperature and pressure divided by the weight
ofanequal volumeofairatthe sametemperature and pressure).
The reference temperature is very often 600P and the reference
pressure atmospheric 14.7 psi absolute. Ifthese are the standard conditions, Equation 4-9 can be simplified to:

(4-10)
Equation 4,.10is applicable to gas ftow only when the pressure
differential is small enough so that gas density does Dot change significantly. A simple rule ofthumb is that the maximum differential in inches
of water should Dot exceed the absolute operating pressure in psi absolut.~. For example, if the gas operating pressure is 22 psi absolute, at a
particular installation, and the maximum differential is 20 inches of
water, Equation 4-10 can be used.
Flow rate 'measurements for gas and steam are more difficult to
make with accuracy than those for liquido The reason is changes in
specific gravity, weight, temperature, pressure, and so on, that mar
occur under operating conditions.
These changes will bave an effect on measurement accuracy and
under certain conditions mar be difficult to predict. An abbreviated set
of tables for the formulas given are included in this book. If more
accuracy is required, more exact equations, along with detailed tables,
such as those found in Principies and Practice oi Flow Meter Engineering
by Spink, should be used.
Another method of performing these ftow calculations is to use a
ftow slide rule. The ftow slide role has the table values incorporated

106 FEEDBACK PROCESSCONTROL

into its scales. If the tables given are used, the resulting accuracy
should be as good as the flow slide rule.
Now several sample problems are given to demonstrate the procedure followed for each type of calculation. Additional problems are
given at the end of tros chapter.
SAMPLEPROBLEMJ A 4-inch schedule 40 pipe carries water that is
measured by a concentric, sharp-edged, orifice plate, d = 2.000 inches,
with flange taps. The differential is measured with an electronic differential pressure transmitter. The transmitter is calibrated Oto 100inches
of water pressure and has an output of 4 to 20 mA dc. If the signal Crom
the transmitter is 18.4 mA dc, find the flow rate.
S/ep 1. Convert the electrical signal to differential pressure.
18.4 -4

x 100 = 90 inches o water

20 -4

Step2. DetermineID of 4-inchschedule40pipe (seeAppendix, table

A-4) = 4.026inches.
2.000 = 0.4968
Step3. Calculatelid = 4:026

Step4. From Table 4-2determineS. This will require interpolation.

S =

0.475

0.1404

0.4968

0.5000

0.1568

(0.5000
0.4968 -0.475
-0.475 )

(0.1568 -0.1404) ~0.1404

S = 0.1547
Step5. Substitute in Equation 4-7:

= 5.667 x 0.1547x 16.21 x 9.487

= 134.83or 135gpm

FLOW MEASUREMENT 107

SAMPLEPROBLEM
2 An elbow is used as a primary device. The taps are
made at 45 degrees. The line is a 6-inch schedule 40 pipe. What is
the water flow rate if the effective radius of curvature is 9 inches
and a differential pressure of 35 inches of water pressure is produced.

= 172.66 .yj"5 =

,021.47gpm

SAMPLEPROBLEM
3 Dry-saturated steam is measured with a ftow nozzle. The d/D is 0.45 and the line size is a S-inch schedule SOpipe. The
static pressure is 335 psi. Calculate the ftow rate at a differential
pressure of 200 inches of water in pounds per hour.
W(pounds per hr.) = 359SD2V:y;

(4-8)

from Table 4-2

s = 0.2026
D = 7.625 inches (Table A-4)
Yf = 0.754 pounds per cubic feet
W(pounds per hour) = 359 x 0.2026 x 7.6252\1200 x 0.754

= 4,228.77x 12.28
W = 51,929 pounds per hour
SAMPLEPROBLEM
4 A 6-inch schedule 40 pipe (1D-6.065) carries fuel
gas with a specific gravity of 0.88. The line pressure is 25 psi. Plowing
temperature is 60oPand the ftow is measured with an orifice plate with
ftange taps. The maximum ftow rate is 2,000,000standard cubic feet per
day. Pind the diameter of the hole to be bored in the concentric orifice
plate if 20 inches of water pressure is full-scale differential.

108 FEEDBACK PROCESSCONTROL

2,000,000
= 7' 727 x S x. 6 0652 V120(25
+-~
24
(460
+ 6~

8,3333.33= 7,727 x S x 36.78 x 1.322

S = 7,727 x83,333.33
36.78 x 1.322= O.18
22
From Table 4-2
.575

.2144

.2218

.2369

.600
~ =
D

( .2218
.2369

-.2144

-.2144

d = (.583)(6.065)

025 + .575 = .5832

= 3.536 inches

SAMPLEPROBLEM5 Gasoline is carried in a 2-inch schedule 40 pipe

(ID = 2.067). A concentric sharp-edged orifice plate with corner taps
is used to measure flow rate. The orifice bare is 1 inch in diameter,
and at full flow, 50 inches of water differential is produced. The
specific gravity of the gasoline is 0.75. What is the flow rate?

Q = 5.667 x 0.1474 x 2.067 x 2.067 x 'V{SO

0-:75

= 5.667 x 0.1474x 2.067 x 2.067 x 8.165

= 29.14 gpm

Fig.

FLOW MEASUREMENT 109

Variable Area Meters (Rotameter)

The variable aTea(Figure 4-12) meter is a form of head meter. In this

flowmeter the aTea of the flow restriction varies so as to maintain a
constant differential pressure. The variable aTeameter, which is often
called a rotameter, consists of a vertical tapered tube through which the
fluid flow being measured passes in an upward direction. A float or
rotor is contained within the tapered tube. This rotor or float is made of
some material-usually metal-more dense than the fluid being measured. As the flow moves up through the tapered tube, it elevates the
float until balance between gravity acting on the float and the upward
force created by the flow is achieved. In achieving this balance of
forces, the area through which the fluid passeshas automatically been
adjusted to accommodate that flow rate.
The tapered tube is often made of transparent material so the float
position can be observed and related to a scale calibrated in units of
flow rate. The rotameter is often used to measure low-flow rate. When
very large flow rates are to be measured, a bypass rotameter may be
used. This consists of an orifice in the main flow line with the variable
aTeameter connected in parallel. An industrial rotameter is useful overthe
upper 90 percent of its scale. The accuracy of the rotameter, like
any head type, depends on many factors. Typical errors may vary
between I and 10 percent of full scale value.
Open-Channel

Flow Rate Measurements

Flow rate measUements

in openchannelsare fundamentalto handling
water and waste. Environmentalconsiderationsbave madethesemea-

4-12. Variablearea(rotarneter)flowmeter.

110 FEEDBACK PROCESSONTROL

surements common in' the typical process plant. Open-channel measurements utilize head meter techniques.
Primary

Devices

The primary devices used in open-channel ftow rate measurements are

weirs and ftumes. There are two basic weirs-the rectangular and the
V-notch. The rectangular weir is made in three varieties (Figure 4-13).
The first has contractions or extensions into the channel that produce a
boxlike opening. The second modification completely suppressesthese
contractions, extending the weir across the entire width of the channel.
The third type, the so-called Cippoletti weir, has end contractions set at
a 4: 1 angle, rather than being perpendicular to the edge of the weir.
Rectangular weirs are primarily designed for larger ftow; their lmit on
range is dictated by the design of the associated channels. Figure 4-14
shows the ranges of weir capacity for the different types and sizes of

welrs.
V-notch weirs are essentially plates (usually metal) that contain a
V-shaped notch (Figure 4-13). The angle of the V can vary, but the
formulas given are for the most common angles 30, 60, and 90 degrees.
V-notch weirs are employed for lower ftow rates than those that would
be measured by a rectangular weir.
In weir measurement (Figure 4-15) the nappe, or profi1e of water
over the weir, must be completely aerated if good accuracy is to be
obtained. All weirs then produce some head loss as the lquid falls free.
If head loss is a problem, a ftume might be a better choice.
Flumes, a further development of the basic weir concept, are designed primarily to reduce the head loss that is experienced with the

CIPPOLETTI

WEIR~S>2H

ECTANGULAR

WEIR-j

MAX
I

: 4~

_L)2H

MAX

PREFERABLY

ANO
>4H

MAX

I
0>3H

MAX

C\PPOLETTI
WEIR

Fig. 4-13. Rectangular.Cippoletti,and V-notchweirs.

GPM
70000
50000

100

40000

80

30000

60
50

20000

40
30

10000
8000
6000
5000

10

4000

3000

6
5

2000

3
1000
800
600
500

1.0

400

De

300

0.6
05
0.4

200

0.3
100
80
60
50

OI

40

0.08

30

006

0.05
20

0.04

FLOW

0.03

10

0.02

8
FLOW

6
5

001

0008

0.006
0005
0004
0003

0.002

Fig. 4-14. Rangeof weir capacityfor differenttypesand sizesof weirs.

111

weir.

112 FEEDBACK PROCESSCONTROL

AERATION'UNDE.

NAPPE

Fig. 4-15. Aerationundernappeoi weir.

Flumes are also able to handle solids suspended in the flowing

liquido
A very popular flume is the Parshali flume. lt is somewhat similar
to one-half of a rectangular Venturi tube. ln flume measurement, head
losses are reduced because it is Dot necessaryto create the nappe. The
Parshall flume is shown in Figure 4-16 and Table 4-3 gives the dimen-

.k.

,..~~--~~~:::::: ~~:.:--::::~
i o
t7::::::;:::~~~t=:=f:~:::::::--i
l CO"".~.,.....c"O" .. TH.OAT...TION!l/.".'N.'.cr' ON c~1

Lc

f ~~~--~:

~.

..PLAN

'. "

.1
M

,\

\\

:~+
.1-0

.1.

Ir:--

'-'0-.,,

;1

'LOW

,
l

oi;;;.
""
,o

.";;,,

''-

SECTION

o-o

Fig. 4.16. Diagramanddimensionsof Parshallflume.

~
:..:.;
c;...
.,
'" ~< '"",:,
~'""

~~~--c

~-c..,

":";oC~.i~~~~@a~

, \oi

~~~t;

a~~=~~:",-c-co",

~~~:,

ci ci ci ci ci ci ci":":"";";

;... ~

;. .., .., .., .., .., .., .., .., .., .., ..,

~ ~

-"""""""""""""""'"

.1~ ...

~ ri:

'"
G)

Q: .:

:;
oc:

~~
~~

~
'"'
~ r;:

'"
~

lo.

cE
'"
~

.9

~
e
=
e

~
t, .t

to.t

'2

~
=
~

'Q

.5

~ .t

e
'~

.!,
~ ...

~
'Q

~
'"
=

NN''~"""":'ob~=~~

~~~~~~~:!:!:!:f:f

e
=

.;!;:-::i
-::i.;!-::.;!
~-'c-~-;'-;';'~;'

~ .;

~"!I!J;)ION
sn!~ ..un;) qloowS

--.N

-..,

'" ~
...

oS

'C

Li:

Qj

:c

.., .., .., .., .., .., .., .., .., ..,

~~~~~~~~~~~~
99~9O999999
N N N ...,J,
...,J,

.
:.:.~..~...
-"'-0-,
= ...g ::-

~.

;:- ::,..,.0--;

O':":'N"'~""""'obd,=

~":-~9~9"'99999
g--NN""",,",.o""~o.
..~.-!!!"~.-!!!
c ;:;:- c '" ~ ;,- ~ ;,- "C>
'9 'i' ";' "f
"f ";' '" -.,.
";'
-NN

b.b

..

'~

..,.

o. o. o. o. o. o. o. o. o.

~~!~~:,:,~~~~:,

5
,5
o
~.

oS

""

.. to.

~ ;;: :::

.:.!::~;:~~~:':~J.
.~

!~;;;~~~~~~:!~~
~,\,O-9'\'99"'9999
OO'--N""""'-o"'~

113

'!~

GPM

MGO
3000

CFS
4000

2000

3000
2000

1000
800
1000
800

600
500
400
300
200

100

so
60
40
50

600
500

400
300
200

80
100

60
50

30

40
20

30
20

10
8
6
5

4
3
2

10
08
06
O~

10
08

04
03

06
05

0.2

03

04

100

-t++

MGO

80

02

OI

60
50
40

004

20

FLOW

10
8

OI

006
QO~

30

FLOW

CFS

008

008
006

OO~
004

002

I
I

I
2

HEAD-INCHES
L-I
2
HEAD-FEET

I I .I.
6

I
8 10

,.,.,.

,..

I~ 20

60

0.03

80 100

002
FLOW

.456810

2/:3

56

8 IQ

001

FIVE
INCHES IS HINIHUN FLA.L SCALE
HEAD III'H
FDXBORD FLOA' ANO CABLE
HE'ER

THIRTY SIX 'NCHES IS ",XIHUM FULL

SCAL' H'AD N'TH FDXBDRD FLDAT AND
CABL' "'T'R

Fig. 4.17. Level and ftow relationship for ParshaIl ftume.

114

FLOW MEASUREMENT 115

sions. By referring to Figure 4-17 the dimensions of a flume capable of

measuring a prescribed flow rate mar be selected. Small flumes mar be
purchased and installed while large flumes are generally fabricated on
site. Several compaDiesspecialize in the construction and calibration of
flumes.
Under ideal conditions, measurements by weirs are very accurate.
In actual plant measurement, however, ibis mar Dot be true because it
is Dot practical to minimize the velocity of approach factor. The correlation between level and flow typically has an error of 3 to 5 percent.
When the error of making the level measurementand relating it to flow
rate is added, the total overall error mar be as high as :t5 percent.
Equations for the more common open-channel flow devices are given in
Table 4-4.
The equations for most open-channel primary devices are nonlinear. To convert head into flow rate it would be necessary to perform
an extensive calculation. This step is generally eliminated by having the
instrument perform the calculation. This is accomplished either with a

Table4.4. Equationsfor Common Open-Channel

Flow Devices
V-Notch Weir
3(1'Qcfs= .6650 H5/2
6(1' Qcfs= 1.432 H5/2
9(1' Qcfs= 2.481 H5/2
Rectangular Weir (with end contractions)
Q = 3.33 (L-0.2H) H3/2
Rectangular Weir (without end contractions)
Q = 3.33 LH3/2
Cippoletti Weir
Q = 3.367 LH3/2

Parshallfiume
Flume
size (in.)
3.
Q = 0.992 HI,547
6
Q = 2.06 HI,58
9
Q = 3.07 HI,53
12
Q = 4.00 HI,522
Q = crs; L = crest length (fi); H

Flume
size (in,)
18
Q = 6.00 H',538
24
Q = 8.00 HI,55.
36
Q = 12.00 HI,566
48
Q = 16.00 Hl.578
= head (fi)

116 FEEDBACK PROCESSCONTROL

carn or by means of a special chart or scale. The head is measured with

any one of several instruments. Several popular types are: float-andcable, ball float, bubble, and pressure-sensing. A typical installation is
shown in Figure 4-18.
Installation and Selection Considerations
To determine the type of primary device to be used for open-channel
flow measurement, the following factors musi be considered:
Probable ftow range
Acceptable head loss
Required accuracy
Condition of liquid
Probably all of the devices will work, but usually one type will be
the most suited to the installation. It must be remembered that a weir
will Dot function properly ifthe weir nappe is Dotaerated, and the flume
must be maintained at criticat depth; otherwise these devices do Dot
function properly. Nappe aeration is largely a question of level in the
receiving channel. If the lquid rises so that clearance is insufficient for
aeration, accurate flow measurement is impossible. When the flow
drops so that the nappe is Dot aerated but is drawn to the weir edge by
capillary attraction, the device does Dot operate properly.
Minimum weir flows are difficult to calculate because they depend
to some degree on the nature of the weir. If the weir is properly sized,
mnimum flow will produce a substantial, measurable amount of head
over the weir. Figure 4-17 shows the relationship between level and
flow for the ParshalI flume (Figure 4-16). Dimensions and maximum
and mnimum flows for the flume are tabulated in Table 4-3.

Fig. 4-18. V-Notchweir with ftoat-and-cable.

FLOW MEASUREMENT 117

Velocity

Flowmeters

Magnetic Flowmeter
The principIe of the magnetic ftowmeter was first stated by Faraday in
1832, but did Dot appear as a practical measurement for the process
plant until the 1950s.Its advantages are no obstruction to ftow, hence
no head loss; it can accommodate solids in suspension; and it has no
pressure connections to plug up. It is very accurate and has a linear
ftow rate to output relationship. Its disadvantages are that measured
material must be lquid; the lquid must bave some electrical conductivity; and it is expensive.
Operation
Operation ofthe magnetic ftowmeter is based on Faraday's well-known
law of electromagnetic induction: The voltage (E) induced in a conductor oflength (D) moving through a magnetic field (H) is proportional
to the velocity (V) of the conductor. The voltage is generated in a plane
that is mutually perpendicular to both the velocity ofthe conductor and
the magnetic field. Stated in mathematical form:

E = CHDV

(4-11)

where C is a dimensional constant. Industrial power generators and

tachometer generators used for speed measurement operate on the
same principIe, electrically conductive process liquid acts in a similar
manner to a rotor in a generator. That is, the liquid passes through a
magnetic field (Figure 4-19) induced by coils (Figure 4-20) built around
a section of pipe. The process liquid is electrically insulated from the
pipe, or flow tube, by the use of a suitable lining when a metal tube is
used, so that the generated voltage is Dot dissipated through the
pipeline. Two metallic electrodes are mounted in the flow tube and
insulated from it; a voltage is developed across these electrodes that is
directly proportional to the average velocity of the liquid passing
through the magnetic field. Because the coils are energized by altemating current power, the magnetic field and resultant induced voltage is
altemating. This generated voltage is protected from interference,
amplified, and transformed into a standard dc current signal by a
transmitter or converter. Line voltage variations are canceled by the
circuits employed.
The magnetic flowmeter measures volume rate at the flowing temperature independent of the effects of viscosity, density, turbulence or

FLOW MEASUREMENT 119

suspended material. Most industrialliquids can be measured by magnetic flowmeters. Exceptions are someorganic chemicals and most refinery products. Water, acids, bases, slurries, liquids with suspended solids, and industrial wastes are commou applications. The limitation is
the electrical conductivity ofthe liquido The magnetic flowmeter offers
no more restriction to flow than an equivalent length of pipe, Figure

4-21.
Structurally, a magnetic flowmeter consists of either a lined metal
tube, usually stainless steel because of its magnetic properties, or an
unlined nonmetallic tube. Linings for the metal tubes can be polytetrafluoroethylene (Teflon), polyurethane, or some other nonmagnetic,
nonconducting material.
The electrodes are suitably insulated from the metal tube. Nonmetallic fiberglass flow tubes do llot require any lining. The electrodes
most be insulated so that the voltage generated can be measured across
the electrodes. The insulation has no bearing on the actual voltage
generation, bot without the insulator, voltage would bleed off through
the metallic walls of the tube.
The coils are similar in design to the deflection coils used on a
television picture tube. Two coils are used and work together to create
a uniform magnetic field. The coils are generally series-connected, but
mar be parallel-connected if the measured flow velocity is low.
The signal output from the flow tube's electrodes is an altemating

Fig. 4-21. Unobstructedftow.

120 FEEDBACK PROCESSCONTROL

voltage at supply frequency. This 10w-Ievel voltage is generally between 1 mV and 30 mV at full flow rate.
This 10w-Ievel altemating voltage must be measured and converted
either into a record or display or into a dc common denominator transmission signal. This signal, typically 4 mA at zero and 20 mA at full
scale, can be fed into a recorder or controller. The device that can
accomplish tros is a special type of transmitter. This transmitter is
located either directly on the flow tube (Figure 4-22), near it, or within
the control room. The preferred location is as near to the flow tube as
possible; temperature and corrosive conditions are the constraints that
dictate location. In some applications a digital or pulse rate signal output Cromthe transmitter mar be required, and this option is available.

Accuracy
The accuracy of most magnetic ftowmeter systems is 1 percent of fullscale measurement. This includes the accuracy of both the meter itself
and its secondary instrument. Because ibis type of meter is inherently
linear, its accuracy at low-ftow rates exceeds the practical accuracy of
such inferential devices as the Venturi tube. Accuracy of the Venturi is
:tO.75 percent, according to ASME Fluid Meters Report, and thai of
the secondary instrument is abolli :tO.5 percent. At the low end ofthe
measurement scale, secondary instrument readability decreases owing
to the square root relationship. The magnetic ftowmeter can be labora-

Fig. 4-22. Foxboromagneticftow tube andtransmitter.

FLOW MEASUREMENT 121

torr calibratedto an accuracyof 0.5 percentof full scaleand is linear

throughout.
Vortex Flowmeter
The Foxboro Vortex ftowmeter, as shown in Figure 4-23, measures
liquid, gas, or steam ftow rates using the principIe of vortex shedding.
The transmitter produces either an electronic analog or polse rate signallinearly proportional to volumetric ftow rate.
Principie or Operation
The phenomenon of vortex shedding can occur whenever a nonstreamlined obstruction is placed in a flowing stream. As the liquid passes
around the obstruction, the stream cannot follow the sharp contours
and becomes separated from the body. High-velocity liquid particles
flow past the lower-velocity (or still) particles in the vicinity ofthe body
to form a shearlayer. There is a large velocity gradient within this shear
layer, making it inherently unstable. The shear layer breaks down arter
some length oftravel into well-defined vortices as shown in Figure 4-24.
These vortices are rotational flow zones that form altemately on
each gide ofthe element with a frequency proportional to the liquid flow

Fie. 4-23. Vortexflowmeter.

122 FEEDBACK PROCESSCONTROL

Fig. 4-24. Vortexsheddingphenomenon.

rate. Differential pressure changes occur as the vortices are formed and
shed. This pressure variation is used to actuate the sealed sensor at a
frequency proportional to vortex shedding.
Thus, a train of vortices generates an altemating voltage output
with a frequency identical to the frequency of vortex shedding. This
frequency is proportional to the flow velocity.
The voltage signal from the detector (Figure 4-25) is conditioned
and amplified for transmission by electronics located in a housing
mounted integral with the flowmeter body. The final output signal is
available either in pulse fonn with each pulse representing a discrete
quantity of fluid for totalizing or, optionally, as a 4 to 20 mA dc analog
signat for flow rate recording or control.
Turbine

Flowmeter

The turbine meler derives its name from its operating principie. A
turbine wheel (rotor) is set in the path ofthe flowing fluid. As the fluid
enters the open volume between the blades of the rotor, it is deflected
by the angle of the blades and imparts a force causing the rotor to turo.
The speed at which the rotor turos is related, over a specified range,
linearly to flow rate.
Several methods are employed to transmit this motion to a readout
device outside ofthe conduit. In some applications a mechanical device
convers the rotor motion directly to a register. In process applications,

Fig. 4-25. Geometryofvortex generator.

FLOW MEASUREMENT 123

however, the usual method is to use an electrical method. A coil containing a permanent magnet is mounted on the meter body. The turbine
flowmeter (Figure 4-26) consists of a section of metal pipe, a multibladed rotor mounted in the center ofthe straight-through passage,and a
magnetic pickup coil mounted outside the fluid passage. A shaft held in
place by fixed radian vanes supports the rotor assembly. As each blade
tip of the rotor passesthe coil it changesthe flux and produces a pulse.
The total number of pulses indicates the volum e of fluid which has
passed through the meter and the rate of the pulses generated becomes
a measure of flow rate.
Turbine flowmeters are frequently employed as sensors for inline
blending systems.
Turbine flowmeters bave excellent accuracy and good rangeability.
They are limited to clean fluids. They are expensive, but do bave
unique features.
Other Flowmeters

Other ftowmetersthat mar be found in someprocessplantsincludethe

ultrasonic,the thermal, a varietyofpositive displacementtypes, metering pumps,and others. The ftowmetersdescribedaccountfor the majority of the ftow rate measuringdevices in everydayuse.

Fig. 4-26. Thrbineflowmeter.

124 FEEDBACK PROCESSCONTROL

Conclusion

Unfortunately, there is no best flowmeter. Factors such as accuracy,

pressure loss, material to be measured, ease of changing capacity, and
ease of installation along with cost must be carefully considered. Once
the details of the problem bave been gathered and the possible alternatives considered, the best selection will usually be clear. Some of the
systems described read correctly to within a few percent. Some are
considerably better. All should bave good repeatability-the all important criterion for control.

Questions
4-1. Match the head meter primary devices with the application (select asingle
rest answer for each):
Orifice plate ~
r.I. T-.
Hgh-pressure~~~._~recovery
Flow nozzle Venturi tube Pitot tube-

Elbow taps-e.

-c.

b. Air ducts
Sedimentin lquid
d. Economyand accuracyare
important
No straightrun available

4-2. A differentialpressuretransmitteris calibratedOto 80 inchesof water

and transmitsa 4 to 20 mA dc signal.This transmitteris placedacrossan
orifice plate which is sizedto create80 inchesof water differentialat 6 gallons
per minute.
Whatis the flow rate whenthe signalis 13 mA dc?
8. 4.5 gpm
c. 3.4 gpm
b. 3.9 gpm
d. 4.9 gpm
4-3. For the equipmentdescribedin the precedingquestion,what will the
signalbe if the flow rate is 4 gpm?
8. 13.3mA dc
c. 8.9 mA dc
b.11.2mAdc
d.6.7mAdc
4-4. Assumethe line used in Questions2 and 3 is a Yz-inchschedule40pipe
(IV = 0.622inches).Pind the orifice boTeusedto satisfythe conditionsgiven
if the measuredfluid is water at 600P.
a. 0.190inches
c. 0.414inches
b. 0.556inches
d. 0.352inches
4.5. The bestchoiceof orifice taps in the precedingproblemwould be:
a. flangetaps
c. pipetaps
b. venacontractataps
d. cornertaps

FLOW MEASUREMENT 125

4-6. A 2-inch schedule 40 line {lD = 2.067) is used to carry gasoline (SP
GR = 0.75). The ftow rate is measured with an orifice plate (d = 1.034)and
pipe taps are used. At full ftow rate a ditIerential pressure of 50 inches of
water is produced. What is the approximate full ftow rate in gpm?
8. 30 gpm
c. 250 gpm
b. 53 gpm
d. 36 gpm
4-7. Assume an 8-inch schedule 160 pipe (ID = 6.813 inches) carries a full
ftow of 40,000pounds per botir of dry saturated steam. The static pressure is
335 psi. Flange taps are used and the ditIerential pressure across the orifice
plate at full-ftow rate is 1OOinches of water. What is the size of the bare in the
orifice plate?
8. 4.357 inches
c. 2.271 inches
b. 3.645 inches
d. 5.109 inches
4-8. Fuel gas is carried in a 6-inch schedule 40 pipe (ID= 6.065 inches).
Flow rate is measured with an orifice plate using fiange taps and the bare in
the orifice is 3.457 inches. The specific gravity is 0.88, the fiowing temperatur~
is 60F, and the static pressure is 25 psi. Maximum fiow rate creates a
ditIerential of 20 inches of water. What is the approximate fiow rate in
cubic feet per day?
8. 1,500,000SCFD
c. 2,500,000SCFD
b. 2,000,000SCFD
d. 3,000,000SCFD
4-9.
8.
b.
c.

The output of a target fiowmeter is:

Proportional to volumetric fiow rate
Proportional to the square of volumetric fiow rate
Proportional to the square root of volumetric ftow rate

4-10.
8.
b.
c.
d.

The output signal or reading of a magnetic fiowmeter is:

Proportional to volumetric fiow rate
Proportional to the square of volumetric fiow rate
Proportional to the square root of volumetric fiow rate
Inversely proportional to volumetric fiow rate

4.11. With suspended solids and/or entrapped gas in a fiowing liquid, the
magnetic fiowmeter will:
8. Read high
b. Read low
c. Read the liquid fiow only
d. Read the correct total volume of the mixture
4-12. A turbine fiowmeter produces an output in the form of pulses. The total
number of pulses is:
8. Inversely proportional to fiow
b. Directly proportional to total ftow
c. Proportional to the square root of fiow
d. Proportional to the square of fiow

Temperature and Humidity

Measurements

Temperature and moisture measurementsare important in process con-trol,

both as direct indications of system or product state and as indirect indications of such factors as reaction rates, energy flow, turbineefficiency,
and lubricant quality.

Temperature

Presenttemperaturescalesbave been in use for about 200years. The

earliest instruments-variations of the common stem thermometer-were
basedon the thermal expansionof gasesand liquids. Suchfilled
systemsare still employedfrequently for direct temperaturemeasurement, although many other types of instrumentsare available. Rep-resentative
temperaturesensorsinclude:
Filled thermal systems
Liquid-in-glassthermometersThermocouplesResistance
temperaturedetectors (RTDs)
126

TEMPERATUREAND HUMIDITY MEASUREMENTS 127

Thermistors
Bimetallic devices
Optical and radiation pyrometers
Temperature-sensitive profitS
Various types of measurement control systems are compared in Table

5-1.
Instrument selection must anticipate overall control requirements.
Low cost often justifies consideration of filled systems for measurements below 1,200oPor 650C. Other advantages of mechanically or
pneumatically transmitted temperature measurements include lowexplosion hazard, simple maintenance requirements, high reliability,
and independence Crom external power. Advantages of electrical systems include higher accuracy and sensitivity, practicality of switching
or scanning several measurement points, larger distances possible between measuring elements and controllers, replacement of components
Table5.1. Comparisonof TemperatureMeasuringSystems
ComparisonFactors
Purchase cost
Long distance transmission
Change or replacement
o components
lnstallation costs
Maintenance
Averaging measurement
Surface measurement
Time constant
(bare bulb and no well)
Temperature difference
Sensitivity

Accuracy
Operating costs

Least
Favorable

Most
Favorable

lntermediate

F
E

p
p

EandP
F
P
T
T

F
P
E

Bo

F
F
R

T
R

E
R
F

Key: F Filled system.

P PneumaticaIly transmitted filled system.
T ElectricaI thermocouple system.

F
F
P
F
E

F
R

T
F
P and T
P

R Electrical RTD system.

E Electrical thermocouple and RTD systems equally rated.
oGeneral purpose wiring only. Explosion-proof
electrical systems cost considerably more than
other types.

128 FEEDBACK PROCESSCONTROL

(rather than complete systems) in the event of failure, fast response,

and ability to measure higher temperatures.

Filled Thermal

Systems

Filled thermal systems, which traditionally bave been used most in

the food, paper, and textile industries, consist of sensors (bulbs) connected through capillary tubing to pressure or volume sensitive elements (Figure 5-1). These systems are simple and inexpensive and
generally bave fast dynamic responses. Their use with pneumatic and
electronic transmitters has removed inherent distance limitations of
filled systems and has minimized the danger of capillary damage.
Moreover, transmitter amplification has made narrow spans practical
and has improved linearity and response.
Application specifications of several types of filled systems are
listed in Table 5-2. These include the Scientific Apparatus Makers Association (SAMA) Classes I (liquid-expansion), 11(vapor-pressure), III
(gas-pressure), and V (mercury-expansion). The SAMA Class 11
classifications also include alphabetical designations, by which A and B
indicate sensorabove and below case ambient, respectively; C indicates
a system in which the sensorcan cross ambient; and D denotes a system
which can operate at ambient conditions.
Liquid-expansion systems are characterized by narrow spans,
small sensors, uniform scales, high accuracy, and capability for difIerential measurements. Class lA devices bave an auxiliary capillary and

Fig. 5-1. Filled measurement

systemsgenerallyare the mostinexpensiveway or
measuringandcontrollingtemperature.

TEMPERATURE AND HUMIDITY

MEASUREMENTS

129

Table5-2. Instrumentsfor Filled ThermalSystemSensors

Gas

Type
Principie
SAMA Class

Liquid
Volumechange

fluids
Lower range limit
Upper range limit

Organic liquido
(Hydro-carbons)
-200"F (- I3IY'C)
+6OO"F(+315C)

Narrowest span (b)

Widest span

40F (25C)
600'F (33IY'C)

71Y'F (4IY'C) (c)

Ambient temperature

lA Full

Not required

Compensation
Sensor size

IB Case

Smallest

Medium

Largest

Typical sensor size

for 1000Cspan

9.5mm (0.375 in.) 00 x

48mm (1.9 in.) long

9.5mm (0.375 in.) OD x

50mm (2 in.) loog

22mm (,. D.) OD x

70mm (6 n.) lang

Overrange capabiJity
Sensor elevation elfect

Medium

Least

None

Class llA, Ves

Class lIB, NODe

Greatest
None

Barometric pressure
effect (altitude)

None

Slightly (greatest on
smaIl spans)

Scale uniformity

Unifonn
"'.5 to "'1.0% ofopan

Non-Uniform

Uniform

;:0.5to;:I.0%ofspan
upper % of scaIe

;:O.5to;:I.0%of

spon

#1-ClasslIA

#2

Accuracy

Response (d)
#1 Fastest

Vapor (a)
Pressure change
11

Pressure change
III

arganic liquido

Puregases

(Hydro-carbons),
water
-425F

(-2SS"C)

+6OO'F (+3IS0C)

400'F (215C)

-45SOF (-2700C)
+l,4OO'F(+7ro'C)
1200F(700C)
1.000'F (5500C)

IIIB case

#3-Class

Slightly (greatest o
small spans)

lIB

#4 Slowest
Cost
Maximum
CapiIlary

standard
length

Highe.1

Lowest

Medium

Clos. lA 30m or 100fI.

30m or 100 fi.

30m or 100I

Clas. IB 6mor 20fi.

(a) Class 11 systems are supplied as either SAMA Class llA or lIB. In Class llA, sensor is aIways
hotter than tubing or instrument case. In Class lIB, sensor is always cooler than tubing or case.
(b) Narrowest spans vary at elevated temperatures.
(c) Smaller spans available in cryogenic regions.
(d) Actual values depend on range, capillary length, sensor dimensions, and type of instrument used.

element to provide ambient temperature compensation and IB systems

often utilize bimetallic techniques. However, fully compensated
liquid-expansion systems are complex and, therefore, expensive.
Vapor-pressure systems are reliable, inherently accurate, and require no compensation for ambient temperature effects. Instruments
follow the vapor-pressure curves ofthe filled fluid, and associated dials
and charts are thus nonunifonn, featuring more widely spaced
increments at high temperatures. Measurements occur at the interface
between liquid and va,porphases of the filling medium. If the temperatufe in the sensorexceeds that in the capillary and indicating element, the
sensoris filled with vaporwhile the capillary and indicatorcontain liquido

130 FEEDBACK PROCESSCONTROL

The converse is true when the relative temperature polarity is reversed.

A transition between liquid and vapor can cause erratic operation, so
that vapor systems mar be unsuitable for ranges that cross the element
and capillary temperatures. Systems mar also be unacceptable if uniform recording or indicating scales are desired.
Gas-pressure systems, which rank second to the vapor-pressure
devices in simplicity and cost, ofIer the widest range of all filled systems. Conventional designs use large-volume sensors, which mar be
shaped to suit particular applications. For example, in duct-temperature averaging, the sensor mar be constructed of a long length of
tubing of small cross section. Conventional recorders are Dot recommended for temperature spans of less than 200F or 110C, but transmitters operating on force-balance principIes can be utilized with spans
as narrow as 50F or 28C. With gas-filled systems, it is difficult to
compensate for ambienttemperature errors, but a sufficiently large sensor
size mar reduce efIects to acceptable limits.
Mercury-expansion systems are classified separately from other
liquid-filled systems because of the unique properties of the fluid. For
example, mercury is toxic and harmful to some industrial processes
and products, and high-liquid density places limitations on sensor-toinstrument elevation difIerences. Sensors used in mercury-expansion
systems are generally larger in diameter and more expensive than those
used in either liquid or vapor systems. For these reasons, mercury is
frequently bypassed in favor of other filling media.

Electrical

Systems

Electrical temperature sensorsbave long been popular in the metal and

paper industries, but increasing use of ele<tronicdevices has stimulated application in a variety of other areas. Thermocouples and resistance temperature detectors (RlDs) are most widely used, and representative application data are given in Table 5-3.
Thermocouples

Thermoelectricity was discovered by Seebeck in 1821.He observed an

electromotive force (emt) generated in a closed circuit oftwo dissimilar
metals when their junctions were at different temperatures. This electricity, produced by the direct action of beat, is used today to measure
temperatures from subzero to high ranges.

TEMPERATURE AND HUMIDITY

MEASUREMENTS

131

Table5.3. Quick-ReferenceSelectorChart for Standard

Thermocouples& RTDs
TEMPERATURE LIMITS
Sensor

Calibration

r -F
oc

RTDs

Nickel
Platinum

SAMA Type 11
SAMA 100 ohm, or

-200 to 315

-320

DIN 43760

-200 to
O to
-210 to
-270 to
-270 to
-270 to
-50 to

-320 to 1,200
32 to 300

Copper
SAMA
Thermocouples Iron-Constantan, ISA Type J
Copper-Constantan, ISA Type T
Chromel-Alumel, ISA Type K
Chromel-Constantan, ISA Type E
Platinum-Platinum Rhodium,
ISA Types R & S
ISA Type B

650
150
760
370
1,260
870
1,480

o to 1,700

to 600

-350
-455
-455
-455
-55

to
to
to
to
to

1,400
700
2,000
1,600
2,700

o to 3,100

A thermocouple consists basically of two dissimiliar metals, such

as iron and constantan wires,joined to produce a thermal electromotive
force when thejunctions are at different temperatures (Figure 5-2). The
measuring, or bot, junction is inserted into themedium where the temperature is to be measured. The reference, or cold, junction is the open
end that is normally connected to the measuring instrument terminals.
The emf of a thermocouple increases as the difference in junction
temperatures increases. Therefore, a sensitive instrument, capable of
measuring emf, can be calibrated and sed to read temperature directly.

<
MEASURING
JUNCTION

Fi2. 5-2. Thennocouple.

THERMOCOUPLE
[JINSTRUMENT
REFERENCE
JUNCTION

132 FEEDBACK PROCESSCONTROL

Figure 5-3 indicates the approximate emf-millivolt output versus

temperature relationship for the most popular thermocouple types. Deviation from the essentially linear relationship is generally less than 1

percent.
Introduction of intermediate metals into a thermocouple circuit will
Dot affect the emf of the circuit, provided the new junctions remain at
the same temperature as the originaljunction. The algebraic sum ofthe
emf' s in a circuit consisting of any number of dissimilar metals is zero,
if all of the circuit is at a uniform temperature. Repeating the law, if in
any circuit of solid conductors the temperature is uniform from Point 1
through all the conducting material to Point 2, the algebraic sum ofthe
emf' s in the entire circuit is totally independent of the intermediate
material and is the same as if Points 1 and 2 were put in contact. If the
individual metals betweenjunctions ofthe circuit are homogeneous,the

Fig. 5-3. Temperature-emf

valuesapproximatethe relationshipof six standard
thermocouples.The conditionsof a particularinstallationmar requireloweringthe
maximumtemperatureto avalue belowthat listed. Temperature/millivolt
curvesare
availablein the appendix.

TEMPERATUREAND HUMIDITY MEASUREMENTS 133

sum ofthe thermal emf's will be zero, provided only that thejunctions
of the metals are all at the same temperature.
The emf in a thermoelectric circuit is independent of the method
employed in forming the functions as long as all the junction is at a
uniform temperature and the two wires make good electrical contact.
The junction mar be made directly, by welding, or by soldering. Furthermore, an instrument for measuring the emf mar be introduced into
a circuit at any point without altering the resultant emf, provided the
junctions that are added to the circuit by introducing the instrument are
all at the same temperature. If the temperatures of the new junctions
are DOt uniform, the effect is that of introducing additional thermocouples into the circuit.
Reference Junction
To make accurate temperature measurements with thermocouples, the
reference junction temperature must remain constant; if it varies, suitable compensation for these variations must be provided.
Should there be an uncompensated variation in the reference junction temperature, there will be a corresponding change in the millivoltage with a resultant error in temperature measurement (Table 5-4).
When used in the laboratory and for other checking and testing
purposes, the thermocouple reference junction can be placed in a vacuum bottle filled with shaved ice saturated with water. This method
provides close temperature control (within a fraction of a degree) and
permits accurate reading.
Table5-4. StandardLimits or Error
Coupleor Wire

Range

Limits

Copper-constantan

-300 to -75F
-75 to +2000F
200 to 700F
-100 to + 530"F

:t2% oCreading
:tl~oF:t~
% oCreading
:t4F

Chromel-alumel

530 to l ,400"F
Oto 530F
530 to 2,300"F

:t~% oCreading
:t4F

Platinumrhodiumplatinum

Oto 1,000F
1,000to 2,700"F

Iron-constantan

:t~% oCreading
:tJoF
:tO.J% oCreading

75
KX

134 FEEDBACK PROCESSCONTROL

~~ WIRE~[]

THERMOCOUPLE

MEASURING
JUNCTION

EXTENSION

<

INSTRUMENT

CONNECTION
HEAD

REFERENCE
JUNCTION

Fig. 5-4. Typical thermocoupleand extensionleads.

To ensure accurate readings, most thermocouples are now installed

with instruments thai provide automatic reference junction compensation. In most instruments, ibis is accomplished by passing current
through a temperature-responsive resistor, which measures the variations in reference temperature and automatically provides the necessary compensating emf by means of the voltage drop produced across

it.
An industrial installation generally consists of a thermocouple with
its connection head, the necessary length of extension wire, and an
indicating, recording, or controlling instrument with internal and automatic reference junction compensation. The extension wires generally
consist of the same materials as the thermocouple elements, or mar be
composed of other materials and alloy wires that generate essentially
the same millivoltage as the thermocouple for application temperatures
up to approximately 400F or 200C(Figure 5-4). The ISA symbols and
color codes for thermocouple lead wires are given in Table 5-5.

Table 5.5. Extension Wire Type Designations and Standard Limits of Error
[SA

Extension

Type

Wire

Color
Code

TX

Copperconstantan
Ironconstantan
Ironcupronel
Chromelalumel
CopperCuNi alloy

+ (blue)
-(red)
+ (white)
-(red)
+ (green)
-(red)
+ (yellow)
-(red)
+ (black)
-(red)

JX

wx

sx

Limits 01
TemperatureRange
F
Error
-75 to +200

:t 1~F

Oto

400

:t4 F

75 to

400

:6 F

to

400

:t4 F

75 to

400

:10 F

Couple
UsedWith
Copperconstantan
Ironconstantan
Chromelalumel
Chromelalumel
Platinum
Rhodiumplatinum

Fig.
<

TEMPERATURE AND HUMIDITY

MEASUREMENTS

135

5-5. Thermocouplesin para11el.

A verage Temperatures
To measure the average temperature across a large duct or vessel, or
around a retort, any number of thermocouples mar be used in parallel
connections. The voltage at the instrument, or at the point of parallel
connection, is the average of that developed by the number of thermocouples used. This voltage is equal to the sum of the individual
voltages divided by the number of thermocouples. For accurate measurement, the resistances of aU thermocouples and extension wire
circuits should be identical. Since the resistance of the actual thermocouple wiU vary with temperature, and since the lengths of extension wires mar also vary, the efIect of these variations can be minimized by using swamping resistors. The values of the swamping
resistors should be high in comparison with the change or difIerence in
resistances encountered. A resistor value of 1,500 ohms generaUy
works weU.
In order to prevent the flow of current through a ground loop, the
thermocouples should not be grounded. AU thermocouples must be of
the same type and must be connected by the correct extension wires
(Figures 5-5 and 5-6).
ParaUel connection of thermocouples for average temperature
measurement is advantageous because the instrument construction and

<

Fig. 5-6. Parallelthennocoupleswith swampingresistors.

136 FEEDBACK PROCESSCONTROL

calibration can be the .sameas that for a single thermocouple. Two or

more thermocouples mar be connected in series to measure average
temperatures; however, this circuit requires that the instrument bave
special reference junction compensation to provide the increased compensating emf for the specific number of thermocouples in the circuit.
The instrument also must be calibrated for the total millivoltage output
of the number of thermocouples used in series. Extension wires Crom
each thermocouple must be extended back to the actual reference junction at the instrument.
Differential Temperatures
Two thermocouples may be used for measuring differential temperatures between two points. Two similar thermocouples are connected
with extension wire of the same material as is used in the thermocouples. The connections are made in such a way that the emf's
developed oppose each other. Thus, if the temperatures of both thermocouples are equal, regardless of the magnitude, the net emf will be
zero. When different temperatures exist, the voltage generated will
correspond to this temperature difference, and the actual difference
may be determined Cromthe proper calibration curve. An instrument
calibrated either for Omillivolts at midscale or Omillivolts at the end of
the scale, depending upon whether thermocouples are operated at high
or low temperatures with respect to each other, may be fumished and
used to read temperature difference directly.
Copper leads may be used between the instrument and the connection box that links the instrument to the thermocouple or extension
wires (Figure 5-7). The instrument should Dot bave the reference junction compensation normally fumished when measuring temperatures
with thermocouples. As in the case of parallel thermocouple connections, the thermocouples should Dot be grounded and the resistance of
both thermocouple circuits should be the same.

Fig. 5-7. Two thermocouplesusedto measuretemperaturedifference.

TEMPERATUREAND HUMIDITY MEASUREMENTS 137

Thermocouple Calibration Curves

Thermocouples develop an extremely small dc voltage, measured in
thousandths of a volt. The millivolt output falls within a range of -11 to
+ 75 millivolts, depending on the type of thermocouple and its temperatufe working range. Calibration curves for thermocouples are included
in the Appendix. Now let us consider some typical thermocouple instruments and investigate their functions.

PotentiometricRecorder
Figure 5-8 shows a simplified circuit diagram of the Foxboro potentiometric recorder (Figure 5-9). An emf input (Ex) is derived Croma
thermocouple or other measuring element. A constant emf (Ez) is
supplied by a Zener diode regulated power supply. As Ex varies, an
error signal is developed, and the circuit becomes unbalanced.
The error signal is converted to an ac voltage by a field effect
transistor chopper and amplified by the integrated circuit amplifier. The
amplified output drives a two-phase balancing motor. The direction of
rotation depends on whether Ex has become greater or smaller. The
motor moves the wiper contact on the slidewire Rsto a position where
the circuit is rebalanced. The slidewire coritact is mechanically con-nected
to the recorder pen, and both are positioned simultaneously.
With a thermocouple input, a temperature sensitive resistor (Rc)
automatically compensates for referenc junction ambient temperature
variations. This resistor changes the balance of the circuit to cancel

138 FEEDBACK PROCESSCONTROL

Fig. 5-9. Foxboro stripchart recorder.

changes in Ex with reference junction temperature variations. For

straight millivolt inputs, Rc is a temperature-stable resistor.
The slidewire is characterized to allow the use of uniform charts
and"scales with thermocouple and certain other nonlinear measurements. Thisis accomplished by varying the resistance per unit length
along the slidewire.
lf the temperature measurement is used in conjunction with an
electrical control system, a thermocouple-to-current transmitter mar
be used. A typical transmitter is shown in Figure 5-10. The transmitter
provides a two-wire output; the same wiring 'is used for both power and
output. The load resistance is connected in series with a dc power
supply (approximately 25 volts), and the current drawn Cromthe supply
is the 4 to 20 mA dc output signal.
Field mounting the transmitter at or near the actual measurement
point eliminates the installation of special thermocouple extension
wires to the receiver. ln some cases, this procedure can also improve
the performapce of the measurement loop, since the possibility of extension wire error is avoided.
As shown in Figure 5-10, the thermocouple/millivolt bridge circuit
consists of a precision-constant current source, a reference-junction
compensator, a bum-out detection current source, and associatedresistors. The amplifier is a very high gain, chopper-stabilized

TEMPERATURE,
AND HUMIDITY MEASUREMENTS 139

Fig. 5-10. Simplifiedcircuit diagramof the Foxboro thermocoupletransmitter.

(Bottom)Foxborothermocoupletransmitter.

amplifier. It functions as a null detector by controlling the 4 to 20 mA dc

output corrent to maintain a null between the compensated thermocouple and feedback volt3;ges. The transmitter is protected from
reverse polarity connection ofthe power supply by means ora diode in
the negative lead of the output circuit. The maximum error for this
transmitter is approximately :t0.25 percent of span.
Many control instruments, such as this transmitter, feature thermocouple bum-out protection. In the event of an open thermocouple, a
s mall voltage applied to the instrument causes it to read full-scale output. This causes the control system to shut off. the beat and prevent
damage.
Resstance Thermal

Detectors

Resistancethermometry is based on the change of electrical conductivitv with temperature.Therefore.a coil of wire can act as a tem-

140 FEEDBACK PROCESSCONTROL

perature sensor, with 'a direct relationship established between resistance and temperature. Standard curves are available, with certified
accuracies within O.IF or oC. Platinum RTDs used as laboratory standards can be obtained with tolerances well within this lmit, and are
capable of precise temperature measurement up to l,650F or 900C. If
an RTD is adjusted to conform to its curve, it may be interchanged with
other RTDs calibrated according to the same curve.

Resistance-thermaldetector temperature measurementmay be

read out on many different types of instruments. If the measurement of
temperature were to be used in conjunction with a pneumatic control
system, the RTD could be attached to a resistance-to-pneumatic convertor, such as the Foxboro Model 34B (Figure 5-11).
The resistance-to-pneumatic convertor converts the temperature
measured by a resistance-temperature element into a proportional 3 to
15 psi or 20 to 100kPa pneumatic output signal. This signal is suitable
for use with many types of pneumatic receiving instruments, such as
controllers, recorders, and vario us computing instruments.
This transmitter has several features. For example, when used with

Fig. 5-11. Foxboro 34BSeriesresistancepneumatictransmitter.

TEMPERATURF;AND HUMIDITY MEASUREMENTS 141

Fig. 5-12. Simplified circuit diagram of 34B Series nickel RTD transmitter.

the appropriate nickel bulb, the converter has an output that is linear
with temperature. Spans as low as 5F or 3C can easily be achieved.
An adjustable range allows a simple field calibration procedure to
change the temperature input range required for a 3 to 15 psi or 20 to
100kPa output. It also operates on all normal supply voltages.
This transmitter is an electromechanical device consisting of a
solid-state, integrated circuit amplifier and an output transducer. Figure 5-12 isa simplified circuit diagram of the nickel RTD version of the
transmitter. The transd ucer is shown in Figure 5-13.
The RTD is wired into a measurement bridge and excited Croma
regulated direct current power supply. The change in resistance of the
RTD causes a bridge output change that is proportional to temperature.
Negative feedback is obtained Cromthe output current and applied to
the opposite side of the measurement bridge. A change in feedback by
the span adjustment changes the gain of the amplifier and thereby
changes the span of measurement.

Fig. 5-13. Outputtransducer.

142 FEEDBACK PROCESSCONTROL

The platinum RTD transmitter incorporates a negative impedance,integrated

circuit amplifier that shunts the platinum RTD to linearizethe
temperature versus the RTD characteristic.
The amplifier section provides the high input impedance and gainrequired
to amplify the low-input signal. The input signal is first converted to ac by an FET (used as a chopper) and then applied to a
high-gain ac amplifier. Th signal is next demodulated and applied to a
voltage follower. The output of this amplifier is applied to a power
transistor to supply the 10to 50 mA dc required by the transducer coil.
A functional diagram applying to the output transducer is shown in
Figure 5-13. The direct current input signal is converted into a proportional air output signal as follows: A coil is positioned in the field of a
fixed, permanent magnet. When a direct current flows through the coil,
the electromagnet and permanent magnet forces are in opposition. Any
increase in the current, which causes a proportional increase in the
force on the coil, tends to close the flapper against the nozzle. The air
output pressure from the pneumatic relay is thereby increased. This
increased pressure is fed to the feedback bellows to exert a force on the
force beam to rebalance the force from the coil. All ofthe forces which
act upon the force beam are balanced in this manner, and the pneumatic output tracks the current input. The capacity tank (C2) and the
associated restrictor provide dynamic compensation to the circuit. During rapid input changes, the restrictor will effectively keep the capacity
tank out of the circuit for faster feedback response through the relay.
For slow changes, the capacity tank acts to damp pressure fluctuations
and to provide a smoother output. The small capacity tank (C.) in the
output circuit provides necessary constant damping.
During alignment or calibration of the output transducer, zero adjustment is made by tuming a screw to advance or retract a spring
which bears on the force beam. Coarse span adjustments are made by
repositioning the feedback bellows with respect to the pivot.
Wheatstone Bridge Recorder
One of the most popular instruments used with RTDs is the slidewire
type Wheatstone bridge recorder. This instrument is almost identical
to the potentiometric recorder previously described and shown in
Figure 5-9.
Figure 5-14 shows a simplified circuit diagram of the Wheatstone
bridge recorder. The resistance temperature detector (RTD) is one arm
of a Wheatstone bridge excited by a Zener diode regulated dc power
supply. Point A (the slidewire contact) and roini B form the input to the
amplifier. When the temperature changes, the resistance of the RTD

TEMPERATURE AND HUMIDITY

MEASUREMENTS

143

Fig. 5-14. SimpIified Wheatstone bridge recorder circuit.

changes. This unbalances the bridge and creates an error signal between Points A and B.
As in the potentiometric recorder, the error signal is converted to
art ac voltage by a field-effect transistor chopper and amplified by the
transistorized amplifier. The output drives a two-phase balancing
motor. The direction of rotation depends on the polarity of the error
signal. The motor moves the wiper contact on the slidewire until the
bridge is rebalanced and no error signal exists. The pen and slidewire
contact are mechanically connected and therefore positioned simultaneously.
With all resistance-temperature measurements, the use of threeconductor RTD cable is recommended. The effect of ambient temperatufe variations on the cable is thereby minimized.
If the cable connecting the RTD to the instrument has only two
conductors, these conductors become part of the resistance being measured. The result then is an error that will vary with ambient temperature. Remember, with all resistance temperature measurements,
three-conductor RTD cable is recommended. The purpose of the threeconductor cable is to stretch out the measuring bridge. Note the threeconductor cable in Figure 5-14. One of the conductors is common to
both sides ofthe bridge while the other two connect one to each side of
the bridge. Any change in cable temperature will be canceled as both
sides of the bridge are changed a like amount.
Occasionally RTD sensors use a four-wire cable. This is generally
in conjunction with a Kelvin double bridge. The four-wire method does
an excellent job of reducing temperature effects on the cable. The
improve ment over the three-wire method, however, is minimal. In
practice, an RTD may be used with as much as 500 feet of threeconductor cable without the cable creating a perceptible error.

Thermistors

144 FEEDBACK PROCESSCONTROL

Thermistors are made of heat-treated metallic oxides, and most thermistors differ from ordinary resistors by having a negative coefficient of
resistance. Thermistors are available with a nearly linear temperature
resistance relationship, and other types are available with a sharp
change in slope at some characteristic temperature.
A thermistor can replace an RTD as a temperature sensor. The
difficulty lies in obtaining units that fit the desired characteristic curve
within acceptable lmits of accuracy. When this is accomplished and
the thermistor is mounted so as to stand up under process conditions, it
performs the same function as the conventional RTD.
One advantage of the thermistor is that it has a greater resistance
change for a given temperature change than that of the conventional
wire RTD. A disadvantage is that the accuracy available, although
good, is slightly inferior to that of the conventional RTD. This presumably accounts for the thermistor's limited application in the process

instrumentationfield.
Radiation pyrometers utilize an optical system to focus energy
radiat~d from a body onto a sensing system. Manual devices are often
l:I~ed;'in which energy at infrared or visible wavelengths is focused on a
tar~t'aridcompared
with the light output of a calibrated optical filamentJ9automated devices, the energy (usually in the infrared band) is
focusedori a series arcar ofthermocouples. This thermopile produces a
millivolt output related to the temperature of the source. Pyrometers
are used where high temperatures are to be measured or where contact
with the object is impossible. Accuracy is influenced by such factors as
reflections, gases present in the radiation path, and surface emissivity
of the body under measurement.
Humidity

Measurements

Humidity, or the amount of moisture in gases,is expressedin several

differentways, including:
1. Relativehumidity-the actual quantity of water vapor presentin a
given spaceexpressedas a percentageof the quantity of water
vapor that would be presentin the same space under saturated
conditions at the existingtemperature.
2. Absolute humidity-mass of vapor per unit magsof dry gas.
3. Dew point-the saturationtemperatureof the mixture at the correspondingvapor pressure.If the gasis cooled at constantpressure
to the dew point, condensationof vapor will begin.

TEMPERATUREAND HUMIDITY MEASUREMENTS 145

A list of the methodsusedto measurehumidity would include the

following:
RelativeHumidity Measurement
Mechanical(hair, wood, skin)
Wet and dry bulb thermometers
Surfaceresistivity devices(including Dunmore)
Crystal frequencychange
AbsoluteMoistureMeasurement
Gravimetry
Electrolysis
Infrared
Conductivity
Capacitance
Color change
Karl Fischer titration
RF power absorption
Neutron reftection
Heat of absorptionor desorption
Nuclear magneticresonance
Dew Point Measurement
Chilled mirrar
Lithium chloride (including DEWCEL)
Wet bulb thermometer
Direct measurement of relati ve humidity may be accomplished by
equating the change in dimension of a hygroscopic substance with the
variation in the moisture content of the air. The hair element (Figure
5-15) consists of a band composed of a number of selected and treated
human hairs. The membrane element consists of a strip of treated
animal membrane.
The instrument will record relative humidity between 20 and 90
percent with a maximum error of :t 5 percent. The calibration should be
adjusted initially by means of a psychrometer. If the hair element is
exposed to humidities over 90 percent or to free moisture, the calibration should be checked.
.Hair elements can be used at temperatures as high as 160For 70C

146 FEEDBACK PROCESSCONTROL

Fig. 5-15. Hair element.

without damage. However, in this event, they should be calibrated

under these temperature conditions.
The hair element is extremely responsive to changes in humidity,
and will follow gradual or small changes with good speed. However,
following a sudden large change, at least 30 minutes should be allowed
for the measuring element to reach a state of equilibrium.
The psychrometer consists of two temperature measurements.
One, called the dry bulb, measures the ambient temperature, while the
second, the wet bulb, is provided with a wick or sleeve saturated with
water that is evaporating. The wet bulb temperature will normally be
lower than that of the dry bulb. The lowering of this wet bulb temperatufe (wet bulb depression) is an indication ofthe moisture content ofthe
surrounding air, or of its humidity. See Appendix for convenient conversions of wet and dry bulb readings into relati ve humidity.
Example: dry bulb reads 90oP; wet bulb 80oP; difference is 10oP.
Pollowing the coordinates to their intersection, we find the relative
humidity to be 65 percent.
A portable version of this instrument, called a sling psychrometer,
provides a convenient way to check the calibration of relative humidity
instruments. Their air velocity past the bulb should be 15 feet (4.6
metres) per second or higher for accurate results.

'-+'2.

TEMPERATURE AND HUMlDffY

"0

MEASUREMENTS

147

.'0
.]

.."

I
l

,
1001

t,
,
,

..
E

~
~
2 00"
~
~
C
E

~~

'

;l

.,~f-.#"

I I

2
C

f-

.fH-~

1--

--'

"- -17' I
--/

-fi

-:eo

-40

-00

00
OEW POINT

40

10
TEMPERATURE

80

100

100

140

.F.

Fig. 5-16. Dewceloperatingcharacteristics.

The Foxboro dew point measuring system consists of an element,

the DEWCEL, a source of power to energize it, and a temperature
instrument to sensethe dew point temperature. The dew point element
consists of a cylinder of woven gIass fabric with two windings of carefully spaced gold wires. The two windings do Dot contact each other.
The fabric is impregnated with lithium chloride. The windings carry
low-voltage power applied through an incandescent lamp ballast resis-

tor.
Moisture determination by this lithium chloride element is based on
the fact that, for every water vapor pressure in contact with a saturated
salt solution, there is an equilibrium temperature at which this solution
neither absorbs noc yields moisture to the surrounding atmosphere.
This equilibrium temperature is shown in Figure 5-16 as the "DEW-

148 FEEDBACK PROCESSCONTROL

CEL@ Characteristic'Line." Below this equilibrium temperature, the

salt solution absorbs moisture. Above this equilibrium temperature, the
saturated salt solution dries out until only dry crystals are left.
The temperature thus determined may be used as a measure of dew
point temperature, absolute humidity, or vapor pressure.
The relative humidity of the atmosphere can be determined from a
knowledge of the dew point and reference to a vapor pressure curve or
chart (see Appendix, Table A-29).
Absolute measurements are less common in process applications
than relative humidity or dew point determinations, but blast fumace
control is one common application of absolute humidity measurement.

Questions

5.1. To obtainthe mostaccuratetemperaturerecord for the range 100to

150F,the filled thermal systemselectedwould be:
8. ClassI
c. ClassIII
b. Class11
d. ClassV
5-2. A resistancethermometerwould be chosenbecauseof:
8. Ability to measurehightemperatures
b. Economy
c. Higher accuracy
d. Simplicity
5-3. Thermocouplesare often chosenbecauseof:
8. High accuracy
b. Ability to measurehightemperatures
c. Economy
d. Ability to measurean extremelynarrow spanof temperature
5.4. The Class11(vapor-pressure)
thermometeremploysa nonlinearscale
because:
8. It is easierto read
b. It maybe ratioedto ftow measurements
c. The vaporpressurecurve is nonlinear
d. The mechanismemployedcausesnonlinearity
5-5. Which filled thermal system,or systems,would you selectfor
measuringthe room temperaturein your house?
8. ClassI
c. ClassIII
b. Class11
d. ClassV
5.6. A ClassIIIB (gas-filled)systememploysa large bulb to:
8. Minimize the ambienteffectson the connectingcapillary
b. Obtainthe optimumdynamicperformance

TEMPERATURE AND HUMIDITY

MEASUREMENTS

149

c. Make a maximumbulb areaavailablefor measurement

d. Reducethe powerproducedin the element
5-7. A temperaturerangebetween300Fand 310For 149Cto 154Cmusi
be measuredwith the greatestpossibleaccuracy.The besichoiceof system
would be:
a. A copper-constantan
thermocouple
b. A copperRTD
c. A nickel RTD
d. A ClasslA filled thermalsystem
5.8.
a.
b.
c.
d.

A hair elementis usedbecauseit:

Measuresabsolutehumidity
Is the mostaccuratetype of humidity measurement
Is simpleand inexpensive
Measuresdew point

5-9. A lithium chloride elementis usuallycalibratedto read:

a. Relative humidity
c. Absolutehumidity
b. Wet bulb temperature
d. Dew point
5-10. Lithium chloride is usedbecause:
a. It is nonpoisonous
b. It is inexpensive
c. It is extremelyhygroscopic
d. It does noi corrodethe equipment
5-11. Whena wet and dry bulb psychrometeris readto determinerelative
humidity:
a. The dry bulb will read lower thanthe wet bulb
b. The two thermometersmayreadthe same
c. The wet bulb will read lower thanthe dry bulb
d. A formula maybe employedto relatethe wet bulbreadingto relative
humidity
5-12. The purposeof using extensionleadwires thai havethe samethermoelectric
characteristicsas the thermocoupleis to:
a. Preventcorrosionat all junctions
b. Extend the referencejunction backto the instrument
c. Preventcreatingan unwantedreferencejunction
d. Make the thermocouplesystemoperatein standardfashion5-13.
Relative humidity is:
a. The moisture presentin a bodyof air expressedas a percentageof
saturationat the existingtemperature
b. The moisturein a body of air, in gramspercubic meter
c. The temperatureat which moisture will condenseCroma bodyof air
d. The ratio of actualmoisturein a volumeof air to the moisturethai
would exist at optimumcomfort in a similarvolume

150 FEEDBACK PROCESSCONTROL

5-14. A psychrometeris:
a. A hair elementinstrument
b. A "wet and dry bulb" humidity instrument
c. An instrumentthat sensespsychologicaldisturbances
d. An instrumentthat readsdirectly in dew point
5-15. A hygrometeris:
a. Convenientfor measuringspecificgravity
b. An instrumentthat measuresgas weight
c. Any instrumentthat measuresmoisturecontent
d. Anothernamefor psychrometer
5-16. A certainthermocouplehasa specifiedtime constantof 2 seconds.If
the processtemperaturechangesabruptly from 800to 900C,the temperature
readouton the indicatorattachedto the thermocouplearter6 secondselapse
will be approximately:
a. 860C
c. 900C
b.835C
d. 895C
5-17. The air velocity pastthe sensorsof a "wet and dry" bulb instrument
shouldbe:
a. 50 feet per second,minimum
b. 2 metresper second,maximum
c. Approximately4.6 metresper second
d. Any value
5-18. The advantageof usinga three-wirecableto connectan RTD to its
associatedinstrumentis that:
a. Referencejunction errors are eliminated
b. The effectof ambienttemperatureon the cablewill be minimized
c. Potentialfailures will be minimal
d. Resistancein the externalcircuit is reduced
5-19. A thermocoupleinstrumenthas an input resistanceof 50,000ohmsand
is usedwith an IC Type J couple.The leadwire is #18 gauge,120feet long.
What is the approximateerror contributedby the leadwire?
a. :t5.0 percent
c. :t0.05 percent
b. :t0.5 percent
d. :tO.1 percent
5-20. The differencebetweenanRTD calibratedto the NR 226curve and one
calibratedto the NR 227curve is:
a. A differentresistance-to-temperature
relationship
b. Nonexistent
c. Greateraccuracywith the NR 226curve
d. Greateraccuracywith the NR 227curve

Analytical measurements seek to define the contents of a process

stream and thus enable control of its composition. Some analytical
measurements are physical; others are electrochemical. The physical
types include humidity, density, differential vapor pressure, boiling
point rise, chromotography, ultraviolet, infrared, and turbidity, some
of which bave been discussed previously. Still others, such as spectroscopy, osmometry, and polarography, are laboratory techniques
and Dot usually associated with automatic control. Some are occasionally used, bot are beyond the scope of this book.
Many other analytical measurements, such as conductivity, pH
(hydrogen iODconcentration), ORP (oxidation-reduction potential), and
specific iODconcentration, are electrochemical. These measurements,
along with capacitance, will be covered in this chapter.

Electrical

Conductivity

While dissociation into ions and the resulting iOD concentration bave
been adequate concepts in the past, the fact is that DOt all the ions
present are necessarily effective. Some ofthem mar be "complexed,"
151

152 FEEDBACK PROCESSCONTROL

that is, "tied up" to'other ions and unavailable for reation. The concept of activity covers this situation. In short, a given compound will
dissociate to some degree, described by the dissociation constant, into
ions, and some portion of these ions will be active, described by the
activity coefficient. However, in many common reactions, the activity
coefficient is so near unity that concentration and activity mar be used
interchangeably. In a growing number of processes, the actual ion activity is different enough Cromthe ion concentration to make it necessary to use the proper term. In general, electrochemical measurements
measure activity rather than concentration, and it is always desirable to
refer to the measurements as ion activity rather than ion concentration.
Electrochemical measurements all rely upon the current-carrying
property of solutions containing ions. Some techniques measure all ions
present (electrolytic conductivity). Others respond mainly to particular
types of ions-hydrogen ions (pH); oxidizing/reducing ions (ORP); selected ions (ion-selective). These will be discussed separately below.
The ability to conduct electricity, or the reciprocal of electrical
resistance, is called conductance.The unit in which it is measured is the
reciprocal ohm, commonly called mho. The conductance of any conductor depends on the nature of the material, the shape of the conducting path, and the temperature. In analytical work, only the nature ofthe
material, in this case the type and activity of the ions present, is important. Thus, the term conductivity, the conductance of a volume of the
material of unit length and area, is generally used. (Conductivity has
largely replaced an earlier term, specific conductance.)
In actual measurement, a conductivity cell of known geometry is
immersed in the material, and the resistance (or conductance) across
the cell is measured. This gives a measurement that can be calibrated
directly in conductivity due to the known shape of the cell. Conductance and conductivity are related as follows: the greater the length
of a material of given conductivity, the higher its resistance (the lower
the tonductance); but the greater the aTeaofthe material, the lower the
resistante (the greater the conductance). That is,
aTea
conductance= conductivity iength

(6-1)

Sincethe unit of conductanceis mho, the unit of conductivity mustbe

conductivity =mho x

centimeters
squarecentimeters

(6-2)

ANALYTICAL MEASUREMENTS 153

or,
conductivity = mho. cm-1
It is unfortunately common practice to omit the dimensional unit,
so that conductivity is then referred to as mhos. This practice leads to
confusion with conductance, and should be avoided.
The conductivity of most electrolytes at the concentrations and the
temperature ranges normally encountered fall well below unity. For
this reason, the micromho per centmetre, the millionth part of the mho
per centmetre, is normally used.
In the SI system of units, the siemens replaces the mho. One mho
equals one siemens. In conductivity units, one micro mho per centimetre (J.mho/cm)equals one microsiemen per centmetre (J.S/cm).
However, the actual SI unit is the microsiemens per metre, since the
unit of length in the SI system is the metre rather than the centmetre.
One J.S/cmequals 100 J.S/m.
The conductivity of material of unit length and area which has a
resistance of 1,000ohms is 0.001 mho. cm-Ior 1,000micromho .cm-1
(sometimes called simply 1,000 micromhos).
Since the measurement depends on the geometry ofthe cell used, a
cell constant (F) has been defined to describe this geometry simply:

F = length (cm)
aTea(cm2)

(6.3)

When the length is I centimetre and the area I square centimetre,

F = 1.0 cm-I. Here, also, the dimensional unit is omitted in common
use, so that the cell is said to bave a cell constant of 1.0. Note that the
critica! dimensions are length and area, Dot volume. Thus while the
above example has a volume of 1 cubic centimetre ifthe length were 0.5
centimetre and the area were 0.5 square centimetre, the cell constant
would still be 1.0. But the volume would only be 0.25 cubic centimetres. Thus it is Dot correct to say that the resistivity is the resistance of unit volume, but only of unit length and cross-sectional area.
A cell with greater area and the same length will bave a lower cell
constant. But for the same solution (that is, for a given conductivity)
the conductance will be greater with this cell. If we solve Equation 6-1
for conductivity,

(6-4)

154 FEEDBACK PROCESSCONTROL

and substituting Equation 6-2,

conductivity = conductancex cell constant

A given measuring instrument will bave a certain conductance (resistance) range. But it may bave a variety of conductivity ranges by
simply using cells of different cell constants. For example, an instrument with a range of O to 100 micromhos per centimeter with a cell
having a constant ofO.Ol will bave a range oro to 1,000micromhos per
centimeter if a cell with a constant of 0.1 is substituted.

Types

ot Calibration

Conductivity instruments can normally be furnished with

1. Calibration in conductivity (,LLSx cm-! or ,LLmhox cm-!)
2. Calibration in terms of concentration of electrolyte
3. Calibration in terms of conductivity difference
Calibration

in Conductivity

Instruments of this type are measured in absolute units, ohms, mhos,

siemens, or microsiemens. Such instruments can be used to measure
the conductivity of any electrolyte at any solution temperature using
a measuring cell with known cell factor. The conductivity of most
solutions increases as the temperature increases. Therefore, if a conductivity instrument calibrated in J-mhox cm-r is used with a solution
of given concentration, the instrument reading will change if the temperature of the solution changes.
Compensation for this effect of temperature on the conductivity of
the solution is possible only if the solution's conductivity temperature
coefficient is known. For example, if an instrument calibrated CromOto
1,000J-mhox cm-r were provided with temperature compensation for
sodium chloride (NaCI), the instrument would Dot correctly compensate in any other solution. For this reason, instruments calibrated in
absolute units are generally Dot furnished with temperature compensation. Instruments ofthis kind are generally the simplest to calibrate and
lend themselves to more ftexibility of application, since they are Dot
limited to any particular electrolyte. If compensation is desired, it can
be supplied, bot only for one particular solution.

ANALYTICAL MEASUREMENTS 155

Calibration in Terms ot Concentration ot Electrolyte

Instruments calibrated in terms of concentration of electrolyte read
percent concentration, grams per litre, parts per million, and the like, for
the range specified. 'This type of calibration is made to the conductivity
values within the specified range of concentration, and at a specified
temperature of a given electrotyte. The instrument can therefore be
used only under the conditions for which it was calibrated.
However, special cases exist which sometimes make conductivity
useful in multiple-electrolyte solutions. If the material of interest is
much more conductive than others in solution, the contaminants mar
bave a negligible effect on readings. For example, contaminants slowly
build up in a sulphuric acid (H2SO4)bath for treating textiles. The
conductance of the contaminants is very low compared with thai of the
acido Thus, the instrument mar read directly in concentration of sulfuric acido Laboratory tests determine when concentration of contaminants is too high and a new bath is then made up.
Figures 6-1 and 6-2 show typical conductivity curves ofNaCl and
H2SO4.Note thai the H2SO4curve reverses itself in the region of 30 to

Fig. 6-1. Percent concentration. Conductivity oCNaCl at lOO"C(212F) and

1SOC(64.4F).

156

FEEDBACK PROCESSCONTROl

Fig. 6-2. PercentConcentration.Conductivityof H2SO.at 70oP(21.IOC)and

40"P (4.4C).

32 percent concentration and again in the vicinity of 84 percent and 93

percent. Obviously, it is impossible to calibrate for ranges that include
these points of inflection. However, in electrolytes having such maxima they shift with temperature. By controlling sample temperature to
a different value, certain ranges mar sometimes be measured which
would be impossible to measure at process temperature.
Polarization

When an electric current is passed through a solution, electrochemical

etfects known as polarization occur. These etfects, if Dot minimized,
wiU res uit in inaccurate measurement. One of the polarization etfects is
electrolysis. Electrolysis generaUy produces a gaseous layer at the
electrode surface, increasing the apparent resistance of the solution.
For tros reason, direct current voltage is Dot practical in conductivity
measurements. If the current is reversed, the layer wiU tend to go back
into solution. Thus, if an altemating current is applied, the polarization
etfect decreases, since the gases and other polarization etfects produced on one-half of the cycle are dissipated on the other half cycle.
Cell Construction

The sensitive portion of the cells shown i~ Figure 6-3 consists of two
platinum electrodes mounted in an H-shaped structure of Pyrex glass
tubing. The electrodes, located in separate sections of the tubing, are

ANALYTICAL MEASUREMENTS 157

Fig. 6-3. Type H conductivityassembly.

concentrically mounted platinum rings and are ftush with the inside
surface of the tubing. Fouling or damaging the electrodes is thus minimized, and the cells mar easily be cleaned chemically or with a bottle
brush. The platinum electrodes in these cells are coated with platinum
black to minimize polarization etfects.
The cell shown in Figure 6-4 employs graphite, rathr than metallic
electrodes. The type of graphite used has the same surface properties
with respect to polarization as metallic electrodes. These
cells require no platinization. They are cleaned chemically or by wiping
the surface of the electrode with a cloth or brush.
Conductivity cells (Figure 6-5) are used for detecting impurities in
boiler feedwater, for concentration of black liquor (in a pulp digester
for kraft paper), for determination of washing etfectiveness by measurement ofpulp wash, and in many other applications where the presence and concentration of a known salt, base, or acid must be deter-

mined.
Electrodeless

Conductivity

Measurements

In addition to the conventionalconductivity measurementsjust described another approachcalled an electrodelesstechnique mar be

employed.This techniquediffers from conventionalconductivity measurementin that no electrodescontactthe processstream.Instead,two

158 FEEDBACK PROCESS

CONTRO~

Fig. 6-4. Conductivitycells.

toroidally wound coils encircle an electrically nonconductive tube that

carries the liquid sample. The first coil is energized by an ultrasonic
oscillator (approximately 20 kHz) which induces an altemating current
in the liquido This current in turo induces a current in the second coil.

Fig. 6-5. Gatevalve insertiontype conductivitycell.

~
~
~.
~

ANALYTICAL MEASUREMENTS 159

The induced current in the second coil is directly proportional to the

electrical conductivity ofthe lquid carried by the tube. No direct contact between the coils and the solution is required, thus eliminating
potential maintenance problems. Figure 6-6 shows a typical electrodeless conductivity measuring system which transmits a 4 to 20 mA dc
signallinearly related to the measured conductivity.
Electrodeless conductivity is especially applicable to the higher
conductivity ranges such as 50 to 1,000 millimhos per centmetre.
Lower conductivity ranges such as 0.01 to 200,000 micromhos per
centmetre are best handled by the conventional techniques previously
described.

Hydrogen
The

term

pH

solution.
of

It

an

means
is,

at

vary

from

any

in

is

tion

of

the

pH

the

to

hydrogen

degree
amo

unt

ofthat

percent).

of
of
ion.

The

potentiometric

ions

acidity
a

(H+)

or

given
Activity

in

alkalinity

ion

actually

values

measurement

can

ofpH,

measurement

that

as
obeys

explanatech-

to
with

ofthe

of

This

an

measurement

along

activity

activity

1 (100
is

equation.

employed

the

effective

called

chapter,

contains

nique

The

percent)

this

of

a measure

time

Nemst

chapter

a measure

solution.

O (O

discussed

(pH)

therefore,

aqueous

present

the

lon Activity

determine
the

Nemst

equation.
To

aid

further

in

understanding
surement,
of
the
properties
some

a thorough

of
pH
meafundamentals
of
aqueous

,---TOROIO

SECONOARY

!'RI MARY
TOROIO
CELL BORE

~~,
solutions

most

be

understood.

CELL

lonization

or Dissociation

;\)'
.1

Stable chemical compounds

are electrically neutral. When
th
. d Wlthsolution,
ter
to ey
forro
areanmlxe
aqueous
wa
o

:;\
\"'
~

!~~
"",,' -'
ELECTRIC ./
"~~~~

they dissociate into positively Fig.6-6. Electrodeless

conductivity
measuring
and negatively charged parsystemo

160 FEEDBACK PROCESSCONTROL

ticles. These charged particles are called ions. Ions travel from one
electrode to the other if a voltage is impressed across electrodes
immersed in the solution.
Positive ions, such as H+, Na+, and so on migrate toward the
cathode, or negative terminal, when a voltage is impressed across the
electrodes. Similarly, negative ions, such as OH-, CI-, S04-2, and so
on, migrate toward the anode, or positive terminal.
The freedom of ions to migrate through a solution is measured as
the electrical conductivity of the solution. Chemical compounds that
produce conducting solutions are called electrolytes. Not all electrolytes completely dissociate into ions. Those that do (strong acids,
strong bases, and salts) are strong electrolytes. Others dissociate, but
produce fewer than one ion for every element or radical in the molecule. These are poor electrical conductors, and, hence, weak electrolytes. All weak acids and weak bases fall into this class.
At a specified temperature, a fixed relationship exists between the
activity of the ions and undissociated molecules. This relationship is
called the dissociation constant (or the ionization constant).
For hydrochloric acid (HCI), the dissociation constant is virtually
infinite, which means that for all practical purposes, the HCI is completely composed of positively charged hydrogen ions and negatively
charged chloride ions. Because of the essentially complete dissociation
into ions, hydrochloric acid is a strong acid:
HCl -+- H+ + Cl-

Conversely,acetic acid has a low-dissociationconstant. It breaks

up in the following way:
CH3COOH~ H+ + CH3COOFew hydrogen ions show up in the solution-less than one for every
100undissociated molecules-therefore, acetic acid is a weak acido
It follows that the strength of an acid solution depends on the
number of hydrogen ions available (the hydrogen ion activity). This
depends not only on the concentration of the compound in water, but
also on the dissociation constant of the particular compound.
When free OH- ions predominate, the solution is basic, or alkaline.
The dissociation of such a compound is illustrated below.
NaOH -+ Na+ + OH-

ANALYTICAL MEASUREMENTS 161

Sodium hydroxide is, for all practical purposes, completely dissociated and is a strong base.
Conversely, ammonium hydroxide, NH4OH, dissociates very little
into NH4+ ions and OH- ions, and is a weak base.
The OH- iODactivity, or strength, of a base depends on the number
of dissociated OH- ions in the solution. The w\mber available depends,
again, DOtonly on the concentration of the compound in water, but also
on the dissociation constant of the particular compound.
Pure water dissociates into H+ and OH- ions, but is very weak in
the senseused above. That is, very little ofthe HOH breaks up into H+
ions and OH- ions. The number of water molecules dissociated is so
small in comparison to those undissociated that the activity of (HOH)
can be considered 1 (100 percent).

HOH ~ H+ + OHAt 77F or 25C, the dissociation constant or water has been
determined to bave a value or 10-14. The product or the activities
(aH+)(aOH-) is then 10-14.
If the activities or hydrogen ions and hydroxyl ions are the same,
the solution is neutral; the H+ and OH- activities must both be 10-7
mols per litre.
It must be remembered that, no matter what compounds are present in an aqueous solution, the prJduct or the activities or the H+ ions
and the OH- ions is always 10-14at 77F or 25C.
If a strong acid, such as HC1, is added to water, many hydrogen
ions are added. This must reduce the number or hydroxyl ions. For
example, if HCl at 25C is added until the H+ activity becomes 10-2,the
OH- activity must become 10-12.
The pH Scale

It is inconvenient to use terms as 10-7, 10-12and 10-2. Therefore, it has

become common to use a special term to represent degree of acidity, or
activity of hydrogen ions. This term is pH, originally derived from the
phrase "the power of hydrogen."
The pH is defined as the negative of the logarithm of the hydrogen
ion activity, or as the log ofthe reciprocat ofthe hydrogen ion activity.
pH = -log aH+ = log

(-::
aH

(6-5)

162 FEEDBACK PROCESSCONTROL

Table 6-1

Temperature oC

-Log

(aH+)(80H-)

14.94
14.00
13.26
12.69
12.26

25
50
75
100

Neutral pH

7.47
7.00
6.63
6.35
6.13

If the hydrogen iOD activity is lO-x, the pH is said to be x. For

example, in pure water (at 25C), the activity of hydrogen iODis 10-7.
Therefore, the pH of pure water is 7 at 25C.
A point frequently overlooked is thai the pH for neutrality varies as
the solution temperature changes due to a change in the dissociation
constant for pure water, as shown in Table 6-1.
An acid solution contains more hydrogen ions than hydroxyl ions.
Table 6-2
Hydrogen [on

Activity
Mols/Litre

Hydroxyl Jan
Activity
Mols/Litre

10.1

0.00000000000001

1
2

0.01

0.001

0.0001

O.0000 1

0.000000001

0.000001

0.00000001

0.0000001

0.0000001

0.00000001

0.000001

pH

Neutral

9
10
11
12
13
14

0.0000000000001

0.000000001
0.0000000001
0.00000000001
0.000000000001
0.0000000000001
0.00000000000001

0.000000000001

0.0000000000I
0.0000000001

0.00001
0.0001
0.001

0.01

0.1
I

10

ANALynCAL MEASUREMENTS 163

Therefore, the activity of hydrogen ions will be greater than 10-1, that
is, 10-6, 10-5,10-4. The pH of an acid solution then, by definition, must
be lower than 7, that is 6, 5, 4.
If the number of OH- exceeds H+ ions, the hydrogen iODactivity
must be less than 10-1, that is, lO-s, 10-9,10-1. Therefore, the pH will
be higher than 7, that is, 8, 9, 10.
Table 6-2 demonstrates that a change of just one pH unit means a
tenfold change in strength ofthe acid or base. The reason for this is that
there is an exponential relationship between pH numbers and hydrogen
iODactivity. With so large a change in acidity or alkalinity taking place
with a change ofjust one pH unit, the need for sensiti ve pH measuring
and control equipment cannot be overemphasized.
Table 6-3 shows the nominal pH values for a number of common
solutions.

Table 6-3
pH
14 ~ Caustic soda 4% (I.ON)
--Calcium hydroxide (sat'd sol.) (Iime)
12 --Caustic soda 0.04% (O.OIN)11
--Ammonia 1.7% (I.ON)
--Ammonia 0.017% (O.OIN)
9 +- Potassium acetate ~.98% (O.lN)
8 +- Sodium bicarbonate 0.84% (O.lN)

6
5 ~ Hydrocyanic acid 0.27% (O.lN)
3 ~ Acetic acid 0.6%

"'
1 +- Hydrochloric acid 0.37% (O.IN)
+- Sulfuric acid 4.9% (I.ON)
O +- Hydrochloric acid 3.7% (I.ON)
-1 +- Hydrochloric acid 37% (lON)

d
-~---~
)LASrIC

164 FEEDBACK PROCESSCONTROL

Fig. 6.7. Schematicdiagramor glass-tippedpH measurement

electrode.
Measurement

ot pH-The

Glass Electrode

A number of different methods for measuring pH are available for

laboratory use, but only one has proved sufficiently accurate and universal for industrial use: the pH glass electrode measuring system.
A special kind of glass, sensitive to hydrogen ions, has been found
to be the most useful medium for measuring pH. Figure 6-7 shows a
INTEGRAL
CABLE

(
r--y=
MOISTURE

SEAL
CAl

SILVERSILVER
CHLORIDE
COLUMN

SHIELDED
!LECTRODE

CABLE

GLASS-TO-METAL
SEAL
; "TOTAL GLASS" CONSTRUCTION)
NTERNAL

SOLUTION

,H-SENSITIVE

Fig. 6-8. Actual measuringelectrode.

MEMBRANE

ANALYTICAL MEASUREMENTS 165

schematic diagram of the measuring electrode in which this special

glass is used. A thin layer of the special glass covers the tip of the
electrode. Figure 6-8 shows an actual measuring electrode.
This glass contains a chamber filled with a solution of constant pH.
An internal electrode conductor element immersed in the internal solution is connected to the electrode lead. The conductor element is discussed in the section on temperature compensation.
If the hydrogen iODactivity is greater in the process solution than
inside the electrode, a positive potential difference will exist across the
glass tip. That is, the external side of the glass will bave a higher
positive potential, and the internal a lower positive potential. If the
process is lesser in hydrogen iODactivity, a negative potential difference will exist.
The relationship between the potential difference and the hydrogen
iODactivity follows the Nernst equation.
E

E'

lo
nF

aH+o~ts~de
aH+
mslde

(6-6)

where:
E = potential difference measured
Eo = a constant for a given electrode system at a specified temR =
T =
n =
F =
aH+ =

perature (25C)
the gas law constant
Absolute temperature
charge on iOD(+ 1)
Faraday's number, a constant
hydrogen iODactivity

In the most common type of pH electrode, that with an internal

buffer solution of7 pH, the voltage across the membrane will be zero at
7pH.
Reference

Electrode

The potential inside the glass is the output o the measuring electrode.
It must be compared to the potential in the solution outside the giassto
determine potential difference and, hence, pH. The sensing o the potential in the solution must be independent o changes in solution composition. Platinum or carbon would act as aRP, or redox-measuring
electrodes. They would be responsive to oxidants or reductants in the
solution. They would Dot yield a true solution potential with solution
composition changes.

166 FEEDBACK PROCESSCONTROL

A reference electrode is the answer. Figure 6-9 is a schematic

diagram of a typical reference electrode.
The connecting wire is in contact with silver that is coated with
silver chloride. This, in turn, is in contact with a solution of potassium
chloride saturated with silver chloride (AgCI). The saturated KCI,
called the salt bridge, in turn, contacts the process solution.
Because the concentration of all of the components Cromthe connecting wire to the KCI solution is fixed, the potential Cromwire to
KCI is fixed. The potential between the KCI and the process solution
(called the liquid junction potential) is normally small, and will vary
only insignificantly with process changes. The overall potential of the
reference electrode is thus essentially constant, virtually independent
of process solution changes, to meet the requirement mentioned earlier.
As the need for high accuracy or repeatability becomes greater,
protection of the reference electrode's internal environment becomes
more important. AIso, prevention of coatings over the electrode tip
becomes more critical. Leaking in of process solution must be prevented. Noxious chemicals, or even distilled water, would change the
concentration, contaminating the electrolyte. The resuIt would be an
unpredictable change in the reference potential. Conductive coatings

Fig. 6-9. Schematicdiagramofreferenceelectrode.

ANALYTICAL MEASUREMENTS 167

on the electrode tip mar cause spurious potentials, and nonconductive

coatings mar literally open the measurement circuit.
Two types of reference electrodes are available-ftowing and nonftowing. The ftowing version (Figure 6-10) is usually selected when the
highest possible accuracy or repeatability is needed. To prevent electrolyte contamination and to minimize coatings or deposits on the tip,
there is provision for a ftow of KCl electrolyte out through the paTOllS
electrode tip. Instrument air is applied inside the electrode. It maintains
a pressure on the electrolyte slightly higher than thai of the process
liquido A 3 psi (20 kPa) differential pressure forces a trickle of 1 or 2 ml
per dar of electrolyte out into the process solution.
The nonftowing version (Figure 6-11)of a reference electrode is often
considered standard. No external pressure is applied to ibis type of
electrode. The electrolyte is a paste of KCl and water thai actually does
ftow, or diffuse, into the process solution. The ftow rate mar be as low
as 0.01 ml per dar; ibis is a function of diffusion properties. After

Fig. 6-10. Flowing referenceelectrode.

168 FEEDBACK PROCESSCONTROL

Plastlc

Body

Internal
Electrode
Element
AgAgCI

Fig. 6 -11. Construction details of nonfiowing reference electrode.

depletion of the electrolyte, the nonflowing electrode is usually either

recharged or replaced.
Tem peratu re Com pensati on

The major potential difference exists in the reference electrode, between the metallic silver and the silver ions in the AgCI solution. It
follows from the Nemst equation:
(6-7)
Because R, n, and F are constants, and aAgoand AAg+ are fixed,
this potential (E) will vary only with the absolute temperature of the
electrode. Since there is a definite chemical relationship between the
ionic silver and the activity of chloride in the KCl electrolyte, the above
expression mar also be written as:
E = Eo -kT

log aCl-

(6-8)

To compensate for ibis temperature sensitivity, another silversilver chloride electrode is inserted into the top of the glass measuring
electrode, the internal conductor element mentioned previously. As the
temperature of the electrode changes, the potentials of the reference
electrode and the conductor element will vary, but will effectively cancel each other, assuming similar values for aAg+ (or aCI-) in each electrode as is usually the case.

ANAl..YT!CAL MEASUREMENTS 169

A remaining temperature effect influences the potential across the

membrane of the glass measuring electrode. This will vary with the
absolute temperature and is greatest at high- and 10w-pH values. At
values around 7, the variation with temperature is zero. This paint is
called the isopotential paint. Figure 6-12 shows the magnitude of the
temperature error, away from the isopotential paint, with glass electractes.
In general, some type of temperature compensation is essential for
accurate, repeatable pH measurement. If the process temperature is
constant, or ifthe measured pH is close to avalue of7, manual temperature compensation mar be used. Otherwise, for good results, automatic temperature compensation will be required.
Reading the Output

ot the pH Electrodes

The high resistance of the giass measuring electrode results in the need
for a millivolt meter with very high internal resistance or sensitivity to
measure the cell output. The pH electrodes mar be compared to a pair
of ftashlight cells in series, but unlike the ftashlight cell, the electrodes

Fig. 6-12. Graph oCpH measurement errors versus pH values at various solution
temperatures due to temperature differences across the measurementelectrode tip.

170 FEEDBACK PROCESSCONTROL

are characterized by xtremely high internal resistance (as high as a

billion ohms). Unless the millivolt meter employed to measure the
voltage created by the electrodes has almost infinite resistance, the
available voltage drop will occur across the internal resistance and
result in no reading, plus possible damage to the electrodes.
In addition to high impedance or sensitivity, the instrument must
bave some provision for temperature compensation. It must also bave
available adjustments that provide for calibration and standardization.
The Foxboro pH transmitter shown in Figures 6-13 and 6-14 is typical
of an instrument designed to make this measurement. Its output is
typically 4 to 20 mA dc which can be fed to a control system. When pH
is measured in the process plant, it is generally for the purpose of
control.
pH Control

Experience has shown that it is virtually impossible to control pH

without considering the process composition as well as more obvious
parameters offtow rate, pressure, temperature, and so on. This consideration is frequently even more critical where plant waste neutralization is being controlled; in such applications, there are usually wide,
relatively uncontrolled variations in composition as opposed to the
normally better defined, and controlled, process stream.
A pH control system can be a relatively simple on/off controlloop
in some batch processes. At the other extreme, it can be a complex,
feedforward/feedback system with multiple sequenced valves for continuous neutralization over a wide, dynamic pH range.

Fig. 6.13. Block diagrarnoftypical transmitter.

ANALYTICAL MEASUREMENTS 171

Fig. 6-14. Actualtransmitter.

The rate of change of inlet pH and fiow is a major consideration in

determining whether feedforward control is necessary. Since changes
in pH refiect logarithmic changes in composition, they can atrect reagent demands tremendously and rapidly. Since reagent addition varies
directly and linearly with fiow, doubling the fiow requires twice as
much reagent at a given pH. But a change of one pH unit at a given fiow
requires a tenfold change in reagent. Flow rate compensation is often
required in complex pH control loops, but usuany only when fiow
changes are greater than 3 : 1.
The amount of reagent addition for butrered and unbutrered solutions, as seen on conventional acid-base titration curves (Figure 6-15),
is also an extremely important consideration in the design of a pH
control system. Reaction time of the various reagents musi also be

,;,

172 FEEDBACK PROCESSCONTROL

005
BASE

001

REAGENT

001

005
AtiO

OI

REAGENT

MOLS REAGENTREQUIRED/LlTER INFLUENT

Fig. 6-15. Typical neutralization curves for unbuffered solutions (strong acid or strong
base) and buffered solutions. Examples of buffered solutions are (I) weak acid and its
sale with the addition of a strong base, and (2) strong acid or base, concentrated, 0.01

M.

given serious considerationin designingboth the processand the control system.

A pH measurementand control system, in conjunction with a
well-designedprocess,canbe a mostsuccessful,reliable systemwhen
proper considerationhas beengivento the various designparameters.
Summary
Continuous pH measurement is avaluable tool for industry and gives
information thai cannot be obtained economically in any other fashion.
It is generally more complex than, say, temperature or pressure measurements, but with constant use, it becomes a routine analytical tool.

Oxidation-Reduction

Potential

Oxidation-reduction potential or redox measurements determine the

oxidizing or reducing properties of a chemical reaction. In this application, the term oxidation is used in its electrochemical senseand applies
to any material which loses electrons in a chemical reaction. By definition, a reduction is the opposite of oxidation, or the gaining of external
electrons. There can be no oxidation without an attending reduction.

ANAL YTICAL MEASUREMENTS 173

For example, a ferrous iODmar lose an electron and become a ferric iOD
(gaining increased positive charge) if a reduction, say, of stannic to
stannous ions (which is the reverse ofthis operation) occurs at the same
time.
This measurement uses electrodes similar to those in pH measurement (except that metal is used instead of glass), but the two types of
measurement should DOtbe confused. The measurement depends on
the oxidizing and reducing chemical properties of reactants (not necessarily oxygen). The inert metal electrode versus the reference electrode
will produce a voltage that is related to the ratio of oxidized to reduced
ions in solution.
This measurement is similar to that of pH in the requirements that
it places on the voltmeter used with it. It is useful for determinations in
waste treatment, bleach production, pulp and paper bleaching, and

others.
The application of ORP measurement or control depends on a
knowledge of what goes on within the particular process reaction. The
ORP measurement can be applied to a reaction only under the following
conditions:
1. There are present in the solution two reacting substances: one that
is being oxidized and one that is being reduced.
2. The speed of reaction, following an addition of one of the substances above, is sufficiently fast for good measurement or control.
3. Contaminating substances are held to a minimum, especially those
capable of causing gide reactions of oxidation or reduction.
4. The pH ofthe solution is controlled in those areas ofthe applicable
curve where variations in pH can affect the ORP measurement.

lon-Selective

Measurement

Certain applications require that the activity of a particular iODin solution be measured. This can be accomplisehd with an electrode designed
to be sensitive to the particular iODwhose concentration is being measured. These electrodes are similar in appearanceto those employed to

measure pH, but are constructed as glass-membraneelectrodes,

solid-membrane electrodes, liquid-ion-exchange-membrane electrodes,
and silicone-rubber-impregnated electrodes.
A common example ofion-selective measurementis found in many
municipal water treatment plants. In this application, a measurement of

174 FEEDBACK PROCESSCONTROL

ftuoride iOD activity can be related to concentration. The measuring

electrode is made of plastic and holds a crystal of lanthanum ftuoride
through which ftuoride ions are conducted. The reference electrode is
the same as thai used for pH. The electrode output is read on a high
impedance voltmeter very similar to thai used with pH electrodes.
Many applications are possible using the ion-selective technique.
At present, measurements in water treatment include hardness,
chlorine content, waste, and pollution. Many other industrial applications bave been suggested and many bave proven effective.

Chromatography

The original application of chromatography consisted of studying the

migration of liquid chemicals through porous material, usually paper.
The word chromatography means color writing, which occurs when
certain extracts or dyes applied to paper produced colored bands.
Modem gas chromatography dates from 1952,when James and Martin
first used the principIe to separate a mixture of volatile fatty acids
having nothing to do with color.
The modem chromatograph is generally used for gas analysis. It
consists of a column, a tube or pipe packed with materials that will
absorb the gases being analyzed at different rates. The gas to be analyzed is carried through the columns by an inert carrier gas, usually
helium. As the gas mixture passes through the column, different components are delayed for varying increments of time. Thus, as the gas
stream leaves the column and is passed through a gas detector, a
chromatogram, or "fingerprint," is formed which mar be used to determine the components in the gas as well as the quantity present. The
packing material for the column must be properly selected to provide
the separation desired for the sample under analysis. The operating
parameters, such as temperature, carrier gas flow and pressure, sample
valve timing and detector sensitivity, all influence output. Therefore, in
a working chromatograph, these factors must be carefully controlled.
Chromatographs are widely used in process work; in fact, they are
among the most popular analytical tools. They are used for both liquids
and vapors, and adaptations are available to provi de an output in conventional analog form such as a 3 to 15 psi (20 to 100 kPa) pneumatic
signal. The details ofthis instrument' s operations are beyond the scope
of this book. Several references suitable for further study are listed at
the end of this chapter.

cable

ANALYTICAL MEASUREMENTS 175

Table6-4. ComparisonofCapaitance,Conductivity,pH, andORPTechniques

Specificity
Sensitivity
Conducting
fluids
Nonconducting
fluids

Maintenance
InstalIation
problems
Cost

Capacitance

Conductivity

pH or
ion-selective

ORP

Poor
Pair
Not
applicable
Good

Poor
Good
Good

Excellent
Excellent
Good

Poor
Excellent
Good

Not

Not

applil

appIiI

Low
Low

Low
Low

High
High

Not
applicable
Medium
Low

Low

Low

Medium

Low

Capacitance

Capacitance is Dot, as applied, an electrochemical measurement. However, a measurement of some characteristic in a nonconducting liquid is
frequently required, and in these applications, capacitance mar provide
the answer.
Electrical capacitance exists between any two conductors separated by an insulator (dielectric). The amount of capacitance depends
on the physical dimensions of the conductors and the dielectric constant of the insulating material. The dielectric constant (K) for a vacllum is 1; all other dielectric materials bave a K greater than 1. For
example, for air, K = 1.00588; for dry paper, K = 2 to 3; for pure
water, K = 80. The Table of Dielectric Constants of Pure Liquids
(NBS Circular 514), available Crom the U.S. Govemment Printing
Office, lists the dielectric constants of nearly all common liquids.
Mixtures of materials bave a composite value of K that can be
directly related to composition. This approach is readily applied to the
determination of water content in materials such as paper and crude oil.
Capacitance is also applicable to level, interface, octane, and other
measurement problems.
In general, the capacitance technique has provided a solution to
many measurement problems that cannot be solved easily by more
conventional means. Table 6-4 compares the application of pH, ORP,
ion-selective, conductivity, and capacitance measurements to process
situations.

176 FEEDBACK PROCESSCONTROL

References
Conductivity cells. Technical Information Sheet 43-10a. Foxboro, MA., The
Foxboro Company.
Fluoride measuring systems-potable water. Technical Information Sheet
43-21a. Foxboro, MA., The Foxboro Company.
Fundamentals of pH measurement. Technical Infor~{ltion Sheet 1-90a.
Foxboro, MA., The Foxboro Company.
pH e1ectrodesand ho1ders. Technical Information Sheet43-11a. Foxboro,
MA., The Foxboro Company.
Theory and app1icationof oxidation-reduction potentials. Technical
Information Sheet 1-61a. Foxboro, MA., The Foxboro Company.
Pneumatic composition transmitter. Technical Information Sheet 37-1.30a.
Foxboro, MA., The Foxboro Company.
Liptk, B. G. Instrument Engineer's Handbook, Sec. 81, Volume I. Radnor,
PA: Chilton.
Shinskey, F. G. pH and pION Control in Process and Waste Streams.
New York: John E. Wiley and Sons, 1973.

Questions

6-1. Conductivityand pH measurements

are:
a. Two differenttechniques
b. Similar in operation
c. ldentical but given differentnames
d. Two techniquesthat usethe sameequipment
6.2.
a.
b.
c.
d.

The three factors that control the conductivityof an electrolyteare:

Specificgravity, density,and volume
Concentration,material in solution,and temperature
Color index, turbidity, and temperature
Hydrogenion concentration,temperature,and pressure

6-3. pH is a measureor:
a. Effective acidity or alkalinity of a liquid
b. The oxidationor reductionpropertiesof a solution
c. Specificconductanceof an electrolyteor total ionic activity
d. Purity in an aqueoussolution
6.4.
a.
b.
c.
d.

A buffer solutionis usedwith pH-measuringinstrumentsto:

Standardizethe equipment
Protectthe equipment
Cleanthe electrodes
Platinizethe referenceelectrode

6-5. Oxidation-reductionpotential (ORP)is a measurement

or:

ANALYTICAL MEASUREMENTS 177

8.
b.
c.
d.

The oxidizing or reducingchemicalpropertiesof a solution

The oxygenpresentin any quantityof a givengasmixture
The hydrogenion concentrationin a given solution
The degreeof ionizationfor a particularsolution

6-6. Capacitancemeasurements
are usuallyappliedto:
8. Conductingliquids
c. Gasmeasurements
b. Nonconductingliquids
d. lonized gases
6.7.
8.
b.
c.
d.

Which of the following are electrochemicalmeasurements:

Humidity and density
Turbidity and differentialvaporpressure
pH and ORP
Dew point and boiling point rise

6-8. A salt (NaCl)solutionmustbe controlled at a concentrationof 12

percent.The bestchoiceof measurement
for the control systemwould be:
8. Conductivity
c. ORP
b. pH
do Capacitance
6-9. An industrial effluentstreamis to be neutralizedby addinga sodium
hydroxide solution.The bestchoiceof analyticalmeasurement
for the control
systemwould be:
8. Conductivity
c. ORP
b. pH
d. Capacitance
6-10. lon-selectivemeasurements:
8. Are similar to conductivity in operationbut usea differentcell
b. Are similarto capacitancemeasurements
but usea differentinstrument
c. Are similarto pH measurements
but use differentelectrodes
d. Are similarto densitybut usemore exacttechniques
6-11. A pH control systemis usedto neutralizea chemicalwastestream
beingdumpedinto a municipalsewagesystem.The desiredpH is 7 or
completeneutrality. An unfortunateaccidentshortcircuits the cable
connectingthe electrodesto the measuringtransmitter.The result will be:
8.
b.
c.
d.

The control valve admittingthe neutralizingagentwill fully open

The control valve will close
The control valve will remainapproximatelyat its hal openposition
The systemwill cycle

6-12. The concentrationof salt in a liquid usedto carry a slurry niust be

monitored.The bestchoice of measurementwill be:
8. Electrodelessconductivity
c. ORP
b. pH
d. An ion-selectivesystem
6-13. A pH systemis to be selected.It is requiredthat the systemfunction
with little or no maintenance.The referenceelectrodeselectedwould be:

178 FEEDBACK PROCESSCONTROL

8. A ftowingtype
b. A pressurizedftowing type

c. A nonftowingtype
d. A pressurizednonftowing
type
6-14. The pH of 8 streamis to be monitoredaccurately.It is discoveredthat
the temperatureof this streamvariesfrom 40 to 60F:
8. A measuringsystemwith automatictemperaturecompensation
is
indicated
b. Temperaturecompensationis not necessary
c. A manualtemperaturecompensationwill be adequate
d. The pH rangewill determinethe needfor temperaturecompensation
6-15. The most popularcarriergasusedin gaschromatographs
is:
8. Helium
c. Hydrogen
b. Air
d. Oxygen

SECTION

II PNEUMATIC

AND ELECTRONIC

CONTROL

SYSTEMS

The feedback controlloop, introduced in Chapter 1, requires further

discussion. This chapter is devoted to a more detailed description of
loop operation and its implicit mathematical relationships.
ln the feedback loop, the variable to be controlled is measured and
compared with the desired value-the set point. The difference between the desired and actual values is called error, and the automatic
controller is designed to make the correction required to reduce or
eliminate this error.
WitQin the controller, an algebraic "summing point" accepts two
inputs-one from measurement, the other from set. The resulting output
represents the difference between these two pieces of information. The
algebraic summing point is shown symbolically in Figure 7-1, where c
represents measurement; r represents set point; and e represents error.
Both rand c must be gi ven their proper signs (+ or -) if the value of e is
to be correct.
A typical closed-loop feedback control system is shown in Figure
7-2. Assume that the algebraic summing point is contained within the
feedback controller (FBC). ln order to select the proper type of feedback controller for a s~cific process application, two factors-time
and gain-must be considered. Time consists of dead time, capacity,
179

180 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

e~r
e=r-c
Fig. 7-1. Algebraic summing point.

and resistance. These factors cause phase changes that will be described in this chapter.
Gain appears in two forms-static and dynamic. Gain is a number
that equals the change in a unit's output divided by the change in input
which caused il.
Gain = AAoutput
input
The static gain of an amplifier is easily computed if, for a given step
change in input, the resulting change in output can be monitored. Case
A in Figure 7-3 shows an amplifier with a static gain of one. In Case B,
the output is magnified when compared with the input by a factor
greater than one. In Case C, the output change is less than the input.
Rere, the input has been multiplied by a number less than one.
The dynamic gain of an amplifier mar be computed by inducing a
giDewave on the input and observing the resulting output. Figure 7-4
illustrates this procedure with an amplifier that has no time lag between
its input and output. If the amplitude of the output is only hali as high as

Fig. 7-2. Closed-loopcontrolsystem.

TUE FEEDBACK CONTROLLOOP 181

Case

20%
5%
10% [ 10%

Output change

10%
10%

Static gain

!Q=1
10

22
10=2

Inputchange

1.--1
10-2

Fig. 7-3. Static gain.

the amplitude of the input giDe wave, the amplifier is said to bave a
dynamic gain of 0.5 for the particular frequency of the input wave, for
example, 1 Hz.
By monitoring the output amplitude for many different input frequencies, a series of dynamic gain numbers mar be found. A plot of the
amplitude ratio (gain) as a function of frequency of the input giDewave
is the gain portion of a Bode diagram, or a frequency-response curve
(Figure 7-4). Note that the higher the frequency, the lower the gain.

Fig. 7-4. Dynamicgain.

)(

182 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

This is true for nearly all processes and instruments. The frequency
scale is normally a logarithmic scale. Gain or amplitude ratio is also
normally expressed in decibels, where l decibel equals 20 times the
logarithm to the base 10 of the amplitude ratio.
As the frequency becomes lower and lower, and finally approaches
zero, a measure of the amplifier's static gain can be obtained.
Just as the static or dynamic gain of an individual amplifier can be
computed, so can the static or dynamic gain of a process controlloop.
Figure 7-5 shows a temperature controlloop. The static loop gain has
been computed by multiplying the static gains of each of the individual
components of the loop. Likewise, the dynamic loop gain could be
calculated by multiplying the dynamic gains of each element at a particular frequency. Pure dead time has a gain of 1.
Each element in the controlloop contributes gain to the totalloop.
Increasing the size of a control valve or narrowing the span of a transmitter has exactly the same effect as increasing the gain of a controller.

Fig. 7-5. ResponsecharacteristicsoCeachelementin a Ceedback

controlloop-heat,
exchangerprocesso
TT Transmittergain = psif'F
TC Controllergain = psi/psi
Diaphragmactuatorgain = inches/psi
Valve gain = Btu/inches(lift)
Reat exchangergain = F/Btu
F
Si ~'
"
.,
-.,
.
~
.~
Static loop gain = l' ~

)(

PS! ,

PS!

)(

mches

)( ~

Btu

The

THE FEEDBACK CONTROLLOOP 183

Proper selection of valve size and transmitter span is as important as

selecting the gain range in the controller.
In practice, the gain of the components in the loop mar nat be
uniform throughout their operating range. For example, control valves
having several characteristics will exhibit linear, equal-percentage
(logarithmic), or other types of gain curves. Unless special precautions
are taken, the gain of many control loops will change if the operating
level shifts. The gain most then be readjusted at the controller to provide satisfactory control.
Let us investigate how a feedback control system functions under
closed-loop conditions.
The Closed-Loop

Control

System

process consists of multiple capacitances (to store energy) andresistances

(to energy flow), as shown in Figure 7-6. Similarly, the
measuring system generally contains multiple capacitances and resis-tances.
The controller also consists of capacitance and resistance net-works.
The final operator, or valve motor, likewise contains capacitance and resistance, and the control valve is essentially a "variable
resistance." Thus, the analysis of the system is simply an analysis of
how signals change as they pass through resistance-capacitance and
dead time networks of various types.

Fig. 7.6. Closed-loopprocesscontrol systemillustratestheRC combinationsthat could

exist arounda feedbackloop. Eachatfectsprocessstability.

184 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Phase Shift Through

RC Networks

Figure 7-7 illustrates the phase shifts that occur when a sine wave is
applied to a network ofresistances and capacitances. Each succeeding
RC combination contributes its own phase shift and attenuation. In
Figure 7-7, bottom, a phase shift of 180degrees (a full half cycle) takes
place. That is, the output (eRa)is 180degrees out of phase with the input
(e applied). The signal has also been attenuated because each resistor
causes a loss in the energy level in the system. This phase change can
lead to instability in a closed-loop system.
Oscillation
In a physical system employing feedback, instability will occur if
energy is fed back in such direction (phase) as to sustain the instability
or oscillation.
Oscillators are often divided into two major categories, those thai
utilize a resonant device and those thai do noto The resonant oscillator
uses a device thai requires a minimum of added energy each cycle to
maintain oscillation. On the other hand, the nonresonant oscillator is
simply an amplifier with some type of phase-shifting network between
output and input. A public address system with the gain so high thai it
howls is a nonresonating type.

I
-"
I
SIMPLIFIEO PHASE SHIFTER

Fig. 7-7. Phase ofsigna! acrossR, R2' andRa is different. At one frequency, the output
signa! can be exact!y 180degrees out of phase with the input signa!, as shown in the

graph.

THE FEEDBACK CONTROLLOOP 185

Academically, both the resonant and nonresonant oscillators follow the same mathematicallaw, but the amplifier gain requirements are
quite different for the two categories. With a nonresonant circuit, oscillation will occur (1) if the feedback is positive (inphase) and (2) if the
gain is unity or greater. These two conditions are necessary for sustained oscillation.
Figure 7-8 shows a simple mechanical oscillator. ln ibis system,
energy stored in the spring is transferred to the weight, back to the
spring, to the weight again, and so on. The oscillations soon die Dut
because some ofthe energy is dissipated in each cycle. lfthe oscillation
is to be continuous, energy musi be added to make up the losses.
ln the bottom part ofthe figure, ibis energy is added by depressing
and releasing the weight. If the energy is to enforce the oscillation, it
musi be in the proper phase with the motion. That is, enforcing energy
musi be added at a time when both the energy and the oscillation are
moving in the same direction.
Figure 7-9 shows a free-running oscillator in which the required
energy is introduced through feedback. This circuit is called a phase
shift oscillator. lt consists of a transistor amplifier and a feedback path
comprised of three resistance-capacitance combinations. This circuit is
important because it is a simple illustration of the manner in which an
ordinary amplifier can be made to oscillate simply by use of feedback.

l@~
Fig. 7-8. At top, the magson the spring received a single downward displacement and
the oscillation is delayed. Below, arter the initial displacement, the magsreceives
periodic small displacements and the oscillation is continuous.

11

186

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

The transistor amplifier Dot only makes up for circuit losses, but
also contributes a phase shift of 180degrees; that is, the collector is 180
degrees out of phase with the base signal. EachRC element contributes
additional phase shift, depending on the values of the components and
the frequency involved. If each RC combination contribute s a phase
shift of 60 degrees, the three together will result in a 180-degreephase
shift. This, added to the 180-degreephase shift through the amplifier,
results in 360 degrees of phase shift. The feedback signal is now in
phase with the input signal. This inphase feedback produces continuous oscillation.
Since only one frequency will be shifted in phase exactly 360 degrees by the RC networks, the oscillator output is of this one fre-

quency.
The closed-loop control system (Figure 7-6) also has phase-shifting
networks, made up of all process and controller <:omponents,together
with an amplifier capable of contributing sufficient gain to overcome the
system's losses. This closed-loop system contains the same basic ingredients as the phase-shift oscillator. Each resistance-capacitance
combination will shift the phase of the energy flowing around the controlloop in the same manner as in Figure 7-9.
Oscillation (instability) will occur whenever (1) the phase relationships through the vario us resistance-capacitance combinations provide
feedback in proper phase, and (2) the system gain is unity, or greater at
the frequency at which the phase shift is 360 degrees.
Figure 7-10 illustrates a plot of system gain (output/input signalBode diagram) and output/input phase shift versus frequency for a
three-mode controller. The phase shift (expressed in degrees of lead or

Fig. 7-9. Shift of phasein threeRC sectionsmaintainsoscillationin phase-shift

oscillator.

THE FEEDBACK CONTROLLOOP 187

o
~

'"o
~
Z
~
c
2

10
"
100
1000
LOGARITHM FREOUENCY-~AOIANS PER SECONO

10000

Fig. 7-10. Bode diagram shows gain of system (top) and phase change (bottom) .at
various frequencies. The gain (output signal/input signal) and phase (output/input) are
measured with the system in open-loop operation.

lag) is plotted linearly against the logarithm of frequency. Note that the
integral is phase lagging and the derivative is phase leading.
The proportional adjustment in the controller adjusts its overall
gain and hence the loop gain. The integral adjustment governs the
low-frequency response, and the derivative adjustment governs the
high-frequency gain of the control mechanism.
In addition to controlling gain, these adjustments also affect the
phase shifts. Thus, an improper setting of any one of the three control
modes can cause the control system to satisfy the two criteria for
oscillation and become unstable,
Stability in the Closed-Loop System
Under normal operating conditions, the control system should produce
stable operation. That is, the controller should retum the system to set
point, in the event o an upset, with mnimum overshoot and oscilla-

tion.
Too much overall gain(too narrow a proportionalband)can cause

188 PNEUMAT1CAND ELECTRON1CCONTROL SYSTEMS

the system to oscillate if the feedback reinforces the oscillation. Too

little gain can cause the system to deviate too far Cromthe desired set
point.
The damped oscillation shown in Figure 7-11 is characteristic ofthe
curve a closed-loop control system will produce when it is subjected to
a step change. If the phase and gain relationships are proper, energy is
Dot fed back to sustain the oscillation, and the cycles die out. The
pattem of decay generally can be controlled by the adjustments available in the control mechanism. Engineering judgment and skill are
required to select the recovery curve best suited to the process being

controlled.
A high-gain setting is desired because this gives the fastest and
most accurate control action. However, too much gain produces oscillation. The best compromise is to use enough gain to produce a damped
oscillation, as shown in the middle of Figure 7-11 (0.25 damping ratio).
Nonlinearities

Thus far it has been assumed that the capacitances and resistances
found in the process control loop bave a fixed value that does not
change with process conditions. This is not always true in practice.
Frequently, process conditions vary the value of the resistances and
capacitances involved and, as a result, the phase and gain relationships
are in constant transition Crom one value to another. At other times,
these values change, limiting or restricting the natural behavior of the
'-PERIOO~
GAI!

MEASURED DR
CDNTRDLLED
VARIABLE
(CLDSED LODP
SYSTEM)

GAIN',

TIMEFig. 7-11. Closed-loopresponseto an upsetdependson gainofloop.

THE FEEDBACK CONTROLLOOP 189

system. Such changes are often considered as a group and are called

nonlinearities.
A thorough understanding of the operation of the linear system is a
prerequisite to an understanding of the nonlinear type. It is common
practice to assume a system is linear for the purpose of basic control,
and then deal with the nonlinear characteristics as problems anse.

Controllers

and Control

Modes

A wide variety of responsecharacteristics involving gain and time

create the different modes of process control. The controller mode
selectedfor a givenprocesswill dependon:
Economics
Precisionof control required
Time responseof the process
Processgaincharacteristics
Safety
Now let us investigatethe most commoncontrol modes.
Two-Position Control
Within this classification there exist several specific methods-on/off,
differential gap, and time-cycle control.
On/off control is the most common. As soon as the measured variable differs from the desired control point, the final operator is driven to
one extreme or the other.
In the usual sequence (Figure 7-12), as soon as the measured variable exceeds the control point, the final operator is closed. It will
remain closed until the measured variable drops below the control
point, at which time the operator will open fully. The measured variable will oscillate about the control point with an amplitude and frequency that depend on the capacity and time response of the processo
As the process lag approaches zero, the curve will tend to become a
straight line. Then the frequency of the final operator open/close cycle
will become high. The response curve will remain constant (amplitude
and frequency) as long as the load on the system does Dot change.
On/off control is found in many household applications, such as the
refrigerator, heating system, and air-conditioning system. A simple

190 PNEUMATIC AND ELECTRONICCONTROLSYSTEMS

MEASURED
VARIA BLE
FINAL
OPERATOR
POSITION

riME'"
FINAL
OPERATOR
POSIT/ON

MEASURED

VARIABLE

TIME -+

Fig. 7-12. Two-positioncontrolacts on operator.

thermostaticallycontrolledelectric heateris anotherfamiliar example.

The room in which the heateris placeddeterminesthe capacityof the
process,and, hence,the responsecurve.
The following requirementsare necessaryfor on/off control to produce satisfactoryresults:
1. Precisecontrol must Dotbe needed.
2. Processmustbave sufficientcapacityto allow the final operatorto
keep up with the measurementcycle.
3. Energy inflow is smallrelative to the energyalreadyexistingin the
processo

Differential-gapcontrol is similar to on/off except that a band, or

gap, is createdaroundthe controlpoint. In Figure7-12,notethat, when

THE FEEDBACK C~TROL LOOP 191

the measured variable exceeds the upper boundary of the gap, the final
operator is closed and remains closed until the measured variable drops
below the lower boundary. It now opens and remains open until the
measured variable again exceeds the upper boundary. In a process
plant, differential-gap control might be used for controlling noncritical
levels or temperatures.
A time-cycle controller is normally set up so thai when the measured variable equals the desired control point, the final operator will
be open for hal the time cycle, and closed the other hal. As the
measured variable drops below the control point, the final operator will
remain open longer than it is closed.
Two-position control is nearly always the simplest and least expensive form of automatic control. Any of the forms discussed can be
implemented with commercially available mechanical, pneumatic, or
electronic instrumentation. On the other hand, two-position control
mar Dot meet the requirements often demanded by today's sophisticated processes. Now let us investigate a typical on/otI controlloop.
Assume we bave a liquid-Ievel process as shown in Figure 7-13.
Throttling

Control

Proportional or throttling control was developed to meet the demand

for a more precise regulation ofthe controlled variable. Throttling control indicates any type of control system in which the final operator is
purposely positioned to achieve a balance between supply to and demand from the processo
The basic type of throttling control is proportional-only control.
This term applied to a control action wherein the position of the final
operator is determined by the relationship between the measured variable and a reference or set point. The basic equation for a proportional
only controller is:

m = Gain (Error) + Bias

(7-1)

where:
m = controller output or valve position
Error = difference between set (r) and measurement (c)

or

e=r-c

(7-2)

v=

Fig. 7-13. PROBLEM: DIFFERENTIAL

GAP CONTROL, CALCULATE PERIOD.

Question I: What is the period of oscillation ofthe controlloop?

Given:
Controller: on/otI with 7 percent differentia! gap.
lnstrumentation response: assume instantaneous.
Load = Q = 60 gpm.
Manipulated Variable = M = O gpm or 90 gpm.
Tank: 6 feet diameter, 12 feet high.
Level transmitter: Oto 8 feet.
Set Point = 50 percent
Question 2: What is the period of oscillation ifthe load is 50 gpm?
Question 3: At which load will the period be shortest?
So/ution:
1. Volume oftransmitted signa! = area x height
1TD2

xH=

2. Volume within 7 percent differential

(0.07)(1692) = 118.4gal.
3. Level cycle:
3a. Rate of rise = (90 -60)

TlIne
" to nse
.118.4gal.
= 30

gpm = 30 gpm

gpm

3b. Rate offall = (O-60)

=.

395

mm.

gpm = -60 gpm

Tirne to fall =

-118.4 gal.
= 1.973min.
-60 gpm
3c. Total tirne = (3.95 + 1.973)min. = 5.92 min.
Solution to no. 2:

3.lD1e
T ' to nse
, = (90118,4
,
-40) gal,
gpm = 2.96 mm.
TlD1e
'
-118.4gal.
'
to "la11 = (O
-50) gpm = 2. 37 mm.
Period oCoscillation = 5.33 min,

Solutionto no. 3:
When Q = Mon + Moll = 90 gpm + O gpm = 45 gpm
2
2

m=
Application

THE FEEDBACK CONTROLLOOP 193

Bias = usually adjusted to p-lacethe valve in its 50 percent open position with zero error.
The term proportional band is simply another way o expressing

gain.
% ProportionalBand =

100

Controller Gain

or
Controller Gain =

100

% Proportional Band

Substituting the error formula (e = r -c),

proportionalband formula ( % PB = ~),

Equation 7-2, and the

Equation7-3, into Equa-

tion 7-1, it mar be expressedas:

100

%PB (r -c)

+ 50%

Another approach to visualizing the effect of varying the proportional band is shown in Figure 7-14. Each position in the proportional
band dictates a controller output. The wider the band, the greater the
input signal (set point -measurement) musi change in order to cause the
output to swing from O to 100percent. Manual adjustment of the bias
shifts the proportional band so thai a given input signat will cause a
different output level.
ot Proportional

Control

Proportional control attempts to retum a measurement to the set point

arter a load upset has occurred. However, it is impossible for a proportional controller to retum the measurement exactly to the set point,
since, by definition (Equation 7-4), the output must equal the bias setting (normally, 50 percent) when measurement equals the set point. If
the loading conditions require a difIerent output, a difIerence between
measurement and set point must exist for this output level. Proportional control mar reduce the efIect of a load change, but it cannot
eliminate it.
The resulting difIerence between measurement and set point, arter
a new equilibrium level has be en reached, is called ofIset. Equation 7-5

194 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

SET POINT
MINUS
MEASUREMENT

GAIN. 1

OUTPUT

TIME-..
OPEN-LOOP

GAIN RESPONSES

0%

1001

CONTROLLER

OUTPUT

MEA5UREMENT,I

"I.

PROPORTIONAL
BANO

OUTPUT
50%

50

100%

0% CONTROLLER

PROPORTIONAL

BANO

-SET

POINT

OUTPUT

.!~I:,_C2~~R2!;~E~
OUTPUT
BIAS (MANUALLY
EFFECT
-

OF

CHANGES

IN

PROPORTIONAL

SET)

= 50%
BAND

Fig. 7-14. Outputresponsereactsto changein gain.

indicatesproportionalresponseto a load upset,and the resultingoffset.

The amount of offset may be calculatedfrom:
Ae =% Proportional Band AM
100

(7-5)

where ~e is changein offset, and ~M is changein measurementrequired by load upset.

From Equation 7-5, it is obvious that, as the proportional band
approacheszero (gainapproachesinfinity), offset will approachzero.
This seemslogical, becausea controller with infinite gain is by definition an on-otI controller that cannot permit a sustainedoffset. Conversely, as the proportional band increases(gain decreases),propor-

___!_'22-:_C2~~~~~!~

THE FEEDBACK CONTROLLOOP 195

tionately more and more offset will exist. Offset mar be eliminated by
manually adjusting the bias until the measured variable equals the set
point.
Narrow-band proportional-only controllers are often used in noncritical, simple temperature loops, such as in maintaining a temperature
in a tank to prevent boiling or freezing. A low, dynamic gain allows a
narrow band to be used. Controllers of ibis nature are typically fieldmounted and pneumatically operated. Many noncritical, level-control
applications having long time constants also use proportional-only control. Now let us apply proportional-only control to the level problem
shown in Figure 7-15.
Proportional-Plus-lntegral

Control

If offset cannot be tolerated, another control mode must be added.

Integral action will integrate any difference between measurement and
set point, and cause the controller's output to change until the difference between measurement and set point is zero.
The response of a pure integral controller to a step change in either
the measurement or the set point is shown in Figure 7-16. The controller output changes until it reaches O or 100 percent of scale, or the
measurement is retumed to the set point. Figure 7-16 assumesan openloop condition where the controller's output is dead-ended in a measuring device, and is Dot connected to the processo
Figure 7-16 also shows an open-loop proportional-plus-integral response to a step change. Integral time is the amount oftime required to
repeat the amount of change caused by the error or proportional action.
In Figure 7-16, integral time equals t r, or the amount of time required to
repeat the amount of output change (xJ. (Some instrument manufacturers define integral as the inverse of the above, or the number of
times per minute the amount of change caused by proportional action is
repeated.)
Another way of visualizing integral action is shown in Figure 7-17.
In this example, a 50 percent proportional band is centered about the
set point. If all elements in the controlloop bave been properly chosen,
the controller's output should be approximately 50 percent. If a load
upset is introduced into the system, the measurement will deviate from
the set point. Proportional response will be immediately seen in the
output, followed by integral action.
Integral action mar be thought of as forcing the proportional band
to shift, and, hence, causing a new controller output for a given rela-

196 PNEUMATIC AND ELECTRON1CCONTROL SYSTEMS

Fig. 7-15. Sample problem: PROPORTIONAL CONTROL, CALCULATE

MEASUREMENT.
Question: What is the level under steady-state control?
Given:
Range of level transmitter: Oto 70 inches
Level controller: Proportional only, PB = 75 percent, bias = 50 percent, Set
Point = 40 inches
Load, q: Fixed at 3.5 gpm
Valve: Pressure drop, ~P, is constant, Linear characteristic, delivers 6 gpm at 100
percent stroke.
Solution:
I. Under steady state, inftow must equal outftow.
6
- 0 gpm = 58.3%
2. Inftow = 3.5
.gpm

3. Inftow % = Controller output

m = 58.3%
.40
4. Set pomt = r= 70 inches
. h = 57.1%

mc es

5. Equation (7-4) for proportional control:

100
m = -(r

%PB

-c)

+ 50%

100(57.1 -c) + 50
58.3 = 75
Solving for c: c = 50.9%
Converting to inches: 0.509 x 70 = 35.6 inches

tionship between rneasurernentand set point. Integral action will continue to shift the proportional band as long as a difference exists between rneasurernentand set point. Integral action has the sarne function
as an operator adjusting the bias in the proportional-only controller.
The width of the proportional band rernains constant, and is shifted in a

THE FEEDBACK CONTROLLOOP 197

MEASUREMENT

CONTROLLER
OUTPUT

TIME-'

Fig. 7-16. Outputrespondsto stepchangein input.

direction opposite to the measurement change. Thus, an increasing

measurement signal results in a decreasing output, and vice versa.
If the controller is lloable to retum the measurement to the set
point, integral action will drive the proportional band until its lower
edge coincides with the set point (see Figure 7-17); hence, a 100 percent
output. Ifthe measurement should then start to corne back within control range, it must cross the set point to enter the proportioning range,
and cause the output to begin throttling. If the system has any substantial capacity, the measurement will overshoot the set point, as shown
on the right gide of Figure 7-17.
This basic limitation is called integral windup. It must be seriously
considered on discontinuous or batch processes where it is common for
the controller output to become saturated and overshoot. A pneumatic
mechanism called a batch switch designed to prevent this overshoot is
described in Chapter 8.

~
.=

LOAD

198 PNEUMATIC AND ELECTRONICCONTROLSYSTEMS

MEASUREMENT
% OF SPAN

50

2~

-~"--""

~;;;;",

'

','~
,

SET paiNT

'~

CHANGE

Fig. 7.17. Integral actionshiftsproportionalband in referenceto differencebetween

input signa!and setpoint.

Integral setting is a function of the dead time associated with the

process and other elements in the controlloop. Integral time should Dot
be set faster than the process dead time. If it is set too fast, the controller's output will be capable of changing faster than the rate at which the
process can respond. Overshoot and cycling will resuIt. An altemate
definition of integral action is the fastest rate at which the process can
respond in astable manner.
Proportional-plus-integral controllers are by far the most common
type used in industrial process control. Pneumatic and electronic
analog, and digital hardware are available with proportional-plusintegral action.
The classical equation or expression for a proportional-plusintegral controller is:

(7-6)

%PB

= gam

e = error (deviation)R
= integraltime
If rou compare this equation with Equation 7-4, the proportionalonly expression, rou will find the bias term replaced by the integral

term.
Fig.

rHE FEEDBACK CONTROLLOOP 199

In operation, then, the output of this controller will continuously

change in the presence of an error signat. Once the error has been
reduced to zero, the output will stop changing and star fixed until an
error redevelops. At this time, the output again will change in such
direction as to eliminate the error once again.
Adding

Derivative

ln controlling

(Rate)

multiple-capacity

processes, a third mode of controls is

often desirable. .
By definition, derivative is the time interval by which derivative
action will advance the effect of proportional action on the final
operator. lt is the time difference required to get to a specified level of
output with proportional-only
action as compared to proportionalplus-derivative action. Figure 7-18 illustrates derivative response to a
camp change; where td equals the controller's derivative time.
Derivative
action occurs whenever the measurement si~nal
changes. On a measurement change, derivative action differentiates the
change and maintains a level as long as the measurement continues to
change at the given rate. Under steady-state conditions, the derivative
acts as a 1 to 1 repeater. lt has no inftuence on a controller's output..By
reacting to a rate of input change, derivative action allows the controller to inject more corrective action than is initially necessary in order to
overcome system inertia. Temperature presents the most common application for derivative. Derivative action should Dot be used on processes that are characterized by predominant dead times, or processes
that bave a high noise content, that is, high-frequenY extraneous signaIs such as are present in the typical ftow application.

MEASUREMENT

,
PROPORTIONAL

-ONLY

ACTION

PROPORTIONAL PLUS
DERIVATIVE ACTION

TIME-

7.18. -Derivative

action increases rate oCcorrection.

DERIVATIVE
TIME, td

Selecting

200 PNEUMATIC AND ELECTRONICCONTROLSYSTEMS

The equationor expressionfor the proportional-plus-integral-plus

derivative controller is:
(7-7)
where:
e = error
R = integral time
D = derivative time
Question: How long will it take for the output to change 5 percent if
the measurement remains constant?

Given:
Proportionalband = 50 percent
Integral time = 3 minutes
Derivative time = 3 minutes
Set point = 45 percent
Measurement= 35 percent
Control action: increase-decrease
So/ution:
1. Error staysconstant,thereforeno output changedue to derivative.
2. No output changedueto proportionalaction. However, the proportional action affectsthe integralresponse.
3. Error = r -c = 45% -35% = 10%
4. R = 3 minutes = time to changean amountequalto the error
5. Becausethe proportionalband = 50%,the output in Step4 is multiplied by

= 2.

Therefore 2(10%)= 20%changein 3 minutes

or

5% changein 3/4minute.
the Controller

Now that the variety of available control modes has been described,the
next logical question is: which should be selected to control a
particular process? Table 7-1 relates process characteristics to the
common control modes. Let us apply the chart to the beat exchanger
processo
The beat exchanger acts as a small-capacity process; that is, a
small change in steam can cause a large change in temperature. Accu-

THE FEEDBACK CONTROLLOOP 201

Table 7-1. Process Characteristics

Control
Mode
On/OlI
Floating
Prop.

Transfer
Lag
Min

Small
Small

Dead
Time

Capacitance

Reaction Load
Rate
Changes

SelfRegulation

Min
Min

High
Low
Moderate

Slow
High
Slow

Any

Slow
Slow

Must bave

SmaIl

Sma11

Prop. +
Integral

Moderate

Moderate

Moderate

Any

Any

Any

Any

Any

Any

Any

Prop. +
Integral +

Deriv.

rate regulation of processes such as this calls for proportional rather

than onloff control.
Variations in water rate cause load changes that produce offset, as
described previously. Thus, the integral mode should also be used.
Whether to include the derivative mode requires additional investigation ofthe process characteristic. Referring to the reaction curve (Figure 7-19), notice that the straight line tangent to the curve at the point
of inftection is continued back to the 150F (starting) level. The time
interval between the start of the upset and the intersection of the tangentialline is marked TA(in Figure 7-12, this was dead time); the time
interval Crom this point to the point of inftection is TB' If time T B
exceeds time TA,some derivative action will prove advantageous. 1fT B
is less than TA, derivative action mar lead to instability because of the
lags involved.

Fig. 7-19. The process reaction curve is obtained by imposing a step change at input.

7-1.
7.3.
7-4.

202

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

Table 7-2.

OFFSET
CA. BEy

~'ES

USE

TO'ERATEO?

P-O."

.s

-'+1

USE

The reaction curve of Figure 7-19 clearly indicates that some derivative action will improve control action. Thus a three-mode controller with proportional, integral, and derivative modes satisfies the needs
of the beat exchanger processoTable 7-2, as applied to this problem,
will yield the same resulto
Ouestions
With adequate gain and inphase feedback. any system will:
8. Drift
c. Increase amplitude
b. Oscillate
d. Degenerate
7-2. The natural frequency at which a closed-loop system will cycle depends

upon:
3. The amplifier gain
b. The attenuation provided by the process
c. The phase shift provided by the resistance-capacitanceand dead time
networks that exist in the system

d. Resonance
8.
b.
c.
d.

The Bode diagram describes:

Gain and phase shift through the usable frequency range
The system's linearity
The reaction to a step change
The recovery curve that will result from a load change

All systems may be assumed to be:

a. Linear
b. Nonlinear

THE FEEDBACK CONTROLLOOP 203

c. Linear for the purpose ofinitial consideration but with full knowledge
that this may not be the case
d. Nonlinear for purposes of analysis with rhe exception that the system
may prove to be linear
7-5. A closed-loop control system that employs a three-mode controller:
a. Can oscillate or cycle at only one frequency
b. Can oscillate or cycle at several frequencies. depending on controller
adjustment
c. Will not oscillate because of the stability provided by derivative
d. Will produce only damped oscillations with a 0.25 damping ratio
7-6. We have a closed-loop control system which is cycling. We should:
a. Increase the proportional band
c. Check and adjust both
b. Increase integral action
d. Immediately shut it down
7-7. A proportional controller is being used to control a process. and the
offset between set point and control point must be held to a minimum. This
would dictate that rhe proportional band:
3. Be as narrow as possible
b. Be as wide as possible
c. Be of moderate value
d. Does not relate to the problem
7-H. The system gain ofthe closed-loop control system:
3. Refers to the process gain
b. Refers to the gain of the measurement and control devices
c. Refers to total gain of all components. including measurement, controller, val ve operator, valve, and process
d. Relates only to the gain ofthe controller
7-9. If a closed-loop control system employs a straight proportional
controller and is under good control. offset:
3. Will vary in magnitude
b. Will not exceed one-half of the proportional band width
c. Will exceed the deviation
d. Is repeated with each reset7-10.
Any closed-loop system with inphase feedback and a gain of one or
more will:
a. Degenerate
c. Exhibit a 0.25 damping ratio
b. Cycle or oscillate
d. Produce square waves7-11.
How long will it take for the Otltput to change 5 percent if the
measurement remains constant?
Gil 'l'li :

Proportional band '" 50 percent

Integral time '" 3 minutes
Derivative time '" 3 minutes

204 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Set paint = 45 percent

Measurement = 35 percent
Control action = increase/decrease7.12.
If a closed-loop control system is adjustd to produce aO.25 damping
ratio when subjected to a step change, the system gain is:
3. 0.1
c. 0.5
b. 0.25
d. 1.07-13.
A straight proportional controller is employed to control a processo
Narrowing the proportional band will:
3. Not change the offset
c. Always cause cycling
b. Decrease the offset
d. Never cause cycling7-14.
The type of process that most often can benefit from derivative is:
3. Flow
c. Temperature
b. Level
d. Pressure
7-15. Pure dead time ina process contributes a gain or:
3. Zero
c. Depends upon dynamics
d. Infinite
d. One
7-16. RefeITing to Figure 7-5, the transmitter span is 200F; the controller
proportional band is adjusted to 150 percent; the equal percentage valve
delivers saturated steam containing 40,000 Btu per minute at full open; and
water enters the exchanger at 50F and is heated to 2000Fat a 40 gpm
maxirnum flow rate. The static gain of this control loop is approximately:
3. 0.5
c. 1.5
b. 1.0
d. 2.0
7-17. Ifthe flow rate ofheated water in Problem 7-16 is reduced to 5 gpm,
you would expect the gain to:
3. Increase
c. Remain the same
b. Decrease
7-18. The integral control mode is:
3. Phase-Ieading
b. Phase-Iagging

c. Inphase
d. Phase-reversing

7-19. The derivative control mode is:

8. Phase-Ieading
b. Phase-Iagging

c. Inphase
d. Phase-reversing

7-20. The most common combination of control modes found in the typical
process plant is:
8. Proportional-only
b. Proportional, integral, and derivative
c. Proportional-plus-integral
d. On/off

Pneumatic Control Mechanisms

The basic pneumatic control mechanism is the ftapper-nozzle unit. This

unit, with amplifying relay and feedback bellows, is a simple, rapidacting control mechanism.
The basic pneumatic mechanism converts a small motion (position)
or force into an equivalent (proportional) pneumatic signal. Since
pneumatic systems mar use a signal of 3 to 15 psi (20 to 100kPa) (3 psi
or 20 kPa at O percent and 15 psi or 100kPa at 100 percent scale), the
instrument must bave the ability to convert a position or small force
into a proportional pneumatic span of 12 psi (3 to 15 psi or 20 to 100kPa).

The Flapper-Nozzle

Unit

Figure 8-1 shows the principIe of the ftapper-nozzIe device. Input air
(reguIated at 20 psi or 138 kPa) is fed to the nozzIe through a reducing
tube. The opening of the nozzIe is Iarger than the tube constriction.
Hence, when the ftapper is moved away from the nozzIe, the pressure
at the nozzIe falls to a Iow vaIue (typicalIy 2 or 3 psi, or 10 or 20 kPa);
when the ftapper is moved cIose to the nozzle, the pressure at the
205

206 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Fig. 8-1. Flapper-nozzleis the basicpneumaticcontrolelement.Flappercan be

positionedby temperature(as shown),pressureelement,or any sensor.

nozzle rises to the supply pressure (20 psi or 138 kPa). Flapper movement of only a few thousandths of an inch (Figure 8-2) produces a
proportional pneumatic signal that mar vary from near zero to the
supply pressure. Some pneumatic control mechanisms use the air at the
nozzle (nozzle pressure) to operate a control valve.
The simple ftapper-nozzle unit shown in Figure 8-1 has several
basic limitations. The output air must all corne through the input constriction if it is used directly to operate a control valve. Hence, the
output pressure can change only slowly, causing sluggish action. Just
as the rate of increase in pressure is limited by the input constriction,
the rate of decrease in pressure is similarly limited by the slow rate of
air passage through the nozzle to the atmosphere.

Fig. 8-2. Flapperneedmove only a fewthousandthsof an inch for full rangeof output
(nozzle)pressure.

The
Fig.

PNEUMATICCONTROLMECHANISMS 207

AIso, the measuring element that positions the ftapper must be

strong enough to overcome the blast of air leaving the nozzle. Since
many measuring sensors are relatively low-force elements, they could
Dot be used to position the ftapper accurately and positively if the
nozzle were made large enough to pass sufficient air to overcome this
limitation.
A second limitation is that the ftapper moves only a few thousandths of an inch for complete action from minimum to maximum nozzle pressure. This small trave I for a change in output from zero to full
value makes the entire element susceptible to vibration and instability.
All these limitations are overcome by employing (1) a relay, or
other amplifier, which amplifies the nozzle pressure; and (2) a feedback
system which repositions the ftapper or nozzle.
Valve Relay or Pneumatic

Amplifier

Figure 8-3 shows a celar connected to a nozzle. The nozzle pressure is

applied to the pneumatic relay, which contains a diaphragm operating a

small ball valve. Because the diaphragm has a large area, small
pressure changes on its surface res uit in a significant force to move the
ball valve. The ball valve, when open, permits the full air supply to
reach the output; when closed, it permits the nozzle pressure to bleed
to atmosphere.
If the ftapper position is changed with respect to the nozzle, the air
pressure acting on the celar diaphragm changes and either opens or
closes the celar ball valve, thus either increasing or decreasingthe ftow
of supply air, which now can ftow directly from the supply to the
output, overcoming the first deficiency (slow action) of the ftapper noz-

zle.
relay is often called a pneumatic amplifier becausea small

8-3. Relay is an arnplifier. That is, small change in nozzle pressure (the input to the
relay) causes a large change in output pressure (to the control valve).

The

208 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

change in nozzle pressure causes a large change in output pressure and

flow. The pneumatic relay shown in Figure 8-3 (used in the Foxboro
Type 12A temperature transmitter and the Model 130 controller)
amplifies the nozzle pressure by a factor of 16; that is, a change in
flapper position of 0.0006 inch, which produces a change in nozzle
pressure of3/4psi, and results in a change arrelar output of 12 psi (from
3 to 15 psi).
The relay shown in Figure 8-3 increases output pressure as nozzle

pressureincreases.Relays can also be constructedto decreasethe

output pressure as nozzle pressure increases.
Linear Aspirating

Relay or Pneumatic

Amplifier

Flapper-nozzle detectors employed in the set-point transmitter and derivative sections of the Model 130 controller use a different type of
pneumatic amplifier. If the flow or volume output requirements are
small, an aspirating relay mar be employed. The pneumatic amplifier,
shown in Figure 8-4, makes use of the Venturi tube principIe and resembles a small Venturi tube. With a 20 psi supply, the throat pressure
of the Venturi can vary 3 to 15 psi or 20 to 100 kPa for a change in
flapper position with respect to the nozzle of less than 0.001 inch. The
aspirating relay accomplishes ibis with almost perfect linearity and
does Dot require parts thai are subject to wear. This recent develop-

.,.SUPP"
,

TO

FLAPPERtNO'

Fig. 8-4. Aspirating cone, showingVenturitube principIe.

Fig.

PNEUMAT1CCONTROLMECHANISMS 209

ment offers advantages over conventional pneumatic amplifiers, provided that a high volume output is Dot required.
Proportional Action
If a controller had only the units described thus"far, ~apper-nozzle and
relay, it would only bave on/otI action. On/otI control is satisfactory for
many applications, such as large-capacity processes. However, if
on/otI action does Dot meet the control requirements, as in low-capacity
systems, the ftapper-nozzle unit can easily be converted into astable,
wide-band proportional device by a feedback positioning system that
repositions the moving ftapper.
An example of a proportional action mechanism is the Foxboro
12A pneumatic temperature transmitter (Figure 8-5). This device,
called a "force-balance pneumatic transmitter," develops a 3 to 15 psi
(20 to 100kPa) output signal proportional to the measured temperature.
Thus, it is functionally a proportional controller.

8-5. Pneumatic temperature transmitter. with tubing and sensor.

210

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

Figure 8-6 shows .the principIe of the Model 12A transmitter. The
forces created by the bellows are automatically balanced as follows.
When the temperature sensor is subjected to an increase in temperatufe:
1. The increased temperature expands the gas contained in the sensor
and increases the force exerted by its bellows. This increases the
moment of force, tending to rotate the force bar clockwise.
2. The flapper now approaches the nozzle and the nozzle pressure
increases. This pressure is applied to the pneumatic relay (Figure
8-3), increasing its output and the pressure applied to the feedback
bellows, thus increasing the counterclockwise moment of force
sufficiently to restore the force bar to equilibrium.
The force bar (flapper) is now repositioned slightly closer (less than
0.001 inch) to the nozzle, and the output pressure has reached a new
levellinearly related to the measured temperature.
In the actual unit, two additional forces act on the force bar. One is
thai applied by the zero elevation spring. Adjustment of tbis spring
determines the constant force it adds to thai supplied by the feedback
bellows. This allows a given span of temperature measurement to
be raised or lowered. The other force is applied by the ambienttemperature- and barometric-pressure-compensating bellows. This
force compensates for ambient temperature or atmospheric pressure
changes acting on the gas-filled thermal system, thus minimizing errors
caJlsedby these changes.

Fig. 8-6. PrincipIeoCpneurnatictemperaturetransmitter.Forcescreatedin bellowsare

balanced.

PNEUMAnc CONTROLMECHANISMS 211

Fig. 8.7. Generalpurposecontroller.

The Model 12A temperature transmitter produces an output proportional to measurement and is a proportional control mechanism.
However, a useful general purpose controller should incorporate additional modes, and also bave the ability to adjust the parameters of the
mechanism along with other convenient features. For these reasons the
unit described, although a proportional controller, is used primarily as a
transmitter to send an input signal to a generalpurpose controller (such
as the 130 Series), as shown in Figure 8-7.

Control

Mechanism

Requirements

A control mechanismshouldbe ableto controlthe processandeasethe

job of the operator. To achievethesegoals, it must meetthe following
requirements:
1. The required control modesmustbe availableand easilyadjustable
throughthe required range.

212

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

2. The controller shouldhave an easilyadjustedset point.

3. The mechanismshould clearly display what is going on for the
benefitof the operator.
4. It shouldbe convenientfor the operatorto switch over to manual
control either for the purpose of operatingthe processmanually
during startupor in otherunusualsituations,or to performmaintellance or adjustmentson the automaticcontrol unit.
5. After manualoperation,it shouldbe a simplematterto switch back
to automatic control smoothly without shockingor bumping the
processo

6. If for any reasonthe unit requires maintenanceor adjustment,it

shouldbe simpleto remove it for this purposewithout interfering
with the processo
7. In manysituations,it mar alsobe convenientand practicalto bave
a continuousrecord of the controllervariableand an alarm system
to signalthe operatorwhensomepredeterminedlmit is exceeded.
Figure 8-8showsthe completeModel 130controller in block diagramformo
The total control mechanismconsistsof the following functionalparts:
1. A set-point unit that produces a 3 to 15 psi or 20 to 100kPa
pneumaticsignalto be fed into the controllermechanism.This unit
is generallylocated within the controller, but it mar be remote.
2. Thederivative unit that acts directly on the measurementsignal.
3. The automatic controller itself-headquarters for the control action.
4. The manualcontrol unit, whichperformstwo functions-switching
Cromautomatic to manual and Crommanual to automatic, and
manualadjustmentof the control valve (regulatedvariable).
5. The automatic balancingunit, consistingof a simple floating controller through which the operator can switch Cromautomaticto
manualand back without any balancingadjustmentsand ret without causinga bump or suddenvalve change.
6. A measurementindicator that displays with pointersand scalethe
measurementbeingfed into the controller, and the set point.
7. An output indicator-a display of controller output that mar be
interpreted in terms of valve position.
The Automatic

Controller

Figure 8-9 develops the mechanism of a proportional controller in

schematic formo This controller, like the Model 12A shown in Figure
8-7, has measurement and feedback bellows and a flapper-nozzle relay

~
~
unit.
Fig.

PNEUMATICCONTROLMECHANISMS 213

Fig. 8-8. Block diagramoCFoxboropneumaticConsotrol130Mcontroller.

The basic added features are a set bellows, which permits adjustment of the set point; and an integral (R) bellows.
ln the actual Model 130 controller (Figure 8-10), a ftoating disk actsas
the ftapper of the ftapper-nozzle system. The resultant moments offorce
due to the four bellows determine the position of the disk with

0f?5!)@
OUTPUT

b=ooa

obOR
b
B
P

b.40

OUTPUT

: o.
o bM
P
b.lo
4

OUTPUT

8.9. Schematicdevelopmentofproportionalcontroller.

PROPORTIONAL
~
=1

214 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

('=I--~ --.

REDUCING
TUBE-

~
~

BAND
DIAL
REVERSING
SWITCH

~.-

ADJUSTING ARM AND

FULCRUM BAR

NU~~LI:

SET (B) OR MEASUREMENT

(A) BELLOWS
--

FORCE BALANCING
FLOATING DISC

-INTEGRAL
BELLDWS
PROPORTIONING BELLOWS --'I

-MEASUREMENT
(B) OR
SET (A) BELLOWS
CONTROLLER ACTION
REVERSING SWITCH

/1

\:

r=

=!
,
r~1
'=--,~_f=

INTEGRAL
RESTRICTOR

Fig. 8-10. Control unit with floating disk flapper nozzle.

respect to the nozzle. Therefore, the relay output pressure varies with
changes in pressure in any ofthe bellows. The mechanism is aligned to
produce a midrange output of9 psi (60 kPa) when the error signal is O.
This is called bias.
The width of the proportional band is adjusted by the position of
the proportional-band-adjusting lever. (The term gain is also used in
place of proportional band. Gain is the reciprocal of proportional
band.) This lever positions the fulcrum abolli which the moments of
force created by the pressure in the four bellows act. These moments of
force tilt the floating disk upward or downward, thereby causing a
change in nozzle pressure. The reguli is an increased or decreased
output pressure Crom the controller. This output pressure, which is
proportional to changes in measurement or set pressure, acts to reposition the control valve, thus influencing the process and bringing abolli a
change in the measurement bellows to establish a new equilibrium in

the system.

PNEUMATICCONTROLMECHANISMS 215

Derivative and Integral

The general purpose controUer needs derivative and integral action, as
described in Chapter 1. These are added to the pneumatic controUer.
The derivative function is added to the measurement signal before it
reaches the controUer (block 4, Figure 8-8). This eliminates derivative
response to set-point change and practicaUy eliminates any interaction
between derivative and the other control modes.
During steady-state conditions, the unit acts as al: 1 repeater, but
upon a change in the measurement the unit adds a derivative inftuence
to the measurement change.
In the derivative unit (Figure 8-11), the force moment (beUows area
times distance from fulcrum) ofbeUows A is 16 times that ofbeUows B,
and the force moment of beUows B plus beUows C equals that of bellows A. As the measurement signal increases, the immediate change in
feedback pressure in B is 16 times the change in pressure in beUows A.
Simultaneously, air starts to ftow through the restrictor to beUows C,
graduaUy reducing the pressure needed in beUows B to restore equilibrium. Thus, the output ofthe derivative unit, which is the signal to the
automatic control unit measurement beUows, reftects the change in

measurementplus a derivative responseadded to that change. The

graph at the bottom of Figure 8-11 shows the signal that the measure-

TIME

Fig. 8-11. Derivative unit, schematicdiagramand responsecurve.

216

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

ment bellows in the automatic control unit receives from the derivative
unit at the measurement change shown.
The integral function takes place within the automatic control unit
itself. The integral bellows opposes the feedback bellows; thus, if the
feedback bellows introduces negative feedback, the force created by
the feedback bellows will act like positive feedback; that is, it tends to
move the flapper in the same direction as error. This would lead to
instability if the integral restriction were Dot used. anly very slow
signals (minutes in duration) can affect the integral bellows because of
the very small integral restriction. The integral action occurs only arter
the proportional and rate actions h~ve affected the processoIf an error
remains (such as offset due to load change), the integral action takes
place slowly. Because integral action will continue in the presence of a
continuous error, such as we mar find in a batch process, it can cause
the output to go to an extreme. This is called integral windup.
Integral time is adjusted by setting an adjustable restrictor, or needle valve. The dial on this pneumatic val ve is calibrated in minutes of
integral time.
Figure 8-12 illustrates the Foxboro Model 130controller. The control rnodes-proportional, integral, and derivative-are in the controller shown. The signal level with respect to time mar be changed by
appropriate adjustments of proportional, integral, and derivative
modes. The proportional band adjustment is calibrated from 5 to 500
percent, while the integral and derivative dials are calibrated from 0.01
to 50 minutes. These adjustments are calibrated in "normal" times.
The effective times are somewhat different.
In any controller, some interaction between control actions mar
exist because change in one action (proportional, integral, or derivative) can create change in the others. This unavoidable interaction,
even if small, should be kept in mind when adjusting or tuning a control
ler.
Manual Control

Unit

Manual operation is achieved by using the manual control unit shown

schematically in Figure 8-13 and depicted in Figure 8-14. With the
transfer switch in the manual position, the thumbwheel is engaged, and
by its action the flapper-nozzle relationship governs the pneumatic output. The operation of the manual control unit is similar to that of any
pneumatic transmitter in that it employs a feedback bellows. Full-scale
manipulation of the thumbwheel corresponds to a 3 to 15 psi (20 to 100
kPa) pneumatic output.

PNEUMATICCONTROLMECHANISMS 217

Fig. 8-12. Pneumatic Consotrol130M controller. (Top) right sicle cover removed and
manual controls. (Battam) the leCtsicle cover is removed to show pneumatic printed
circuit board.

Transfer

Automatic to Manual
With the transfer switch in the automatic position, the thumbwheel in
the manual control unit is disengaged and the automatic controller's
pneumatic output is fed into the manual control unit bellows. The manual control unit remains balanced at this output. At the time of transfer
to manual, the manual unitinstantfy starts transmitting this same output. Thus, the response record of the transfer from automatic to manual (Figure 8-15) is smooth and bumpless.
Manual to Automatic
The transfer Crom manual to automatic requires the use of an
automatic-balancing unit, which consists of a singie-pivoted diaphragm
with four air pressure compartments. This is shown in schematic Corm

AIR
SWITCH
SUPPLY SIGNAL

OUTPUT

Fig. 8-13. Schematicof manualcontrolunit.

Fig. 8-14. Manual controlunit;

TRANSFER

TIME-Fig. 8-15. Responserecord oftransfer Cromautomaticto manual.

218

PNEUMATICCONTROLMECHANISMS 219

in Figure 8-16. At the left, the unit is in the automatic position; at the
right, it is in manual. It will act as a simple proportional controller with
a fixed proportional band of approximately 30 percent. The basic probIem in transferring from manual to automatic lies in the fact that the
output of the automatic controller must equal that of the manual unit at
the moment of transfer, and then a change at a predetermined integral
rate is necessaryto bring the measurement to the set paint. The output
ofthe manual station is the control unit' s output. Full air supply is sent
to the three pneumatic switches, closing two and opening one, as
shown in Figure 8-16.
The proportional bellows of the automatic control unit is disconnected from the output of the controller. The integral restrictor is
bypassed, and the integral bellows is disconnected from the proportional bellows. The input signals to the automatic balancing unit shown
represent the automatic control unit proportional bellows pressure in
bellows D, and the manual control output in bellows A. Bellows B is
the balancing pressure and bellows C is the output to the integral bel-

lows.
If either the measurement or the set paint to the automatic control
unit changes, the pressure in the proportional bellows must also
change, because it is operating as a proportional-only control unit.
When the change in pressure in the proportional bellows is sensed by
the balancing unit, the unit' s output will change the pressure in the
controller's integral bellows. This, in turo, will cause the proportional
bellows pressure to change in the opposite direction until it once again
equals the output of the manual control relay, or until the supply
pressure limits are reached.
Any diiference between the set and measurement bellows is thus
balanced by the difference between the integral and proportional bel-

lows.
If the output of the manual control unit changes, a similar action
occurs, forcing the proportional bellows pressure to equal the manual
control unit output.
When transferring from manual to automatic (Figure 8-17), the
output will remain at the level determined by the operator when the
controller is in manual. If the measurement input is equal to the set
paint, the output remains constant until corrective action is required. If
the measurement does Dot equal the set paint at the moment of transfer ,
the output will camp from the level of manual operation to the level
necessary to make the measurement equal to set paint a function of the
controller's integral rate.

220

PNEUMATIC ANDELECTRONIC CONTROLSYSTEMS

Fig. 8-16. Schematic of auto-manual mechanism with transfer switch in automatic

position (above) and manual position (opposite page).

Set Po nt

The set-point knob is attachedto a pneumatictransmitter, which relates the transmitter's output positionor the set-pointpointer (Figures
8-18and 8-19).The output or the transmitteris applied to the set bellows or the controller. If the automaticcontroller is operating, normal

PNEUMATICCONTROLMECHANISMS 221

changes can be made by turning the set-paint knob. Proportional-plusintegral action will occur. Since the derivative amplifier exists only in
the measurementcircuit, derivative action will nat occur. If it is desired
to bring the process slowly to the new set paint with integral action
only, the controller is simply switched to manual, the set-paint change
is made, and the controller is switched back to automatic. Now the
measurement will approach the new set paint, with integral action only,
and no overshoot will occur.

222

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

SET paiNT

LTIME-i

TIME-

Fig. 8-17. TransCerCrommanual to automatic ([eft) when measurementequals set point

and (right) when measurement does not equal set point.

Fig. 8-18. FoxboropneumaticConsotrol130Mcontrolletpulled out to show

access-to-mode
adjustments.

PNEUMATICCONTROLMECHANISMS 223

LlNEARIZING
ASPIRATOR
RELAY

iC

TO REVERSE
FROMREMOTE
SWITCH AND
SET POINT
AUTOMATIC
SIGNAL
L.A.R.
CONTROL U 1T
.L

'j

LOCAL
I ~.
SETPOINT1n-"1
TRANS

AlS \~~-.
r-rr

II

REMOTE
SET
POINT

RECEIVERI

SET KNOB

Fig. 8-19. Schematicoflocal and remote set-pointmechanisms.

The Closed-Loop

Pneumatic

Controlled

System

Now let us apply the pneumatic controller to a process controlloop.

The objective of a control system is to maintain a balance between
supply and demand over time. (See Chapter 1 for a discussion of supply
and demand.) The closed-loop control system achieves this balance by
measuring the demand and regulating the supply to maintain the desired balance.
Figure 8-20 illustrates a typical and familiar controlled system in
which the temperature of heated water (controlled variable) is regulated by control of i~put steam. The control system comprises a

IIPULATED

224

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

VARIABLE 1

Fig. 8.20. The processto becontrolledoccursin a beatexchanger.All elementsofthe

pneumaticcontrol systemare shown-transmitter, controller, valve, input water,
output water, and steam.

pneumatic temperature transmitter (Foxboro Model 12A), a forcebalance pneumatic controller (Foxboro 130 Series), a pneumatic (diaphragm) valve actuator, and a control val ve (wide-range V-port).
Figure 8-21 shows an accepted manner of representing a system in
a block diagram. All the elements of the actual system are included.

CONTROLLER

,'

:.--J ERROR 1 fAMPLIFIER'

"

iDETECTORM (RELAY)t=l

L-

-INTERNALFEEDBACKLOOPjI

SET

II M~.A_S.u.~.!.NG
II
MEANS
~EXTERNAL

CONTROLLED

FEEDBACK

LOOP

MANIPULATEO

Fig. 8-21. Block diagramshowingall the elementsor Figure8-20.

PNEUMATICCONTROLMECHANISMS 225

The Process

The process to be controlled is a shell-and-tube beat exchanger. Lowpressure steam applied to the shell beats water flowing through the
tubes. The exchanger has 33.5 square feet ofheat-transfer surface, and
the time required for the beat exchange to take place across this surface causes the exchanger to bave a delayed response. The response
characteristic is that of a multiple-capacitance, multiple-resi&tance
circuit.

The response or reaction curve of the beat exchanger was 'obtained

by (1) allowing the beat exchanger to stabilize with a constant flow of
both steam and water, (2) holding the water flow constant, (3) increasing the steam flow suddenly by opening the control valve, and (4)
plotting the increase in water temperature. The resulting curve is the
reaction curve illustrated in Figure 8-22.
'
Since the water temperature was measured with the pneumatie
temperature transmitter, the transmitter output depends nat only on the
response of the beat exchanger, but also on the response of the transmitter, which is nat instantaneous.
The Controller

Set Point

The set point o the controller is established by setting an air pressure

in a "set" bellows by means o an air regulator. The setting dial (Figure
8-23) is spread over the span o the measured (controlled) variable.

Fig. 8-22. Responseof outlet temperatureto stepsin steamvalve positiontest results.

226 PNEUMATIC AND ELECTRONICCONTROLSYSTEMS

OUTPUT

SET KNOB

Fig. 8-23. Schematic of set-point mechanism.

Valve and Actuator

The controller output positions a control valve which, in turn, governs

the energy or material inflow to the processoThe speed of response of
the valve depends on the type of valve actuator, that is, the device that
transduces the controller output signal into a valve position.
The actuator can be pneumatic or electric. The system under study
uses a familiar pneumatic actuator, which consists of a large diaphragm
(110 square inches) to which the pneumatic signal is-applied. When 3
psi (20 kPa) is applied to an area of 110 square inches, a force of 330
pounds is developed; when 15 psi (100 kPa) is applied, the thrust will be
1,650 pounds. This downward diaphragm force is opposed by a large
coil spring that pushes upward with a force of 330 to 1,650 pounds.
Increasing or decreasing the pneumatic signat to the diaphragm will
cause the diaphragm to move until it reaches a position where diaphragm force and spring force are in equilibrium. Since the relationship
between spring thrust and excursion is a linear one (Hooke's law),
valve trave} is related linearly to controller output.
Valve actuators mar be arranged either to open the valve or close it
on increasing air pressure. The choice of action is usually dictated by
the process being controlled. For a beat exchanger process, the air-toopen valve usually is selected because the spring will close the val ve
and cause the temperature to drop in event of air failure. This is called
fail-safe action. If the process were one that became hazardous when
the valve closed, the action would be reversed to make the valve open
on air failure.
The valve actuator has a response time because it has the capacity

PNEUMATICCONTROLMECHANISMS 227

to hold air and the connectingtubing offers resistanceto air ftow. The
time constant for a pneumatic operator is typically several seconds
with normallengths of connectingtubing. Adding tubing lengthensthe
time constant.Thus, the valve actuatorcontributestime lag or phase
shift to the controlloop, dependingon the lengthof connectingtubing

used.
Final Control

Element

The function of the pneumaticvalve actuatoris to positionthe control

valve. The valve canbe one of manytypes, dependinguponthe process
to be controlled. For the beat exchangerprocessbeing discussed,a
single-seatwide-rangeequalpercentagevalve is used (Figure 8-24).
The single-seatequalpercentagevalve hasa contouredinner valve
that provides an exponentialrelationship betweenvalve stroke and
valve capacity.

Fig. 8-24. Cutawayview of single-seat,wide-range,equalpercentvalve.

228 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Fig. 8-25. Closed-loopcontrol system,showingall basiccomponents.

Dynamic Behavior ot Closed-Loop Control Systems

When all components are connected to form the closed-loop control
system (Figure 8-25), each component contributes to control system
operation. For example, the total gain (output signal/input signal) is
affected by the gain of every one ofthe loop components. However, the
only loop component that has an adjustable gain is the controller. Adjustment ofthe proportional band results in adjustment ofthe totalloop
or system gain.
Let us assume that the control system has stabilized with the outftow temperature at 1500F,66C and that the system is subjected to an
upset by raising the set paint suddenly to 1600F(71C).

~SE

PNEUMATICCONTROLMECHANISMS 229

".,3."CON.'.,.
'" ,. 6.

,.

The action around the controlloop (Figure 8-26) will be as follows:

1. Increasing the set point increases air pressure in the set bellows,
thereby causing the ftapper to approach the nozzle.
2. Increased nozzle pressure is amplified and becomes output to the
control valve. Because the valve actuator is a resistancecapacitance element, the actual change in valve position will be
delayed behind the change in output signal.

230 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

3. The addedbeat (Btu) will be nearly in stepor phasewith the valve

position.
4. The added beat causesthe temperatureof the beat exchangerto
rise, and as it Tises,the pressurein the measurement
bellows will
increase.
5. Any unbalanceof the momentsof force contributed by the measurementand set bellows will causethe integral bellows to continue to change through the integral restricter, and the output
pressureto increaseuntil the temperaturereachesthe set point.
The momentsof force are now balanced,and the ftapperis within
its throttling range(a band of travelless than0.001 inch wide).
6. After approximately2 minutesbaveelapsed,the forcesexertedby
the various bellows will be in equilibrium; the ftapperis within its
throttling range; and the processhas stabilized at 160F,71C.
Thesestepsare illustrated in Figure 8-26.
(Note: If the upsetwere causedby a load change,derivative action
would bave beenaddedto the functions described.)
Adjusting the Controller
To accomplish effective control and cause the process to feaci in an
optimum fashion, the controller behavior, with respect to time, musi be
matched to thai of the processoThis is done by adjusting the proportional band, integral time, and derivative time.
Thning or adjusting the controller for optimum performance can be
achieved either by trial-and-error, or by more exacting methods. The
usual trial-and-error procedure in adjusting the controller settings to
the process conditions is to set the integral restrictor at maximum and
the derivative restrictor at minimum, and then to adjust the proportional band to produce the minimum process stabilization time. Then
the derivative restrictor is increased gradually, and the proportional
band narrowed, until a combination of proportional and derivative action is obtained which produces a shorter stabilization time than normal
proportional, and with less upset to the processo
To eliminate offset, the integral restrictor is then set to avalue thai
will bring the control point to the set point in a minimal time without
upsetting the stability of the system.
Other methods of adjustment by mathematical and analog analysis
are discussed in Chapter 12.
Suppose a set point of 20 percent scale change is made in the
Foxboro 130 Series controller with the controller on automatic. Note

PNEUMATICCONTROLMECHANISMS 231

OUTFLOW

WATER
TEMPERATUREI

Fig. 8-27. Controlled set-point change made in automatic mode (top) and in manual mode,
then switched to automatic (bottom).

the damped oscillations that occur as the system Testabilizes (Figure

8-27, top). The instability lasts for several minutes. If the controller is
first switched to manual for the set-point change, a reaction will occur
when switched back to automatic. This is shown in the lower part of
Figure 8-27 and discussed previously. The controlled variable retums
to the new set point at the integral rate. Any adjustments tend to cause
a temporary upset; therefore, the technique of switching to manual,
making any required adjustments and switching back to automatic, can
be used to advantage with the Foxboro 130Series controller.

Batch Controli er

In Chapter 7, reference was made to integral saturation or integral

windup. This occurs when the integral mode is used on a batch or
discontinuous processo During the time the process is shut down, the
integral circuit of a conventional two- or three-mode controller will
saturate at the supply pressure.
After the process is restarted, measurement overshoot of the set
point can occur unless precautions are taken. The batch switch is a
special pneumatic mechanism designed to prevent this overshoot.
During the time a process is otI control, the measurement will be

-Fig.

232 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

riME

CuRVE

8-28. (Left) Recovery from sustained deviation; proportional plus integral controller
without batch feature. (Right) With batch feature.

below the set point and the controller's output will be at its maximum
value. Ifthe time oftbis deviation is long enough, the integral circuit of
a conventional controller will also reach ibis value. When process conditions retum to normal, no change in controller output can occur until
the measurementreaches the set point. Figure 8-28, curve A illustrates
ibis control action.
The batch switch eliminates tbis integral circuit saturation and
conditions the integral bellows to permit output to start to change
before the measurement reaches theset point. If the controller output
starts to change before the measurement reaches the set point, overshoot can be prevented and the measurement can retum to the set point
smoothly. Figure 8-28, curve B illustrates tbis control action.
Principie

ot Operation

A schematic diagram of a proportional-plus-integral batch controller is

shown in Figure 8-29. The only addition to a conventional two-mode
controller is a specially designed pressure switch between the relay
output and the integral circuit.
The switch is actuated by the output pressure of the controller
relay. Its trip point is adjusted by a spring, or an external pneumatic
signal, to trigger at 15 psi (100 kPa). When the controller's output is
below this pressure, the ball valve closes the vent port and permits
passage of the relay output to the integral circuit. This allows normal
integral response as long as the measurement remains within the proportional band (below 15 psi or 100 kPa). Should the measurement
reach the 15 psi or 100 kPa limit of the proportional band throttling
range, the force on the batch switch diaphragm will seat the ball valve
in the opposite direction. This cuts off the relay output pressure to the
integral circuit and simultaneously vents the circuit to the atmosphere.

PNEUMATICCONTROLMECHANISMS 233

Fig. 8-29. SchematicoCproportional-plus-integraJ

batchcontroller.

This causes the pressure in the circuit to dropo As long as the relay
output pressure is above the trip point, the integral circuit pressure will
drop until there is no longer any pressure in the integral bellows. At this
point, the proportional band throttling range will have shifted completely below the set point as indicated in Figure 8-28B, curve B. During this operation, the controller's output pressure is at 15 psi or 100
kPa (or more) and the valve is wide open. When the batch process is
restarted, the control valve will begin to throttle as soon as the measurement reaches the 15 psi or 100kPa lmit of the proportional band
throttling range. Note that as soon asthe controller output drops below
15 psi or 100kPa, normal integral response is restored. This overcomes
the tendency to overshoot the set point.
Load Bias (Preload)

Due to the time characteristics of certain processes, shifting the proportional band completely below the set point will cause an intolerable
delay in bringing the batch to the set point. In such cases, an adjustable
back pressure mar be applied to the vent port to "bias" the batch
switch. Thus, the pressure in the integral circuit mar be prevented from
dropping below a preselected amount and the proportional band throttling range will shift only partially below the set point. Although this
will allow faster recovery, a slight overshoot will occur if the "bias" is

234 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

increased too much. Increasing the bias to 15 psi or 100 kPa would
obviously completely eliminate batch action.
One limitation thai applies to all pneumatic control systems is the
distance thai mar be accommodated between components. The maximum distance depends on two major factors: the pneumatic signal
travels at the speed of sound, and the loop components along with the
tubing all bave capacity and resistance. Thus, an RC time constant
musi exist. Volume boosters often are applied to lengthen ibis working
distance. Unfortunately, a volume booster will do nothing to speed up
signal velocity. A booster will help if a large volume, such as thai found
in a pneumatic valve actuator, is involved.
The distance lmit for pneumatic control systems is approximately
200 feet (60 meters). Iftbis distance factor is ignored, the dynamics ofthe
control system will suffer and reguli in poor control operation. If it
becomes necessary to lengthen these distances substantially, the only
practical solution is to utilize electrical signals thai mar work over an
almost unlimited distance.
Questions

8.1. The integraldial in a pneumaticcontrolleris calibratedin:

8. Minutes or repeats
d. Percentage
b. Integralunits
e. OtIset

c. Gain
8-2. True or false: In a proportional-onlycontroller,ifthe measurement
equals
the set point the output will equalthe bias.
8-3. True or false: In an integralcontroller,the rate of changeof the output
is proportionalto the error.
8.4. True or false: The largerthe numberon the integraldial the greaterthe
effectof the integralaction.
8-5. True or false: In a batchoperation,if a controllerhaswound up, it is
quite possiblethat the valve may stay in an extremepositionuntil the
measurementactuallygoesbeyondthe set point beforethe valve beginsto
changeits position.
8-6. True or false: What somemanufacturerscall rate others call derivative.
8-7. Indicate all the correctstatements:
8. Gainis the reciprocatof the proportionalband.
b. The proportionalbandis the reciprocatof gain.

PNEUMATICCONTROLMECHANISMS 235

c.
d.
e.
f.

The
The
The
The

proportional band times the gain equals 1.

gain divided by the proportional band equals 1.
narrower the proportional band, the higher the gain.
wider the proportional band, the lower the gain.

8-8. In a process controlled by a proportional-plus-integral controller, the

measurement was at the set point and the output was 9 psi (60 kPa). The
measurement then quickly decreased to a certain value below the set point
and leveled out there. The output responded by changing to 8 psi (55 kPa). As
time progresses, what would you expect of the output pressure?
a. Increase to 9 psi (60 kPa) to bring the measurement back to the set

point.
b. Remain at 8 psi (55 kPa) as long as the measurement stays where it is.
c. Decrease and continue to decrease to 3 psi (20 kPa).
d. Continue to decrease until the measurementreaches the set point, or if
it does not retum to the set point, decrease to Opsi.
8-9. The range of the temperature measuring system used in conjunction
with a Model 130proportional-only contro1ler is O to 150oP(66C). The output
is 9 psi (60 kPa) when the set point and indicator are both at 75P (24C). If
the proportional band is 200 percent, what is the output when the
measurement is 150oP(66C)?
a. 9 psi (60 kPa)
c. 12 psi (83 kPa)
b. 3 psi (20 kPa)
d. 15 psi (100 kPa)8-10.
If the span of a measuring transmitter in a control system is made
one-half of its value, the proportional-band adjustment in the controller must
be -to
maintain the same quality of control.
a. Cut in half
d. Narrowed
b. Doubled
e. None of the above.

c. Squared
8-11. With a proportional-plus-integral controller, a sustained error will result

in:
a.
b.
c.
d.
e.

Windup
A fixed offset
A temporarynarrowingof the proportionalband
A delay in the process
None of the above

8-12. By locating the derivative function in the input measurement

circuit,
which of the following advantagescanbe realized?
a. Smoothbumplesstransfer
b. No derivative bump with a set-pointchange
c. No proportionalresponseto a set-pointchange
d. Integral adjustmentis isolated from response

8-13.

236 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

a.
b.
c.
d.

If proportional-plus-integral
controlis good,the additionof derivative:
Will anticipatechangesand speedup corrections
Will alwaysimprove control
Will makethe controlleradjustmentseasierto accomplish
May create stability problemsin somesystems

8-14. The advantageof addingderivativeto a controlleris always:

a. Increasedstability
b. The ability to overcomea big pure deadtime lag
c. The ability to reactmore quickly to measurement
change
d. A decreasein the pure deadtime of the process
8-15. For fail-safeactionthe control valve should,uponenergy(air)failure:
a. Open
b. Close
c. Move in suchdirectionas to makethe processnonhazardous
d. Stay in its previous position
8-16. If the closed-loopcontrol systemhastoo muchgain, it will cycle. The
only loop componentthat hasconvenientlyadjustablegainis the:
a. Measuringtransmitter
c. Process
b. Valve operator
d. Controller
8-17. Adjusting the controller for optimumperformance:
a. Is not required,becauseit adjustsitself
b. Requiresa specialtool
c. Is usuallydone by trial and error
d. Always requiresa very involved mathematicalanalysisof the process
8-18. A processis to be controlledusingan all pneumaticsystem.The
maximumdistancebetweenloop componentswill be:
a. 1,000feet
c. 200feet
b. 500feet
d. 20 feet
8-19. If the distancebetweenloop componentsin the all pneumaticcontrol
systemmustbe increasedit will require:
a. A pneumaticvolume booster
b. Largersizedtubing
c. Smallersizedtubing
d. Conversionto an electricalsignal
8-20. PneumaticsignalstraveI throughthe signaltubing al:
a. 100feet per second
b. Approximatelythe speedof sound
c. A rate that dependson tubing size
d. Approximatelythe speedof light

237

Electronic control systems,already widely accepted,are gainingrapidly in popularity for a numberof reasons:
1. Electrical signalsoperateover greatdistanceswithout contributing
time lags.
2. Electrical signals can easily be made compatible with a digital
computer.
3. Electronic units can easily handle multiple-signalinputs.
4. Electronic devices can be designedto be essentiallymaintenance
free.
s. Intrinsic safety techniques bave virtually eliminated electrical
hazards.
6. Generally, electricalsystemsare lessexpensiveto install, take up
less space,and can handle almost all processmeasurements.
7. Electronic devices are more energy efficient than comparable
pneumatic equipment.

8. Specialpurposedevices suchas nonlinearcontrollers and analogcomputingunits are simplified.

This chapter will describethe electronic instrumentcomponents
that make up a typical multiloop system. The processis one found
within the boiler room of most processplants-a feedwater control
systemfor a boiler drum.

238 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Feedwater

Control

Systems

A sample electronic control system is shown in Figure 9-1. The water

level in the drum is to be controlled. This is accomplished by measuring
Dot only the water level, but the steam ftow out of the boiler and the
water ftow into the drum. The three measured variables are fed to an
electronic control system, which controls the water to the drum.
Transm itters

Two ditIerential pressure transmitters (Fig. 9-2) are used, one to measure water flow; the other to measure drum water level.
Referring to Fig. 9-3, in operation the ditIerence in pressure between the high and low gide of the transmitter body is sensed by a twin
diaphragm capsule (1) which transforms the ditIerential pressure into a
force equal to the ditIerential pressure times the etIective area of the
diaphragm. The resultant force is transferred through the C-flexure (2)
to the lower end of the force bar (3). Attached to the force bar is a
cobalt-nickel alloy diaphragm which serves as a fulcrum point for the
force bar and also as a seal to the process in the low-pressure cavity
gide of the transmitter body. As a result of the force generated, the

Fig. 9-1. Peedwater control system. (I) Drum water level (2) Steam ftow from boiler,
via turbine steam pressure. and (3) Water ftow into drum.

Fig. 9-2. Differentialpressuretransmitter.

Fig. 9-3. Operationof electronicforce balancedifIerentialpressuretransmitter.

239

force

240 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

bar pivots abolli the CoNi alloyseal, transferring a force to the

vector mechanism (5).
The force transmitted by the vector mechanism to the lever system
(11) is dependent on the adjustable angle. Changing the angle adjuststhe
span of the instrument. At point (6), the lever system pivots and
moves a femte disk, part ora differential transformer (7) which serves
as a detector. Any position change of the ferrite disk changes the output of the differential transformer determining the amplitude output of
an oscillator (8). The oscillator output is rectified to a d-c signat and
amplified, resulting in a,4-20 mA d-c transmitter output signal. A feedback motor (9) in series with the output signal, exerts a force proportional to the error signal generated by the differential transformer. This
force re balances the lever system. Accordingly, the output signal ofthe
transmitter is directly proportional to the applied differential pressure
at the capsule.
Any given applied differential pressure within the calibrated measurement range will result in the positioning of the detector' s femte
disk, which, in turn, will maintain an output signal from the amplifier
proportional to measurement, thus keeping the force balance in equilibrium. A simplified schematic of the electronic circuitry is shown in Fig.

9-4.
Steam ftow out ofthe drum is measured in terms of steam pressure,
which is measured with a pressure transmitter. The operation of the
pressure transmitter is similar to the operation of the differential

Fig. 9-4. Working schematic oCelectronic transmitter.

ELECTRONIC(ANALOG) CONTROLSYSTEMS 241

pressure transmitter_,The same feedback technique used in the differential pressure transmitter is employed in the pressure transmitter. The
major difference lies in the size and construction of the sensor.
The pressure transmitter measures the pressure in the first stage of
the power turbine. Since there is a linear relationship between first
stage turbine pressure and steam flow, the output signal from the
pressure transmitter is linear with steam flow out of the boiler,
A variety of electronic differential pressure transmitters are available from a number of manufacturers. Some of these make use of strain
gauge detectors, capacitive detectors, resonant wire detectors, and inductive detectors. Many are motion or open loop devices which are still
capable of accuracies within a fraction of a percent. In the application
of any electronic transmitter, the user must guard against subjecting the
electronics to temperatures which might result in damage. This probIem generally can be avoided by observing the precautions recommended by the manufacturer.

The Controllers

The controllers described befe are the Foxboro SPEC 200 type. The
SPEC 200 is generally a split-architecture system. In this system two
areas may be used, a display area and a oest area. Field equipment,
such as measuring transmitters, electrical valve actuators, and the like,
generally operates on 4 to 20 mA dc. Within the oest, SPEC 200 operates on O to 10 volts.
The display area contains control stations, manual stations, recorders, and indicators to provide the necessary operator displays and
controls. These units are shelf-mounted and contain only the electronic
circuitry required to communicate the display and adjustments necessary for an operator to control and monitor a processo
The nest area contains the analog control, computing, alarm, signal
conditioning, and input and output signal converter units. These units
are in the form of "modules" and "circuit cards." The oest itself is
basically an enclosure provided for the mounting of "modules."
System power is supplied to the oest. This power supply must
deli ver + 15 and -15 V dc for operation of the display and nestmounted instruments. Recorder chart drives and alarm lights require 24
V ac.
When a single location is required, the oest and display areas may

RAONaR,

THIRD EDITION

for

NORMAN

CHILTON COMPANY

A. ANDERSON

PENNSYLVANIA

CONTENTS

Preface

SECTION

I FEEDBACK

1 Introduction

PROCESS

to Process

Control

CONTROL

Types of Processes 5 Processes with More Than One Capacity and Resistance 6 Dead Time 6 Measurement 7 Symbols 1.0 The Feedback
Loop 1.0 Feedback Control 1.4 ControIling the Process 1.4 Selecting
ControIler Action 1.6 Up sets 1.7 Process Characteristics and ControIlability 1.7 Controller Responses 1.8 On/Off Control 1.9 Proportional
Action 20 Integral Action (Reset) 23 Derivative Action 25 Selecting
the ControIler 27 Conclusion 29 Questions 30
2 Process/Pressure

Measuring

Instruments

33

What Is Pressure? 33 Units of Measurement 35 Pressure Measurement 35 The Pascal 36 Bar Versus Pascal 37 Gauge, Absolute, and
Differential Pressure 37 Understanding the Effects of Gravity 38 GravityDependent Units 38 Gravity-Independent Units 39 Pressure Standards 39 Plant Instruments That Measure Pressure Directly 44 Bell

vi

CONTENTS

Instrument 44 Slack .or Limp-Diaphragm 44 Pressure Gauges 46

Liquid or Steam Pressure Measurement 48 Seals and Purges 48 Pulsation Dampener 48 Metallic Bellows 49 Pressure Transmitters 52 Signal Transmissions 52 Pneumatic Recorders and Indicators 53 Mechanical
Pressure Seals 55 Calibration Techniques 59 Field Standard 59 Portable Pneumatic Calibrator 59 Force-Balance Pneumatic Pressure Transmitter 60 Pneumatic Relay 62 Principle of Operation 63 Absolute
Pressure Transmitter 64 Questions 65
3 Level and Density Measurements 69
Level Measurement Methods 69 Float-and-Cable 70' Displacement
(Buoyancy) 71 Head or Pressure 73 Capacitance 78 Conductance 79
Radiation 79 Weight 79 Ultrasonic 80 Thermal 81 Density Measurement Methods for Liquids and Liquid glumes 81 Hydrostatic Head 82
Radiation 84 Vibration 84 Temperature Eft'ects and Considerations 84
Dift'erential Pressure Transmitter 85 Questions 87
4 Flow Measurement 90
Constriction or Differential Head Type 91 Primary Devices 94 Secondai-y
Devices 99 Relating Flow Rate to Differential 100 Effect of Temperature
on Flow Rate 103 Variable Area Meters (Rotameter) 109 Open-Channel
Flow Rate Measurements 109 Primary Devices 110 Installation and
Selection Considerations 116 Velocity Flowmeters 117 Magnetic Flowmeter 117 Vortex Flowmeter 121 Turbine Flowmeter 122 Other Flowmeters 123 Conclusion 124 Questions 124
5 Temperature and Humidity Measurements

126

Temperature 126 Filled Thermal Systems 128 Electrical Systems 130

Thermocouples 130 Resistance Thermal Detectors 139 Thermistors 144
Humidity Measurements 144 Questions 148
6 Analytical Measurements 151
Electrical Conductivity 151 Types of Calibration 154 Calibration in
Conductivity 154 Calibration in Terms of Concentration of
Electrolyte 155 Polarization 156 Cell Construction 156 Electrodeless
Conductivity Measurements 157 Hydrogen Ion Activity (pH) 159 Ionization or Dissociation 159 The pH Scale 161 Measurement of pH-The
Glass Electrode 164 Reference Electrode 165 Temperature Compensation 168 Reading the Output of the pH Electrodes 169 pH Control 170
Summary 172 Oxidation-Reduction Potential 172 Ion-Selective Measurement 173 Chromatography 174 Capacitance 175 References 176
Questions 176

CONTENTS

vii

SECTION II PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS 179

7 The Feedback Control Loop 179

The Closed-Loop Control System 183 Phase Shift Through Re Networks 184 Oscillation 184 Stability in the Closed-Loop System 187
Nonlinearities 188 ControUers and Control Modes 189 Two-Position
Application of Proportional ConControl 189 Throttling Control 191
trol 193 Proportional-Plus-Integral Control 195 Adding Derivative
(Rate) 199 Selecting the Controller 200 Questions 202
8 Pneumatic Control Mechanisms 205
The Flapper-Nozzle Unit 205 Valve Relay or Pneumatic Amplifier 207
The Linear Aspirating Relay or Pneumatic Amplifier 208 Proportional
The Automatic ConAction 209 Control Mechanism Requirements 211
troller 212
Derivative and Integral 215 Manual Control Unit 216
Transfer 217 Set Point 220 The Closed-Loop Pneumatic Control
System 223 The Process 225 The Controller Set Point 225 Valve and
Actuator 226 Final Control Element 227 Dynamic Behavior of ClosedLoop Control Systems 228 Adjusting the Controller 230 Batch Controller 231
Principle of Operation 232 Load Bias (Preload) 233
Questions 234
9 Electronic (Analog) Control Systems 237
Feedwater Control Systems 238 Transmitters 238 The Controllers 241
Principle of Operation 243 Increase/Decrease Switch 244 Deviation
Signal Generation 244 Derivative Action 244 Proportional Band
Action 245
Final Summing and Switching 245 High and Low Limits 245
External Integral (- R) Option 246 External Summing (- S) Option 246
Power Supply Fault Protection Circuit 246 Automatic/Manual Switch 247
Controller Adjustments 247 General Description of the Feedwater Control
System 247 Control Station 248 Ml2AX + A4-R Drum Level Control
Unit 249 Ml2AX + A4 Feedwater Control Unit 250 Ml2AP + SUM
Feedwater/Steam Flow Computing Unit 250 Square Root Extractor 250
Feedwater Control 252 Closed-Loop Operation 252 Comparison of Electronic and Pneumatic Systems 254 Questions 256
SECTION III ACTUATORS AND VALVES 259

10 Actuators 259
Valve Actuator 259 Valve Positioner 261
Electrical Signals 263 Plinciple of Operation 265 Electric Motor Actuators 267 Questions 268

viii

CONTENTS

11 Control Valves 270

Capacity of a Control Valve 272 Valve Sizing 272 Determining Pressure

Drop Across the Valve 273 Cavitation and Flashing 274 Valve Rangeability 276 Selection Factors 278 Sequencing-Control Valves 279
References 292 Questions 292
Viscosity Corrections 281
SECTION IV CONTROL LOOP ADJUSTMENT AND ANALYSIS 295
12 Controller Adjustments 295

Proportional-Only Controller 295

Proportional-Pius-Integral Controller 2%
Adding Derivative Action 297 Tuning Maps 298 Adjusting a Controller
with the Proportional, Integral, and Derivative Mode 299 Closed-Loop
Cycling Method 301

Questions 303

13 Step-Analysis Method of Finding Time Constant 304

Block Diagrams 304 Step Analysis of Single-Time-Constant System 308

Step Analysis of Two-Time-Constant System 309 Percent-Incomplete
Method 310 Multicapacity System 313 Finding Control Modes by Step
Analysis 315 On/Off Control Action 316 Proportional Control Action 316
Proportional-Plus-Integral Action 318 Proportional-Plus-Derivative Control Action 319 Proportional-Plus-Derivative-Plus-Integral Control
Action 320 Using the Reaction Curve to Determine Controller Adjustments 320 References 324 Questions 325
14 Frequency Response Analysis 328

Finding the Time Constant from the Bode Diagram 331

System 331
Integral 337

Response 339

Testing a

Control Objectives 335 Adding Derivative 336 Adding

Adjusting the Proportional Band 339 Closed-Loop

Conclusion 340

Questions 341

SECTION V COMBINATION CONTROL SYSTEMS 343

15 Split-Range, Auto-Selector, Ratio, and Cascade Systems 343

Duplex or Split-Range Control 343 Auto-Selector or Cutback Control 345

Nonpipeline Application of Auto-Select/Cutback Control Systems 348 Cascade Controller 352 Saturation in Cascaded Loops 355 Questions 357
16 Feedforward Control 360
Definition 360

History 360

Advantages 360

Technique 361

cation to Heat Exchanger 361

Distillation Control 369
References 370 Questions 371

Appli-

Conclusion 370

CONTENTS

ix

SECTION VI PROCESS COMPUTERS AND SIMULATION 372

17 Computer Interface and Hardware 372

Basic Elements of the Computer 372

Input 373

Central Processing

Unit (CPU) 373 Output 374 Types of Memory 374 Core 374
Drum 375
Disk 375
Semiconductor (Solid-State) 376 Operator Com-

munication Devices 377 Process Interface Equipment 377

sion 385 Questions 385

Conclu-

18 Computer Software and Operation 388

A Brief Introduction 388 Basic Computer Operation 390 Real-Time

Clock and Power FaiVRestart Logic 394 Control Software 394 Achieving
Process Control 396 Questions 400
19 Programmed Control Systems 402

Description of Circuit 403 Sequential Control Systems 405 Programmable Controllers 406 Programming Language 406 Programming 408
Logic System Automation 409 System Control Displays 410 Conclusion 411
Questions 411
20 Constructing and Instrumenting Real and Simulated Processes 413

Simulated Processes 413 Constructing an Electric Process Simulator 414

The Foxboro Electronic Process Simulator 415 Pneumatic Process Simulator 420 Operation of the Pneumatic Simulator 423 Time Base 424
Real Processes 427 Questions 433
Appendix 436

Units and Conversion Tables for Process Control 436 Specific Gravities of
Common Liquids 451
Standard Pipe Dimensions 452 Properties of
Saturated Steam and Saturated Water 453 Resistance Values 454
Velocity and Pressure Drop 455 Thermocouple Temperature 458 Relative Humidity (% saturation) Tables 476
Glossary 480
Answers to Questions 488
Bibliography 491
Index 492

FEEDBACK PROCESSCONTROL

PROCESS

REACTIONCURVE
EC

(a)

EDc
TIME

(b)

TIME

PRESSURE
IN TANK

(cI

SUPPLY
TIME

TEMP.

(dI

TIME

Fig. 1-2. Four types or systems: (a) electric, (b) hydraulic, (c) pneumatic, and (d)
thermal. Each has a single capacity and a single resistance and all bave identical
response characteristics.

regulate a process and certain aspects of process behavior will be discussed in this text. Examples of some completely instrumented process
systems will be given to demonstrate the practical application of the
instrument components.
The physical system to be controlled mar be electrical, thermal,
hydraulic, pneumatic, gaseous,mechanical, or any other physical type.
Figure 1-2 and Table 1-1 compare severa! common systems. All followthe
same basic laws of physics and dynamics.

INTRoDucnON TO PROCESSCONTROL 5

The behavior of a process with respect to time defines its dynamic

characteristics. Behavior Dot involving time defines its static characteristics. Both static (steady) and dynamic (changing with time) responses must be considered in the operation and understanding of a
process control system.
Types

ot Processes

The simplest process contains a single capacity and a single resistance.

Figure 1-2 illustrates a single-capacity, single-resistance process in (a)
electrical, (b) hydraulic, (c) pneumatic, and (d) thermal forms. To show
how these behave with respect to time, we can impose a step upset
(sudden change) in the input to the process and examine the output.
The resulting change in process::variable with respect to time is plotted
in Figure 1-3. The reaction curve of all four types of systems will be
identical.
This type of curve (exponential) is basic to automatic control. It
can be obtained easily with an electrical capacitor and resistor arranged
as in Figure l-2a.

Fig. 1-3. Universaltirne-constantchart, showingexponentialrise anddecay.

FEEDBACK PROCESSCONTROL

Figure l-2a shows a simple RC circuit-a resistor, capacitor, and

battery source in series. The instant the circuit is closed, the capacitor
starts to charge to the voltage of the battery. The rate at which the
capacitor charges gradually decreases as the capacitor voltage approaches the battery voltage (voltage curve A in Figure 1-3). Although
the rate varies, the time it takes the capacitor to charge to 63 percent of
the battery voltage is a constant for any one value of Rand C. Thus, no
matter what the voltage of the battery, the capacitor charges to 63
percent of the battery voltage in a time interval called the time constant, or characteristic time (T) ofthe circuit. The value orT in seconds
is the product of the resistance (in ohms) and the capacitance (in
farads).
Note thai the charging time increases with an increase in either R
orC.
~
This simple RC circuit is often used to produce the transient
waveform shown, which is called an exponential-rise transient.
On discharge, the circuit reacts similarly. For example, if the battery in Figure l-2a is replaced by a solid conductor, the charged
capacitar discharges 63 percent of its charge in RC seconds.
The simple RC circuit shown in Figure l-2a symbolizes many real
physical situations. It is important to examine the circuit in detail.
InRC seconds, the capacitar charges to 63.2 percent ofthe applied
voltage. In the nextRC seconds, the capacitar charges to 63.2 percent
ofthe remaining voltage, or to 87 percent ofthe applied voltage. In the
third interval of RC seconds, the capacitar charges to 95 percent of the
applied voltage. Although the capacitar never charges to exactly 100
percent of the applied voltage, it does charge to 99 percent in 4.6 RC
seconds as shown in Figure 1-3, which is a curve of capacitar voltage
(or current) versus time. Note that time is plotted inRC (time-constant)
units.
Processes with More Than One Capacity and Resistance
In practice, a process will contain many capacitance and resistance
elements. Figure 1-4 illustrates a process containing two resistance
elements and two capacitance elements. Figure 1-5 shows the resulting
process reaction curve. Note that the additional capacitance and resistance essentially affect the initial curve shape, adding a delay to the
processo
Dead Time
Dead time is a delay between two related actions. For example, assume
that the temperature sensor shown in Figure 1-1 was located 10 feet

I~RODUCTION TO PROCESSCONTROL 7

Fig. 1-4. Multicapacitysystem.

(3.048 m) away from the beat exchanger. lf the liquid travels at a

velocity of 10 feet (3.048 m) per second, a dead time of one second will
occur. ln some process control situations, dead tim e becomes the most
difficult factor in the equation'. Dead time mar also be called pure delay,
transport lag, or distance/velocity lag. Dead time is rarely found in its
pure form, but occurs frequently in combination with resistancecapacitance and other types of lags. Dead time is a difficult factor to
equate when applying control to the processo
Measurement

To employ feedback control, we musi first measure the condition we

wish to maintain at the desired standard. The condition (variable) mar
be temperature, pressure, ftow, level, conductivity, pH, moisture content, or the like.

Fig. l-S. Characteristiccurve oCmulticapacitysystem.

FEEDBACK PROCESSCONTROL

The measuring element is connected to the control element. ln

many installations, the measurement is located far from the controller.
This problem is solved by using a measuring transmitter (Figure 1-6).
The measuring transmitter usually develops an electrical signal for an
electronic controller or a pneumatic signal for a pneumatic controller.
Measuring transmitters bave attained great popularity in the process industries. They perform the measurementand develop a pneumatic or electric signal proportional to the variable in one unit. This signal
can be transmitted long distances. Pneumatic transmitters generally
produce an air pressure change of 3 to 15 psi or 20 to 100kPa (see p. 36
for definition of pascal unit) for measurement change of O to 100per-

Fig. 1.6. Measuringtransmittersconvertthe variableto be measuredinto a

proportionalpneumaticor electricalsignal.

INTRODUCTIONTO PROCESSCONTROL 9

Table 1.2. StandardISA and SAMA FunctionalDiagramElements

,
/

~
FLOW
\:-..:I TRANSMITTER

~
L!-J

SQUARE ROOT
EXTRACTOR

r:-
~

1:"';\ LEVEL
\!:!I TRANSMITTER

r:
~

MUL TIPlIER

L.LJ

~
~

PRESSURE
TRANSMITTER

~
L.:::...J DIVIDER

r'd:"":':1DERIVATIVE
~
CONTROL ACTION

l1:r"
~

TEMPERATURE
TRANSMITTER

r::-,
BIAS, ADDITION
L..=..J OR SUBTRACTION

r::;-,
~

TIME FUNCTION
CHARACTERIZER

1::\
~

POSITION
TRANSMITTER

rA""l
~

COMPARATOR
DIFFERENCE'

~
~

UNSPECIFIEDOR
NONLINEAR FUNCTION
CHARACTERIZER

~
L!.J

ADDER
SUMMR

O /,LIGHT
PANEL
1';"\
\.!oI

INDICA

t:'\
~

RECORDER

t:;:'\

RELAY

COlL

TOR

rJ1 CONTROL
INTEGRAL (RESEn
ACTION

r:-::l
~
*

A VERAGER

~
~

INTEGRATOR

-11-

rl..
\J

NORMALLYOPEN

PROPORTIONAL
CONTROL ACTION

"*

RELAYCONTACT

QUOTATION
ITEM NUMBER
MOUNTED
FRONT

ON
PANEL

OF

REGULATED
PROCESS

THE

AIR

NORMALLYCLOSED
RELAYCONTACT

~
V

AUTO/MANUAL
TRANSFERSWITCH

A\.
'W

MANUAL SIGNAL
GENERATOR

/::'~

ANALOG SIGNAL
GENERATOR

m
L-:-J

TRANSFEROR
TRIP RELAY

m
~

SOLENOID
ACTUATOR

~
~

ELECTRIC
MOTOR

r;
~

HIGH SIGNAL
SELECTOR

m
~

HIGH SIGNAL
lIMITER

~
~

HIGH SIGNAL
MONITOR

m
LOW SIGNAL
L.::::..J SELECTOR

m
~

LOW SIGNAL
lIMITER

r:-:--l
L!:!-J

LOW SIGNAL
MONITOR

~
~

~
~

HIGH AND LOW

lIMITER

~
~

HIGH AND LOW

SIGNAL MONITOR

VELOCITY OR
RATE lIMITER

~
ANALOG TO
~DIGITALCONV.

r::-1
RESISTANCE TO
~CURRENTCONV.

I;ViVITHERMOCOUPLETO

~VOLTAGE

~VOLTAGECONV.

L.:::..JCURRENTCONV.

~VOLTAGECONV.

r-viV1 VOLTAGE TO
~VOLTAGECONV.

rp;j1 PNEUMATIC TO
~CURRENTCONV.

~
PNEUMATIC TO
~VOLTAGECONV.

r;;;::\
MOTORIZED
~OPERATOR

r:;:-,
CURRENT TO
~PNEUMATICCONV.

~
~

f";O\ HYDRAUlIC
~OPERATOR

r:::::\
UNSPECIFIED
~OPERATOR

..JA
"'D'"

TO

r::';::"1 RESISTANCE TO
~VOLTAGECONV.
~CURRENTTO

VOL TAGE TO
PNEUMAnCCONV.

PNEUMATIC
OPERATOR

h..L.. STEM ACTION

~
(GLOBE)VALVE

THREE-WAY
SELECTOR VAL VE

ROTARYACnON
(BALLI VALVE

10

FEEDBACK PROCESSCONTROL

cent; that is, O percent of measurement yields an output pressure of 3

psi or 20 kPa, 50 percent of measurementyields 9 psi or 60 kPa and 100
percent yields 15 psi or 100 kPa output. Electronic transmitters produce either voltage or current signat outputs. For instance, the output
of analog transmitters is commonly 4 to 20 mA dc.
Symbols
A set of symbols has beenadopted to show instrumentation layouts and
to make these layouts more uniformo Once rou become familiar with
these symbols, it will become easy to visualize the system.
At present, two sets of symbols are in use. One set is provided by
the Scientific Apparatus Makers Association (SAMA) and the other by
Instrument Society of America (ISA). In ibis book the ISA symbols
will be used where applicable. Figure 1-7 and Tables 1-2 and 1-3 describe the symbols and identification letters often used. If rou are involved in the preparation or use of instrument loop diagrams, it is
suggested thai rou obtain the publication thai defines the standards
employed. A loop diagram musi contain the information needed for
both engineering and construction. This includes identification, description, connections and location, as well as energy sources.

The Feedback

Loop

The objective of a control system is to maintain a balance between

supply and demand over a period of time. As noted previously, supply
and demand are defined in terms of energy or material into (the manipuINSTRUMENT

FOR SINGLE MEASURED

INSTRUMENT

INSTRUMENT
MOUNTED ON
BOARD

LOCALLY
MOUNTED

VARIABLE

INSTRUMENT
MOUNTED
BEHIND
BOARD

NSTRUMENT

FOR TWO

MEASURED

Fig. 1-7. Instrumentfor measuredvariables.

VARIABLES

INTRODUCTIONTO PROCESSCONTROL II

Table 1-3. Meanings of Identification Letters

SUCCEEDINGLE1TERS

FIRST LE1TER
Measured or
Inilialing
Variable
A

Readoulo,
Passive
Funclion

Modijier

User'scooice

Voltage (EMF)

F
G

F1ow rate

User', choice

User',cboic.

Control

Conductivity
(electrical)
Density (mass)
or specitX:

Modi]ier

Alarrn

Analysis
Burner flame

Output
Function

Differential

gravity
Primary element
Ratio (fraction)

Glas.

Gaging
(dimensiooal)
Hand (manually
initiated)

Current

(electrical)
Power

Time or time

High

Indicat.
Controlstation

scheduIe
L

Level

Moisture or

humidity
User's cboice

User's cboic.

Oritice
(restriction)

Pressure or

Paint (test
connection)

Middle or inter-

mediate
User's cboice

vacuum
Q

Quantity or
event

Radioactiviy

Speed or

Low

Light (pilot)

Integrate

User'schoice

User's cboice

or

totalize
Recordor pnn!
Switch

SafelY

frequency
Transmit

Temperature

MuItivariable

Viscosity

w
x

Weight or force

Well

Unclassified

Unclassified

Multifunction

Multifunction

Multifunction

Valve, damper,
or louver
Unclassified

User's cboice

Relay or compute

Position

Drive, actuate
or unclassified
final control

element

Unclassified

FEEDBACK PROCESSCONTROL

Fig. }-8. Heatexchanger.

lated variable) and out of (the controlled variable) the processo The
closed-loop control system achieves this balance by measuring the demand and regulating the supply to maintain the desired balance over
time.
The basic idea of a feedback controlloop is most easily understood
by imagining what an operator would bave to do if automatic control
did Dot exist. Figure 1-8 shows a common application of automatic
control found in many industrial plants: a beat exchanger that uses
steam to beat cold water. In manual operation, the amount of steam
entering the beat exchanger depends on the air pressure to the valve,
which is set on the manual regulator. To control the temperature manually, the operator would watch the indicated temperature, and by comparing it with thedesired temperature, would open or close the valve to
admit more or less steam. When the temperature had reached the desired value, the operator would simply hold that output to the valve to
keep the temperature constant. Under automatic control, the temperatufe controller performs the same function. The measurement signat to
the controller from the temperature transmitter is continuously compared to the set-point signat entered into the controller. Based on a
comparlson of the signats, the automatic controller can tell whether
the measurement signat is above or below the set point and move the
valve accordingly until the measurement (temperature) comes to its
final value.
The simple feedback controlloop shown in Figure 1-9 illustrates
the four major elements of any feedback controlloop.

13

INTRODUCTION TO PROCESS CONTROL

+
~;~OL~:'-=- --.or
I MEASUREMENT

SUPPLV-*
FINAL ACTUATOR

CONTROLLED
VARIABLE

Fig. 1-9. Feedbackcontrolloop.

I. Measurementmust be made to indicate the current value of the

variable controlled by the loop. Commonmeasurementsused in
industry include ftow rate, pressure,level, temperature,analytical
measurementssuchas pH, ORPand conductivity; and manyotherg particular to specificindustries.
2. For every processthere mustbe afina! actuator that regulatesthe
supply of energyor materialto the processand changesthe measurementsignal. Most oftenthis is somekind of valve, but it might
also be a belt or motor speed,louver position, and so on.3.
The kinds of processesfound in industrial plants are as varied as
the materials they produce. They range from the commonplace,
suchas loops to control ftow rate, to the large and complex, such
as distillation columns in the petrochemicalindustry. Whether
simple or complex, they all consistof somecombinationof capacity resistanceand deadtime.4.
The last elementof the loop is the automaticcontroller. Its job is to
control the measurement.To "control" meansto keep the measurement at a constant, acceptablevalue. In this chapter, the
mechanismsinsidethe automaticcontroller will Dotbe considered.
Therefore, the principIesto be discussedmar be applied equally
well to both pneumaticand electronic controllers and to the controllers from any manufacturer.All automaticcontrollers use the
samegeneralresponses,althoughthe internalmechanismsand the
definitions given for these responsesmar differ slightly from one
another.

14 FEEDBACK PROCESSCONTROL

One basic concept is thai for automatic feedback control to exist,

the automatic controlloop musi be closed. This means thai information
musi be continuously passed around the loop. The controller musi be
able to move the val ve, the valve musi be able to atrect the measurement, and the measurement signal musi be reported to the controller. If
ibis path is broken at any point, the loop is said to be open. As soon as
the loop is opened-for example, when the automatic controller is
placed on manual-the automatic unit in the controller is no longer able
to move the valve. Thus, signats from the controller in response to
changing measurementconditions do noi atrect the valve and automatic
control does noi exist.
Feedback Control
Several principIes associated with feedback control can be observed by
considering a familiar control situation-adjusting the temperature of
water in a bathtub. This is obviously a manually controlled system.
One hand feels the water in the tub while the other manipulates the
inftow to reach the desired temperature. If a thermometer were used to
measure the temperature, greater accuracy would resulto Improved
measurement generally results in improved control.
The bathtub also illustrates the important effect of process capacity. Capacity (Figure 1-2)is a measure ofthe amount ofenergy it takes
to change a system a unit amount; thermal capacity is Btu/F, or the
amount of beat required to increase the temperature 1F. Since the
bathtub has a large capacity, it can be controlled in any of several
ways-by partially filling the tub with cold water, for example, and
then adding enough bot water to reach the desired temperature; or by
mixing the bot and cold to get the same resulto

Controlling

the Process

In performing the control function, the automatic controller uses the

difference between the set-point and the measurement signats to develop the output signat to the valve. The accuracy and responsiveness
of these signats is a basic limitation on the ability of the controller to
control the measurement correctly. If the transmitter does Dot send an
accurate signat, or ifthere is a lag in the measurementsignat, the ability
of the controller to manipulate the process will be degraded. At the
same time, the controller must receive an accurate set-point signat. In

INTRODUCTIONTO PROCESSCONTROL 15

controllers using pneumatic or electronic set-point signals generated

within the controller, miscalibration of the set-point transmitter will
develop the wrong value. The ability of the controller to position the
valve accurately is ret another limitation. If there is friction in the
valve, the controller mar Dot be able to move the valve to a specific
stem position to produce a specific flow, and this will appear as a
difference between measurement and set point. Repeated attempts to
position the valve exactly mar lead to hunting in the valve and in the
measurement. Or, if the controller is able only to move the valve very
slowly, the ability of the controller to control the process will be degraded. One way to improve the response of control valves is to use a
valve positioner, which acts as a feedback controller to position the
valve at the exact position corresponding to the controller output signal. However, positioners should be avoided in favor of volume boosters on fast-responding loops such as flow and liquid pressure.
For proper process control, the change in output Cromthe controller must be in such a direction as to oppose any change in the measurement value. Figure 1-10 shows a direct-connected valve to control

CONTROLLER
SPAN OF
MEASUREMENT

--rINLET FLOW

Lr

,-',
,,"

[[],
=

". ,'~~~""'-1

...,...,-

~}f~

-"..,"~ 2--=~.;:

-~-=-~_.=~.

f
OUTLET

_E/
I
100

I
50

I
O

PERCENT OPENING OF VALVE

FLOW

Fig. 1-10. In proportionalcontrol, the controllingvalve's positionis proportionalto the

controlled variable(level).

,:--=('i
~--=~
~I

16 FEEDBACK PROCESSCONTROL

level in a tank at midscale. As the level in the tank rises, the fioat acts to
reduce the fiow rate coming in. Thus, the higher the lquid level, the
more the fiow will be reduced. In the same way, as the level falls, the
fioat will open the valve to add more lquid to the tank. The response of
this system is shown graphically. As the level moves from O to 100
percent, the valve moves from fully open to fully closed. The function
of an automatic controller is to produce this kind of opposing response
over varying ranges. In addition, other responses are available to control the process more efficiently.
Selecting

Controller

Action

Depending on the action of the valve, increases in measurement mar

require either increasing or decreasing outputs for control. All controllers can be switched between direct and reverse action. Direct action
means that, when the controller sees an increasing signal from the
transmitter, its output will increase. For reverse action, increasing
measurement signals cause the controller output to decrease. To determine which of these responses is correct, an analysis of the loop is
required. The first step is to determine the action of the valve.
In Figure 1-1, for safety reasons the valve must shut if there is a
failure in the plant air supply. Therefore, this valve must be air-toopen, or fail-closed. Second, consider the effect of a change in measurement. For increasing temperature, the steam ftow to the beat exchanger should be reduced; therefore, the valve must close. To close
this valve, the signal from the automatic controller to the valve must

decrease. Therefore, this controller requires reverse, or increase/

decrease, action. Ifdirect action is selected, increasing signals from the
transmitter will result in a larger steam ftow, causing the temperature to
increase further. The result would be a runaway temperature. The
same thing will occur on any decrease in temperature, causing a falling
temperature. Incorrect selection of the action of the controller always
results in an unstable controlloop as soon as the controller is put into
automatic.
Assuming that the proper action is selected on the controller, how
does the controller know when the proper output has been reached? In
Figure 1-10, for example, to keep the level constant, a controller must
manipulate the ftow in to equal the ftow out. Any difference will cause
the level to change. In other words, the ftow in, or supply, must balance
the ftow out, or demandoThe controller performs its job by maintaining
this balance at a steady rate, and acting to restore this balance between
supply and demand whenever it is upset.

INTRODUCTION TO PROCESS CONTROL

17

U psets

There are three conditions that require different ftows to maintain the
level in the tank. First, if the position of the output hand valve is
opened slightly, more ftow leaves the tank, causing the level to fall.
This is a change in demand, and to restore balance, the inlet ftow valve
must be opened to supply a greater ftow rate. A second type of unbalanced condition is a change in the set point. Maintaining any other level
besides midscale in the tank causes a different ftow out. This change in
demand requires a different input valve position. The third type of
up set is a change in the supply. If the pressure output of the pump
increases, even though the inlet valve remains in the same position, the
increased pressure causes a greater ftow, which at first causes the level
to begin to rise. Sensingthe increased measurement, the level controller must close the val ve on the inlet to hold the level at a constant value.
In the same way, any controller applied to the beat exchanger shown in
Figure 1-1 must balance the supply ofheat added by the steam with the
beat removed by the water. The temperature remains constant if the
ftow of beat in equals the ftow of beat out.

Process

Characteristics

and Controllability

The automatic controller uses changes in the position of the final actuator to control the measurement signal, moving the actuator to oppose any change it sees in the measurement signal. The controllability
of any process depends on the efficiency of the measurement signal
response to these changes in the controller output. For proper control,
the measurement should begin to respond quickly, but then Dot change
too rapidly. Because of the tremendous number of applications of automatic control, characterizing a process by what it does, or by industry, is an almost hopeless task. However, all processes can be described by the relationship between their inputs and outputs. Figure
1-11 illustrates the temperature response of the beat exchanger when
the control valve is opened by manually increasing the controller output signal.
At first, there is no immediate response at the temperature indication. Then the temperature begins to change, steeply at first, then approaching a final, constant level. The process can be characterized by
the two elements of its response. The first element is the dead time, or
the time before the measurement begins to respond. In this example, a
delay arises because the beat in the steam must be conducted to the

18 FEEDBACK PROCESSCONTROL

STEAM FLOW

(MANUAL)
I

OUTLET

WATER

TEMPERATURE

L~,I"/"""'
TIME

Fig. l-Il. Responseof beatexchangerto stepupset.

water before it can affect the temperature, and then to the transmitter
before the change can be seen. Dead time is a function of the physical
dimensions of a process and such things as belt speeds and mixing
rates. Second, the capacity of a process is the material or energy that
must enter or leave the process to change the measurements-for example, the gallons necessary to change level, the Btu's necessary to
change temperature, or the standard cubic feet of gas necessary to
change pressure. The measure of a capacity is its response to a step
input. Specifically, the size of a capacity is measured by its time constant, which is defined as the time necessary to complete 63 percent of
its total response. The time constant is a function of the size of the
process and the rate of material or energy transfer. For this example,
the larger the tank and the smaller the ftow rate of the steam, the longer
the time constant. These numbers can be as short as a few seconds, or
as long as several hours. Combined with dead time, they define the time
it takes the measurement signal to respond to changes in the valve
position. A process will begin to respond quickly, but then not change
too rapidly, if its dead time is small and its capacity is large. In short,
the larger the time constant of capacity compared to the dead time, the
better the controllability of the processo

Controli er Responses

The first and most basic characteristico the controller responsehas

beenshownto be either direct or reverseaction. Oncethis distinction

-..TIME

INTRODUCTION TO PROCESS CONTROL

19

has been made, severa! types o responses are used to control a processoThese are (1) on/otI, two-position, control, (2) proportional action, (3) integral action (reset), and (4) derivative action.
On/off Control

On/off control is illustrated in Figure 1-12for a reverse-acting controller

and an air-to-close valve. An on/off controller has only two outputs,
either full maximum or full minimum. For this system, it has been
determined that, when the measurement falls below the set point, the
valve must be closed to cause it to increase. Thus, whenever the signal
to the automatic controller is below the set point, the controller output
will be 100 percent. As the measurement crosses the set point, the
controller output goes to O percent. This eventually causes the measurement to decrease, and as the measurement again crosses the set
point, the output goes to maximum. This cycle will continue indefinitely because the controller cannot balance the supply against the
load. This continuo us oscillation mar or mar Dot be acceptable, depending on the amplitude and length ofthe cycle. Rapid cycling causes
freqent upsets to the plant supply system and excessive valve wear.
The time of each cycle depends on the dead time in the process because
the dead time determines the time it takes for the measurement signal
to reverse its direction once it crosses 'the set point and the output of
the controller changes. The amplitude of the signal depends on how
rapidly the measurement signal changes during each cycle. On large
capacity processes, such as temperature vats, the large capacity causes

MEASUREMENT

100
% SIGNAL TO
VALVE

Fig. 1-12. On-off control for reverse-actingcontrollerand air-to-closevalve.

20 FEEDBACK PROCESSCONTROL

Fig. 1-13.Automatic controller with artificial signal.

a long time constant. Therefore, the measurement can change only

slowly. As a result, the cycle occurs within a very narrow band around
the set point, and this control mar be quite acceptable, if the cycle is
Dot too rapid. The on/off control is the one used most frequently in
commercial and domestic installations. However, if the process measurement is more responsive to changes in the supply, the amplitude
and frequency ofthe cycle begins to increase. At some point, this cycle
will become unacceptable and some Corm of proportional control will
be required.
In order to study the remaining three modes of automatic control,
open-loop responses will be used. Open-Ioop means that only the response ofthe controller will be considered. Figure 1-13shows an automatic controller with an artificial signal Croma manual regulator introduced as the measurement. The set point is introduced normally and
the output is recorded. With this arrangement, the specific controller
responses to any desired change in measurement can be observed.
Proportional

Action

Proportional response is the basis for the three-mode controller. If the

other two, integral and derivative are present, they are added to the
proportional response. "Proportional" means that the percent change
in the output of the controller is some multiple of the percent change in
the measurement.
This multiple is caUedthe "gain" ofthe controUer. For some controUers, proportional action is adjusted by such a gain adjustment,
while for others a "proportional band" adjustment is used. Both bave
the same purposes and effect. (See Appendix for table showing the
controUer adjustments from one manufacturer to another.) Figure 1-14
iUustrates the response of a proportional controUer from an inputl
output pointer pivoting on one ofthree positions. With the pivot in the
center between the input and the output graph, 100 percent change in

INTRODucnON TO PROCESSCONTROL 21

% OUTPUT

Fig. 1-14.Responseof proportionalcontroller, input to output, for three proportional

band values.

measurement is required to obtain 100percent change in output, or full

va1vetrave1. A controller adjusted to respond in this way is said to bave
a 100 percent proportiona1 band. When the pivot is moved to the righthand position, the measurement input wou1d need to change by 200
percent in order to obtain full output change from Oto 100percent. This
is c'alled a 200 percent proportiona1 band. Finally, if the pivot were in
the 1eft-hand position, and if the measurement moved over on1y 50
percent of the sca1e,the output wou1dchange over 100 percent of the
scale. This is called a 50 percent proportiona1 band. Thus, the smaller
the proportional band, the smaller amount the measurement must
change to cause full valve travel. In other words, the smaller the proportional band, the greater the output change for the same size measurement change. This relationship is illustrated in Figure 1-15. This
graph shows how the controller output will respond as a measurement

Fig. 1-15. ProportionaI diagram.

22 FEEDBACK PROCESSCONTROL

deviates from set paint. Each line on the graph represents a particular
adjustment of the proportional band. Two basic properties of proportional control can be observed from this graph:
1. For every value of proportional band, whenever the measurement
equals the set paint, the normal output is 50 percent.
2. Each value of the proportional band defines a unique relationship
between measurement and output. For every measurement value
there is a specific output value. For example, using the 100percent
proportional band line, whenever the measurement is 25 percent
above the set paint, the output from the controller must be 25
percent. The output from the controller can be 25 percent only if
the measurement is 25 percent above the set paint. In the same
way, whenever the output from the controller is 25 percent, the
measurement will be 25 percent above the set paint. In short, there
is one specific output value for every measurement value.
For any process controlloop, only one value of the proportional
band is the best. As the proportional band is reduced, the controller
response to any change in measurementbecomes increasingly greater.
At some paint, depending on the characteristic of each particular process, the response in the controller will be large enough to drive the
measurement back so far in the opposite direction as to cause constant
cycling of the measurement. This proportional band value, known as
the ultimate proportional band, is a lmit on the adjustment of the
controller in that loop. On the other hand, if too wide a proportional
band is used, the controller response to any change in measurement is
too small and the measurement is Dot controlled as tightly as possible.
The determination of the proper proportional band for any application
is part of the tuning procedure for that loop. Proper adjustment of the
proportional band can be observed by the response ofthe measurement
to an upset. Figure 1-16shows several examples ofvarying the proportional band for the beat exchanger.
Ideally, the proper proportional band will produce one-quarter amplitude damping, in which each hal cycle is one-half the amplitude of
the previous hal cycle. The proportional band that will cause onequarter wave damping will be smaller, thereby yielding tighter control
over the measured variable, as the dead time in the process decreases
and the capacity increases.
One consequence of the application of proportional control to the
basic controlloop is offset. Offset means that the controller will maintain the measurement at avalue different from the set paint. This is
most easily seen in Figure 1-10. Note that ifthe load valve is opened,

INTRODUCTIONTO PROCESSCONTROL 23

WATER FLOW
(LOAO)

i
I

1~
TEMPERATURE

-j/'\/,~"-",,,
I

,-~

PROPORTI ONA L BAND

TOOWIOE

CHANGE IN
UEMANO

B
C

TOO NARROW
CORRECT

TIME

Fig. 1-16. Examples of varying proportional band for beat exchanger.

ftow will increase through the valve and the val ve would bave to open.
But note that, because of the proportional action of the linkage, the
increased open position can be achieved only at a lowered level. Stated
another way, in order to restore balance between the ftow in and the
ftow out, the level must stabilize at avalue below the set point. This
difference, which will be maintained by the control loop, is called
offset, and is characteristic of the application of proportional-only control to feedback loops. The acceptability ofproportional-only control
depends on whether this offset can be tolerated. Since the error necessary to produce any output decreases with the proportional band, the
narrower the proportional band, the less the offset. For large capacity,
small dead time applications accepting a very narrow proportional
band, proportional-only control will probably be satisfactory, since the
measurement will remain within a small percentage band around the set
point.
If it is essential that there be no steady state difference between
measurement and set point under all load conditions, an additional
function must be added to the controller. This function is called integral
action (an older term is reset).
Integral

Action

(Reset)

The open-loop response of the integral mode is shown in Figure 1-17,

which indicates a step change in the artificial measurement away from
the set point at some instant in time. As long as the measurement
remains at the set point, there is no change in the output due to the
integral mode in the controller. However, when any error exists between measurement and set point, the integral action will cause the
output to begin to change and continue to change as long as the error

24 FEEDBACK PROCESSCONTROL

% MEASUREMENT

L_-

SET
paiNT

I
I
I

...riME
Fig. l-I?

Open-loop response oCintegral mode.

exists. This function, then, causes the output to change until the proper
output is achieved in order to hold the measurement at the set paint at
various loads. This response is added to the proportional response of
the controller as shown in Figure 1-17. The step change in the measurement first causes a proportional response, and then an integral
response, which is added to the proportional. The more integral action
there is in the controller, the more quickly the output changes due to
the integral response. The integral adjustment determines how rapidly
the output changesas a function oftime. Among the various controllers
manufactured, the amount of integral action is measured in one of two
ways-either in minutes per repeat, or the number of repeats per minute. For controllers measuring integral action in minutes per repeat,
the integral time is the amount of time necessary for the integral mode
to repeat the open-loop response caused by proportional mode, for a
step change in error. Thus, for these controllers, the smaller the integral number, the greater the action of the integral mode. On controllers
that measure integral action in repeats per minute, the adjustment indicates how many repeats of the proportional action are generated by the
integral mode in one minute. Table 12-1 (p. 300) relates the controller
adjustments from one manufacturer to another. Thus, for these controllers, the higher the integral number, the greater the integral action.
Integral time is shown in Figure l-IS. The proper amount of integral
action depends on how fast the measurement can respond to the additional valve trave 1 it causes. The controller must nat drive the valve
faster than the dead time in the process allows the measurement to
respond, or the valve will reach its lmits before the measurement can
be brought back to the set paint. The valve will then remain in its
extreme position until the measurement crosses the set paint, whereupon the controller will drive the valve to its opposite extreme, where it

INTRpDUCTION TO PROCESSCONTROL 25

% MEASUREMENT

~
,

:+-RT

% OUTPUT

(I/D)

SET paiNT

-+1

RT. RESETTIME MIN./RPT.

::::::""J--r-

b~I~

I
I

a. b

TIME

Fig. l-IS. Open-loopresponseoCproportional plus integralmodes.

will remain until the measurementcrosses the set point in the opposite
direction. The result will be an integral cycle in which the valve travels
from one extreme to another as the measurementoscillates around the
set point. When integral action is applied in controllers on batch processes, where the measurement is away from the set point for long
periods between batches, the integral may drive the output to its maximum, resulting in "integral wind-up." When the next batch is started,
the output will Dot corne off its maximum until the measurement
crosses the set point, causing large overshoots. This problem can be
prevented by including a "batch function" in the controller, a function
specifically designed to prevent "wind-up."

Derivative

Action

The third response found on controllers is the derivative mode.

Whereasthe proportional mode respondsto the size of the error and
the integral mode respondsto the size and time duration of the error,
the derivative moderespondsto how quickly the error is changing.ln
Figure 1-19,two derivative responsesare shown.The first is a response
to a stepchangeof the measurementaway from the set point. For a
step, the measurementis changinginfinitely fast, and the derivative
mode in the controller causesa considerablechangeor spike in the
output, which dies immediatelybecausethe measurement
hasstopped
changingarterthe step. The secondresponseshowsthe responseofthe
derivative mode to a measurementthat is changingat a constantrate.
The derivative output is proportionalto the rate of changeofthis error.
The greaterthe rate of change,the greaterthe output dueto the deriva-

...TIME

26 FEEDBACK PROCESSCONTROL

% MEASUREMENT

-_J==~

--~

"~!

-t__J-

% OUTPUT

(IlO)

+TIME J...

~
Fig. 1-19. Two derivative responses.

tive response. The derivative holds this output as long as the measurement is changing. As soon as the measurementstops changing, regardless owhether it is at the set point, above or below it, the response due
to derivative action will cease. Among all brands o controllers, derivative response is commonly measured in minutes, as shown in Figure
1-20. The derivative time in minutes is the time that the open-loop,
proportional-plus-derivative response, is ahead o the response due to
proportional action alone. Thus, the greater the derivative number, the
greater the derivative response. Changes in the error are the result o
changes in either the set point or the measurement, or both. To avoid a
large output spike caused by step changes in the set point, most modem controllers apply derivative action only to ehanges in the measurement. Derivative action in controllers helps to control processes with
especially large time constants. Derivative action is unnecessary on

""

% MEASUREMENT

DT' DERIVATIVE TIME, MINUTES

,,

DT""
I

:.:::~J
% OUTPUT

r-

::::::::::~'
.f"

PROPORTIONAL ONLY

PROPORTIONAL + DERIVATIVE

Fig. 1-20. Open-loop response oCproportional plus derivative modes.

INTRODUCTIONTO PROCESSCONTROL 27

processes thai respond fairly quickly to valve motion, and cannot be

used at all on processes with noise in the measurement signal, such as
flow, since the derivative in the controller will respond to the rapid
changes in measurement it sees in the noise. This will cause large and
rapid variations in the controller output, which will keep the valve
constantly moving up and down, wearing the valve and causing the
measurement to cycle.
As previously described, the response of an element is commonly
expressed in terms of a time constant, defined as the time thai will
elapse until an exponential curve reaches 63.2 percent ora step change
in input. Although the transmitter does nat bave an exact exponential
response, the time constant for the pneumatic temperature transmitter
and its associated thermal system is approximately 2 seconds.
The reaction curve shown in Figure 1-21 incorporates the response
of the beat exchanger and measuring system, and it represents the
signal thai will actually reach the controller. This curve indicates thai,
given a sudden change inside the beat exchanger, more than 30 seconds
will elapse before the controller receives a signal thai is a true representation of thai change. From the reaction curve, or characteristic, we
can determine the type of controller required for satisfactory control
under ibis difficult, but common, delayed response characteristic.
Selecting

the Controller

The beat exchanger acts as a small-capacity process; thai is, a small

change in steam can cause a large change in temperature. Accurate

Fig. 1-21.The processreactioncurve is obtainedby imposinga stepchangeat input.

28 FEEDBACK PROCESSCONTROL

regulation of processes such as this calls for proportional rather than

on/otI control.
Variations in water rate cause load changes that produce offset, as
described previously. Thus, the integral mode should also be used.
Whether or Dot to include the derivative mode requires additional
investigation of the process characteristic. Referring to the reaction
curve (Figure 1-19), notice that the straight line tangent to the curve at
the point of inftection is continued back to the 150For 66C (starting)
level. The time interval between the start of the upset and the intersection of the tangentialline is marked TA;the time interval from this point
to the point of inftection is TB.IfTB exceeds TA,some derivative action
will prove advantageous. If TBis less than TA,derivative action mar
lead to instability because of the lags involved.
The reaction curve of Figure 1-21 clearly indicates that some derivative will improve control action. Thus, a three-mode controller with
proportional, integral, and derivative modes satisfies the needs of the
beat exchanger processo
Figure 1-22shows the combined proportional, integral, and deriva-

Fig. 1-22. Open-loopresponseofthree-modecontroller.

INTRODUCTIONTO PROCESSCONTROL 29

tive responsesto a simulated' beat exchanger temperature measurement

thai deviates from the set point due to a load change. When the measurement begins to deviate from the set point, the tirst response from
the controller is a derivative response proportional to the rate of change
of measurement thai opposes the movement of the measurement away
from the set point. This derivative response is combined with the proportional response. In addition, as the integral mode in the controller
sees the error increase, it drives the valve further still. This action
continues until the measurement stops changing, at which point the
derivative response ceases. Since there is still an error, the measurement continues to change due to integral action, until the measurement
begins to move back toward the set point. As soon as the measurement
begins to move back toward the set point, there is a derivative response
proportional to the rate of change in the measurement opposing the
return of the measurement toward the set point. The integral response
continues, because there is still error, although its contribution decreases with the error. AIso, the output due to proportional is changing.
Thus, the measurement comes back toward the set point. As soon as
the measurement reaches the set point and stops changing, derivative
response again ceases and the proportional output returns to 50 percent. With the measurement back at the set point, there is no longer any
changing response due to integral action. However, the output is at a
new value. This new value is the reguli ofthe integral action during the
time thai the measurement was away from the set point, and compengates for the load change thai caused the original upset.

Conclusion

This chapter has describedthe responsesof a three-modecontroller

when it is used in the feedbackcontrol of industrial measurements.
The
readershouldbave a clear understandingof the following points:
1. In order to achieve automatic control, the control loop musi be
closed.
2. In order to maintainastable feedbackcontrolloop, the most important adjustmentto the controller is the selectionof the proper
action, either reverseor direct, on the controller. Properselection
of tbis action will causethe controller output to changein sucha
way thai the movementofthe valve will opposeany changein the
measurementseenby the controller.
3. The proper value of the settingsof the proportionalband, the integral mode, and derivative time dependson the characteristicsof

30 FEEDBACK PROCESSCONTROL

the processoThe proportional band is the basic tuning adjustment on

the controller. The more narrow the proportional band, the more
the controller reacts to changes in the measurement. If too narrow
a proportional band is used, the measurement cycles excessively.
If too wide a proportional band is used, the measurement will
wander and the offset will be too large.4.
The function of the integral mode is to eliminate offset. If too much
integral action is used, the res uIt will be an oscillation of the measurement as the controller drives the valve from one extreme to the
other. If too little integral action is used, the measurement will
retum to the set point too slowly.5.
The derivative mode opposes any change in the measurement. Too
little derivative action has no significant effect. Too much derivative action causes excessive response of the controller and cycling
in the measurement.

Questions

l-I. All control systemsthat fit into the usual patternare:

a. Open-Ioop
c. Closed-loop
b. Nonself-regulating
d. On/off
1-2. If operatingproperly,automaticcontrol will always:
a. Reducemanpower
b. Reducecosts
c. Make the processoperatemore uniformly
d. Decreasemaintenance
1-3. Automatic controllersoperateon the differencebetweenset point and
measurement,which is called:
a. Offset
c. Error
b. Bias
d. Feedback
1-4. A two-positioncontroller (on/oil) always:
a. Controls with a fixed offset
b. Controlsarounda point
c. Automaticallyadjustsits integraltime
d. Requiresprecisetuning
1-5. Gainand proportionalbandsare:
a. Reciprocallyrelated
b. Two differentcontrol modes
c. Adjusted independentlyof one another
d. Controllerfunctionscalibratedin time units

INTR9DUCTION TO PROCESSCONTROL 31

1-6. Whenwe adjustintegraltime in a controller:

a. We determineanRC time constantin the controller's internalfeedback
path
b. We adjustthe time it will take for integralto equalderivative
c. We setthe processtime constantso that it will alwaysequal1
d. What happensspecificallydependson the type of controller,pneumatic
or electronic
1-7. Match the following:
Controller, two-position-a.
Derivative -b.
Deviation-c.
Final-controllingelement-d.
Proportionalband
adjustment-f.
Regulatedby control valve -g.
Reset-h.
Set point-

Gain
Rate
Integral
Controller, on/otI
e. Valve
Desiredvalue
Manipulatedvariable
Error

1-8. A proportionalcontroller will bave an otIsetditIerencebetweenset

point and control point:
a. At all times
b. Equal to the proportionalbandsetting
c. That dependsuponprocessload
d. That will eventuallyvanish
1-9. If it were possiblefor a proportionalcontrollerto bavea true Opercent
proportionalband, the controller gainwould bave to be:
a. Unity
c. 100
b. O
d. Infinite
1-10. If the proportionalbandof the controlleris adjustedto minimum
possiblevalue,the control actionis likely to be:
a. On/otI
c. Excellent
b. With maximumotIset
d. Inoperative
I-Il. The following symbol @
representsa:
8. Flow rate controller
b. Fixed control point

appearsin an instrumentdiagram.It
c. Frequencyconverter
d. Final control element

1-12. With a proportional-onlycontroller if measurement

equalsset point, the
output will be:
8. O

c. 50 percent

b. 100percent

d. Impossibleto define

1-13. If in a proportional-plus-integral
controller measurement
is away from
the set point for a long period,the controller's output will be:

32 FEEDBACK PROCESSCONTROL

8. Oor 100percent,dependingon actionselected

b. Unknown
c. O

d. 100percent
1-14. In the modemcontroller,derivative actionis appliedonly to the:
8. Error
c. Setpoint
b. Measurement
d. Integral circuit
1-15. The functionof the integral(reset)modeis to:
8. Opposechangein measurement
b. Automaticallyadjustthe controller's gain
c. Eliminate offset
d. Stabilizethe controlloop

Process/Pressure Measuring
Process/Pressure
Instruments
Instruments

Pressure is a universal
universal processing
processing condition.
condition. It is also
also a condition
oflife
Pressure
conditionoflife
on this planet:
at the
an atmospheric
that
atmospheric ocean
ocean that
on
planet: we live at
the bottom
bottom of an
extends upward
upward for many
miles. This mass
mass of air has
has weight,
weight, and
and this
extends
manymiles.
weight
pressing downward
downward causes
causes atmospheric
atmospheric pressure.
pressure. Water,
Water, a funfunweight pressing
damental
necessityof life, is supplied
most of us
supplied to most
us under
under pressure.
pressure. In
damental necessity
the typical process
process plant,
plant, pressure
pressure influences
influences boiling
boiling point
point temperatemperathe
tures, condensing
temperatures, process
process efficiency,
efficiency, costs,
costs, and
and
tures,
condensingpoint
point temperatures,
other important
important factors.
measurementand
control of pressure,
other
factors. The
The measurement
and control
pressure, or
lack of it-vacuum-in
it-vacuum-in the
plantis critical. Instruments
lack
the typical process
process plant
Instruments
are
How these
are available
available to measure
measure a wide
wide range
range of pressures.
pressures. How
these instruinstruments function
chapter.
ments
function is the subject
subjectof this chapter.
What
Is Pressure?
Pressure?
What Is

Pressure
Pressure is force dividel
divided by the area over which it is applied. Pressure
Pressure
is often defined in terms of
of "head."
"head." For example, assume
assume that we bave
have
a water column 1 foot square
square and 23
23 feet tall. We
We want to find the
pressure in the bottom
boUom ofthe
mar be
ofthe column. The weight ofthe
ofthe column may
calculated by first finding the volume of water. This is the area
of the
area of
33

34

FEEDBACK PROCESSCONTROL

base multiplied by height, or 1 times 23 equals 23 cubic feet. Water

weighs 62.43 pounds per cubic foot. So the weight of 23 cubic feet will
be 23 times 62.43, or 1,435.89pounds. The area ofthe base is 1 square
foot, or 12 inches times 12 inches, or 144square inches. The pressure
equals 1,434.89 divided by 144 equals 9.9715, or approximately 10
pounds per square inch. In practice, we find that only the height ofthe
water confits. It mar be present in a small pipe or beneath the surface
of a pond. In any case, at a depth of 23 feet, the pressure will amo unt to
approximately 10 pounds per square inch. If in your home the water
pressure is 50 pounds per square inch and the system uses a gravity
feed, the water tank, or reservoir, holds the water at a height of 50
divided by 10, or 5 times 23 equals 115 feet above the point where the
pressure measurement is made. Head and pressure, then, mar mean
the same thing. We must be able to convert from one to the other. You
mar encounter reference to inches of mercury for pressure measurement. Mercury is 13.596times as heavy as an equal volume of water.
Therefore, ahead ofmercuryexerts apressure 13.596times greaterthan
an equivalent head of water. Because it is hazardous, mercury no
longer is used commonly in manometers.
The head or pressure terms cited thus far are called, collectively,
"gauge pressure." For example, if a tire gauge is used to check the
pressure in your automobile tires, it measures gauge pressure. Gauge
pressure makes no allowance for the fact that on earth we exist under a
head of air, or an atmosphere. The height of this head of air varies with
elevation, and also to some degree with weather conditions. If rou ride
an elevator from the bottom to the top floor of a tall building, rou will
likely feel your ears "pop." This is caused by the change in atmospheric pressure.
A simple method of measuring atmospheric pressure would be to
take a length of small diameter (0.25 inches) glass tubing about 35
inches long, sealed at one end. Fill the tube entirely with mercury and
temporarily seal the end. Invert this end into a deep dish of mercury
and remo ve the seal. The result will be a column of mercury as shown
in Figure 2-1 with some space remaining at the top. Atmospheric
pressure on the surface of the exposed mercury will balance the height
of mercury in the tube and prevent it from running out of the tube. The
height of the mercury above the level in the dish is, then, a measure of
atmospheric pressure. At sea level, this would amount to approximately 29.9 inches, or 14.7pounds per square inch. When the etfect of
the atmosphere is included in our measurement, we then must use
absolute pressure (gauge pressure plus atmospheric pressure).

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 35

Fig. 2-1. Mercurybarometer.

Units ot Measurement

Every major country has adopted its own favorite units of measurement. The United States has traditionally employed the English system. However, international trade has made it necessaryto standardize
units of measurement throughout the world. Fortunately, during this
standardization, there has be en rationalization ofthe measurementsystem. This has led to the adoption of the System lnternatonal d'Untes
(SI), a metric system of units. The force of common usage is so strong
that the familiar English system will undoubtedly persist for many
years, but the changeover is definitely underway. The time will soon
corne when process industries will deal exclusively with SI units.

Pressure

Measurement

Perhaps the area that has caused the most concern in the change to SI
units is pressure measurement. The new unit of pressure, the pascal, is
unfamiliar even to those who have worked in the older CGS (centimetre, gram, second) metric system. Once it is accepted and understood,
it willlead to a great simplification of pressure measurement Cromthe
extremes of full vacuum to ultrahigh pressure. It will reduce the multiplicity ofunits now common in industry to one standard that is compatible with other measurements and calculations. To understand the pascal and its relationship to other units of pressure measurement, we
must return to a basic understanding of pressure.
As noted previously, pressure is force per unit area.

36 FEEDBACK PROCESSCONTROL

From Newton's laws, force is equal to magstimes acceleration. In

the English system, the distinction between mags and force became
blurred with common usage of terms such as weight and mags. We live
in an environment in which every object is subject to gravity. Every
object is accelerated toward the center of the earth, unless it is restrained. The force acting on each object is proportional to its mags. In
everyd ay terms ibis force is called the weight of the objecto
W= m' G

where:
W = weight of the object
m = its mass
G = accelerationdue to gravity
Because gravity on the earth's surface is roughly constant, it has
been easy to talk abolli a weight of 1 pound and a mass of 1 pound
interchangeably. However, in faci, force and mass, as quantities, are as
different as apples and pears, as the astronauts bave observed. A numbec of schemes bave be en devised to overcome ibis problem. For example, a quantity called the pound-force was invented and made equal
to the force on a mass of one pound under a specified acceleration due
to gravity. The very similarity between these two units led to more
confusiott. The pascal, by its definition, removes all these problems.
The Pascal
The SI unit of pressure is defined as the pressure or stress that arises
when a force of one newton (N) is applied uniformly over an area of one
square metre (m2). This pressure has been designated one pascal (Pa).
Thus, Pa = N/m2. This is a small unit, but the kilopascal (KPa),
1,000 pascals, and the megapascal (MPa), one million pascals, permit
easy expression of common pressures. The definition is simple, because gravity has been eliminated. The pascal is exactly the same at
every point, even on the moon, despite changes in gravitational acceleration.
In SI units, the unit offorce is derived from the basic unit for mass,
the kilogram (kg), and the unit of acceleration (metres per second per
second, mls2). The product of mass times acceleration is force and is

Gauge,

PROCESS/PRESSURE
MEASURING INSTRUMENTS 37

designated in newtons. One newton would be the force of one kilogram

accelerating at one metre per second per second (N = kg' mls2).
Bar versus Pascal

After the introduction of SI units, the use of the "bar" (lOSPa) gained
favor, especially in European industry, where it closely resembles the
CGS unit of kg/cm2 (kilograms per square centimetre). At thai time, the
SI unit was called the "newton per square metre." As well as being
quite a mouthful, it was found to be inconveniently small (one N/m2
equals 0.000145psi). The use ofthe millibar in meteorology lent weight
to the acceptance of the bar. However, the use of a multiple like (lOS)in
such an important measurement and the resulting incompatibility of
stress and pressure units led to the adoption ofthe N/m2, giving it a new
name, the pascal (Pa), in October 1971. The kilopascal (kPa), 1,000
pascals, equals 0.145 psi and most common pressures are thus expressed in kPa. The megapascal(mPa) equals 145 psi and is convenient
for expressing high pressures.
Absolute,

and Differential

Pressure

The pascal can be used in exactly the same way as the English or CGS
metric units. The pascal may be regarded as a "measuring gauge," the
size ofwhich has been defined and is constant. This gauge can be used
to measure pressure quantities relative to absolute vacuum. Used in
this way, the results will be in pascal (absolute). The gauge may also be
used to measure pressures relati ve to the prevailing atmospheric
pressure, and the results will be pascal (gauge). Ifthe gauge is used to
measure the difference between pressures, it becomes pascal (differen-

tial).
The use of gauge pressure is extremely important in industry, since
it is a measure ofthe stress within a vesseland the tendency offtuids to
leak auto It is really a special case of differential pressure measurement,
inside versus outside pressure. Where there is any doubt about whether
a pressure is gauge, differential, or absolute, it should be specified in
full. However, it is common practice to shaw gauge pressure without
specifying, and to specify by saying "absolute" or "differential" only
for absolute or differential pressures. The use of "g" as in psig is
disappearing, and the use of "a" as in psia is frowned upon. Neither g
flor a is recognized in SI unit symbols. However, M is recognized for
differential pressure in all units.

38 FEEDBACK PROCESSCONTROL

Understanding

the Effects

of Gravity

Before discussing the effects of gravity on pressure measurement, it is

well to keep in mind the size of error thai can arise if gravity is regarded
as constant.
The "standard" gravitational acceleration is 9.806650rn/S2.This is
an arbitrary figure selected as a near average of the actual acceleration
due to gravity found all over the earth. The following are typical values

at different places:
Melbourne (Australia)
Foxboro (USA)
Soest (The Netherlands)

9.79966 rn/S2
9.80368 rn/S2
9.81276 rn/S2

Hence the difference around the world is approximately :t 0.1 percent from the average. This is oflittle practicat importance in industrial
applications. However, with some transmitters being sold with a rated
accuracy of :t0.25 percent, it is well to consider the effect of a gravityinduced difference of more than hal the tolerance thai can arise if the
transmitter was calibrated in Europe and tested in Australia.
Gravity-Dependent

Units

Units such as psi, kglcm2, inches of water, and inches of mercury (Hg)
are all gravity dependent. The English unit pounds per square inch (psi)
is the pressure generated when the force of gravity acts on a mass of
one pound distributed aveT one square inch. Consider a dead weight
tester and a standard mass of one pound which is transported around
the earth's surface: the pressure at each paint on the earth will vary as
the gravitational acceleration varies. The same applies to units such as
inches of water and inches of mercury. The force at the bottom of each
column is proportional to the height, density, and gravitational acceleration.
Dead weight testers are primary pressure standards. They generate
pressure by applying weight to a piston that is supported by a fluid,
generally oil or air. By selecting the weights and the cross-sectional
aTea of the piston, the pressure generated in any gravity field can be
calculated. Therefore, dead weight testers are gravity dependent. For
accurate laboratory work, the gravity under which the tester was calibrated and that at the place of use must be taken into account. Similarly,
the pressure obtained by a certain height of fluid in a manometer depends on density and gravity. These factors must be corrected for the

PROCESS/PRE:SSURE
MEASURINGINSTRUMENTS 39

existing conditions ifprecision regulis are to be obtained. Factors given

in the conversion tables in the Appendix, it should be noted, deal with
units of force, noi weight. Dead weight testers will be discussed in
more detaillater in ibis chapter.
Gravity-lndependent

Units

While gravity plays no part in the definition of the pascal, it has the
same value wherever it is measured. Units such as pounds-force per
square inch and kilogram-force per square centimetre are also independent of gravity because a specific value of gravitational acceleration
was selected in defining these units.
Under equal gravity conditions, the pound-mass and pound-force
are numerically equal (which is the cause of considerable confusion).
Under nonstandard gravity conditions (the usual case), correction factors are required to compensate for the departure from standard. It
should be noted that the standard value of actual gravity acceleration is
llot recognized as such in the SI unit system, where only the SI unit of
acceleration of one metre per second per second is used. In the future,
only the measured actual gravity at the location of measurement (G)
will be used when gravity plays a part in the system under investigation. The pascal is a truly gravity-independent unit and will be used to
avoid the presently confusing question of whether a stated quantity is
gravity dependent.

Pressure

Standards

Now let us consider the calibration standards that are employed with
pressure-measuringinstruments and the basic instruments that are used
to measure pressure. It mar help to look at the ways in which the
standards for pressure calibration are established. You will recalI that
head is the same as pressure. A measure of head, then, can be a dependable measure of pressure. Perhaps the oldest, simplest, and, in
many respects, one of the most accurate and reliable ways of measuring pressure is the liquid manometer. Figure 2-2 shows a differential
manometer. When only a visual indication is needed and static
pressures are in a range that does Dot constitute a safety hazard, a
transparent tube is satisfactory. When conditions for the visual manometer are unsuitable, a variety of ftoat-type liquid manometers are
often employed.

40 FEEDBACK PROCESSCONTROL

HIGH
PRESSURE

Fig. 2-2. Simple U-tube manometer.

The simplest differential gauge is the liquid-filled manometer: it is

basic in its calibration and free of frictional effects. It is often used to
calibrate other instruments. The most elementary type is the U-gauge,
which consists of a glass tube bent in the form of a U, or two straight
glass tubes with a pressure connection at the bottom. When a differential pressure is applied, the difference between the heights of the two
columns of lquid is read on a scale graduated in inches, millimetres, or
other units. In the more advanced designs, vertical displacement of one
side of the manometer is suppressed by using a chamber of large surface area on that side. Figure 2-3 shows such a manometer. If the area
ratio is in the vicinity of 1,600 to 1, the displacement in the large
chamber becomes quite small and the reading on the glass tube will
become extremely close to true inches or true millimetres. The large
side would have to be of infinite area for the reading in the glass tube to
be exact. This problem is sometimes overcome with a special calibration of the scale. However, if the glass tube becomes broken and must
be replaced, the scale must be recalibrated.
A more common and quite reliable design features a zeroing gauge
glass as shown in Figure 2-4. The scale mar be adjusted to zero for each
differential pressure change, and the reading mar be taken from a scale
graduated in actual units of measurement arter rezeroing. The filling
LOW
PRESSURE

Fi2. 2-3. Well or reservoirmanometer.

PROCESS/PR;:SSURE
MEASURINGINSTRUMENTS 41

Fig. 2-4. Well manometerwith zeroingadjustment.

lquid is usually water or mercury, or some other stable fluid especially

compounded by the manometer manufacturers for use with their product. Incline manometer tubes, such as those shown in Figure 2-5, will
give magnified readings, but must be made and mounted carefully to
avoid errors due to the irregularities ofthe tube. It is also essential that
the manometers be precisely positioned to avoid errors due to level.
Still other types of manometers for functions other than simple
indication, including those used with high pressure and hazardous
fluids, employ a float on one leg of the manometer. When reading a
manometer, there are several potential sources of error. One is the
effect of gravity, and another is the effect of temperature on the material contained within the manometer. Correction tables are available
which provide the necessary correction for the conditions under which
the manometer is to be read. Perhaps even more important is the
meniscus correction (Figure 2-6). A meniscus surface should always be
read at its center-the bottom, in the case of water, and the top, in the
case of mercury. To be practical, gravity and temperature corrections
are seldom made in everyday work, but the meniscus correction, or
proper reading, must always be taken into account.

HIGH
PRESSURE

Fig. 2-5. Inclined manometer.

Fig.

42

FEEDBACK PROCESSCONTROL

LIQUID
GLASS

DOES NOT WEl

(MERCURY)

2-6. Readinga manometer.

The dead weight tester is shown in Figure 2-7. The principIe of a

dead weight is similar to thai of a balance. Gravity acts on a calibrated
weight, which in tom exerts a force on a known area. A known
pressure then exists throughout the fluid contained in the system. This
fluid is generally a suitable oil. Good accuracy is possible, bot requires
thai several factors be well established: (1) the piston area; (2) the
weight precision; (3) gravity corrections for the weight if the measurement is to be made in an elevation quite different Crom the original
calibration elevation; (4) buoyancy, since as each weight displaces its
volume of air, the air weight displaced should also be taken into consideration; (5) the absence offriction; (6) head oftransmitting fluid and (7)
operation technique (the weight should be spun to eliminate frictional
effect); unless the instrument being calibrated and the tester are at
precisely the same level, the head of material can contribute an appreciable error. Perhaps the most important part of the procedure is to
keep the piston floating. This is accomplished generally by spinning the
weight platform.

--- WEIGHT
DEAD WEIGHT

TESTER

Fig. 2-7. Ranges30psiandup: increasepressurewith crank unti! pressuresupportsan

accuratelyknownweight. An accuratetestgaugemar beusedwith hydraulic pumpin a
similarsetup.

TO

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 43

A properly operated dead weight tester should bave a pressure

output accurate to a fraction of a percent (actually, 0.1 percent of its
calibration or calibrated reading).
Still another type of dead weight tester is the pneumatic dead
weight tester. This is a self-regulating primary pressure standard. An
accurate calibrating pressure is produced by establishing equilibrium
between the air pressure on the underside of the ball against weights of
known mass on the top. A diagram of an Ametek pneumatic tester is
shown in Figure 2-8. In this construction, a precision ceramic ball is
ftoated within a tapered stainless steel nozzle. A ftow regulator introduces pressure under the ball, lifting it toward the annulus between the
ball and the nozzle. Equilibrium is achieved as soon as the ball beginsto
lift. The ball ftoats when the vented ftow equals the fixed ftow from
the supply regulator. This pressure, which is also the output pressure,
is proportional to the weight load. During operation, the ball is centered
with a dynamic film of air, eliminating physical contact between the ball
and the nozzle. When weights are added or removed fro~ the weightcarner,
the ball rises or drops, affecting the air ftow. The regulator
sensesthe change in ftow and adjusts the pressure beneath the ball to
bring the system into equilibrium, changing the output pressure accordingly. Thus, regulation of output pressure is automatic with change of
weight mass on the spherical piston or ball. The pneumatic dead weight

INSTRUMENT
UNOER TEST

Fig. 2-8. Pneumatictester. Courtesyof AMETEK Mfg. CD.

44 FEEDBACK PROCESSCONTROL

tester has an accuracy o :t 0.1 o 1 percent o indicated reading. It is

commonly used up to a maximum pressure o 30 psi or 200 kPa gauge.

Plant Instruments

That Measure

Pressure

Directly

Thus far in tbis chapter we bave been concemed with the detinition of
pressure, and some of the standards used bave been described. In the
plant, manometers and dead weight testers are used as standards for
comparison and calibration.
The working instruments in the plant usually include simple mechanical pressure gauges, precision pressure recorders and indicators,
and pneumatic and electronic pressure transmitters. A pressure transmitter makes a pressure measurementand generates either a pneumatic
or electrical signal output thai is proportional to the pressure being
sensed. We will discuss transmitters in detaillater in ibis chapter. Now
we will deal with the basic mechanical instruments used for pressure
measurement, how they operate and how they are calibrated. When the
amount of pressure to be measured is very small, the following instruments might be used.
Bell Instrument

This instrument measures the pressure difIerence in the compartment

on each gide ora bell-shaped chamber. Ifthe pressure to be measured is
gauge pressure, the lower compartment is vented to atmosphere. Ifthe
lower compartment is evacuated, the pressure measured will be in
absolute units. If the difIerential pressure is to be measured, the higher
pressure is applied to the top of the chamber and the lower pressure to
the bottom.
The bell chamber is shown in Figure 2-9. Pressure ranges as low as
Oto 1 inch (Oto 250 Pa) of water can be measured with this instrument.
Calibration adjustments are zero and span. The difficulty in reading a
manometer accurately to fractions of an inch are obvious, ret the manometer is the usual standard to which the bell difIerential instrument is
calibrated. The bell instrument finds applications where very low
pressures must be measured and recorded with reasonable accuracy.
Slack or Limp-Diaphragm

The slack or limp-diaphragminstrument is used when very small

pressuresare to be sensed.The most commonapplicationofthis gauge

Fig.

PROCESS/PRE$SURE
MEASURINGINSTRUMENTS 45

2.9. Bell instrument.

is measurement of fumace draft. The range of this type instrument is

CromO to 0.5 inches (125 Pa) of water to O to 100 inches (25 kPa) of
water above atmospheric pressure.
To make this instrument responsive to very small pressures, a large
aTea diaphragm is employed. This diaphragm is made of very thin,
treated nonporous leather, plastic, or rubber, and requires an extremely small force to deftect it. A spring is always used in combination
with the diaphragm to produce a deftection proportional to pressure.
Let us assume the instrument shown in Figure 2-10 is being used to
measure pressure. The low-pressure chamber on the top is vented tothe
atmosphere. The pressure to be measuredis applied to the high-pressure
chamber on the bottom. This causesthe diaphragm to move upward. It

Fig. 2-10. Slack or limp-diaphragm instrument.

46 FEEDBACK PROCESSCONTROL

will move until the force developed by pressure on the diaphragm is

equal to that applied by the calibrated spring. ln the process of achieving balance, the lever attached to the diaphragm is tipped and this
motion transmitted by the sealed link to the pointer. This instrument is
calibrated by means of zero and span adjustments. The span adjustment allows the ridged connection to the spring to be varied, thus
changing the spring constant. The shorter the spring the greater the
spring constant; thus, a greater force is required to deftect it. As the
spring is shortened, a higher pressure is required for full deftection of

the pointer. The zero adjustmentcontrols the spring's pivot point,

thereby shifting the Creeend of the calibrated spring and its attached
linkages. Thrning the zero adjustment changes the pointer position on
the scale, and it is normally adjusted to read zero scale with both the
high and low pressure chambers vented to the atmosphere. lf pressure
, is applied to both the high- and low-pressure chambers, the resultant
reading will be in differential pressure. This instrument is extremely
sensitive to overrange, and care should be taken to avoid this problem.
The other long-term difficulty that can develop is damage to the diaphragm. The diaphragm mar become stiff, develop leaks, or become
defective, which results in error. These difficulties mar be observed by
periodically inspecting the diaphragm.
Pressure Gauges
In the process plant, we frequently find simple pressure ganges scattered throughout the process and used to measure and indicate existing
pressures. Therefore, it is appropriate to devote some attention to the
operation of a simple pressure gauge.
The most common of all the pressure ganges utilizes the Bourdon
tube. The Bourdon tube was originally patented in 1840. In bis patent,
Eugene Bourdon stated: "I bave discovered that if a thin metallic tube
be ftattened out then bent or distorted Croma straight line, it has the
property of changing its Cormconsiderably when exposed to variations
of internal or external pressure. An increase of interna! pressure tends
to bring the tube to a straight cylindrical Corm, and the degree of
pressure is indicated by the amount of alteration in the Corm of the

tube."
Figure 2-11A shows such a tube. As pressure is applied internally,
the tube straightens out and returns to a cylindrical Corm. The excursion of the tube tip moves linearly with internal pressure and is
converted to pointer position with the mechanism shown. Once the

PROCESS/PRE;SSURE
MEASURINGINSTRUMENTS 47

Fig. 2-11. (A) Bourdontube. (B) Typical pressuregauge.

internal pressure is removed, the spring characteristic of the material

returns the tube to its original shape.
Ifthe Bourdon tube is overranged, that is, pressure is applied to the
point where it can no longer return to its original shape, the gauge mar
take a new set and its calibration becomes distorted. Whether the gauge
can be recalibrated depends on the extent ofthe overrange. A severely
overranged gauge will be ruined, whereas one that has been only
slightly overranged mar be recalibrated and reused.
Most gauges are designed to handle approximately 35 percent of
the upper range value as overrange without damage. Typically, a gauge
will exhibit some amo unt of hysteresis, that is, a difference due to
pressure moving the tube in the upscale direction versus the spring
characteristic of the tube moving it downscale. A typical gauge mar
also exhibit some amount of drift, that is, a departure from the true
reading due to changes over a long period in the physical properties of
the materials involved. All of these sources of error are typically included in the manufacturer' s statement of accuracy for the gauge. A
typical Bourdon tube gauge, carefully made with a Bourdon tube that
has been temperature-cycled or stress-relieved, will bave an accuracy
of :t 1 percent of its upper range value. A carefully made test gauge will
bave an accuracy of :t0.25 to 0.50 percent ofits upper range value. The
range of the gauge is normally selected so that it operates in the upper
part of the middle third of the scale.

48 FEEDBACK PROCESSCONTROL

In addition to the gauges used for visual observations of pressure

throughout the plant, the typical instrument shop wi11bave a number of
test gauges which are used as calibration standards. The gauge must be
vertical to read correctly. A typical pressure gauge is shown in Figure

2-11 B.
Liquid

or Steam Pressure

Measurement

Whena liquid pressureis measured,the piping is arrangedto prevent

entrappedvaporswhich mar causemeasurement
error. If it is impossible to avoid entrappingvapors, vents should be provided at all high
points in the line.
When steampressureis measured,the steamshouldbe prevented
from enteringthe Bourdontube. Otherwise,the hightemperaturemar
damagethe instrument.Ifthe gaugeis belowthe paint of measurement,
a "siphon" (a singleloop or pigtail in a vertical plane)is provided in the
pressureline to the gauge.A cock is installed in the lme betweenthe
loop and the gauge.In operation,the looptraps condensate,preventing
the steamfrom enteringthe gauge.
Seals and Purges
If the instrument is measuring a viscous, volatile, corrosive, or extremely bot or cold fluid, the use of a pressure seal or a purge is
essential to keep the fluid out of the instrument. Liquid seals are used
with corrosive fluids. The sealing fluid should be nonmiscible with the
measured fluids. Liquid purges are used on bot, viscous, volatile, or
corrosive lquids, or lquids containing solids in suspension. The purging lquid must be of such a nature that the small quantities required (in
the order of I gph) (4 litres per hour) will not injure the product or
processoMechanical seals will be described in detaillater in this chap-

ter.
Pulsation Dampener
If the instrument is intended for use with a fluid under pressure and
subject to excessive fluctuations or pulsations, a deadener or damper
should be installed. This will provide a steady reading and prolong the
life of the gauge.
Two other elements that use the Bourdon principIe are the spiral
(Figure 2-12) and helical (Figure 2-13). The spiral and helical are, in
effect, multitube Bourdon tubes. Spirals are commonly used for

Fig.

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 49

2-12. Spiral.

Fig. 2-13. Helical.

pressure ranges up to o to 200 psi or 1.4 MPa, and helicals are made to
measure pressures as high as Oto 80,000 psi or 550 MPa. The higher the
pressure to be measured, the thicker the walls ofthe tubing Cromwhich
the spiral or helical is constructed. The material used in the construction mar be bronze, beryllium copper, stainless steel, or a special NiSpan C alloy. Spirals and helicals are designed to provide a lever motion of approximately 45 degrees with full pressure applied. If this
motion is to be translated into pen or pointer position, it is common
practice to utilize a four-bar linkage, and this necessitates a special
calibration technique. If, instead of measuring gauge pressure, it is
necessary to read absolute pressure, the reading must make allowances
for the pressure of the atmosphere. This mar be done by utilizing an
absolute double spiral element. In tros element, two spirals are used.
One is evacuated and sealed; the second has the measured pressure
applied.
The evacuated sealed element makes a correction for atmospheric
pressure as read on the second element. Thus, the reading can be in
terms of absolute pressure, which is gauge plus the pressure of the
atmosphere, rather than gauge.
An absolute double spiral element of this type mar be used to
measure pressures up to 100 psi or 700 Kpa absolute. This element is
shown in Figure 2-14.
Metallic

Bellows

A bellows is an expandableelementmade up of a series of folds or

corrugationscalled convolutions(Figure 2-15).Wheninternalpressure
is appliedto the bellows, it expands.Becausea sizablearcais involved.

SO FEEDBACK PROCESS CONTROL

Fig. 2-14. Absolutedoublespiral.

the applied pressure develops a sizable force to actuate an indicating or

recording mechanism. (In some instruments, the bellows is placed in a
sealed can and the pressure is applied externally to the bellows. The
bellows will then compress in a fashion similar to the expansion just
described. Such an arrangement is shown in Figure 2-16).
A variety of materials are used to fabricate bellows, including
brass, beryllium copper, copper nickel alloys, phospher bronze, Monel,
steel, and stainless steel. Brass and stainless steel are the most commonly used. Metallic bellows are used from pressure ranges of a few

Fig. 2-15. Bellows receiver unit.

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 51

Fig. 2-16. Bellows in sealedcan.

ounces to many pounds per square inch. A bellows will develop many
times the power available from a helical, spiral, or Bourdon tube.
A bellows is typically rated in terms of its equivalent square inch
area. To create a linear relationship between the excursion of the bellows and the applied pressure, it is common practice to bave the bellows work in conjunction with a spring, rather than with the spring
characteristic of the metal within the bellows itself. Each bellows and
spring combination has what is called a spring rate. The springs used
with the bellows are usually either helical or spiral. Typically, the
spring rate of the helical spring is ien times or more thai of the bellows
material itself.
Using a spring with a bellows has several advantages over relying
on the spring characteristics of the bellows alone. The calibration procedure is simplified, since adjustments are made only on the spring.
Initial tension becomes zero adjustment and the number of active turos
becomes span adjustment. A spring constructed of stable material will
exhibit long-term stability thai is essential in any component.
.When a measurement of absolute pressure is to be made, a special
mechanism employing two separate bellows mar be used. It consists of
a measuring bellows and a compensating bellows, a mounting support,
and an output lever assembly (Figure 2-17). The measuring and compensating bellows are fastened to opposite ends of the fixed mounting
support: the free ends of both bellows are attached to a movable plate
mounted between them. The motion of ibis movable plate is a measure
of the difference in pressure between the two bellows.

52 FEEDBACK PROCESSCONTROL

Fig. 2-17. Absolutepressurebellows.

Since the compensating bellows is completely evacuated and

sealed, ibis motion is a measure of pressure above vacuum, or absolute
pressure applied to the measuring bellows. This movable plate is attached to the output lever assembly which, in turo, is linked to the
instrument pen, or pointer.
In many applications, the bellows expands very little, and the force
it exerts becomes its significant output. This technique is frequently
employed in force-balance mechanisms which will be discussed in
some detail in Chapter 8.

Pressure

Transmitters

Signal Transmissions

In the process plant, it is impractical to locate the control instruments

out in the plant near the processoIt is also true thai mostmeasurements
are Dot easily transmitted from some remote location. Pressure measiIrement is an exception, but if a high pressure measurement of some
dangerous chemical is to be indicated or recorded several hundred feet
from the point of measurement, a hazard will be created. The hazard
mar be from the pressure and/or from the chemical carried in the line.
To eliminate ibis problem, a signal transmission system was developed. In process instrumentation, ibis system is usually either pneumatic (air pressure) or electrical. Becauseibis chapter deals with pressure,
the pneumatic, or air pressure system, will be discussed first. Later it
will become evident thai the electrical transmitters perform a similar
function.

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 53

U sing the transmission system, it will be possible to install most of

the indicating, recording, and control instruments in one location. This
makes it practical for a mnimum number of operators to rUll the plant
efficiently.
When a transmission system is employed, the measurementis converted to a pneumatic signal by the transmitter scaled from O to 100
percent ofthe measured value. This transmitter is mounted close to the
point of measurement in the processo The transmitter output-air
pressure for a pneumatic transmitter-is piped to the recording or control instrument. The standard output range for a pneumatic transmitter
is 3 to 15 psi, 20 to 100kPa; or 0.2 to 1.0 bar or kg/cm2. These are the
standard signals that are almost universally used.
Let us take a closer look at what this signal means. Suppose we
have a field-mounted pressure transmitter that has been calibrated to a
pressure range of 100 psi to 500 psi (689.5 to 3447.5 kPa).
When the pressure being sensedis 100psi (69 kPa), the transmitter
is designed to produce an output of 3 psi air pressure (or the approximate SI equivalent 20 kPa). When the pressure sensedris es to 300 psi
(2068.5 kPa), or midscale, the outputwill climb to 9psi, or 60kPa, and at
top scale, 500 psi (3447.5 kPa), the signal output will be 15 psi, or 100kPa.
This signal is carried by tubing, usually V4-inchcopper or plastic, to
the control room, where it is either indicated, recorded, or fed into a
controller.
The receiving instrument typically employs a bellows element to
convert this signal into pen or pointer position. Or, if the signal is fed to
a controller, a bellows is also used to convert the signal for the use of
the controller. The live zero makes it possible to distinguish between
true zero and a dead instrument.
The top scale signal is high enoughto be useful without the possibility of creating hazards.
Pneumatic Recorders and Indicators
Pressure recorders and indicators were described earlier in this chapter. Pneumatic recorders and indicators differ only in that they always
operate from a standard 3 to 15 psi or 20 to 100 kPa signal. The
indicator scale, or recorder chart, mar be 1abeledO to 1,000 psi. This
would represent the pressure sensed by the measuring transmitter and
converted into the standard signat that is transmitted to the receiver.

54 FEEDBACK PROCESSCONTROL

The receiver converts the signat into a suitable pen or pointer position.
Because the scale is labeled in proper units, it is possible to read the

measuredpressure.
A typical pneumatic indicator is shown in Figure 2-18 (top) and its
operation mar be visualized by studying Figure 2-18 (bottom, right).
The input signal passes through an adjustable needle valve to provide damping, then continues to the receiver bellows. This bellows,

Fig. 2-18. Pneumaticindicator.

PROCESS/PRESSURE MEASURING INSTRUMENTS

55

acting in expansion, moves a fQrce plate. A spring opposing the bellows

provides zero and span adjustments. Regulating the amo unt of spring
used (its effective length) provides a span adjustment. Setting the initial
tension on the spring provides a zero adjustment.
The force plate is connected to a link thai drives the pointer arbor
assembly. Changing the length ofthe link by turning the Dut on the link
provides an angularity adjustment. The arm connecting the pointer and
the arbor is a rugged crushed tube designed to reduce torsional effects.
A takeup spring is also provided to reduce mechanical hysteresis.
Overrange protection is provided to prevent damage to the bellows or
pointer movement assembly for pressures up to approximately 30 psi or
about 200 kPa.
A pneumatic recorder also uses a receiver bellows assembly as
shown in Figure 2-19 (top). This receiver provides an unusually high
torque due to an effective bellows area of 1.1 square inches.
The input signal tirst passesthrough an adjustable damping restrictor. The output of this signat is the input to the receiver bellows assembly. The receiver consists of a heavy duty impact extruded aluminum
can containing a large brass bellows working in compression. The large
effective area of the bellows assures an extremely linear pressure-topen position relationship. The motion, created by the input signal, is
picked up by a conventionallink and transferred to an arbor. Calibration is easily accomplished by tuming zero, span, and angularity adjustments on ibis simple linkage system. All calibration adjustments are
accessible through a door (Figure 2-19) on the gide of the instrument
and mar be made while instrument is in operation. The arm interconnecting the pen and the arbor incorporates a rugged tubular member to
reduce torsional effects and takeup spring to reduce mechanical hysteresis. Overrange protection is provided to prevent damage to the
bellows or pen movement assembly for pressures up to 30 psi or about
200 kPa.
Mechanical

Pressure

Seals

Application
A sealedpressuresystemis usedwith a pressuremeasuringinstrument
to isolate corrosive or viscousproducts, or products that tend to solidify, from the measuringelementand its connectivetubing.

Definition

A sealedpressuresystemconsistsof a conventionalpressuremeasuring elementor a force-balancepressuretransmittercapsuleassembly

PROCESS/PRESSURE
MEASURINGINSTRUMENTS 57

connected, either directly orby capillary tubing, to a pressure seal as

seen in Figure 2-20. The system is solidly filled with a suitable lquid
transmission medium.
The seat itself mar take many forms, depending on process conditions, but consists of a pressure sensitive flexible member, the diaphragm, functioning as an isolating membrane, with a suitable method of
attachment to a process vessel or line.
Principie or Operation
Process pressure applied to the flexible member of the seal assembly
forces some of the filling fluid out of the seal cavity into the capillary
tubing and pressure measuring element, causing the element to expand
in proportion to the applied process pressure, thereby actuating a pen,
pointer, or transmitter mechanism. A sealed pressure system offers
high resolution and rapid response to pressure changes at the diaphragm. The spring rate ofthe flexible member musi be low when compared with the spring rate of the measuring element to ensure thai the
fill volume displacement will full stroke the measuring element for the
required pressure range. A low-diaphragm spring rate, coupled with
maximum fill volume displacement, is characteristic of the ideal system.

Fig. 2-20. SeaIconnected to 6-inch pressure gauge.

58 FEEDBACK PROCESSCONTROL

A sealed pressure system is somewhat similar to a liquid-filled

thermometer, the primary differences being a flexible rather than rgid
member at the process, and no initial filling pressure. The flexible
member ofthe seal should ideally accommodate any thermal expansion
of the filling medium without perceptible motion of the measuring element. A very stiff seal member, such as a Bourdon tube, combined with
a low-pressure (high-volume change) element, will produce marked
temperature effects Crom both varying ambient or elevated process

temperatures.
The Foxboro l3DMP Series pneumatic d/p Cell transmitters with
pressure seals (Figure 2-21) measure differential pressures in ranges of
O to 20 to O to 850 inches of water or O to 5 to O to 205 kPa at static
pressures Cromfull vacuum up to flange ralingo They transmit a proportional 3 to 15 psi or 20 to 100kPa or 4 to 20 mA dc signat to receivers
located up to several hundred yards or melers Cromthe point of measurement.

Fig. 2-21. Sealsconnectedto differentialpressuretransmitter.

PROCESS/PRESSURE MEASURING INSTRUMENTS

S9

FiUing Fluids
Ideally, the filling fluid used in a sealed pressure system should be
noncompressible, bave a high-boiling point, a low-freezing point, a low
coefficient of thermal expansion, and low viscosity. It should be noninjurious to the diaphragm and containing parts, and should Dot cause
spoilage in the event ofleakage. Silicone-based liquid is the most popular filling fluid.
The system is evacuated before the filling fluid is introduced. The
system musi be completely filled with fluid and free from any air pockets thai would contract or expand during operation, resulting in erroneous indications at the pen or pointer or in an output signat. The
degree of accuracy of any filled pressure system depends on the perfection of the filling operation.
Calibration

Techniques

The procedure for calibration of a pressure instrument consists of comparing the reading of the instrument being calibrated with a standard.
The instrument under calibration is then adjusted or manipulated to
make it agree with the standard. Success in calibration depends Dot
only on one' s ability to adjust the instrument, but on the quality of the
standard as well.
Field Standards
Field standards must be reasonably convenient to use and must satisfy
the accuracy requirements for the instrument under calibration. A 100inch water column, for example, is extremely accurate but Dot practical
to set up out in the plant. For practical reasons, we find that most field
standards are test ganges.The test gaugeis quite similar, in most cases,
to the regular Bourdon ganges. However, more care has gone into its
design, construction, and calibration, making it very accurate. A good
quality test gauge will be accurate to within ::\:0.25percent of its span.
This is adequate for most field use.
Under some conditions, a manometer mar be used in the field. This
usually occurs when a low-pressure range is to be calibrated and no
other suitable standard is readily available.
Portable Pneumatic Calibrator
All of the ingredients required to perform a calibration bave been combined into a single unit called a portable pneumatic calibrator. This unit

60

FEEDBACK PROCESSCONTROL

Fig. 2-22. Portablepneumaticcalibration.Courtesyo Wallace& Tieman.

contains an accurate pressure gauge or standard, along with a pressure

regulator and suitable manifold connections. The portable pneumatic
calibrator will accurately apply, hold, regulate, and measure gauge
pressure, differential pressure, or vacuum. The gauge case mar be
evacuated to make an absolute pressure gauge. The pointer in the
operation of the gauge in this calibrator differs in that it makes almost
two full revolutions in registering full scale, providing a scale length of
45 inches. This expanded scale makes the gauge very easy to read.
The calibrator is available in seven pressure ranges (SI unit
ranges). Of these, three are sized for checking 3 to 15 psi or 20 to 100
kPa pneumatic transmission instrumentation. The calibrator is shown
in both picture and schematic form in Figures 2-22 and 2-23. The air
switching arrangement makes it possible to use the calibrator as both a
source or signal and a precise gauge for readout. This portable pneumatic calibrator is manufactured by Wallace and Tieman, Inc., Belleville, New Jersey.
Force-Balance

Pneumatic

Pressure

Transmitter

A pneumaticpressuretransmitter sensesa pressureand converts that

pressureinto a pneumaticsignatthat mar be transmittedover a reason-

0/1"

PROCESS/PRE~SURE
MEASURINGINSTRUMENTS 61

~~

o~
~~
;c
~"
~m
~C

~~z
mO~
,,~~
cm~

z
~2
~~
~'"

~~~;~:~
;

l ,t'i'

" ",'
~, \c:
\\'i",:"

'

Fig. 2-23. Connectionsfor differentpressurereadouts(courtesyof Wallace& Tieman):

For GaugePressure:testpressureis appliedto the capsulethroughthe appropriateP
connection;the caseis opento atmospherethroughS.
For DifferentialPressure:high-testpressureis appliedto the capsulethroUghthe
appropriateP connection.Low-testpressureis appliedto the casethroughS.
For AbsolutePressure:testpressureis appliedto the capsulethroughthe appropriateP
connectionand the caseis continuouslysubjectedto full vacuumthroughS.
For Vacuum:the capsuleis opento atmospherethroughconnectionP; the caseis
connectedto test vacuumat S.
For Positiveand NegativePressures:testpressureis appliedto the capsulethroughthe
appropriateP connectionandthe caseis opento atmospherethroughS.

able distance. A receiving instrument is then employed to convert the

signat into a pen or pointer position or a measurement input signat to a
controller. The Foxboro Model llGL force-balance pneumatic
pressure transmitter is an example of a simple, straightforward device
that performs this service (Figure 2-24). Before its operation is discussed, the operation of two vital components must be described: the
ftapper and the nozzle, or detector; and the pneumatic amplifier, or
relay. These two mechanisms are found in nearly every pneumatic
instrument.
The ftapper and nozzle unit converts a small motion (position) or

62 FEEDBACK PROCESSCONTROL

FLE'~"

Fig. 2-24. ModelllGM pressure transmitter is a force-balance instrument that

measures pressure and transmits it as a proportional 3 to 15 psi or 20to 100kPa
pneumatic signa!.

force into an equivalent (proportional) pneumatic signal. Flapper

movement of only six ten-thousandths of an inch (0.0015 cm) will
change the nozzle pressure by 0.75 psi or 5.2 kPa. This small pressure
change applied to the pneumatic amplifier or relay becomes an
amplified change of 3 to 15 psi or 20 to 100kPa in the amplifier output.
Let us take a closer look at the operation of the alplifier or relay.
It is shown in a cross-sectional view in Figure 2-25.
Pneumatic

Relay

A relay is a pneumatic amplifier. Like its electronic counterpart, the

function of the relay is to convert a small change in the input signal (an
air pressure signal) to a large change in the output signal. Typically, a 1
psi or 7 kPa change in input will produce approximately a 12 psi or 80
kPa change in output.
The supply enters the relay through a port on the surface of the

PROCESS/PRE$SURE
MEASURINGINSTRUMENTS 63

INPUT
DIAPHRAGM
SUPPLY

STEM

VALVE

BALL VALVE
OUTPUT
EXHAUST

SPRING

Fig. 2-25. Cross-sectionalviewoCamplifier.

instrument on which the Telar is mounted. The input signal (nozzle

pressure) enters the Telar through another port and acts on the diaphragm. Snce the diaphragm is in contact with a stem valve, the two
move in unson.
As the input signal increases, the stem pushes against a ball valve
which in turo moves a flat spring, allowing the supply of air to enter the
Telar body. Further motion of the stem valve causes it to close off the
exhaust port. Thus, when the input pressure increases, the stem
(exhaust) valve closes and the supply valve opens; when the input
decreases, the stem valve opens and the supply valve closes. This
varies the pressure to the output.
Principie

ot Operation

The 11GM Pneumatic Transmitter (Figure 2-24) is a force-balance instrument that measures pressure and transmits it as a proportional 3 to
15 psi pneumatic signal (20 to 100kPa).
The pressure is applied to a bellows, causing the end of the bellows
to exert a force (through a connecting bracket) on the lower end of the
force bar. The metal diaphragm is a fulcrum for the force bar. The force
is transmitted through the ftexure connector to the range rod, which
pivots on the range adjustment wheel.
Any movement of the range rod causes a minute change in the
clearance between the ftapper and nozzle. This produces a change in
the optput pressure from the relay to the feedback bellows until the
force of the feedback bellows balances the pressure on the measure-

64 FEEDBACK PROCESSCONTROL

ment bellows. The output pressure which is established by this balance

is the transmitted signal and is proportional to the pressure IlPplied to
the measurement bellows.
Ifthe pressure
be measured
is high, such as 5,000 psi or 35.MPa,
a different
sensing to
element
is employed.
The pressure being measured is applied to a Bourdon tube. This
pressure tends to straighten the tube and causes a horizontal force to be
applied to the lower end of the force bar. The diaphragm seat serves as
both a fulcrum for the force bar and as a seal for the pressure chamber.
The force is transmitted through a ftexure connector to the range rod,
which pivots on the range adjustment wheel.
Any movement of the rangerod causes a minute change in the
clearance between the ftapper and nozzle. This produces a change in
the output pressure from the rely to the feedback bellows until the
force in the bellows balances the force created by the Bourdon tube.
The output pressure, which establishes the force-batancing, is the
transmitted pneumatic signal that is proportionat to the pressure being
measured. This signat is transmitted to a pneumatic receiver to record,
indicate, or control.
The calibration procedure for this transmitter is the same as for the
low pressure transmitter, but the clllibration pressure is developed with
a dead weight tester. For safety, oil or liquid, never air, should be used
for high pressure calibrations.
Pressure measurements found in some applications require that
absolute, rather than gauge pressure, be determined. To handle these
applications, the absolute pressure transmitter mar be used.
Absolute

Pressure Transmitter

The pressure being measured is applied to one side of a diaphragm in a

capsule. The space on the other side of the diaphragm is evacuated,
thus providing a zero absolute pressure reference (Figure 2-26).
The pressure exerts a force on the diaphragm thai is applied to the
lower end of the force bar. The diaphragm seal serves both as a fulcrum
for the force bar and as a seal for the pressure chamber. The force is
transmitted through the ftexure connector to the range bar, which
pivots on the range adjustment wheel.
Any movement of the range bar causes a minute change in the
clearance between the ftapper and nozzle. This produces a change in
the output pressure from the relay (Figure 2-26) to the feedback bellows

PROCESS/PRES:sURE
MEASURINGINSTRUMENTS 65

i'ig. 2-26. Pneumaticabsolutepressuretransmitter.

until the force in the bellows balances the force on the diaphragm

capsule.
The output pressure, which establishes the force-balane, is the
transmitted pneumatic signal, which is proportional to the 'absolute
pressure being measured. This signal is transmitted to a pneumatic
receiver to record, indicate, or control.

Ouestions
2-1. An ordinary commercial Bourdon gauge has a scale of O psi to 250 psi,
and an accuracy of:t 1.percent of span. If the gauge reads 1.75psi, within
what maximum and minimum values will the correct pressure fall?
8. 1.74to 1.76psi
c. 1.76.5to 1.78.5psi
b. 1.72.5to 1.77.5psi
d. 1.79to 1.80psi
2-2.
8.
b.
c.

A manometer, read carefully:

Always has zero error
Has an error caused by the liquid's impurity unless corrected
May bave an error caused by temperature and gravity effects on the
various components and so on.
d. Has an error that varies only with altitude

66 FEEDBACK PROCESSCONTROL

2-3.
a.
b.
c.
d.

Absolutepressuris:
Gaugepressureplus atmosphericpressure
Gaugepressurelessatmosphericpressure
Gaugepressureplus atmosphericpressuredivided by two
Always referencedto a point at the peakof Mt. Washington,NH

2-4. The pressureat the bottomof a pondwhere it is exactly46 feet deep

will be:
a. 100psi
c. 20 psi
b. 46 psi
d. 20 psi absolute
2-5. The advantagesof makingabsolutepressuremeasurements
rather than
gaugeare:
a. Greateraccuracy
b. Eliminateserrors introducedby barometricvariations
c. Is more indicativeof true processconditions
d. Is more relatedto safetythangaugepressure
2-6. A bellows elementis usedas a receiverfor a 3 to 15psi pneumatic
signaland by error a 30-psipressureis appliedto it. The result will be:
a. A damagedbellows
b. An instrumentin needof recalibration
c. No damage
d. A severezero shift
2-7. A pressureinstrumentis calibratedfrom 100to 600psi. The spanof this
instrumentis:
a.600
c. 500
b. 100
d.400
2-8. Instrumentsthat measurepressureare generallyclassifiedas:
a. nonlinear
c. free of hysteresis
b. linear
d. Doneof the above
2-9. Whenreadinga manometer,it is goodpracticeto:
a. Readat the bottomof the meniscus
b. Readat the top of the meniscus
c. Read at the centerof the meniscus
d. Carefullyestimatethe averageof the meniscus
2-10. If a pressure range of O to 1 inch of water is to be measured and
recorded, an instrument capable of doing this would be:
a. A bellows
c. A Bourdon tube
b. A bell with a lquid seal
d. A mercury manometer

2-11. Whenmeasuringa pressurethat fluctuatesseverely:

a. A large-capacitytank shouldbe installed
b. A pulsationdampenershouldbe employed

PROCESSlPRESSURE
MEASURINGINSTRUMENTS 67

c. No problemis created
d. Readthe peakand minimumand divide by two to obtainthe true
pressure
2-12. A measurement
of absolutepressureis to be madeusinga mechanism
employingtwo separatebellows. Measurementis appliedto one bellowsand
the other:
8. Is sealedat an atmosphericpressureof 14.7psi
b. Is completelyevacuatedand sealed
c. Containsalcoholfor temperaturecompensation
d. Has an activeareatwice that of the measuringbellowsand is sealedat a
pressuretwice atmospheric
2-13. The dangerof havinga high-pressureline carryinga dangerous
chemicalrupture in the control room is:
8. Eliminated by usingspecialduty piping
b. Ignored
c. Eliminated throughthe useof a transmissionsystem
d. Minimized by placingthe line within a protectivebarrier
2-14. The standardpneumatictransmissionsignalmost generallyused in the
United Statesis:
8. 3 to 27 psi
c. 3 to 15psi
b. 10to 50 psi
d. 2 to 12psi
2-15. A sealedpressuresystem:
8. Is similar in someways to a liquid-filled thermometer
b. Sealsthe processmaterialin the instrument
c. Must be usedat a fixed temperature
d. Is alwaysusedwith manualtemperaturecompensation
2-16. A pneumaticrelay:
8. Is a setof electricalcontactspneumaticallyactuated
b. Is a signalbooster
c. Is a pneumaticamplifier
d. Containsa regulatoractuatedby a bellows
2-17. Whenthe clearancebetweenflapper and nozzlechangesby 0.0006
inchesthe output of the transmitterwill changeby:
8. 3 psi
c. 12psi
b. 15psi
d. 6 psi
2-18. An instrumentis to be calibratedto measurea rangeof Oto 6,000psi.
For sucha calibration:
8. oil or lquid shouldbe used
b. Air mustbe used
c. An inert gas suchas nitrogenis required
d. Any sourceof highpressureis acceptable

68 FEEDBACK PROCESSCONTROL

2-19. The kilopascal(kPa),an SI Ilnit of pressure:

a. Is always equivalentto psi x 6.895
b. Isgravity dependent
c. Is a force of 1.000newtons (N) applied uniformly over an areaof one
squaremetre
d. Changessubstantiallyfrom placeto place
2-20. A pneumaticpressuretransmitteris calibratedto a pressurerangeof
100to 500psi. The signaloutput is 10.2psi. Whatis the measuredpressurein
psi?
8. 272psi
c. 267psi
b. 340psi
d. 333psi
2-21. A viscoseline is located 100feet from the control roomwherea record
of line pressuremustbe made.The maximumpressureis approximately85
psi. Selectthe instrumentationthat canhandlethis job:
8. A sealedsystemwith 100feet of capillaryconnectedto a special
elementin a recordinginstrument.
b. A sealedsystemconnectedto a nearbypneumatictransmittersendinga
signalto a pneumaticreceivingrecorder
c. A direct connectedspiral elementrecorder
d. A pneumatictransmitterdirectly connectedto the line
2-22. A pressurerecorderreads38 psi and the barometerreads30.12inches
of mercury. The absolutepressureis
8. 23.21 psi absolute
c. 52.79psi absolute
b. 68.12psi absolute
d. 38 psi absolute
2-23. The lowestrangesof pressuremar be measuredby
8. a watercolumn
c. a bell meter
b. a bellows
d. a slackor limp-diaphragm
2-24. A slack diaphragmindicatoris to be calibratedOto 1 inch of water. The
standardwould likely be:
8. a water column
b. a columnof kerosene
c. a mercury-filledinclined manometer
d. a water-filledinclined manometer

Level and Density Measurements

Level Measurement
Measurement Methods
Methods
Level

The
The typical process
process plant
plant contains
contains many
many tanks,
tanks, vessels,
vessels, and
and reservoirs.
reservoirs.
measurement
store or process
process materials.
materials. Accurate
Accurate measurement
Their function
function is to store
of thesecontainers
materialin the tanks
of the contents
contents ofthese
containers is vital. The
The material
tanks is
usually
occasionallyit may
mar consist
consistof solids.
usually liquid, but occasionally
solids.
Initially,
level measurement
measurementappears
presenta simple
Initially, level
appears to present
simple problm.
problem.
However,
closerlook soon
soonreveals
variety of problems
that must
mustbe
However, a closer
reveals a variety
problems that
be
resolved.
materialmay
mar be
very corrosive;
mar tend
resolved. The material
be very
corrosive; it may
tend to solidify; it
mar tend
mar contain
contain solids;
mar create
other
may
tend to vaporize;
vaporize; it may
solids; or it may
create other
difficulties.
difficulties.
The
commonmethods
The common
methods employed
employed for automatic
automatic continuous
continuous liquid
level measurements
level
measurements are
are as
as follows
follows (see
(see also
also Table
Table 3-1):
3-1):
1.
Float-and-eable
1. Float-and-cable
2.
2. Displacement
Displacement(buoyancy)
(buoyancy)
3. Head
Head (pressure)
(pressure)
4. Capacitance
Capacitance
5. Conductance
5.
Conductance
6. Radiation
Radiation(nucleonic)
(nucleonic)
7. Weight
7.
Weight
8. Ultrasonic
8.
Ultrasonic
9. Thermal
9.
Thermal
69

70 FEEDBACK PROCESSCONTROL

Table3-1. Liquid Level Measurement

Closedor
Method
Bubbletube
Diaphragm
box
Head Pressure
Diff. Pressure
Displacement
(buoyancy)
Float-andCable
Weight
Radiation
(nucleonic)
Capacitance
Ultrasonic
Conductance

Available Upper
Range Values

Open

10 in to 250 ft
0.25 to 75 m
4 in to 250 ft
0.lt075m
2 in to many ft.
50 mm to many metres
5 in to many ft.
0.25 to many metres
6 in to 12 ft
0.15 to 3.6 m
3 in to 50 ft.
75 mm to 15 m
lnches to Feet
mm to m
Depends on Tank Dim.

Open

Any type including

Open

corrosive or dirty
Clean

Tank

Open
Both
Both

Condition
of Liquid

Any type with proper

selection
Any type with proper
selection
Any type with proper
selection

Open

Any type

Both

Any type

Wide

Both

Any type

Wide
Wide
One or more Points

80th

Nonconductive

80th
80th

Any type
Conductive

Float-and-Cable
A ftoat-and-cable or ftoat-and-tape instrument (Figure 3-1) measures
liquid level by transmitting to a mechanism the rise and fall of a ftoat
thai rides on the surface of the liquido Mechanisms are available to
accommodate level variations ranging from a few inches to many feet.
Float-and-cable devices are used primarily in open tanks, whereas ftoat
level switches mar be designed to operate in a pressurized tank.
Float devices bave the advantage of simplicity and are insensitive
to density changes. Their major disadvantage is their limitation to reasonably clean liquids. Thrbulence mar also create measurement problems. The ftoat and cable technique does Dot lend itselfto the transmitter concept as well as do some of the following techniques.

LF;VEL AND DENSITY MEASUREMENTS 71

Fig. 3-1. Float-and-cablerecorder.

Displacement

(Buoyancy)

The displacement, or buoyancy, technique is a type of force-balance

transmitter (Figure 3-2). It mar be used to measure liquid level, interface, or density by sensingthe buoyant force exerted on a displacer by
the liquid in which it is immersed. The buoyant force is converted by a
force-balance pneumatic or electronic mechanism to a proportional 3 to
15 psi, 20 to 100kPa, 4 to 20 mAldc or 10 to 50 mAldc signal.

Fig. 3-2. Buoyancy type level measuring transmitter.

72

FEEDBACK PROCESSCONTROL

Archimedes' principIe states that a body immersed in a liquid will

be buoyed upward by a force equal to the weight of the liquid displaced. The displacer element is a cylinder of constant cross-sectional
area and heavier than the liquid displaced. It must be slightly longer
than the level change to be measured (Figure 3-3). Since the displacer
must accurately sense the buoyant force throughout the full range of
measurement, displacer lengths and diameters wiU vary according to
the particular process requirements.
The formula to determine the buoyant force span for liquid level
applications is as foUows:

where:
f
V
Lw
L
B

=
=
=
=
=

buoyant force span (lbf or N)

total displacer volume (cubic inches or cubic centimeters)
working length of displacer (inches or millimeters)
total displacer length (inches or millimeters)
a constant (weight of unit volume of water) (0.036 Ibf/in3
or 9.8 x 10-3N/cm3
SG = specific gravity of the lquid
The minimum buoyant force span for the buoyancy level transmitters is 1.47 pounds-force or 6.7 newtons. The maximum mass of the
displacer-plus-hanger cable must not exceed 12 pounds or 5.4 kilograms (or buoyant force span x 6, whichever is smaller).
For a cylindrical displacer, the formula for the volume is:

Fig. 3-3. Buoyancytype level transmitterinstallation.

LEVE.L AND DENSITY MEASUREMENTS

73

In the above formula, as well as the sizing of displacers, the working

volume considered is .1imitedto the cylindrical volume of the displacer
and does not include the volume of the dished end pieces or hanger
connection.
U nless certain precautions are taken, buoyancy transmitters are
not recommended for extremely turbulent process conditions. The turbulence mar cause the displacer to swing erratically, resulting in unpredictable measurement effects or physical damage to the displacer,
transmitter, or vessel. In such cases, some form of displacer containment should be used.
Buoyancy transmitters mar be applied readily to glass-lined vessels, vessels in which a lower connection is not permissible or possible,
density applications with ftuctuating pressures or levels, and high-

temperature service. Interface level measurementmar be accomplished by permitting the interface level to vary over the length of the
displacer.
f = V (B) (SG difi)
SG difI = SG lower liquid minus the SG upper liquid
Head or Pressure

Measurementof head, or pressure,to determine level is the most

common approach.The ways in which level mar be determinedby
measuringpressureare many and varied. Some of the more common
techniques are presented befe.
Hubble Thbe Method
In the air purge, or hubble tube system (Figure 3-4), liquid level is
determined by measuring the pressure required to force a gas into the
liquid at a point beneath the surface. In this way, the level mar be
obtained without the liquid entering the piping or instrument.
A source of clean air or gas is connected through a restriction to a
hubble tube immersed a fixed depth in the tank. The restriction reduces
the air flow to a minute amount, which builds up pressure in the hubble
tube until it just balances the fluid pressure at the end of the hubble
tube. Thereafter, pressure is kept at this value by air bubbles escaping
through the liquido Changes in the measured level cause the air pressure in the hubble tube to build up or dropo A pressure instrument
connected at this point can be made to register the level or volume of
liquido A small V-notch is filed in the bottom of the tube so that air

74 FEEDBACK PROCESSCONTROL

Fig. 3-4. Usingbubblepipe.

emerges in a steady stream of small bubbles rather than in intermittent

large bubbles (Figure 3-5).
An advantage ofthe hubble tube method is that corrosive or solids
bearing lquids can damage only an inexpensive, easily replaced pipe.
Diaphragm Box
The diaphragm box is shown in Figure 3-6. It is similar to the diaphragm seals used for pressure gauges (Figure 2-20, p. 57) except that
the fill is air and the diaphragm is very slack, thin, and flexible. The

DETAlL

OF NOTCH

IN BUBBLE PIPE

Fig. 3-5. Detail or notch in bubblepipe.

LEYEL AND DENSITY MEASUREMENTS 75

Fig. 3-6. Diaphragrn boxo

diaphragmbox is usuallysuspendedfrom a chain. The instrumentthat

sensesthe pressurechangesand relatesthemto level mar be mounted
above or below the vessel.The diaphragmbox is used primarily for
water level measurementin openvessels.
Pressure and Differential Pressure Methods Using Differential Pressure
Transmitters
These are the most popular methods of measuring liquid level and are
shown in Figure 3-7. With an open tank, the pressure at the high
pressure side of the cell is a measure of the liquid level. With a closed
tank, the etfect of tank pressure on the measurement is nullified by
piping this pressure to the opposite side of the cell.
When a sealing fluid is used, it musi possess a specific gravity
higher than that of the liquid in the vessel. The sealing fluid must also
be immiscible with the liquid in the vessel.

76 FEEDBACK PROCESSCONTROL

OPEN TANK

Fig. 3-7.
Span = xGL
Suppression= yGL+ zG,

Span = xGL
Suppression= yGL+ zG,

Span = xGL
Elevation =/dG. -yGL

Where GL = specific gravity oCliquid in tank

G. = specific gravity oCliquid in outside filled line or tines
IC transmitter is at level oClower tank tap, or iC air purge is used, z = O.
(Note: The density oCthe gas in the tank has been disregarded in these calculations.)

EXAMPLE: Assume an open tank withX = 80 inches,y = 5 inches, and

z = 10 inches. The specific gravity ofthe tank liquid is 0.8; the specific
gravity of the liquid in the connecting leg is 0.9.
Span = 80 (0.8) = 64 inches head of water
Suppression = 5 (0.8) + 10 (0.9)
= 4 + 9 = 13 inches head of water
Range = 13 to 77 inches head of water
EXAMPLE:Assume a closed tank with X = 70 inches, y = 20 inches,
and d = 100 inches. The specific gravity of the tank liquid is 0.8; a
sealing liquid with a specific gravity of 0.9 is Ilsed.
Span = 70 (0.8) = 56 inches head of water
Elevation = 100 (0.9) -20
= 90 -16

(0.8)

= 74 inches head of water

Range = -74 to -18 inches head of water

LEVEL AND DENSITY MEASUREMENTS 77

Fig. 3-8. The repeateris a flanged,force-balance,pneumaticinstrumentthat measures

pressureand delivers an outputequalto the measuredpressure.

(minus sign indicates thai the higher pressure is applied to the low side
of the cell.)
At times the liquid to be measured possessescharacteristics thai
create special problems. Assume, for example, thai the liquid being
measured will solidify if it is applied to a wet leg and it is virtually
impossible to keep the leg dry. This type of application could use
a pressure repeater. The repeater (Figure 3-8) is mounted above
the maximum level of the liquid, and the liquid level transmitter is
mounted near the bottom ofthe tank. The pressure in the vapor section
is duplicated by the repeater and transmitted to the instrument below,
(Figure 3-9). Thus, the complications of a wet leg are avoided, and a
varying pressure in the tank will Dot atrect the liquid level measurement.
The pressure range thai mar be repeated is O to 100 psi or Oto 700
kPa. The supply pressure musi exceed the pressure to be repeated by

78 FEEDBACK PROCESSCONTROL

Fig. 3-9. Liquid-level transmitter(electronicor pneumatic)used with a repeater.

20 psi or 140kPa. A somewhatsimilar configurationmar be arranged

by usinga sealedsystemas describedin Chapter2 and shownin Figure
2-21,p. 58.
Capacitance
If a probe is inserted into a tank and the capacitance measured between
it and the tank, a sizable change in capacitance will occur with liquid
level. This phenomenon is due primarily to the substantial difIerence
between the dielectric constant o air and that o the liquid in the tank.
This technique is best applied to nonconductive liquids, since it is best
to avoid the problems generated by conducting materials like acids

(Figure 3-10).

Fig. 3-10. Capacitance.

LEVEL AND DENSITY MEASUREMENTS 79

Conductance
Conductivity level sensors consist of two electrodes inserted into the
vessel or tank to be measured. When the level rises high enough to
provide a conductive path from one electrode to the other, a relay (solid
state or coil) is energized. The relay mar be used for either alarm or
control purposes. Conductivity then becomes either point control or an
alarm point. The lquid involved must be a conductor and must Dot be
hazardous if a spark is created. Level by conductivity finds occasional
applications in process plant s (Figure 3-11).
Radiation

A radiation level measurement generally consists of a radioactive

source on one gide of the tank and a suitable detector on the other. As
the radiation passes through the tank, its intensity varies with the
amount of material in the tank and can be related to level. One advantage is that nothing comes in contact with the liquido Among the
disadvantages are the high cost and the difficulties associated with
radioactive materials. Radioactive techniques do bave the ability to
solve difficult level-measuring problems (Figure 3-12).
Weight

Occasionally the measurement or the contents or a tank is so difficult

that Done or the usual schemes will work. When this occurs it may be
advantageous to consider a weighing system. Weight cells, either hy-

LlNE

Fig. 3-11. Conductivity.

80 FEEDBACK PROCESSCONTROL

DETECTOR

~~~Jr;~-~~'-==-L__~
\~::::-

--

\~"
\ ::;"""""
\ \'

\\"

\"

INDICATOR

\,

-+

~-g

i~CONVERTER

NUCLEONIC(RADIATION)
Fig. 3-12. Radiationtechnique.

draulic or strain gauge, are used to weigh the vessel and its contents.
The tare weight of the tank is zero adjusted out of the reading, which
will result in a signal proportional to tank contents (Figure 3-13).
One advantage of the weight system is that there is no direct contact with the contents of the tank and the sensor. However, the system
is not economical, and varying densities mar confuse the relationship
between signal and true level.
Ultrasonic

The ultrasonic level sensor(Figure 3-14) consists of an ultrasonic

generator or oscillator operating at a frequency of approximately

Fig. 3-13. Level measurement using weight method.

LEVEL AND DENSITY MEASUREMENTS 81

Fig. 3-14. Ultrasonictechnique.

20,000 Hz and a receiver. The time required for the sound waves to
travel to the liquid and back to the receiver is carefully measured. The
time is a measure of level. This technique has excellent reliability and
good accuracy. Furthermore, nothing comes in contact with liquid in
the tank, which minimizes corrosion and contamination. The only generallimitation is economic.
Thermal

One approach employing a thermal sensorto determine level relies on a

difference in temperature between the lquid and air above it. When the
lquid contacts the sensor, a point determination o level is made.
Another approach depends on a difference in thermal conductivity
between the lquid level sensedand the air or vapor above it. Thermal
techniques are inexpensive, but bave achieved only modest popularity
in process applications.
Density

Measurement

Methods

for Liquids

and Liquid

Slurries

Density is defined as mags per unit volume. Specific gravity is the

unitless ratio of the density of a substance to the density of water at
some standard temperature. In Europe, this is known as relative den-

sity.
The measurement and control of liquid density are critical to a great
number of industrial processes. Although density can be of interest, it
is usually more important as an inference of composition, of concentration of chemicals in solution, or of solids in suspension.

82 FEEDBACK PROCESSCONTROL

Many methods are available to determine concentration, including

measurement of changes in electrolytic conductivity and of rise in boiling paint. This section deals with common industrial methods of
measuring density continuously (Table 3-2).
Agitation in a process tank where density is being measured musi
be sufficient to ensure uniformity ofthe lquid. Velocity effects musi be
avoided; agitation should be sufficient to maintain a uniform mixture
without affecting head pressure at measurement points.
Nearly aU industrialliquid density instruments utilize the effect of
density changes on weight as a measurement. Such instrumentation
falls into three major categories, defined by the principIes employed.
These include two other categories, radiation and vibration methods.
For density measurement, a cylindrical displacer is located so as to
be fuUy submerged. The buoyant force on the displacer changes only as
a function of changing lquid density, and is independent of changing
lquid level in the vessel. Buoyancy transmitters based upon the Archimedes principIe may be readily adapted to the measurement of
lquid density or specific gravity.
Hydrostatic Head
Proved suitable for many industrial processes is the method that continuously measures the pressure variations produced by a fixed height
of liquid (Figure 3-15). Briefly, the principIe is as follows: The ditIerence in pressure between any two elevations below the surface (A and
B) is equal to the ditIerence in liquid head pressure between these
elevations. This is true regardless of variation in level above elevation

TABLE 3-2. Liquid Density Measurement

Minimum
Method

Span

Hydrometer

0.1
0.005
0.05
0.05
O.O~
0.05

Displacer
Hydrostatic head
Radiation
Weight oCfixed volume
Vibrating U-Thbe

Condition
oi Liquid

Accuracy

Clean
Clean

:!:1%

Any
Any

Clean
Clean

as % Span

:!:1%
~ to 1%
1%
1%
1-3%

Fig.

LEVEL AND DENSITY MEASUREMENTS 83

lz=ZZarZz=
3-15. Fixed heightofliquid for densitymeasurement.

A. This difference in elevation is represented by dimensionH. Dimension

(H) must be multiplied by the specific gravity (G) ofthe liquid to obtain
the difference in head in inches or metres of water, which are the
standard units for instrument calibration.
To measure the change in head resulting from a change in specific
gravity from GI to G2, it is necessary only to multiply H by the difference between G2 and GI. Expressed mathematically:
P = H (G2 -GJ

(3-3)

The change in head (P) is differential pressure in inches or metres of

water. GI is mnimum specific gravity and G2 is maximum specific
gravity.
It is common practice to measure only the span of actual density
changes. Therefore, the measurement zero is suppressed; thai is, the
instrument "zero," or lower range value, is elevated to the minimum head pressure to be encountered. This allows the entire instrument measurement span to be devoted to the differential caused by
density changes. For example, ifG1 is 1.0 andH is 100inches, the span
of the measuring device musi be elevated 100 inches of water
(H x G). For a second example, if GI is 0.6 and H is 3 metres, the
zero suppression should be 1.8 metres of water.
The two principal relationships thai musi be considered in selecting
a measuring device are:

Vibration

84 FEEDBACK PROCESSCONTROL

Span = H X (G2,- GJ
Zero suppression = H x GI,

(3-4)

For a given instrument span, a low gravity span requires a greater H

dimension (a deeper tank).
Radiation

Density measurements by this method are based on the principIe that

absorption of gamma radiation increases with the increasing specific
gravity of the material measured.
The principal instrumentation includes a constant gamma source
(usually radium), a detector, and an indicating or recording instrument.
Variations in radiation passing through a fixed volume of ftowing process liquids are converted iota a proportional electrical signal by the

detector.

Damping a vibrating object in contact with the process fluid increases

as the density ofthe process fluid increases. This principIe is applied to
industrial density measurement. The object that vibrates (from externally applied energy) is usually an immersed reed or plate. A tube or
cylinder that conducts, or is filled with, the process fluid can also be
used. Density can be inferred from one of two measurements: changes
in the natural frequency of vibration when energy is applied constantly,
or changes in the amplitude of vibration when the object is "rung"
periodically, inthe "bell" mode.
Temperature

Effects and Considerations

The density of a liquid varies with expansion due to rising temperature,

but Dot allliquids expand at the same rate. Although a specific gravity
measurement must be corrected for temperature effects to be completely accurate in terms of standard reference conditions for density
and concentration, in most cases this is Dot a practical necessity.
For applications in which specific gravity measurement is extremely critical, it m~y be necessary to control temperature at a constant value. The necessary correction to the desired base temperature
can then be included in the density instrument calibration.

LEVEL AND DENSITY MEASUREMENTS 85

Differential

Pressure Transmitter

There are a variety of system arrangements for hydrostatic head density measurements with d/p Cell transmitters. Although ftange-mounted
d/p Cell transmitters are often preferred, pipe-connected transmitters
can be used on lquids where crystallization or precipitation in stagnant
pockets will noi occur.
These d/p Cell transmitter methods are usually superior to those
using bubble tubes. They can be applied wherever the vessel is high
enough to satisfy the mnimum transmitter span. They are also well
suited for pressure and vacuum applications.
Constant level overftow tanks permit the simplest instrumentation
as shown in Figure 3-16. Only one d/p Cell transmitter is required. With
H as the height of lquid above the transmitter, the equations are still:
Span = H X (G2

G}. zero suppression = H X GI

Applications with level and/or static pressure variations require compensation. There are three basic arrangements for density measurement
under these conditions. FiTSt, when a seal fluid can be chosen that is
always heavier than the process fluid and will Dot mix with it, the
method shown in Figure 3-17 is adequate. This method is used extensively on hydrocarbons with water in the wet leg.

Fig. 3-16. Constant level, open overftow tanks require only one d/p Cell transmitter for
density measurement.

Fig.

86 FEEDBACK PROCESSCONTROL

y-:'--=(

\
H

DENSITY
~
SIGNA L

A I

=
UQUID

a':iJ-l

PURGE

(SP.GR.=GS)
3-17. In an open or closed tank with varying level and/or pressure. a wet leg can be
filled with seal fluid heavier than the process liquido

For a wet leg fluid o specific gravity Gs. an elevated zero transmitter
musi be used. The equations become:
Span = H x (G2 -GJ,

zero elevation = H x (GB -GJ

When no seal or purge lquid can be tolerated, there are ways to provide a "mechanical seal" for the low-pressure leg, or for both legs, if
needed. Figure 3-18 shows the use of a pressure repeater for the upper
connection.
The repeater transmits the total pressure at elevation B to the low
pressure side of the d/p Cell transmitter. In this way, the pressure at
elevation B is subtracted from the pressure at elevation A. Therefore,
the lower transmitter measures density (or H x G, where G is the
specific gravity of the lquid). The equations for the lower transmitter

are:
Span = H X (G2 -GJ
Zero suppression = H x G1

The equationfor the upper repeateris:

Output (maximum) = d 1 (max) x G2 + P (max),

Fig.
whered1

LEVEL AND DENSITY MEASUREMENTS 87

3-18. In an open or closed tank with varying level and/or pressure where seal fluid
or purge is Rot suitable, a pressure repeater can be used.

is the distance from elevation B to the lquid surface, andP is

the static pressure on the tank, if any. When there is no pressure on the
tank, the upper repeater output is a measurement of level.
Transmitter locations should be as high as possible above the bottom ofthe tank. The process lquid then can be drained below this level
for removal and maintenance of the transmitter.

Questions

3-1. An opentank containsa liquid of varying densityand the level within

the tank mustbe accuratelymeasured.The bestchoiceof measuringsystem
would be:
a. Hubble tube
b. Diaphragmbox
c. Float and cable
d. Head type with differentialpressuretransmitter
3-2. A chain-suspended
diaphragmbox and pressureinstrumentare usedto
measureliquid level in a tank which is 10feet in diameterand 15feetdeep.
The tank containsa liquid that hasa specificgravity of 0.9 at ambient
temperature.The spanof the pressureinstrumentrequiredis approximate1y:
a. 10 psi
c. 5 psi
b. 25 psi
d. 15psi

88 FEEDBACK PROCESSCONTROL

3-3. A pressurized tank contains a lquid, SG = 1.0, and the level measuring
pressure taps are 100inches apart. A pneumatic differential pressure
transmitter is used to measure level. The leg that connects the top of the tank
to the transmitter is filled with tank lquid. The top tap should be connected

to:
a. The low-pressuretap on the transmitter.
b. The high-pressuretap on the transmitter.
, Either the high- or the low-pressuretap.
d. The high-pressuretap but througha sealchamber.
3.4. In Qu~stion3-3the bottom connectionis madeto the low-pressuretap
on the transmitterand the top is madeto the high-pressure
tap. The signal
output, if the transmitteris calibratedOto 100inches,whenthe tank is full
will be:
a. Topscalevalue
c. O
b. Bottom scalevalue
d. A measureof density
3.5. A plant hasa water tower mountedon top of an 80-footplatform. The
tank is 30 feet high. Whatis the heightof water in the tank if a pressuregauge
on the secondfloor, height 15feet,reads 40 psi.
a. Full
c. 4.74feet
b. 12.42feet
d. 27.42feet
3-6. A displaceris 5 inchesin diameterand 30 incheslong. If it is submerged
to a depthof 20 inchesin lquid sa = 0.8, whatforce will it exert on the top
works?
a. 11.3pounds
c. 35.5pounds
b. 426.3pounds
d. 1.47pounds
3.7. A closed tank level (Sa = 0.83)is measuredwith a differentialpressure
transmitter.The level mar vary from 10to 100inches.The high-pressuretap
is 10inchesabovethe transmitterand a watersealfluid is used.A pressure
repeateris usedfor the top tank pressure.The differentialpressure
transmittershouldbe calibrated:
a. 10to 100inches
b. 10to 84.7inches
c. 8.3 to 74.7inches
d. 8.3 to 98.3inches
3.8. To resistthe corrosiveeffectsof a very unusual,highlyexplosive
chemical,a storagetank is lead-lined.Level measurement
is difficult because
the materialalso solidifiesonce it entersa measuringtap. To measurethe level
with accuracythe bestchoicewould be:
a. Weighthe tank and its contentsand zero out the tare weightof the tank.
b. Use a radioactivelevel measurement.
c. Install a conductivitylevel measurement.
d. Use a thermal conductivitylevel detector.

LEVEL AND DENSITY MEASUREMENTS 89

3-9. A displacer48 incheslong and 1%inchesin diameteris usedto measure

densityon a 4 to 20 mA dc transmitter.Its total submergedweightchanges
from 1.42poundsto 5.7pounds.If the densityis checkedand determinedto
be 0.4 whenthe output is 4 mA dc, the densityat an outputof 12mA dc is:
a. 0.6
c. 1.0
b. 0.8
d. 1.6
3-10. A densitymeasuringsystemis set up as shownin Figure3-18.The H
distanceis 100inchesand the lower transmitteris calibratedto a 120-inch
span.If the transmitteris pneumaticand deliveringa 12-psisignal,the
specificgravity of the tank's contentsis:

a. 1.0

c. 0.9

b. 0.83

d. 1.2

Flow Measurement

The process industries by their very nature deal constantly with flowing
fluids, and measurement of these flows is essential to the operation of
the plant. These measurements are indicated, recorded, totalized, and
used for control. Flow measurement is generally the most common
measurement found in the process plant.
Fluid flow measurement is accomplished by:
Ao Displacement
lo Positive displacement meters
20 Metering pumps
Bo Constriction Type, Differential Head
lo Closed conduit or pipe
ao Orifice plate

bo Venturi tube
Co Flow nozzle
do Pitot tube
eo Elbow
fo Target (drag force)
go Variable aTea (rotameter)
20 Open channel
ao Weir

bo Flume
90

FLOWMEASUREMENT 91

C. Velocity Flowmeters
1. Magnetic
2. Turbine
3. Vortex or swirl
4. Ultrasonic
5. Thermal
D. Mass Flowmeters
1. Weighttypes
2. Head and magnetictypes compensatedfor temperature,pressure,
and density
3. Gyroscopeprecisiontypes
4. Centrifugalforce (torque)types
Positive displacementmeters and meteringpumps measurediscrete quantities of flowing fluid. This flow is indicated in terms of an
integrated or totalized flow volume (gallons,cubic feet, litres, cubico
metres, and the like). A typical applicationof this type of flow measurementis custodytransfer, and familiar examplesaredomesticwater
metres and gasolinepumps. The other types of meters listed above
measureflow rate.
lt is flow rate-quantity per time, suchas gallons per minute or
liters per second-which is most generallyused for measurementand
related control applications in the processplant. The most common
rate meter is the constriction or headtype.
Constriction

or Differential

Head Type

In this type of flow measurement a primary device or restriction in the

flow line creates a change in fluid velocity that is sensedasa differential
pressure (head). The instrument u~ed to sensethis differential pressu~e
is called the secondary device, and this measurementis related to flow
rate. The flow-measuring system, orlen called a flowmeter, consists of
primary and secondary devices properly connected.
The operation of the head-type flow-measuring system depends on
a theorem first proposed by Daniel Bernoulli (1700-1782). According to
this theorem the total energy at a point in a pipeline is equal to the total
energy at a second point if friction between the points is neglected.
The energy balance between Points 1 and 2 in a pipeline can be
expressed as follows:

+ Z2 = ~p + ~
(VmJ2+ Zl

92 FEEDBACK PROCESSCONTROL

where:
P
Vm
Z
p
G

=
=
=
=
=

Static pressure, absolute

Fluid stream velocity
Elevation of center line of the pipe
Fluid density
Acceleration due to gravity

If the energy is divided into two forms, static head and dynamic or
velocity head, some of the inferential flow devices can be more easily
studied.
For example, with an orifice plate, the change in cross-sectional
aTeabetween the pipe and the orifice produces a change in flow velocity
(Figure 4-1). The flow increases to pass through the orifice. Since total
energy at the inlet to, and at the throat or, the orifice remains the same
(neglecting losses), the velocity head at the throat must in.crease,causing a corresponding decrease in static head. Therefore, there is a head
difference between a point immedlately ahead of the restriction and a
point within the restriction or downstream from it. The resulting differential head or pressure is a function of velocity that can then be related
to flow. Mechanical flow-measuring instruments use some device or
restriction in a flow line that results in such a differential head.
The following should clarify these relationships. Assume a tank as
shown in Figure 4-2. A flow line enters the tank and replaces the out-

Fig. 4-1. Orifice plate differential producer. The difference in head (pressure) atP, and
P 2is a function of velocity which can be related to ftow.

FLOW MEASUREMENT 93

-+
a=A.J2;

-..Q =AV2-;;h

Fig. 4-2. Free and tank-to-tankflow. usedto simplify conceptofflow formula(4-1).

ftow through the orifice located near the bottom of the tank. If the level
in the tank is H, the velocity of outftow will be:
y2 = 2GB

or

This is the lawof falling bodies that is developed in most physics

books. The volume of liquid discharged per time unit through the
orifice c3;nbe calculated by geometry.

=AV
Substituting the expression for velocity in this equation:

(4-1)
This expression may also be used to calculate the rate offtow past a
point in a pipe. The actual ftow rate will be less than this equation will

V=Y2GH
Q
Q=A\f2Gii

O)
Fig.

94

FEEDBACK PROCESS CONTROL;

calculate. The reduct"ionis caused by several factors, including friction

and contraction of the stream within the pipe. Fortunately all common
primary devices bave been extensively tested and their coefficients
determined. These data become part of the ftow calculations.
Primary Devices

The orifice plate is the most popular primary device found in most
process plants. Orifice plates are applicable to all clean fluids, but are
Dot generally applicable to fluids containing solids in suspension(dirty
fluids).
A conventional orifice plate consists of a thin circular plate containing a concentric hole (Figure 4-3). Th plate is usually made of stainless
steel but other materials, such as monel, nickel, hastelloy, steel, glass,
and plastics, are occasionally used. The most popular orifice plate is the
sharp-edged type. The upstream face of the plate usually is polished
and the downstream gide is often counterbored to prevent any interference with the flowing fluid. The bore in the plate is held to a tolerance
of a few ten-thousandths of an inch in small sizes and to within a few
thousandths in sizes above 5 inches in diameter.
ln addition to the conventional, sharp-edged, concentric plate,
there are others that bave been designed to handle special situations.
These special types constitute only a small percentage of the total, but
they do occasionally solve a difficult problem.
There are two plates (Figure 4-3) designed to accommodate limited
amounts of suspended solids. The eccentric plate has a bote that is
bored off-center, usually tangent to the bottom of the flow line (inside
periphery of the pipe). The segmentat orifice plate has a segment removed from the lower hal of the orifice plate. ln addition, there are

CONCENTRIC

4-3. Orificeplate types.

ECCENTRIC

SEGMENTAL

Reynolds

FLOW MEASUREMENT 95

quadrant-edged or special-purpose orifice plates with rounded edges.

This design minimizes the effect of low Reynolds number (see below).
Orifice plates along with other primary devices operate most satisfactorily at high ftow rates. The reason is the change in ftow coefficient
that occurs when the ftow rate changes from high to very low. When
this happens the ftow velocity distribution changes from uniform to
parabolic. This mar be related to a ratio of inertia forces to viscous
forces called Reynolds number.
Number = Average Fluid Velocity x Orifice Diameter
Absolute Viscosity
In most process plants the Reynolds numbers encountered will be
high-between 10,000 and 1,000,000. Under these conditions the ftow
coefficients stay fixed and head-type ftow measurementsare reasonably
accurate. To achieve the ultimate in accuracy, a Reynolds number
correction is made and this is described in the text Principies and Practices o/ Flowmeter Engineering by L. K. Spink. At low Reynolds number, the ftow changes from turbulent to parabolic or laminar and the
coefficients undergo a change. The changeover cannot be precisely
defined, which makes the application of head-type ftow measurements
difficult at low-ftow velocities. The quadrant-edged orifice has the ability to perform well at lower velocities than the square-edged plate.
However, it does Dot perform as well at high Reynolds numbers (velocities).
Orifice plates that are to handle lquids containing small amounts of
dissolved air should contain a small vent hole bored at the top to permit
the air to pass through the plate. Specifications bave been established
for the size of this vent hole in relation to the size of the orifice. Plates
for use with steam should bave a similar hole drilled at the bottom to
drain any condensed steam. In certain ftow measurements, the orifice
plate should also bave a bottom drain hole to permit the condensation

to drain.
Table 4-1 summarizes application factors to help in deciding
whether the orifice plate can do a particular job.
Another primary device that is frequently found in the process
plant is the Venturi tube shown in Figure 4-4. The Venturi tube produces a large differential with a mnimum permanent pressure loss. It
has the added advantage of being able to measure ftows containing
suspended solids. The most significant disadvantage is its cost which,
when compared with other primary devices, is high.

~~
G
VP
Fig.

96 FEEDBACK PROCESSCONTROL

Table4-1. Primary Flow MeasurementDevices

" t
:

o:
Accuracy (empirical
data)

None

Pressure loss #

, P

Use on dirty service

Differenlial produced
for given ftow and size

p
E

E
P
P

E
P
F

G
G
E

~
.9"

.s~

".."
.~
~~~~"to"~
~5

"='"
" E

E
EG

None
E
E

GG

F
None

Non.

VP

G
G
F

U
E

F
P

F
F

VP
E

P
F

For liquido containing

vapors

For vapors containing

condensate

For viscous ftows

First cost

EGG *
F
E

E
U
G

Ease of changing
capacity

E
G

Ease of instalIalion

Excellent

G
F

Good
FairP
Poor

VP
U
/I

Very poor
Unknown
E indicates lowest loss, P highest, etc.

EG

EG
p
E
G

None
E

u
uVP

E
F

G
F

.For measuring velocity at oRe point in conduit the

well-designed Pitot tube is reliable. For measuring
total ftow, accuracy depends on velocity traverse.
t Excellent in verticalline if ftow is upward.
4;Excellent in verticalline if ftow is downward.
Requires a velocity traverse.

Several other primary devices mar be classified as modifications of

the Venturi tube. The Lo-Loss and Dall tubes are similar in many ways
to the Venturi. The flow nozzle (Figure 4-5) has some similarity but
does not bave a diffuser cone, and this lmits its ability to minimize
permanent pressure loss. The Venturi family ofprimary devices is often

4.4. The HerschelVenturitube,developedin 1888,to producea largedifferentialpressure

with a smallheadloss.

Fig.

FLOW MEASUREMENT 97

4-5. Flownozzleassembly.

chosen to minimize permanent pressure loss or to cope with solids

suspended in the flowing material. Table 4-1 provides aid in making a
selection.
The Pitot tube is another primary device. It has the advantage of
practically no pressure dropo Its limitations are its inability to handle
solids carried by the flowing material and somewhat limited accuracy.
Solids tend to plug the openings in the tube, and the classical Pitot tube
senses impact pressure at only one point, thus decreasing accuracy. A
multi-impact opening type (annubar tube) is available which improves
the potential accuracy. In the process plant Pitot tubes are used more
for testing than for continuous use. It is possible to install a Pitot tube
into a flow line under pressure if the required equipment is available.
Occasionally a pipe elbow (Figure 4-6) mar be used as a primary
device. Elbow taps bave an advantage in thai most piping configurations contain elbows thai can be used. If an existing elbow is used, no
additional pressure drop is created and the expense involved is minimal. The disadvantages are thai accuracy will be Iacking and dirty
flows mar tend to plug the taps. Repeatability should be reasonably
good. Two taps locations are used at either 45 or 22Y2degrees from the

Fig.

98 FEEDBACK PROCESSCONTROL

rb

4-6. Elbow(usedas prirnarydevice).

inlet face ofthe elbow. At low-ftow velocities, the differential produced

is inadequate for good measurement, thus the elbow is a usable choice
only for high top-scale velocities.
The target or drag body ftow device (Figure 4-7) is in many ways a
variation of the orifice plate. Instead of aplate containing a bore
through which the ftow passes,this device contains a solid circular disc
and the ftow passes around il. An essential part of the measurement is
the force-balance transducer that converts this force into a signal. This
signal is related to ftow squared, just as it wouid be with any other
primary device connected to a differential pressure transmitter. The

4-7. Targetflowmeter.

FLOW MEASUREMENT 99

target meter mar be generally applied to the measurement of lquids,

vapors, and gases. Condensates pass freely along the bottom of the
pipe, white gases and vapors pass easily along at the top. The target
flow device has demonstrated the ability to handle many difficult assignments Dot possible with other primary elements. Both pneumatic
and electronic force-balance transducers can be used, making either
type signal, pneumatic or electrlc, available.
Secondary

Oevices

Secondaryinstrumentsmeasurethe differential produced at the primary devices and convert it into a signal for transmissionor into a
motion for indication, recording, or totalization.
Secondaryinstrumentsinclude the mercury manometer,the socalled mercurylessor diaphragm(beUows)meterand various types of
force-balanceand motion-balancepneumaticand electronictransmitters.
In the past the mercury manometer was an extremely popular secondary device, but the high price and danger of mercury bave all but
eliminated it from everyday use. However, a few mercury manometers
are still employed for ftow rate measurements in gas pipelines. The
bellows or diaphragm secondary devices are popular when a direct
indication or record is desired. The most popular secondary device is
the force-balance transmitter. The reasons for this popularity are virtually unlinited adjustability, no harm from extensive overrange, and
the a vailability of a standard signal that can be fed into a recorder, a
controller, or other instruments that may be combined to form a sys-

tem.
Still another detail is the mounting of the secondary device (locations are shown in Figure 4-8). With liquid ftow, the secondary device is

LIQUID FLOW

Fia. 4.8. Flow.

GAS FLOW

STEAM FLOW

100 FEEDBACK PROCESSCONTROL

below the pipeline being measured to ensure that the connecting lines
attached to the sicle of the pipe remain liquid filled. With steam, the
lines should always remain filled with condensate; but with gas, the
secondary device is mounted above the ftow line to drain away any
liquid that mar be present.
When any primary device is installed in a pipeline, the accuracy
will be improved if as much straight run as possible precedes il.
Straight run beyond the primary device is of far less concem.
Still another important consideration is the location of the pressure
taps. With nozzles, Venturi, Dall and Pitot tubes, and elbows, the
pressure tap locations are established. For orifice plates, however, a
variety of tap locations are used. These are ftange, corner, vena contracta, or D and D/2, along with full ftow or pipe taps. Figure 4-9
describes these tap locations, and Figure 4-10 shows the importance of
tap location. In general, ftange taps are preferred except when physical
limitations make pipe taps advantageous. Corner taps require a special
ftange and vena contracta tap locations relate to orifice openings.
Corner taps are advantageous for pipe sizes under 2 inches.
Figure 4-11 describes the permanent head loss caused by the primary device selected.
Relating

Flow Rate to Differential

The two basic formulas bave already been introduced. In the English
system, G = 32 feet per second squared, and in the metric system,
G = 980 centimetres per second squared. The height (h) would be expressed in feet in the English system and centimetres in the metric

Fig. 4-9. Taplocationsfor orifices.

FLOW MEASUREMENT 101

Fig. 4-10. Differentialpressureprofile with orifice plate.

t

system. In Equation 4-1, area (A) would be expressed in square feet

and in the metric system in square centimetres yielding cubic feet per
second or cubic centimetres per second. In the United States, the
English system is the most common. This system is in the process of
being changed to the SI metric, but it appears that the total conversion

Fig. 4-11. Pennanent head los s in a ditferential producer may be plotted as a percent of
measured ditferential for the several types of primary devices. These values must be
interpreted according to the acceptable head loss limit for any particular application.

102 FEEDBACK PROCESSCONTROL

will take a long time to complete. For tros reason the sample calculations that follow will use the English system.
In Equation 4-1, V is in fe et per second; the acceleration due to
gravity (G) is in feet per second squared; H is the height in feet of a
column of the fluid caused by the differential pressure across a primary
device. To express this equation in terms of an equivalent differential
(h), in inches of water, H is replaced by (h/12)G) where G is the
specific gravity of the fluid at flowing temperature.
Substituting V into Equation 4-1:

(4-2)
where:
Q = Volumetric flow (cubic feet per second)
A = Cross-sectional aTea of the orifice or throat of the primary
device (square feet)
h = Differential across the primary device (inches of water)
G, = Specific gravity of the fluid (dimensionless)
G = Acceleration due to gravity (a constant: 32.17feet per second2)
For lquids, it is more useful to express Q in gallons per minute.
AIso, it is convenient to express the area of the orifice or throat interms
of its diameter (d in inches). Substituting in Equation 4-2:

x~

(4-3)

Equation 4-3 most be modified to take into account several factors,

such as the contraction of the jet, frictionallosses, viscosity, and the
velocity of approach. This modification is accomplished by applying a
discharge coefficient (K) to the equation. K is defined as the actual
ftow rate divided by the theoretical ftow rate through the primary
device. Basic texts carry K factors for most primary devices. For
instance, K is listed according to vario us pressure tap locations, and
di D ratios. Where extremely accurate results are required, however,
the K factor must be determined in the laboratory by actual ftow
calibration of the primary device.

FLOW MEASUREMENT 103

Applying K to Equatiort 4-3 gives:

Since both K and d are unknown, another value is defined. The

quantity (S) is set equal to K (d/D)2. Since Kd2 then equals SD2,
substitution produces:

S values (sizingfactor) are tabulated againstbeta ratios (di D) for various

differential devices in Table 4-2.
j
Effect of Temperature on Flow Rate
The above equations give the flow through the primary device at conditions that exist at the time the measurement is made. In most industries, it is more desirable to know the equivalent volume flow at a stated
reference temperature, usually 600P(15.6C).
Por lquids, this correction can be applied by multiplying the equation by the ratio ofthe lquid specific gravity at the flowing temperature
(G,) divided by the lquid specific gravity at the reference temperature
(G). Thus Q gallons per minute at the reference temperature:
Q

= 5.667

SD2

fE::
V G;

Gi

(4-6)

which simplifiesto:
Q = 5.667 SIJ2 YG;l
GI

Similar equations can be developed for steam or vapor ftow in weight

units (such as pounds per ho ur) or for gas ftow in volume units such
as standard cubic feet per hour (scfh).

W (poundsper hour) = 359SD2v'h:y;

Q (scfh) = 218.4SD2~ ~

Pb" T;G

104 FEEDBACK PROCESSCONTROL

Table4-2. Sizing Factors

S'ta
or
d/D

Ratio
0.100
0.125
0.150
0.175
0.200
0.225
0.250
0.275
0.300
0.325
0.350
0.375
0.400
0.425
0.450
0.475
0.500
0.525
0.550
0.575

Square-Edged
Orifice, Flange
Corner or
Radius Taps

Ful/-Flow
Quadrant-

2\-2& 8D

Nozz/e
and

Lo-Loss

Dall

Taps

Venturi

Tub.

Tube

(PipJ

0.005990

0.006100

0.009364

0.009591

0.01349

0.01389

Edged
Orifice

0.01839

0.01902

0.02402

0.02499

0.0305

0.03044

0.03183

0.0390

0.03760

0.03957

0.0484

0.04558

0.04826

0.0587

0.05432

0.05796

0.08858

0.06390

0.06874

0.1041

0.07429

0.08068

0.1210

0.1048

0.08559

0.09390

0.1392

0.1198

0.09776

0.1085

0.1588

0.1356

0.1170

0.1267

0.1109

0.1247

0.1800

0.1527

0.1335

0.1443

0.1251

0.1426

0.2026

0.1705

0.1500

0.1635

0.1404

0.1625

0.2270

0.1900

0.1665

0.1844

0.1568

0.1845

0.2530

0.2098

0.1830

0.207

0.1745

0.2090

0.2810

0.2312

0.2044

0.232

0.1937

0.2362

0.3110

0.2539

0.2258

0.260

0.0700
0.0824
0.0959
0.1106

0.2144

0.2664

0.3433

0.2783

0.2472

0.292

O.(X)()

0.2369

0.3002

0.3781

0.3041

0.2685

0.326

0.625
0.650
0.675
0.700
0.725
0.750
0.775
0.800
0.820

0.2614

0.3377

0.4159

0.3318

0.2956

0.364

0.2879

0.3796

0.4568

0.3617

0.3228

0.3171

0.4262

0.5016

0.3939

0.3499

0.3488

0.4782

0.5509

0.4289

0.3770

0.3838

0.6054

0.4846

0.4100

0.4222

0.6667

0.5111

0.4430

O.4646

0.5598

0.4840

0.5113

0.6153

0.5250

0.6666

0.5635

FLOW MEASUREMENT 105

In Equations 4-8 and 4-9: .

'Yf= Specific weight of the steam or vapor at operating conditions
in pounds per cubic foot
T b = Reference temperature (absolute); (i. e., 460 plus the reference
temperature in F.)
Pb = Reference pressure (psi absolute)
Tf = Operating temperature at the primary device (absolute); (i.e.,
460 plus operating temperature in F.)
Pf = Operating pressure (psi absolute)
G = Specific gravity ofthe gas (molecular weight ofthe gas divided
by the molecular weight of air or the weight of a volume ofthe
gas at a given temperature and pressure divided by the weight
ofanequal volumeofairatthe sametemperature and pressure).
The reference temperature is very often 600P and the reference
pressure atmospheric 14.7 psi absolute. Ifthese are the standard conditions, Equation 4-9 can be simplified to:

(4-10)
Equation 4,.10is applicable to gas ftow only when the pressure
differential is small enough so that gas density does Dot change significantly. A simple rule ofthumb is that the maximum differential in inches
of water should Dot exceed the absolute operating pressure in psi absolut.~. For example, if the gas operating pressure is 22 psi absolute, at a
particular installation, and the maximum differential is 20 inches of
water, Equation 4-10 can be used.
Flow rate 'measurements for gas and steam are more difficult to
make with accuracy than those for liquido The reason is changes in
specific gravity, weight, temperature, pressure, and so on, that mar
occur under operating conditions.
These changes will bave an effect on measurement accuracy and
under certain conditions mar be difficult to predict. An abbreviated set
of tables for the formulas given are included in this book. If more
accuracy is required, more exact equations, along with detailed tables,
such as those found in Principies and Practice oi Flow Meter Engineering
by Spink, should be used.
Another method of performing these ftow calculations is to use a
ftow slide rule. The ftow slide role has the table values incorporated

106 FEEDBACK PROCESSCONTROL

into its scales. If the tables given are used, the resulting accuracy
should be as good as the flow slide rule.
Now several sample problems are given to demonstrate the procedure followed for each type of calculation. Additional problems are
given at the end of tros chapter.
SAMPLEPROBLEMJ A 4-inch schedule 40 pipe carries water that is
measured by a concentric, sharp-edged, orifice plate, d = 2.000 inches,
with flange taps. The differential is measured with an electronic differential pressure transmitter. The transmitter is calibrated Oto 100inches
of water pressure and has an output of 4 to 20 mA dc. If the signal Crom
the transmitter is 18.4 mA dc, find the flow rate.
S/ep 1. Convert the electrical signal to differential pressure.
18.4 -4

x 100 = 90 inches o water

20 -4

Step2. DetermineID of 4-inchschedule40pipe (seeAppendix, table

A-4) = 4.026inches.
2.000 = 0.4968
Step3. Calculatelid = 4:026

Step4. From Table 4-2determineS. This will require interpolation.

S =

0.475

0.1404

0.4968

0.5000

0.1568

(0.5000
0.4968 -0.475
-0.475 )

(0.1568 -0.1404) ~0.1404

S = 0.1547
Step5. Substitute in Equation 4-7:

= 5.667 x 0.1547x 16.21 x 9.487

= 134.83or 135gpm

FLOW MEASUREMENT 107

SAMPLEPROBLEM
2 An elbow is used as a primary device. The taps are
made at 45 degrees. The line is a 6-inch schedule 40 pipe. What is
the water flow rate if the effective radius of curvature is 9 inches
and a differential pressure of 35 inches of water pressure is produced.

= 172.66 .yj"5 =

,021.47gpm

SAMPLEPROBLEM
3 Dry-saturated steam is measured with a ftow nozzle. The d/D is 0.45 and the line size is a S-inch schedule SOpipe. The
static pressure is 335 psi. Calculate the ftow rate at a differential
pressure of 200 inches of water in pounds per hour.
W(pounds per hr.) = 359SD2V:y;

(4-8)

from Table 4-2

s = 0.2026
D = 7.625 inches (Table A-4)
Yf = 0.754 pounds per cubic feet
W(pounds per hour) = 359 x 0.2026 x 7.6252\1200 x 0.754

= 4,228.77x 12.28
W = 51,929 pounds per hour
SAMPLEPROBLEM
4 A 6-inch schedule 40 pipe (1D-6.065) carries fuel
gas with a specific gravity of 0.88. The line pressure is 25 psi. Plowing
temperature is 60oPand the ftow is measured with an orifice plate with
ftange taps. The maximum ftow rate is 2,000,000standard cubic feet per
day. Pind the diameter of the hole to be bored in the concentric orifice
plate if 20 inches of water pressure is full-scale differential.

108 FEEDBACK PROCESSCONTROL

2,000,000
= 7' 727 x S x. 6 0652 V120(25
+-~
24
(460
+ 6~

8,3333.33= 7,727 x S x 36.78 x 1.322

S = 7,727 x83,333.33
36.78 x 1.322= O.18
22
From Table 4-2
.575

.2144

.2218

.2369

.600
~ =
D

( .2218
.2369

-.2144

-.2144

d = (.583)(6.065)

025 + .575 = .5832

= 3.536 inches

SAMPLEPROBLEM5 Gasoline is carried in a 2-inch schedule 40 pipe

(ID = 2.067). A concentric sharp-edged orifice plate with corner taps
is used to measure flow rate. The orifice bare is 1 inch in diameter,
and at full flow, 50 inches of water differential is produced. The
specific gravity of the gasoline is 0.75. What is the flow rate?

Q = 5.667 x 0.1474 x 2.067 x 2.067 x 'V{SO

0-:75

= 5.667 x 0.1474x 2.067 x 2.067 x 8.165

= 29.14 gpm

Fig.

FLOW MEASUREMENT 109

Variable Area Meters (Rotameter)

The variable aTea(Figure 4-12) meter is a form of head meter. In this

flowmeter the aTea of the flow restriction varies so as to maintain a
constant differential pressure. The variable aTeameter, which is often
called a rotameter, consists of a vertical tapered tube through which the
fluid flow being measured passes in an upward direction. A float or
rotor is contained within the tapered tube. This rotor or float is made of
some material-usually metal-more dense than the fluid being measured. As the flow moves up through the tapered tube, it elevates the
float until balance between gravity acting on the float and the upward
force created by the flow is achieved. In achieving this balance of
forces, the area through which the fluid passeshas automatically been
adjusted to accommodate that flow rate.
The tapered tube is often made of transparent material so the float
position can be observed and related to a scale calibrated in units of
flow rate. The rotameter is often used to measure low-flow rate. When
very large flow rates are to be measured, a bypass rotameter may be
used. This consists of an orifice in the main flow line with the variable
aTeameter connected in parallel. An industrial rotameter is useful overthe
upper 90 percent of its scale. The accuracy of the rotameter, like
any head type, depends on many factors. Typical errors may vary
between I and 10 percent of full scale value.
Open-Channel

Flow Rate Measurements

Flow rate measUements

in openchannelsare fundamentalto handling
water and waste. Environmentalconsiderationsbave madethesemea-

4-12. Variablearea(rotarneter)flowmeter.

110 FEEDBACK PROCESSONTROL

surements common in' the typical process plant. Open-channel measurements utilize head meter techniques.
Primary

Devices

The primary devices used in open-channel ftow rate measurements are

weirs and ftumes. There are two basic weirs-the rectangular and the
V-notch. The rectangular weir is made in three varieties (Figure 4-13).
The first has contractions or extensions into the channel that produce a
boxlike opening. The second modification completely suppressesthese
contractions, extending the weir across the entire width of the channel.
The third type, the so-called Cippoletti weir, has end contractions set at
a 4: 1 angle, rather than being perpendicular to the edge of the weir.
Rectangular weirs are primarily designed for larger ftow; their lmit on
range is dictated by the design of the associated channels. Figure 4-14
shows the ranges of weir capacity for the different types and sizes of

welrs.
V-notch weirs are essentially plates (usually metal) that contain a
V-shaped notch (Figure 4-13). The angle of the V can vary, but the
formulas given are for the most common angles 30, 60, and 90 degrees.
V-notch weirs are employed for lower ftow rates than those that would
be measured by a rectangular weir.
In weir measurement (Figure 4-15) the nappe, or profi1e of water
over the weir, must be completely aerated if good accuracy is to be
obtained. All weirs then produce some head loss as the lquid falls free.
If head loss is a problem, a ftume might be a better choice.
Flumes, a further development of the basic weir concept, are designed primarily to reduce the head loss that is experienced with the

CIPPOLETTI

WEIR~S>2H

ECTANGULAR

WEIR-j

MAX
I

: 4~

_L)2H

MAX

PREFERABLY

ANO
>4H

MAX

I
0>3H

MAX

C\PPOLETTI
WEIR

Fig. 4-13. Rectangular.Cippoletti,and V-notchweirs.

GPM
70000
50000

100

40000

80

30000

60
50

20000

40
30

10000
8000
6000
5000

10

4000

3000

6
5

2000

3
1000
800
600
500

1.0

400

De

300

0.6
05
0.4

200

0.3
100
80
60
50

OI

40

0.08

30

006

0.05
20

0.04

FLOW

0.03

10

0.02

8
FLOW

6
5

001

0008

0.006
0005
0004
0003

0.002

Fig. 4-14. Rangeof weir capacityfor differenttypesand sizesof weirs.

111

weir.

112 FEEDBACK PROCESSCONTROL

AERATION'UNDE.

NAPPE

Fig. 4-15. Aerationundernappeoi weir.

Flumes are also able to handle solids suspended in the flowing

liquido
A very popular flume is the Parshali flume. lt is somewhat similar
to one-half of a rectangular Venturi tube. ln flume measurement, head
losses are reduced because it is Dot necessaryto create the nappe. The
Parshall flume is shown in Figure 4-16 and Table 4-3 gives the dimen-

.k.

,..~~--~~~:::::: ~~:.:--::::~
i o
t7::::::;:::~~~t=:=f:~:::::::--i
l CO"".~.,.....c"O" .. TH.OAT...TION!l/.".'N.'.cr' ON c~1

Lc

f ~~~--~:

~.

..PLAN

'. "

.1
M

,\

\\

:~+
.1-0

.1.

Ir:--

'-'0-.,,

;1

'LOW

,
l

oi;;;.
""
,o

.";;,,

''-

SECTION

o-o

Fig. 4.16. Diagramanddimensionsof Parshallflume.

~
:..:.;
c;...
.,
'" ~< '"",:,
~'""

~~~--c

~-c..,

":";oC~.i~~~~@a~

, \oi

~~~t;

a~~=~~:",-c-co",

~~~:,

ci ci ci ci ci ci ci":":"";";

;... ~

;. .., .., .., .., .., .., .., .., .., .., ..,

~ ~

-"""""""""""""""'"

.1~ ...

~ ri:

'"
G)

Q: .:

:;
oc:

~~
~~

~
'"'
~ r;:

'"
~

lo.

cE
'"
~

.9

~
e
=
e

~
t, .t

to.t

'2

~
=
~

'Q

.5

~ .t

e
'~

.!,
~ ...

~
'Q

~
'"
=

NN''~"""":'ob~=~~

~~~~~~~:!:!:!:f:f

e
=

.;!;:-::i
-::i.;!-::.;!
~-'c-~-;'-;';'~;'

~ .;

~"!I!J;)ION
sn!~ ..un;) qloowS

--.N

-..,

'" ~
...

oS

'C

Li:

Qj

:c

.., .., .., .., .., .., .., .., .., ..,

~~~~~~~~~~~~
99~9O999999
N N N ...,J,
...,J,

.
:.:.~..~...
-"'-0-,
= ...g ::-

~.

;:- ::,..,.0--;

O':":'N"'~""""'obd,=

~":-~9~9"'99999
g--NN""",,",.o""~o.
..~.-!!!"~.-!!!
c ;:;:- c '" ~ ;,- ~ ;,- "C>
'9 'i' ";' "f
"f ";' '" -.,.
";'
-NN

b.b

..

'~

..,.

o. o. o. o. o. o. o. o. o.

~~!~~:,:,~~~~:,

5
,5
o
~.

oS

""

.. to.

~ ;;: :::

.:.!::~;:~~~:':~J.
.~

!~;;;~~~~~~:!~~
~,\,O-9'\'99"'9999
OO'--N""""'-o"'~

113

'!~

GPM

MGO
3000

CFS
4000

2000

3000
2000

1000
800
1000
800

600
500
400
300
200

100

so
60
40
50

600
500

400
300
200

80
100

60
50

30

40
20

30
20

10
8
6
5

4
3
2

10
08
06
O~

10
08

04
03

06
05

0.2

03

04

100

-t++

MGO

80

02

OI

60
50
40

004

20

FLOW

10
8

OI

006
QO~

30

FLOW

CFS

008

008
006

OO~
004

002

I
I

I
2

HEAD-INCHES
L-I
2
HEAD-FEET

I I .I.
6

I
8 10

,.,.,.

,..

I~ 20

60

0.03

80 100

002
FLOW

.456810

2/:3

56

8 IQ

001

FIVE
INCHES IS HINIHUN FLA.L SCALE
HEAD III'H
FDXBORD FLOA' ANO CABLE
HE'ER

THIRTY SIX 'NCHES IS ",XIHUM FULL

SCAL' H'AD N'TH FDXBDRD FLDAT AND
CABL' "'T'R

Fig. 4.17. Level and ftow relationship for ParshaIl ftume.

114

FLOW MEASUREMENT 115

sions. By referring to Figure 4-17 the dimensions of a flume capable of

measuring a prescribed flow rate mar be selected. Small flumes mar be
purchased and installed while large flumes are generally fabricated on
site. Several compaDiesspecialize in the construction and calibration of
flumes.
Under ideal conditions, measurements by weirs are very accurate.
In actual plant measurement, however, ibis mar Dot be true because it
is Dot practical to minimize the velocity of approach factor. The correlation between level and flow typically has an error of 3 to 5 percent.
When the error of making the level measurementand relating it to flow
rate is added, the total overall error mar be as high as :t5 percent.
Equations for the more common open-channel flow devices are given in
Table 4-4.
The equations for most open-channel primary devices are nonlinear. To convert head into flow rate it would be necessary to perform
an extensive calculation. This step is generally eliminated by having the
instrument perform the calculation. This is accomplished either with a

Table4.4. Equationsfor Common Open-Channel

Flow Devices
V-Notch Weir
3(1'Qcfs= .6650 H5/2
6(1' Qcfs= 1.432 H5/2
9(1' Qcfs= 2.481 H5/2
Rectangular Weir (with end contractions)
Q = 3.33 (L-0.2H) H3/2
Rectangular Weir (without end contractions)
Q = 3.33 LH3/2
Cippoletti Weir
Q = 3.367 LH3/2

Parshallfiume
Flume
size (in.)
3.
Q = 0.992 HI,547
6
Q = 2.06 HI,58
9
Q = 3.07 HI,53
12
Q = 4.00 HI,522
Q = crs; L = crest length (fi); H

Flume
size (in,)
18
Q = 6.00 H',538
24
Q = 8.00 HI,55.
36
Q = 12.00 HI,566
48
Q = 16.00 Hl.578
= head (fi)

116 FEEDBACK PROCESSCONTROL

carn or by means of a special chart or scale. The head is measured with

any one of several instruments. Several popular types are: float-andcable, ball float, bubble, and pressure-sensing. A typical installation is
shown in Figure 4-18.
Installation and Selection Considerations
To determine the type of primary device to be used for open-channel
flow measurement, the following factors musi be considered:
Probable ftow range
Acceptable head loss
Required accuracy
Condition of liquid
Probably all of the devices will work, but usually one type will be
the most suited to the installation. It must be remembered that a weir
will Dot function properly ifthe weir nappe is Dotaerated, and the flume
must be maintained at criticat depth; otherwise these devices do Dot
function properly. Nappe aeration is largely a question of level in the
receiving channel. If the lquid rises so that clearance is insufficient for
aeration, accurate flow measurement is impossible. When the flow
drops so that the nappe is Dot aerated but is drawn to the weir edge by
capillary attraction, the device does Dot operate properly.
Minimum weir flows are difficult to calculate because they depend
to some degree on the nature of the weir. If the weir is properly sized,
mnimum flow will produce a substantial, measurable amount of head
over the weir. Figure 4-17 shows the relationship between level and
flow for the ParshalI flume (Figure 4-16). Dimensions and maximum
and mnimum flows for the flume are tabulated in Table 4-3.

Fig. 4-18. V-Notchweir with ftoat-and-cable.

FLOW MEASUREMENT 117

Velocity

Flowmeters

Magnetic Flowmeter
The principIe of the magnetic ftowmeter was first stated by Faraday in
1832, but did Dot appear as a practical measurement for the process
plant until the 1950s.Its advantages are no obstruction to ftow, hence
no head loss; it can accommodate solids in suspension; and it has no
pressure connections to plug up. It is very accurate and has a linear
ftow rate to output relationship. Its disadvantages are that measured
material must be lquid; the lquid must bave some electrical conductivity; and it is expensive.
Operation
Operation ofthe magnetic ftowmeter is based on Faraday's well-known
law of electromagnetic induction: The voltage (E) induced in a conductor oflength (D) moving through a magnetic field (H) is proportional
to the velocity (V) of the conductor. The voltage is generated in a plane
that is mutually perpendicular to both the velocity ofthe conductor and
the magnetic field. Stated in mathematical form:

E = CHDV

(4-11)

where C is a dimensional constant. Industrial power generators and

tachometer generators used for speed measurement operate on the
same principIe, electrically conductive process liquid acts in a similar
manner to a rotor in a generator. That is, the liquid passes through a
magnetic field (Figure 4-19) induced by coils (Figure 4-20) built around
a section of pipe. The process liquid is electrically insulated from the
pipe, or flow tube, by the use of a suitable lining when a metal tube is
used, so that the generated voltage is Dot dissipated through the
pipeline. Two metallic electrodes are mounted in the flow tube and
insulated from it; a voltage is developed across these electrodes that is
directly proportional to the average velocity of the liquid passing
through the magnetic field. Because the coils are energized by altemating current power, the magnetic field and resultant induced voltage is
altemating. This generated voltage is protected from interference,
amplified, and transformed into a standard dc current signal by a
transmitter or converter. Line voltage variations are canceled by the
circuits employed.
The magnetic flowmeter measures volume rate at the flowing temperature independent of the effects of viscosity, density, turbulence or

FLOW MEASUREMENT 119

suspended material. Most industrialliquids can be measured by magnetic flowmeters. Exceptions are someorganic chemicals and most refinery products. Water, acids, bases, slurries, liquids with suspended solids, and industrial wastes are commou applications. The limitation is
the electrical conductivity ofthe liquido The magnetic flowmeter offers
no more restriction to flow than an equivalent length of pipe, Figure

4-21.
Structurally, a magnetic flowmeter consists of either a lined metal
tube, usually stainless steel because of its magnetic properties, or an
unlined nonmetallic tube. Linings for the metal tubes can be polytetrafluoroethylene (Teflon), polyurethane, or some other nonmagnetic,
nonconducting material.
The electrodes are suitably insulated from the metal tube. Nonmetallic fiberglass flow tubes do llot require any lining. The electrodes
most be insulated so that the voltage generated can be measured across
the electrodes. The insulation has no bearing on the actual voltage
generation, bot without the insulator, voltage would bleed off through
the metallic walls of the tube.
The coils are similar in design to the deflection coils used on a
television picture tube. Two coils are used and work together to create
a uniform magnetic field. The coils are generally series-connected, but
mar be parallel-connected if the measured flow velocity is low.
The signal output from the flow tube's electrodes is an altemating

Fig. 4-21. Unobstructedftow.

120 FEEDBACK PROCESSCONTROL

voltage at supply frequency. This 10w-Ievel voltage is generally between 1 mV and 30 mV at full flow rate.
This 10w-Ievel altemating voltage must be measured and converted
either into a record or display or into a dc common denominator transmission signal. This signal, typically 4 mA at zero and 20 mA at full
scale, can be fed into a recorder or controller. The device that can
accomplish tros is a special type of transmitter. This transmitter is
located either directly on the flow tube (Figure 4-22), near it, or within
the control room. The preferred location is as near to the flow tube as
possible; temperature and corrosive conditions are the constraints that
dictate location. In some applications a digital or pulse rate signal output Cromthe transmitter mar be required, and this option is available.

Accuracy
The accuracy of most magnetic ftowmeter systems is 1 percent of fullscale measurement. This includes the accuracy of both the meter itself
and its secondary instrument. Because ibis type of meter is inherently
linear, its accuracy at low-ftow rates exceeds the practical accuracy of
such inferential devices as the Venturi tube. Accuracy of the Venturi is
:tO.75 percent, according to ASME Fluid Meters Report, and thai of
the secondary instrument is abolli :tO.5 percent. At the low end ofthe
measurement scale, secondary instrument readability decreases owing
to the square root relationship. The magnetic ftowmeter can be labora-

Fig. 4-22. Foxboromagneticftow tube andtransmitter.

FLOW MEASUREMENT 121

torr calibratedto an accuracyof 0.5 percentof full scaleand is linear

throughout.
Vortex Flowmeter
The Foxboro Vortex ftowmeter, as shown in Figure 4-23, measures
liquid, gas, or steam ftow rates using the principIe of vortex shedding.
The transmitter produces either an electronic analog or polse rate signallinearly proportional to volumetric ftow rate.
Principie or Operation
The phenomenon of vortex shedding can occur whenever a nonstreamlined obstruction is placed in a flowing stream. As the liquid passes
around the obstruction, the stream cannot follow the sharp contours
and becomes separated from the body. High-velocity liquid particles
flow past the lower-velocity (or still) particles in the vicinity ofthe body
to form a shearlayer. There is a large velocity gradient within this shear
layer, making it inherently unstable. The shear layer breaks down arter
some length oftravel into well-defined vortices as shown in Figure 4-24.
These vortices are rotational flow zones that form altemately on
each gide ofthe element with a frequency proportional to the liquid flow

Fie. 4-23. Vortexflowmeter.

122 FEEDBACK PROCESSCONTROL

Fig. 4-24. Vortexsheddingphenomenon.

rate. Differential pressure changes occur as the vortices are formed and
shed. This pressure variation is used to actuate the sealed sensor at a
frequency proportional to vortex shedding.
Thus, a train of vortices generates an altemating voltage output
with a frequency identical to the frequency of vortex shedding. This
frequency is proportional to the flow velocity.
The voltage signal from the detector (Figure 4-25) is conditioned
and amplified for transmission by electronics located in a housing
mounted integral with the flowmeter body. The final output signal is
available either in pulse fonn with each pulse representing a discrete
quantity of fluid for totalizing or, optionally, as a 4 to 20 mA dc analog
signat for flow rate recording or control.
Turbine

Flowmeter

The turbine meler derives its name from its operating principie. A
turbine wheel (rotor) is set in the path ofthe flowing fluid. As the fluid
enters the open volume between the blades of the rotor, it is deflected
by the angle of the blades and imparts a force causing the rotor to turo.
The speed at which the rotor turos is related, over a specified range,
linearly to flow rate.
Several methods are employed to transmit this motion to a readout
device outside ofthe conduit. In some applications a mechanical device
convers the rotor motion directly to a register. In process applications,

Fig. 4-25. Geometryofvortex generator.

FLOW MEASUREMENT 123

however, the usual method is to use an electrical method. A coil containing a permanent magnet is mounted on the meter body. The turbine
flowmeter (Figure 4-26) consists of a section of metal pipe, a multibladed rotor mounted in the center ofthe straight-through passage,and a
magnetic pickup coil mounted outside the fluid passage. A shaft held in
place by fixed radian vanes supports the rotor assembly. As each blade
tip of the rotor passesthe coil it changesthe flux and produces a pulse.
The total number of pulses indicates the volum e of fluid which has
passed through the meter and the rate of the pulses generated becomes
a measure of flow rate.
Turbine flowmeters are frequently employed as sensors for inline
blending systems.
Turbine flowmeters bave excellent accuracy and good rangeability.
They are limited to clean fluids. They are expensive, but do bave
unique features.
Other Flowmeters

Other ftowmetersthat mar be found in someprocessplantsincludethe

ultrasonic,the thermal, a varietyofpositive displacementtypes, metering pumps,and others. The ftowmetersdescribedaccountfor the majority of the ftow rate measuringdevices in everydayuse.

Fig. 4-26. Thrbineflowmeter.

124 FEEDBACK PROCESSCONTROL

Conclusion

Unfortunately, there is no best flowmeter. Factors such as accuracy,

pressure loss, material to be measured, ease of changing capacity, and
ease of installation along with cost must be carefully considered. Once
the details of the problem bave been gathered and the possible alternatives considered, the best selection will usually be clear. Some of the
systems described read correctly to within a few percent. Some are
considerably better. All should bave good repeatability-the all important criterion for control.

Questions
4-1. Match the head meter primary devices with the application (select asingle
rest answer for each):
Orifice plate ~
r.I. T-.
Hgh-pressure~~~._~recovery
Flow nozzle Venturi tube Pitot tube-

Elbow taps-e.

-c.

b. Air ducts
Sedimentin lquid
d. Economyand accuracyare
important
No straightrun available

4-2. A differentialpressuretransmitteris calibratedOto 80 inchesof water

and transmitsa 4 to 20 mA dc signal.This transmitteris placedacrossan
orifice plate which is sizedto create80 inchesof water differentialat 6 gallons
per minute.
Whatis the flow rate whenthe signalis 13 mA dc?
8. 4.5 gpm
c. 3.4 gpm
b. 3.9 gpm
d. 4.9 gpm
4-3. For the equipmentdescribedin the precedingquestion,what will the
signalbe if the flow rate is 4 gpm?
8. 13.3mA dc
c. 8.9 mA dc
b.11.2mAdc
d.6.7mAdc
4-4. Assumethe line used in Questions2 and 3 is a Yz-inchschedule40pipe
(IV = 0.622inches).Pind the orifice boTeusedto satisfythe conditionsgiven
if the measuredfluid is water at 600P.
a. 0.190inches
c. 0.414inches
b. 0.556inches
d. 0.352inches
4.5. The bestchoiceof orifice taps in the precedingproblemwould be:
a. flangetaps
c. pipetaps
b. venacontractataps
d. cornertaps

FLOW MEASUREMENT 125

4-6. A 2-inch schedule 40 line {lD = 2.067) is used to carry gasoline (SP
GR = 0.75). The ftow rate is measured with an orifice plate (d = 1.034)and
pipe taps are used. At full ftow rate a ditIerential pressure of 50 inches of
water is produced. What is the approximate full ftow rate in gpm?
8. 30 gpm
c. 250 gpm
b. 53 gpm
d. 36 gpm
4-7. Assume an 8-inch schedule 160 pipe (ID = 6.813 inches) carries a full
ftow of 40,000pounds per botir of dry saturated steam. The static pressure is
335 psi. Flange taps are used and the ditIerential pressure across the orifice
plate at full-ftow rate is 1OOinches of water. What is the size of the bare in the
orifice plate?
8. 4.357 inches
c. 2.271 inches
b. 3.645 inches
d. 5.109 inches
4-8. Fuel gas is carried in a 6-inch schedule 40 pipe (ID= 6.065 inches).
Flow rate is measured with an orifice plate using fiange taps and the bare in
the orifice is 3.457 inches. The specific gravity is 0.88, the fiowing temperatur~
is 60F, and the static pressure is 25 psi. Maximum fiow rate creates a
ditIerential of 20 inches of water. What is the approximate fiow rate in
cubic feet per day?
8. 1,500,000SCFD
c. 2,500,000SCFD
b. 2,000,000SCFD
d. 3,000,000SCFD
4-9.
8.
b.
c.

The output of a target fiowmeter is:

Proportional to volumetric fiow rate
Proportional to the square of volumetric fiow rate
Proportional to the square root of volumetric ftow rate

4-10.
8.
b.
c.
d.

The output signal or reading of a magnetic fiowmeter is:

Proportional to volumetric fiow rate
Proportional to the square of volumetric fiow rate
Proportional to the square root of volumetric fiow rate
Inversely proportional to volumetric fiow rate

4.11. With suspended solids and/or entrapped gas in a fiowing liquid, the
magnetic fiowmeter will:
8. Read high
b. Read low
c. Read the liquid fiow only
d. Read the correct total volume of the mixture
4-12. A turbine fiowmeter produces an output in the form of pulses. The total
number of pulses is:
8. Inversely proportional to fiow
b. Directly proportional to total ftow
c. Proportional to the square root of fiow
d. Proportional to the square of fiow

Temperature and Humidity

Measurements

Temperature and moisture measurementsare important in process con-trol,

both as direct indications of system or product state and as indirect indications of such factors as reaction rates, energy flow, turbineefficiency,
and lubricant quality.

Temperature

Presenttemperaturescalesbave been in use for about 200years. The

earliest instruments-variations of the common stem thermometer-were
basedon the thermal expansionof gasesand liquids. Suchfilled
systemsare still employedfrequently for direct temperaturemeasurement, although many other types of instrumentsare available. Rep-resentative
temperaturesensorsinclude:
Filled thermal systems
Liquid-in-glassthermometersThermocouplesResistance
temperaturedetectors (RTDs)
126

TEMPERATUREAND HUMIDITY MEASUREMENTS 127

Thermistors
Bimetallic devices
Optical and radiation pyrometers
Temperature-sensitive profitS
Various types of measurement control systems are compared in Table

5-1.
Instrument selection must anticipate overall control requirements.
Low cost often justifies consideration of filled systems for measurements below 1,200oPor 650C. Other advantages of mechanically or
pneumatically transmitted temperature measurements include lowexplosion hazard, simple maintenance requirements, high reliability,
and independence Crom external power. Advantages of electrical systems include higher accuracy and sensitivity, practicality of switching
or scanning several measurement points, larger distances possible between measuring elements and controllers, replacement of components
Table5.1. Comparisonof TemperatureMeasuringSystems
ComparisonFactors
Purchase cost
Long distance transmission
Change or replacement
o components
lnstallation costs
Maintenance
Averaging measurement
Surface measurement
Time constant
(bare bulb and no well)
Temperature difference
Sensitivity

Accuracy
Operating costs

Least
Favorable

Most
Favorable

lntermediate

F
E

p
p

EandP
F
P
T
T

F
P
E

Bo

F
F
R

T
R

E
R
F

Key: F Filled system.

P PneumaticaIly transmitted filled system.
T ElectricaI thermocouple system.

F
F
P
F
E

F
R

T
F
P and T
P

R Electrical RTD system.

E Electrical thermocouple and RTD systems equally rated.
oGeneral purpose wiring only. Explosion-proof
electrical systems cost considerably more than
other types.

128 FEEDBACK PROCESSCONTROL

(rather than complete systems) in the event of failure, fast response,

and ability to measure higher temperatures.

Filled Thermal

Systems

Filled thermal systems, which traditionally bave been used most in

the food, paper, and textile industries, consist of sensors (bulbs) connected through capillary tubing to pressure or volume sensitive elements (Figure 5-1). These systems are simple and inexpensive and
generally bave fast dynamic responses. Their use with pneumatic and
electronic transmitters has removed inherent distance limitations of
filled systems and has minimized the danger of capillary damage.
Moreover, transmitter amplification has made narrow spans practical
and has improved linearity and response.
Application specifications of several types of filled systems are
listed in Table 5-2. These include the Scientific Apparatus Makers Association (SAMA) Classes I (liquid-expansion), 11(vapor-pressure), III
(gas-pressure), and V (mercury-expansion). The SAMA Class 11
classifications also include alphabetical designations, by which A and B
indicate sensorabove and below case ambient, respectively; C indicates
a system in which the sensorcan cross ambient; and D denotes a system
which can operate at ambient conditions.
Liquid-expansion systems are characterized by narrow spans,
small sensors, uniform scales, high accuracy, and capability for difIerential measurements. Class lA devices bave an auxiliary capillary and

Fig. 5-1. Filled measurement

systemsgenerallyare the mostinexpensiveway or
measuringandcontrollingtemperature.

TEMPERATURE AND HUMIDITY

MEASUREMENTS

129

Table5-2. Instrumentsfor Filled ThermalSystemSensors

Gas

Type
Principie
SAMA Class

Liquid
Volumechange

fluids
Lower range limit
Upper range limit

Organic liquido
(Hydro-carbons)
-200"F (- I3IY'C)
+6OO"F(+315C)

Narrowest span (b)

Widest span

40F (25C)
600'F (33IY'C)

71Y'F (4IY'C) (c)

Ambient temperature

lA Full

Not required

Compensation
Sensor size

IB Case

Smallest

Medium

Largest

Typical sensor size

for 1000Cspan

9.5mm (0.375 in.) 00 x

48mm (1.9 in.) long

9.5mm (0.375 in.) OD x

50mm (2 in.) loog

22mm (,. D.) OD x

70mm (6 n.) lang

Overrange capabiJity
Sensor elevation elfect

Medium

Least

None

Class llA, Ves

Class lIB, NODe

Greatest
None

Barometric pressure
effect (altitude)

None

Slightly (greatest on
smaIl spans)

Scale uniformity

Unifonn
"'.5 to "'1.0% ofopan

Non-Uniform

Uniform

;:0.5to;:I.0%ofspan
upper % of scaIe

;:O.5to;:I.0%of

spon

#1-ClasslIA

#2

Accuracy

Response (d)
#1 Fastest

Vapor (a)
Pressure change
11

Pressure change
III

arganic liquido

Puregases

(Hydro-carbons),
water
-425F

(-2SS"C)

+6OO'F (+3IS0C)

400'F (215C)

-45SOF (-2700C)
+l,4OO'F(+7ro'C)
1200F(700C)
1.000'F (5500C)

IIIB case

#3-Class

Slightly (greatest o
small spans)

lIB

#4 Slowest
Cost
Maximum
CapiIlary

standard
length

Highe.1

Lowest

Medium

Clos. lA 30m or 100fI.

30m or 100 fi.

30m or 100I

Clas. IB 6mor 20fi.

(a) Class 11 systems are supplied as either SAMA Class llA or lIB. In Class llA, sensor is aIways
hotter than tubing or instrument case. In Class lIB, sensor is always cooler than tubing or case.
(b) Narrowest spans vary at elevated temperatures.
(c) Smaller spans available in cryogenic regions.
(d) Actual values depend on range, capillary length, sensor dimensions, and type of instrument used.

element to provide ambient temperature compensation and IB systems

often utilize bimetallic techniques. However, fully compensated
liquid-expansion systems are complex and, therefore, expensive.
Vapor-pressure systems are reliable, inherently accurate, and require no compensation for ambient temperature effects. Instruments
follow the vapor-pressure curves ofthe filled fluid, and associated dials
and charts are thus nonunifonn, featuring more widely spaced
increments at high temperatures. Measurements occur at the interface
between liquid and va,porphases of the filling medium. If the temperatufe in the sensorexceeds that in the capillary and indicating element, the
sensoris filled with vaporwhile the capillary and indicatorcontain liquido

130 FEEDBACK PROCESSCONTROL

The converse is true when the relative temperature polarity is reversed.

A transition between liquid and vapor can cause erratic operation, so
that vapor systems mar be unsuitable for ranges that cross the element
and capillary temperatures. Systems mar also be unacceptable if uniform recording or indicating scales are desired.
Gas-pressure systems, which rank second to the vapor-pressure
devices in simplicity and cost, ofIer the widest range of all filled systems. Conventional designs use large-volume sensors, which mar be
shaped to suit particular applications. For example, in duct-temperature averaging, the sensor mar be constructed of a long length of
tubing of small cross section. Conventional recorders are Dot recommended for temperature spans of less than 200F or 110C, but transmitters operating on force-balance principIes can be utilized with spans
as narrow as 50F or 28C. With gas-filled systems, it is difficult to
compensate for ambienttemperature errors, but a sufficiently large sensor
size mar reduce efIects to acceptable limits.
Mercury-expansion systems are classified separately from other
liquid-filled systems because of the unique properties of the fluid. For
example, mercury is toxic and harmful to some industrial processes
and products, and high-liquid density places limitations on sensor-toinstrument elevation difIerences. Sensors used in mercury-expansion
systems are generally larger in diameter and more expensive than those
used in either liquid or vapor systems. For these reasons, mercury is
frequently bypassed in favor of other filling media.

Electrical

Systems

Electrical temperature sensorsbave long been popular in the metal and

paper industries, but increasing use of ele<tronicdevices has stimulated application in a variety of other areas. Thermocouples and resistance temperature detectors (RlDs) are most widely used, and representative application data are given in Table 5-3.
Thermocouples

Thermoelectricity was discovered by Seebeck in 1821.He observed an

electromotive force (emt) generated in a closed circuit oftwo dissimilar
metals when their junctions were at different temperatures. This electricity, produced by the direct action of beat, is used today to measure
temperatures from subzero to high ranges.

TEMPERATURE AND HUMIDITY

MEASUREMENTS

131

Table5.3. Quick-ReferenceSelectorChart for Standard

Thermocouples& RTDs
TEMPERATURE LIMITS
Sensor

Calibration

r -F
oc

RTDs

Nickel
Platinum

SAMA Type 11
SAMA 100 ohm, or

-200 to 315

-320

DIN 43760

-200 to
O to
-210 to
-270 to
-270 to
-270 to
-50 to

-320 to 1,200
32 to 300

Copper
SAMA
Thermocouples Iron-Constantan, ISA Type J
Copper-Constantan, ISA Type T
Chromel-Alumel, ISA Type K
Chromel-Constantan, ISA Type E
Platinum-Platinum Rhodium,
ISA Types R & S
ISA Type B

650
150
760
370
1,260
870
1,480

o to 1,700

to 600

-350
-455
-455
-455
-55

to
to
to
to
to

1,400
700
2,000
1,600
2,700

o to 3,100

A thermocouple consists basically of two dissimiliar metals, such

as iron and constantan wires,joined to produce a thermal electromotive
force when thejunctions are at different temperatures (Figure 5-2). The
measuring, or bot, junction is inserted into themedium where the temperature is to be measured. The reference, or cold, junction is the open
end that is normally connected to the measuring instrument terminals.
The emf of a thermocouple increases as the difference in junction
temperatures increases. Therefore, a sensitive instrument, capable of
measuring emf, can be calibrated and sed to read temperature directly.

<
MEASURING
JUNCTION

Fi2. 5-2. Thennocouple.

THERMOCOUPLE
[JINSTRUMENT
REFERENCE
JUNCTION

132 FEEDBACK PROCESSCONTROL

Figure 5-3 indicates the approximate emf-millivolt output versus

temperature relationship for the most popular thermocouple types. Deviation from the essentially linear relationship is generally less than 1

percent.
Introduction of intermediate metals into a thermocouple circuit will
Dot affect the emf of the circuit, provided the new junctions remain at
the same temperature as the originaljunction. The algebraic sum ofthe
emf' s in a circuit consisting of any number of dissimilar metals is zero,
if all of the circuit is at a uniform temperature. Repeating the law, if in
any circuit of solid conductors the temperature is uniform from Point 1
through all the conducting material to Point 2, the algebraic sum ofthe
emf' s in the entire circuit is totally independent of the intermediate
material and is the same as if Points 1 and 2 were put in contact. If the
individual metals betweenjunctions ofthe circuit are homogeneous,the

Fig. 5-3. Temperature-emf

valuesapproximatethe relationshipof six standard
thermocouples.The conditionsof a particularinstallationmar requireloweringthe
maximumtemperatureto avalue belowthat listed. Temperature/millivolt
curvesare
availablein the appendix.

TEMPERATUREAND HUMIDITY MEASUREMENTS 133

sum ofthe thermal emf's will be zero, provided only that thejunctions
of the metals are all at the same temperature.
The emf in a thermoelectric circuit is independent of the method
employed in forming the functions as long as all the junction is at a
uniform temperature and the two wires make good electrical contact.
The junction mar be made directly, by welding, or by soldering. Furthermore, an instrument for measuring the emf mar be introduced into
a circuit at any point without altering the resultant emf, provided the
junctions that are added to the circuit by introducing the instrument are
all at the same temperature. If the temperatures of the new junctions
are DOt uniform, the effect is that of introducing additional thermocouples into the circuit.
Reference Junction
To make accurate temperature measurements with thermocouples, the
reference junction temperature must remain constant; if it varies, suitable compensation for these variations must be provided.
Should there be an uncompensated variation in the reference junction temperature, there will be a corresponding change in the millivoltage with a resultant error in temperature measurement (Table 5-4).
When used in the laboratory and for other checking and testing
purposes, the thermocouple reference junction can be placed in a vacuum bottle filled with shaved ice saturated with water. This method
provides close temperature control (within a fraction of a degree) and
permits accurate reading.
Table5-4. StandardLimits or Error
Coupleor Wire

Range

Limits

Copper-constantan

-300 to -75F
-75 to +2000F
200 to 700F
-100 to + 530"F

:t2% oCreading
:tl~oF:t~
% oCreading
:t4F

Chromel-alumel

530 to l ,400"F
Oto 530F
530 to 2,300"F

:t~% oCreading
:t4F

Platinumrhodiumplatinum

Oto 1,000F
1,000to 2,700"F

Iron-constantan

:t~% oCreading
:tJoF
:tO.J% oCreading

75
KX

134 FEEDBACK PROCESSCONTROL

~~ WIRE~[]

THERMOCOUPLE

MEASURING
JUNCTION

EXTENSION

<

INSTRUMENT

CONNECTION
HEAD

REFERENCE
JUNCTION

Fig. 5-4. Typical thermocoupleand extensionleads.

To ensure accurate readings, most thermocouples are now installed

with instruments thai provide automatic reference junction compensation. In most instruments, ibis is accomplished by passing current
through a temperature-responsive resistor, which measures the variations in reference temperature and automatically provides the necessary compensating emf by means of the voltage drop produced across

it.
An industrial installation generally consists of a thermocouple with
its connection head, the necessary length of extension wire, and an
indicating, recording, or controlling instrument with internal and automatic reference junction compensation. The extension wires generally
consist of the same materials as the thermocouple elements, or mar be
composed of other materials and alloy wires that generate essentially
the same millivoltage as the thermocouple for application temperatures
up to approximately 400F or 200C(Figure 5-4). The ISA symbols and
color codes for thermocouple lead wires are given in Table 5-5.

Table 5.5. Extension Wire Type Designations and Standard Limits of Error
[SA

Extension

Type

Wire

Color
Code

TX

Copperconstantan
Ironconstantan
Ironcupronel
Chromelalumel
CopperCuNi alloy

+ (blue)
-(red)
+ (white)
-(red)
+ (green)
-(red)
+ (yellow)
-(red)
+ (black)
-(red)

JX

wx

sx

Limits 01
TemperatureRange
F
Error
-75 to +200

:t 1~F

Oto

400

:t4 F

75 to

400

:6 F

to

400

:t4 F

75 to

400

:10 F

Couple
UsedWith
Copperconstantan
Ironconstantan
Chromelalumel
Chromelalumel
Platinum
Rhodiumplatinum

Fig.
<

TEMPERATURE AND HUMIDITY

MEASUREMENTS

135

5-5. Thermocouplesin para11el.

A verage Temperatures
To measure the average temperature across a large duct or vessel, or
around a retort, any number of thermocouples mar be used in parallel
connections. The voltage at the instrument, or at the point of parallel
connection, is the average of that developed by the number of thermocouples used. This voltage is equal to the sum of the individual
voltages divided by the number of thermocouples. For accurate measurement, the resistances of aU thermocouples and extension wire
circuits should be identical. Since the resistance of the actual thermocouple wiU vary with temperature, and since the lengths of extension wires mar also vary, the efIect of these variations can be minimized by using swamping resistors. The values of the swamping
resistors should be high in comparison with the change or difIerence in
resistances encountered. A resistor value of 1,500 ohms generaUy
works weU.
In order to prevent the flow of current through a ground loop, the
thermocouples should not be grounded. AU thermocouples must be of
the same type and must be connected by the correct extension wires
(Figures 5-5 and 5-6).
ParaUel connection of thermocouples for average temperature
measurement is advantageous because the instrument construction and

<

Fig. 5-6. Parallelthennocoupleswith swampingresistors.

136 FEEDBACK PROCESSCONTROL

calibration can be the .sameas that for a single thermocouple. Two or

more thermocouples mar be connected in series to measure average
temperatures; however, this circuit requires that the instrument bave
special reference junction compensation to provide the increased compensating emf for the specific number of thermocouples in the circuit.
The instrument also must be calibrated for the total millivoltage output
of the number of thermocouples used in series. Extension wires Crom
each thermocouple must be extended back to the actual reference junction at the instrument.
Differential Temperatures
Two thermocouples may be used for measuring differential temperatures between two points. Two similar thermocouples are connected
with extension wire of the same material as is used in the thermocouples. The connections are made in such a way that the emf's
developed oppose each other. Thus, if the temperatures of both thermocouples are equal, regardless of the magnitude, the net emf will be
zero. When different temperatures exist, the voltage generated will
correspond to this temperature difference, and the actual difference
may be determined Cromthe proper calibration curve. An instrument
calibrated either for Omillivolts at midscale or Omillivolts at the end of
the scale, depending upon whether thermocouples are operated at high
or low temperatures with respect to each other, may be fumished and
used to read temperature difference directly.
Copper leads may be used between the instrument and the connection box that links the instrument to the thermocouple or extension
wires (Figure 5-7). The instrument should Dot bave the reference junction compensation normally fumished when measuring temperatures
with thermocouples. As in the case of parallel thermocouple connections, the thermocouples should Dot be grounded and the resistance of
both thermocouple circuits should be the same.

Fig. 5-7. Two thermocouplesusedto measuretemperaturedifference.

TEMPERATUREAND HUMIDITY MEASUREMENTS 137

Thermocouple Calibration Curves

Thermocouples develop an extremely small dc voltage, measured in
thousandths of a volt. The millivolt output falls within a range of -11 to
+ 75 millivolts, depending on the type of thermocouple and its temperatufe working range. Calibration curves for thermocouples are included
in the Appendix. Now let us consider some typical thermocouple instruments and investigate their functions.

PotentiometricRecorder
Figure 5-8 shows a simplified circuit diagram of the Foxboro potentiometric recorder (Figure 5-9). An emf input (Ex) is derived Croma
thermocouple or other measuring element. A constant emf (Ez) is
supplied by a Zener diode regulated power supply. As Ex varies, an
error signal is developed, and the circuit becomes unbalanced.
The error signal is converted to an ac voltage by a field effect
transistor chopper and amplified by the integrated circuit amplifier. The
amplified output drives a two-phase balancing motor. The direction of
rotation depends on whether Ex has become greater or smaller. The
motor moves the wiper contact on the slidewire Rsto a position where
the circuit is rebalanced. The slidewire coritact is mechanically con-nected
to the recorder pen, and both are positioned simultaneously.
With a thermocouple input, a temperature sensitive resistor (Rc)
automatically compensates for referenc junction ambient temperature
variations. This resistor changes the balance of the circuit to cancel

138 FEEDBACK PROCESSCONTROL

Fig. 5-9. Foxboro stripchart recorder.

changes in Ex with reference junction temperature variations. For

straight millivolt inputs, Rc is a temperature-stable resistor.
The slidewire is characterized to allow the use of uniform charts
and"scales with thermocouple and certain other nonlinear measurements. Thisis accomplished by varying the resistance per unit length
along the slidewire.
lf the temperature measurement is used in conjunction with an
electrical control system, a thermocouple-to-current transmitter mar
be used. A typical transmitter is shown in Figure 5-10. The transmitter
provides a two-wire output; the same wiring 'is used for both power and
output. The load resistance is connected in series with a dc power
supply (approximately 25 volts), and the current drawn Cromthe supply
is the 4 to 20 mA dc output signal.
Field mounting the transmitter at or near the actual measurement
point eliminates the installation of special thermocouple extension
wires to the receiver. ln some cases, this procedure can also improve
the performapce of the measurement loop, since the possibility of extension wire error is avoided.
As shown in Figure 5-10, the thermocouple/millivolt bridge circuit
consists of a precision-constant current source, a reference-junction
compensator, a bum-out detection current source, and associatedresistors. The amplifier is a very high gain, chopper-stabilized

TEMPERATURE,
AND HUMIDITY MEASUREMENTS 139

Fig. 5-10. Simplifiedcircuit diagramof the Foxboro thermocoupletransmitter.

(Bottom)Foxborothermocoupletransmitter.

amplifier. It functions as a null detector by controlling the 4 to 20 mA dc

output corrent to maintain a null between the compensated thermocouple and feedback volt3;ges. The transmitter is protected from
reverse polarity connection ofthe power supply by means ora diode in
the negative lead of the output circuit. The maximum error for this
transmitter is approximately :t0.25 percent of span.
Many control instruments, such as this transmitter, feature thermocouple bum-out protection. In the event of an open thermocouple, a
s mall voltage applied to the instrument causes it to read full-scale output. This causes the control system to shut off. the beat and prevent
damage.
Resstance Thermal

Detectors

Resistancethermometry is based on the change of electrical conductivitv with temperature.Therefore.a coil of wire can act as a tem-

140 FEEDBACK PROCESSCONTROL

perature sensor, with 'a direct relationship established between resistance and temperature. Standard curves are available, with certified
accuracies within O.IF or oC. Platinum RTDs used as laboratory standards can be obtained with tolerances well within this lmit, and are
capable of precise temperature measurement up to l,650F or 900C. If
an RTD is adjusted to conform to its curve, it may be interchanged with
other RTDs calibrated according to the same curve.

Resistance-thermaldetector temperature measurementmay be

read out on many different types of instruments. If the measurement of
temperature were to be used in conjunction with a pneumatic control
system, the RTD could be attached to a resistance-to-pneumatic convertor, such as the Foxboro Model 34B (Figure 5-11).
The resistance-to-pneumatic convertor converts the temperature
measured by a resistance-temperature element into a proportional 3 to
15 psi or 20 to 100kPa pneumatic output signal. This signal is suitable
for use with many types of pneumatic receiving instruments, such as
controllers, recorders, and vario us computing instruments.
This transmitter has several features. For example, when used with

Fig. 5-11. Foxboro 34BSeriesresistancepneumatictransmitter.

TEMPERATURF;AND HUMIDITY MEASUREMENTS 141

Fig. 5-12. Simplified circuit diagram of 34B Series nickel RTD transmitter.

the appropriate nickel bulb, the converter has an output that is linear
with temperature. Spans as low as 5F or 3C can easily be achieved.
An adjustable range allows a simple field calibration procedure to
change the temperature input range required for a 3 to 15 psi or 20 to
100kPa output. It also operates on all normal supply voltages.
This transmitter is an electromechanical device consisting of a
solid-state, integrated circuit amplifier and an output transducer. Figure 5-12 isa simplified circuit diagram of the nickel RTD version of the
transmitter. The transd ucer is shown in Figure 5-13.
The RTD is wired into a measurement bridge and excited Croma
regulated direct current power supply. The change in resistance of the
RTD causes a bridge output change that is proportional to temperature.
Negative feedback is obtained Cromthe output current and applied to
the opposite side of the measurement bridge. A change in feedback by
the span adjustment changes the gain of the amplifier and thereby
changes the span of measurement.

Fig. 5-13. Outputtransducer.

142 FEEDBACK PROCESSCONTROL

The platinum RTD transmitter incorporates a negative impedance,integrated

circuit amplifier that shunts the platinum RTD to linearizethe
temperature versus the RTD characteristic.
The amplifier section provides the high input impedance and gainrequired
to amplify the low-input signal. The input signal is first converted to ac by an FET (used as a chopper) and then applied to a
high-gain ac amplifier. Th signal is next demodulated and applied to a
voltage follower. The output of this amplifier is applied to a power
transistor to supply the 10to 50 mA dc required by the transducer coil.
A functional diagram applying to the output transducer is shown in
Figure 5-13. The direct current input signal is converted into a proportional air output signal as follows: A coil is positioned in the field of a
fixed, permanent magnet. When a direct current flows through the coil,
the electromagnet and permanent magnet forces are in opposition. Any
increase in the current, which causes a proportional increase in the
force on the coil, tends to close the flapper against the nozzle. The air
output pressure from the pneumatic relay is thereby increased. This
increased pressure is fed to the feedback bellows to exert a force on the
force beam to rebalance the force from the coil. All ofthe forces which
act upon the force beam are balanced in this manner, and the pneumatic output tracks the current input. The capacity tank (C2) and the
associated restrictor provide dynamic compensation to the circuit. During rapid input changes, the restrictor will effectively keep the capacity
tank out of the circuit for faster feedback response through the relay.
For slow changes, the capacity tank acts to damp pressure fluctuations
and to provide a smoother output. The small capacity tank (C.) in the
output circuit provides necessary constant damping.
During alignment or calibration of the output transducer, zero adjustment is made by tuming a screw to advance or retract a spring
which bears on the force beam. Coarse span adjustments are made by
repositioning the feedback bellows with respect to the pivot.
Wheatstone Bridge Recorder
One of the most popular instruments used with RTDs is the slidewire
type Wheatstone bridge recorder. This instrument is almost identical
to the potentiometric recorder previously described and shown in
Figure 5-9.
Figure 5-14 shows a simplified circuit diagram of the Wheatstone
bridge recorder. The resistance temperature detector (RTD) is one arm
of a Wheatstone bridge excited by a Zener diode regulated dc power
supply. Point A (the slidewire contact) and roini B form the input to the
amplifier. When the temperature changes, the resistance of the RTD

TEMPERATURE AND HUMIDITY

MEASUREMENTS

143

Fig. 5-14. SimpIified Wheatstone bridge recorder circuit.

changes. This unbalances the bridge and creates an error signal between Points A and B.
As in the potentiometric recorder, the error signal is converted to
art ac voltage by a field-effect transistor chopper and amplified by the
transistorized amplifier. The output drives a two-phase balancing
motor. The direction of rotation depends on the polarity of the error
signal. The motor moves the wiper contact on the slidewire until the
bridge is rebalanced and no error signal exists. The pen and slidewire
contact are mechanically connected and therefore positioned simultaneously.
With all resistance-temperature measurements, the use of threeconductor RTD cable is recommended. The effect of ambient temperatufe variations on the cable is thereby minimized.
If the cable connecting the RTD to the instrument has only two
conductors, these conductors become part of the resistance being measured. The result then is an error that will vary with ambient temperature. Remember, with all resistance temperature measurements,
three-conductor RTD cable is recommended. The purpose of the threeconductor cable is to stretch out the measuring bridge. Note the threeconductor cable in Figure 5-14. One of the conductors is common to
both sides ofthe bridge while the other two connect one to each side of
the bridge. Any change in cable temperature will be canceled as both
sides of the bridge are changed a like amount.
Occasionally RTD sensors use a four-wire cable. This is generally
in conjunction with a Kelvin double bridge. The four-wire method does
an excellent job of reducing temperature effects on the cable. The
improve ment over the three-wire method, however, is minimal. In
practice, an RTD may be used with as much as 500 feet of threeconductor cable without the cable creating a perceptible error.

Thermistors

144 FEEDBACK PROCESSCONTROL

Thermistors are made of heat-treated metallic oxides, and most thermistors differ from ordinary resistors by having a negative coefficient of
resistance. Thermistors are available with a nearly linear temperature
resistance relationship, and other types are available with a sharp
change in slope at some characteristic temperature.
A thermistor can replace an RTD as a temperature sensor. The
difficulty lies in obtaining units that fit the desired characteristic curve
within acceptable lmits of accuracy. When this is accomplished and
the thermistor is mounted so as to stand up under process conditions, it
performs the same function as the conventional RTD.
One advantage of the thermistor is that it has a greater resistance
change for a given temperature change than that of the conventional
wire RTD. A disadvantage is that the accuracy available, although
good, is slightly inferior to that of the conventional RTD. This presumably accounts for the thermistor's limited application in the process

instrumentationfield.
Radiation pyrometers utilize an optical system to focus energy
radiat~d from a body onto a sensing system. Manual devices are often
l:I~ed;'in which energy at infrared or visible wavelengths is focused on a
tar~t'aridcompared
with the light output of a calibrated optical filamentJ9automated devices, the energy (usually in the infrared band) is
focusedori a series arcar ofthermocouples. This thermopile produces a
millivolt output related to the temperature of the source. Pyrometers
are used where high temperatures are to be measured or where contact
with the object is impossible. Accuracy is influenced by such factors as
reflections, gases present in the radiation path, and surface emissivity
of the body under measurement.
Humidity

Measurements

Humidity, or the amount of moisture in gases,is expressedin several

differentways, including:
1. Relativehumidity-the actual quantity of water vapor presentin a
given spaceexpressedas a percentageof the quantity of water
vapor that would be presentin the same space under saturated
conditions at the existingtemperature.
2. Absolute humidity-mass of vapor per unit magsof dry gas.
3. Dew point-the saturationtemperatureof the mixture at the correspondingvapor pressure.If the gasis cooled at constantpressure
to the dew point, condensationof vapor will begin.

TEMPERATUREAND HUMIDITY MEASUREMENTS 145

A list of the methodsusedto measurehumidity would include the

following:
RelativeHumidity Measurement
Mechanical(hair, wood, skin)
Wet and dry bulb thermometers
Surfaceresistivity devices(including Dunmore)
Crystal frequencychange
AbsoluteMoistureMeasurement
Gravimetry
Electrolysis
Infrared
Conductivity
Capacitance
Color change
Karl Fischer titration
RF power absorption
Neutron reftection
Heat of absorptionor desorption
Nuclear magneticresonance
Dew Point Measurement
Chilled mirrar
Lithium chloride (including DEWCEL)
Wet bulb thermometer
Direct measurement of relati ve humidity may be accomplished by
equating the change in dimension of a hygroscopic substance with the
variation in the moisture content of the air. The hair element (Figure
5-15) consists of a band composed of a number of selected and treated
human hairs. The membrane element consists of a strip of treated
animal membrane.
The instrument will record relative humidity between 20 and 90
percent with a maximum error of :t 5 percent. The calibration should be
adjusted initially by means of a psychrometer. If the hair element is
exposed to humidities over 90 percent or to free moisture, the calibration should be checked.
.Hair elements can be used at temperatures as high as 160For 70C

146 FEEDBACK PROCESSCONTROL

Fig. 5-15. Hair element.

without damage. However, in this event, they should be calibrated

under these temperature conditions.
The hair element is extremely responsive to changes in humidity,
and will follow gradual or small changes with good speed. However,
following a sudden large change, at least 30 minutes should be allowed
for the measuring element to reach a state of equilibrium.
The psychrometer consists of two temperature measurements.
One, called the dry bulb, measures the ambient temperature, while the
second, the wet bulb, is provided with a wick or sleeve saturated with
water that is evaporating. The wet bulb temperature will normally be
lower than that of the dry bulb. The lowering of this wet bulb temperatufe (wet bulb depression) is an indication ofthe moisture content ofthe
surrounding air, or of its humidity. See Appendix for convenient conversions of wet and dry bulb readings into relati ve humidity.
Example: dry bulb reads 90oP; wet bulb 80oP; difference is 10oP.
Pollowing the coordinates to their intersection, we find the relative
humidity to be 65 percent.
A portable version of this instrument, called a sling psychrometer,
provides a convenient way to check the calibration of relative humidity
instruments. Their air velocity past the bulb should be 15 feet (4.6
metres) per second or higher for accurate results.

'-+'2.

TEMPERATURE AND HUMlDffY

"0

MEASUREMENTS

147

.'0
.]

.."

I
l

,
1001

t,
,
,

..
E

~
~
2 00"
~
~
C
E

~~

'

;l

.,~f-.#"

I I

2
C

f-

.fH-~

1--

--'

"- -17' I
--/

-fi

-:eo

-40

-00

00
OEW POINT

40

10
TEMPERATURE

80

100

100

140

.F.

Fig. 5-16. Dewceloperatingcharacteristics.

The Foxboro dew point measuring system consists of an element,

the DEWCEL, a source of power to energize it, and a temperature
instrument to sensethe dew point temperature. The dew point element
consists of a cylinder of woven gIass fabric with two windings of carefully spaced gold wires. The two windings do Dot contact each other.
The fabric is impregnated with lithium chloride. The windings carry
low-voltage power applied through an incandescent lamp ballast resis-

tor.
Moisture determination by this lithium chloride element is based on
the fact that, for every water vapor pressure in contact with a saturated
salt solution, there is an equilibrium temperature at which this solution
neither absorbs noc yields moisture to the surrounding atmosphere.
This equilibrium temperature is shown in Figure 5-16 as the "DEW-

148 FEEDBACK PROCESSCONTROL

CEL@ Characteristic'Line." Below this equilibrium temperature, the

salt solution absorbs moisture. Above this equilibrium temperature, the
saturated salt solution dries out until only dry crystals are left.
The temperature thus determined may be used as a measure of dew
point temperature, absolute humidity, or vapor pressure.
The relative humidity of the atmosphere can be determined from a
knowledge of the dew point and reference to a vapor pressure curve or
chart (see Appendix, Table A-29).
Absolute measurements are less common in process applications
than relative humidity or dew point determinations, but blast fumace
control is one common application of absolute humidity measurement.

Questions

5.1. To obtainthe mostaccuratetemperaturerecord for the range 100to

150F,the filled thermal systemselectedwould be:
8. ClassI
c. ClassIII
b. Class11
d. ClassV
5-2. A resistancethermometerwould be chosenbecauseof:
8. Ability to measurehightemperatures
b. Economy
c. Higher accuracy
d. Simplicity
5-3. Thermocouplesare often chosenbecauseof:
8. High accuracy
b. Ability to measurehightemperatures
c. Economy
d. Ability to measurean extremelynarrow spanof temperature
5.4. The Class11(vapor-pressure)
thermometeremploysa nonlinearscale
because:
8. It is easierto read
b. It maybe ratioedto ftow measurements
c. The vaporpressurecurve is nonlinear
d. The mechanismemployedcausesnonlinearity
5-5. Which filled thermal system,or systems,would you selectfor
measuringthe room temperaturein your house?
8. ClassI
c. ClassIII
b. Class11
d. ClassV
5.6. A ClassIIIB (gas-filled)systememploysa large bulb to:
8. Minimize the ambienteffectson the connectingcapillary
b. Obtainthe optimumdynamicperformance

TEMPERATURE AND HUMIDITY

MEASUREMENTS

149

c. Make a maximumbulb areaavailablefor measurement

d. Reducethe powerproducedin the element
5-7. A temperaturerangebetween300Fand 310For 149Cto 154Cmusi
be measuredwith the greatestpossibleaccuracy.The besichoiceof system
would be:
a. A copper-constantan
thermocouple
b. A copperRTD
c. A nickel RTD
d. A ClasslA filled thermalsystem
5.8.
a.
b.
c.
d.

A hair elementis usedbecauseit:

Measuresabsolutehumidity
Is the mostaccuratetype of humidity measurement
Is simpleand inexpensive
Measuresdew point

5-9. A lithium chloride elementis usuallycalibratedto read:

a. Relative humidity
c. Absolutehumidity
b. Wet bulb temperature
d. Dew point
5-10. Lithium chloride is usedbecause:
a. It is nonpoisonous
b. It is inexpensive
c. It is extremelyhygroscopic
d. It does noi corrodethe equipment
5-11. Whena wet and dry bulb psychrometeris readto determinerelative
humidity:
a. The dry bulb will read lower thanthe wet bulb
b. The two thermometersmayreadthe same
c. The wet bulb will read lower thanthe dry bulb
d. A formula maybe employedto relatethe wet bulbreadingto relative
humidity
5-12. The purposeof using extensionleadwires thai havethe samethermoelectric
characteristicsas the thermocoupleis to:
a. Preventcorrosionat all junctions
b. Extend the referencejunction backto the instrument
c. Preventcreatingan unwantedreferencejunction
d. Make the thermocouplesystemoperatein standardfashion5-13.
Relative humidity is:
a. The moisture presentin a bodyof air expressedas a percentageof
saturationat the existingtemperature
b. The moisturein a body of air, in gramspercubic meter
c. The temperatureat which moisture will condenseCroma bodyof air
d. The ratio of actualmoisturein a volumeof air to the moisturethai
would exist at optimumcomfort in a similarvolume

150 FEEDBACK PROCESSCONTROL

5-14. A psychrometeris:
a. A hair elementinstrument
b. A "wet and dry bulb" humidity instrument
c. An instrumentthat sensespsychologicaldisturbances
d. An instrumentthat readsdirectly in dew point
5-15. A hygrometeris:
a. Convenientfor measuringspecificgravity
b. An instrumentthat measuresgas weight
c. Any instrumentthat measuresmoisturecontent
d. Anothernamefor psychrometer
5-16. A certainthermocouplehasa specifiedtime constantof 2 seconds.If
the processtemperaturechangesabruptly from 800to 900C,the temperature
readouton the indicatorattachedto the thermocouplearter6 secondselapse
will be approximately:
a. 860C
c. 900C
b.835C
d. 895C
5-17. The air velocity pastthe sensorsof a "wet and dry" bulb instrument
shouldbe:
a. 50 feet per second,minimum
b. 2 metresper second,maximum
c. Approximately4.6 metresper second
d. Any value
5-18. The advantageof usinga three-wirecableto connectan RTD to its
associatedinstrumentis that:
a. Referencejunction errors are eliminated
b. The effectof ambienttemperatureon the cablewill be minimized
c. Potentialfailures will be minimal
d. Resistancein the externalcircuit is reduced
5-19. A thermocoupleinstrumenthas an input resistanceof 50,000ohmsand
is usedwith an IC Type J couple.The leadwire is #18 gauge,120feet long.
What is the approximateerror contributedby the leadwire?
a. :t5.0 percent
c. :t0.05 percent
b. :t0.5 percent
d. :tO.1 percent
5-20. The differencebetweenanRTD calibratedto the NR 226curve and one
calibratedto the NR 227curve is:
a. A differentresistance-to-temperature
relationship
b. Nonexistent
c. Greateraccuracywith the NR 226curve
d. Greateraccuracywith the NR 227curve

Analytical measurements seek to define the contents of a process

stream and thus enable control of its composition. Some analytical
measurements are physical; others are electrochemical. The physical
types include humidity, density, differential vapor pressure, boiling
point rise, chromotography, ultraviolet, infrared, and turbidity, some
of which bave been discussed previously. Still others, such as spectroscopy, osmometry, and polarography, are laboratory techniques
and Dot usually associated with automatic control. Some are occasionally used, bot are beyond the scope of this book.
Many other analytical measurements, such as conductivity, pH
(hydrogen iODconcentration), ORP (oxidation-reduction potential), and
specific iODconcentration, are electrochemical. These measurements,
along with capacitance, will be covered in this chapter.

Electrical

Conductivity

While dissociation into ions and the resulting iOD concentration bave
been adequate concepts in the past, the fact is that DOt all the ions
present are necessarily effective. Some ofthem mar be "complexed,"
151

152 FEEDBACK PROCESSCONTROL

that is, "tied up" to'other ions and unavailable for reation. The concept of activity covers this situation. In short, a given compound will
dissociate to some degree, described by the dissociation constant, into
ions, and some portion of these ions will be active, described by the
activity coefficient. However, in many common reactions, the activity
coefficient is so near unity that concentration and activity mar be used
interchangeably. In a growing number of processes, the actual ion activity is different enough Cromthe ion concentration to make it necessary to use the proper term. In general, electrochemical measurements
measure activity rather than concentration, and it is always desirable to
refer to the measurements as ion activity rather than ion concentration.
Electrochemical measurements all rely upon the current-carrying
property of solutions containing ions. Some techniques measure all ions
present (electrolytic conductivity). Others respond mainly to particular
types of ions-hydrogen ions (pH); oxidizing/reducing ions (ORP); selected ions (ion-selective). These will be discussed separately below.
The ability to conduct electricity, or the reciprocal of electrical
resistance, is called conductance.The unit in which it is measured is the
reciprocal ohm, commonly called mho. The conductance of any conductor depends on the nature of the material, the shape of the conducting path, and the temperature. In analytical work, only the nature ofthe
material, in this case the type and activity of the ions present, is important. Thus, the term conductivity, the conductance of a volume of the
material of unit length and area, is generally used. (Conductivity has
largely replaced an earlier term, specific conductance.)
In actual measurement, a conductivity cell of known geometry is
immersed in the material, and the resistance (or conductance) across
the cell is measured. This gives a measurement that can be calibrated
directly in conductivity due to the known shape of the cell. Conductance and conductivity are related as follows: the greater the length
of a material of given conductivity, the higher its resistance (the lower
the tonductance); but the greater the aTeaofthe material, the lower the
resistante (the greater the conductance). That is,
aTea
conductance= conductivity iength

(6-1)

Sincethe unit of conductanceis mho, the unit of conductivity mustbe

conductivity =mho x

centimeters
squarecentimeters

(6-2)

ANALYTICAL MEASUREMENTS 153

or,
conductivity = mho. cm-1
It is unfortunately common practice to omit the dimensional unit,
so that conductivity is then referred to as mhos. This practice leads to
confusion with conductance, and should be avoided.
The conductivity of most electrolytes at the concentrations and the
temperature ranges normally encountered fall well below unity. For
this reason, the micromho per centmetre, the millionth part of the mho
per centmetre, is normally used.
In the SI system of units, the siemens replaces the mho. One mho
equals one siemens. In conductivity units, one micro mho per centimetre (J.mho/cm)equals one microsiemen per centmetre (J.S/cm).
However, the actual SI unit is the microsiemens per metre, since the
unit of length in the SI system is the metre rather than the centmetre.
One J.S/cmequals 100 J.S/m.
The conductivity of material of unit length and area which has a
resistance of 1,000ohms is 0.001 mho. cm-Ior 1,000micromho .cm-1
(sometimes called simply 1,000 micromhos).
Since the measurement depends on the geometry ofthe cell used, a
cell constant (F) has been defined to describe this geometry simply:

F = length (cm)
aTea(cm2)

(6.3)

When the length is I centimetre and the area I square centimetre,

F = 1.0 cm-I. Here, also, the dimensional unit is omitted in common
use, so that the cell is said to bave a cell constant of 1.0. Note that the
critica! dimensions are length and area, Dot volume. Thus while the
above example has a volume of 1 cubic centimetre ifthe length were 0.5
centimetre and the area were 0.5 square centimetre, the cell constant
would still be 1.0. But the volume would only be 0.25 cubic centimetres. Thus it is Dot correct to say that the resistivity is the resistance of unit volume, but only of unit length and cross-sectional area.
A cell with greater area and the same length will bave a lower cell
constant. But for the same solution (that is, for a given conductivity)
the conductance will be greater with this cell. If we solve Equation 6-1
for conductivity,

(6-4)

154 FEEDBACK PROCESSCONTROL

and substituting Equation 6-2,

conductivity = conductancex cell constant

A given measuring instrument will bave a certain conductance (resistance) range. But it may bave a variety of conductivity ranges by
simply using cells of different cell constants. For example, an instrument with a range of O to 100 micromhos per centimeter with a cell
having a constant ofO.Ol will bave a range oro to 1,000micromhos per
centimeter if a cell with a constant of 0.1 is substituted.

Types

ot Calibration

Conductivity instruments can normally be furnished with

1. Calibration in conductivity (,LLSx cm-! or ,LLmhox cm-!)
2. Calibration in terms of concentration of electrolyte
3. Calibration in terms of conductivity difference
Calibration

in Conductivity

Instruments of this type are measured in absolute units, ohms, mhos,

siemens, or microsiemens. Such instruments can be used to measure
the conductivity of any electrolyte at any solution temperature using
a measuring cell with known cell factor. The conductivity of most
solutions increases as the temperature increases. Therefore, if a conductivity instrument calibrated in J-mhox cm-r is used with a solution
of given concentration, the instrument reading will change if the temperature of the solution changes.
Compensation for this effect of temperature on the conductivity of
the solution is possible only if the solution's conductivity temperature
coefficient is known. For example, if an instrument calibrated CromOto
1,000J-mhox cm-r were provided with temperature compensation for
sodium chloride (NaCI), the instrument would Dot correctly compensate in any other solution. For this reason, instruments calibrated in
absolute units are generally Dot furnished with temperature compensation. Instruments ofthis kind are generally the simplest to calibrate and
lend themselves to more ftexibility of application, since they are Dot
limited to any particular electrolyte. If compensation is desired, it can
be supplied, bot only for one particular solution.

ANALYTICAL MEASUREMENTS 155

Calibration in Terms ot Concentration ot Electrolyte

Instruments calibrated in terms of concentration of electrolyte read
percent concentration, grams per litre, parts per million, and the like, for
the range specified. 'This type of calibration is made to the conductivity
values within the specified range of concentration, and at a specified
temperature of a given electrotyte. The instrument can therefore be
used only under the conditions for which it was calibrated.
However, special cases exist which sometimes make conductivity
useful in multiple-electrolyte solutions. If the material of interest is
much more conductive than others in solution, the contaminants mar
bave a negligible effect on readings. For example, contaminants slowly
build up in a sulphuric acid (H2SO4)bath for treating textiles. The
conductance of the contaminants is very low compared with thai of the
acido Thus, the instrument mar read directly in concentration of sulfuric acido Laboratory tests determine when concentration of contaminants is too high and a new bath is then made up.
Figures 6-1 and 6-2 show typical conductivity curves ofNaCl and
H2SO4.Note thai the H2SO4curve reverses itself in the region of 30 to

Fig. 6-1. Percent concentration. Conductivity oCNaCl at lOO"C(212F) and

1SOC(64.4F).

156

FEEDBACK PROCESSCONTROl

Fig. 6-2. PercentConcentration.Conductivityof H2SO.at 70oP(21.IOC)and

40"P (4.4C).

32 percent concentration and again in the vicinity of 84 percent and 93

percent. Obviously, it is impossible to calibrate for ranges that include
these points of inflection. However, in electrolytes having such maxima they shift with temperature. By controlling sample temperature to
a different value, certain ranges mar sometimes be measured which
would be impossible to measure at process temperature.
Polarization

When an electric current is passed through a solution, electrochemical

etfects known as polarization occur. These etfects, if Dot minimized,
wiU res uit in inaccurate measurement. One of the polarization etfects is
electrolysis. Electrolysis generaUy produces a gaseous layer at the
electrode surface, increasing the apparent resistance of the solution.
For tros reason, direct current voltage is Dot practical in conductivity
measurements. If the current is reversed, the layer wiU tend to go back
into solution. Thus, if an altemating current is applied, the polarization
etfect decreases, since the gases and other polarization etfects produced on one-half of the cycle are dissipated on the other half cycle.
Cell Construction

The sensitive portion of the cells shown i~ Figure 6-3 consists of two
platinum electrodes mounted in an H-shaped structure of Pyrex glass
tubing. The electrodes, located in separate sections of the tubing, are

ANALYTICAL MEASUREMENTS 157

Fig. 6-3. Type H conductivityassembly.

concentrically mounted platinum rings and are ftush with the inside
surface of the tubing. Fouling or damaging the electrodes is thus minimized, and the cells mar easily be cleaned chemically or with a bottle
brush. The platinum electrodes in these cells are coated with platinum
black to minimize polarization etfects.
The cell shown in Figure 6-4 employs graphite, rathr than metallic
electrodes. The type of graphite used has the same surface properties
with respect to polarization as metallic electrodes. These
cells require no platinization. They are cleaned chemically or by wiping
the surface of the electrode with a cloth or brush.
Conductivity cells (Figure 6-5) are used for detecting impurities in
boiler feedwater, for concentration of black liquor (in a pulp digester
for kraft paper), for determination of washing etfectiveness by measurement ofpulp wash, and in many other applications where the presence and concentration of a known salt, base, or acid must be deter-

mined.
Electrodeless

Conductivity

Measurements

In addition to the conventionalconductivity measurementsjust described another approachcalled an electrodelesstechnique mar be

employed.This techniquediffers from conventionalconductivity measurementin that no electrodescontactthe processstream.Instead,two

158 FEEDBACK PROCESS

CONTRO~

Fig. 6-4. Conductivitycells.

toroidally wound coils encircle an electrically nonconductive tube that

carries the liquid sample. The first coil is energized by an ultrasonic
oscillator (approximately 20 kHz) which induces an altemating current
in the liquido This current in turo induces a current in the second coil.

Fig. 6-5. Gatevalve insertiontype conductivitycell.

~
~
~.
~

ANALYTICAL MEASUREMENTS 159

The induced current in the second coil is directly proportional to the

electrical conductivity ofthe lquid carried by the tube. No direct contact between the coils and the solution is required, thus eliminating
potential maintenance problems. Figure 6-6 shows a typical electrodeless conductivity measuring system which transmits a 4 to 20 mA dc
signallinearly related to the measured conductivity.
Electrodeless conductivity is especially applicable to the higher
conductivity ranges such as 50 to 1,000 millimhos per centmetre.
Lower conductivity ranges such as 0.01 to 200,000 micromhos per
centmetre are best handled by the conventional techniques previously
described.

Hydrogen
The

term

pH

solution.
of

It

an

means
is,

at

vary

from

any

in

is

tion

of

the

pH

the

to

hydrogen

degree
amo

unt

ofthat

percent).

of
of
ion.

The

potentiometric

ions

acidity
a

(H+)

or

given
Activity

in

alkalinity

ion

actually

values

measurement

can

ofpH,

measurement

that

as
obeys

explanatech-

to
with

ofthe

of

This

an

measurement

along

activity

activity

1 (100
is

equation.

employed

the

effective

called

chapter,

contains

nique

The

percent)

this

of

a measure

time

Nemst

chapter

a measure

solution.

O (O

discussed

(pH)

therefore,

aqueous

present

the

lon Activity

determine
the

Nemst

equation.
To

aid

further

in

understanding
surement,
of
the
properties
some

a thorough

of
pH
meafundamentals
of
aqueous

,---TOROIO

SECONOARY

!'RI MARY
TOROIO
CELL BORE

~~,
solutions

most

be

understood.

CELL

lonization

or Dissociation

;\)'
.1

Stable chemical compounds

are electrically neutral. When
th
. d Wlthsolution,
ter
to ey
forro
areanmlxe
aqueous
wa
o

:;\
\"'
~

!~~
"",,' -'
ELECTRIC ./
"~~~~

they dissociate into positively Fig.6-6. Electrodeless

conductivity
measuring
and negatively charged parsystemo

160 FEEDBACK PROCESSCONTROL

ticles. These charged particles are called ions. Ions travel from one
electrode to the other if a voltage is impressed across electrodes
immersed in the solution.
Positive ions, such as H+, Na+, and so on migrate toward the
cathode, or negative terminal, when a voltage is impressed across the
electrodes. Similarly, negative ions, such as OH-, CI-, S04-2, and so
on, migrate toward the anode, or positive terminal.
The freedom of ions to migrate through a solution is measured as
the electrical conductivity of the solution. Chemical compounds that
produce conducting solutions are called electrolytes. Not all electrolytes completely dissociate into ions. Those that do (strong acids,
strong bases, and salts) are strong electrolytes. Others dissociate, but
produce fewer than one ion for every element or radical in the molecule. These are poor electrical conductors, and, hence, weak electrolytes. All weak acids and weak bases fall into this class.
At a specified temperature, a fixed relationship exists between the
activity of the ions and undissociated molecules. This relationship is
called the dissociation constant (or the ionization constant).
For hydrochloric acid (HCI), the dissociation constant is virtually
infinite, which means that for all practical purposes, the HCI is completely composed of positively charged hydrogen ions and negatively
charged chloride ions. Because of the essentially complete dissociation
into ions, hydrochloric acid is a strong acid:
HCl -+- H+ + Cl-

Conversely,acetic acid has a low-dissociationconstant. It breaks

up in the following way:
CH3COOH~ H+ + CH3COOFew hydrogen ions show up in the solution-less than one for every
100undissociated molecules-therefore, acetic acid is a weak acido
It follows that the strength of an acid solution depends on the
number of hydrogen ions available (the hydrogen ion activity). This
depends not only on the concentration of the compound in water, but
also on the dissociation constant of the particular compound.
When free OH- ions predominate, the solution is basic, or alkaline.
The dissociation of such a compound is illustrated below.
NaOH -+ Na+ + OH-

ANALYTICAL MEASUREMENTS 161

Sodium hydroxide is, for all practical purposes, completely dissociated and is a strong base.
Conversely, ammonium hydroxide, NH4OH, dissociates very little
into NH4+ ions and OH- ions, and is a weak base.
The OH- iODactivity, or strength, of a base depends on the number
of dissociated OH- ions in the solution. The w\mber available depends,
again, DOtonly on the concentration of the compound in water, but also
on the dissociation constant of the particular compound.
Pure water dissociates into H+ and OH- ions, but is very weak in
the senseused above. That is, very little ofthe HOH breaks up into H+
ions and OH- ions. The number of water molecules dissociated is so
small in comparison to those undissociated that the activity of (HOH)
can be considered 1 (100 percent).

HOH ~ H+ + OHAt 77F or 25C, the dissociation constant or water has been
determined to bave a value or 10-14. The product or the activities
(aH+)(aOH-) is then 10-14.
If the activities or hydrogen ions and hydroxyl ions are the same,
the solution is neutral; the H+ and OH- activities must both be 10-7
mols per litre.
It must be remembered that, no matter what compounds are present in an aqueous solution, the prJduct or the activities or the H+ ions
and the OH- ions is always 10-14at 77F or 25C.
If a strong acid, such as HC1, is added to water, many hydrogen
ions are added. This must reduce the number or hydroxyl ions. For
example, if HCl at 25C is added until the H+ activity becomes 10-2,the
OH- activity must become 10-12.
The pH Scale

It is inconvenient to use terms as 10-7, 10-12and 10-2. Therefore, it has

become common to use a special term to represent degree of acidity, or
activity of hydrogen ions. This term is pH, originally derived from the
phrase "the power of hydrogen."
The pH is defined as the negative of the logarithm of the hydrogen
ion activity, or as the log ofthe reciprocat ofthe hydrogen ion activity.
pH = -log aH+ = log

(-::
aH

(6-5)

162 FEEDBACK PROCESSCONTROL

Table 6-1

Temperature oC

-Log

(aH+)(80H-)

14.94
14.00
13.26
12.69
12.26

25
50
75
100

Neutral pH

7.47
7.00
6.63
6.35
6.13

If the hydrogen iOD activity is lO-x, the pH is said to be x. For

example, in pure water (at 25C), the activity of hydrogen iODis 10-7.
Therefore, the pH of pure water is 7 at 25C.
A point frequently overlooked is thai the pH for neutrality varies as
the solution temperature changes due to a change in the dissociation
constant for pure water, as shown in Table 6-1.
An acid solution contains more hydrogen ions than hydroxyl ions.
Table 6-2
Hydrogen [on

Activity
Mols/Litre

Hydroxyl Jan
Activity
Mols/Litre

10.1

0.00000000000001

1
2

0.01

0.001

0.0001

O.0000 1

0.000000001

0.000001

0.00000001

0.0000001

0.0000001

0.00000001

0.000001

pH

Neutral

9
10
11
12
13
14

0.0000000000001

0.000000001
0.0000000001
0.00000000001
0.000000000001
0.0000000000001
0.00000000000001

0.000000000001

0.0000000000I
0.0000000001

0.00001
0.0001
0.001

0.01

0.1
I

10

ANALynCAL MEASUREMENTS 163

Therefore, the activity of hydrogen ions will be greater than 10-1, that
is, 10-6, 10-5,10-4. The pH of an acid solution then, by definition, must
be lower than 7, that is 6, 5, 4.
If the number of OH- exceeds H+ ions, the hydrogen iODactivity
must be less than 10-1, that is, lO-s, 10-9,10-1. Therefore, the pH will
be higher than 7, that is, 8, 9, 10.
Table 6-2 demonstrates that a change of just one pH unit means a
tenfold change in strength ofthe acid or base. The reason for this is that
there is an exponential relationship between pH numbers and hydrogen
iODactivity. With so large a change in acidity or alkalinity taking place
with a change ofjust one pH unit, the need for sensiti ve pH measuring
and control equipment cannot be overemphasized.
Table 6-3 shows the nominal pH values for a number of common
solutions.

Table 6-3
pH
14 ~ Caustic soda 4% (I.ON)
--Calcium hydroxide (sat'd sol.) (Iime)
12 --Caustic soda 0.04% (O.OIN)11
--Ammonia 1.7% (I.ON)
--Ammonia 0.017% (O.OIN)
9 +- Potassium acetate ~.98% (O.lN)
8 +- Sodium bicarbonate 0.84% (O.lN)

6
5 ~ Hydrocyanic acid 0.27% (O.lN)
3 ~ Acetic acid 0.6%

"'
1 +- Hydrochloric acid 0.37% (O.IN)
+- Sulfuric acid 4.9% (I.ON)
O +- Hydrochloric acid 3.7% (I.ON)
-1 +- Hydrochloric acid 37% (lON)

d
-~---~
)LASrIC

164 FEEDBACK PROCESSCONTROL

Fig. 6.7. Schematicdiagramor glass-tippedpH measurement

electrode.
Measurement

ot pH-The

Glass Electrode

A number of different methods for measuring pH are available for

laboratory use, but only one has proved sufficiently accurate and universal for industrial use: the pH glass electrode measuring system.
A special kind of glass, sensitive to hydrogen ions, has been found
to be the most useful medium for measuring pH. Figure 6-7 shows a
INTEGRAL
CABLE

(
r--y=
MOISTURE

SEAL
CAl

SILVERSILVER
CHLORIDE
COLUMN

SHIELDED
!LECTRODE

CABLE

GLASS-TO-METAL
SEAL
; "TOTAL GLASS" CONSTRUCTION)
NTERNAL

SOLUTION

,H-SENSITIVE

Fig. 6-8. Actual measuringelectrode.

MEMBRANE

ANALYTICAL MEASUREMENTS 165

schematic diagram of the measuring electrode in which this special

glass is used. A thin layer of the special glass covers the tip of the
electrode. Figure 6-8 shows an actual measuring electrode.
This glass contains a chamber filled with a solution of constant pH.
An internal electrode conductor element immersed in the internal solution is connected to the electrode lead. The conductor element is discussed in the section on temperature compensation.
If the hydrogen iODactivity is greater in the process solution than
inside the electrode, a positive potential difference will exist across the
glass tip. That is, the external side of the glass will bave a higher
positive potential, and the internal a lower positive potential. If the
process is lesser in hydrogen iODactivity, a negative potential difference will exist.
The relationship between the potential difference and the hydrogen
iODactivity follows the Nernst equation.
E

E'

lo
nF

aH+o~ts~de
aH+
mslde

(6-6)

where:
E = potential difference measured
Eo = a constant for a given electrode system at a specified temR =
T =
n =
F =
aH+ =

perature (25C)
the gas law constant
Absolute temperature
charge on iOD(+ 1)
Faraday's number, a constant
hydrogen iODactivity

In the most common type of pH electrode, that with an internal

buffer solution of7 pH, the voltage across the membrane will be zero at
7pH.
Reference

Electrode

The potential inside the glass is the output o the measuring electrode.
It must be compared to the potential in the solution outside the giassto
determine potential difference and, hence, pH. The sensing o the potential in the solution must be independent o changes in solution composition. Platinum or carbon would act as aRP, or redox-measuring
electrodes. They would be responsive to oxidants or reductants in the
solution. They would Dot yield a true solution potential with solution
composition changes.

166 FEEDBACK PROCESSCONTROL

A reference electrode is the answer. Figure 6-9 is a schematic

diagram of a typical reference electrode.
The connecting wire is in contact with silver that is coated with
silver chloride. This, in turn, is in contact with a solution of potassium
chloride saturated with silver chloride (AgCI). The saturated KCI,
called the salt bridge, in turn, contacts the process solution.
Because the concentration of all of the components Cromthe connecting wire to the KCI solution is fixed, the potential Cromwire to
KCI is fixed. The potential between the KCI and the process solution
(called the liquid junction potential) is normally small, and will vary
only insignificantly with process changes. The overall potential of the
reference electrode is thus essentially constant, virtually independent
of process solution changes, to meet the requirement mentioned earlier.
As the need for high accuracy or repeatability becomes greater,
protection of the reference electrode's internal environment becomes
more important. AIso, prevention of coatings over the electrode tip
becomes more critical. Leaking in of process solution must be prevented. Noxious chemicals, or even distilled water, would change the
concentration, contaminating the electrolyte. The resuIt would be an
unpredictable change in the reference potential. Conductive coatings

Fig. 6-9. Schematicdiagramofreferenceelectrode.

ANALYTICAL MEASUREMENTS 167

on the electrode tip mar cause spurious potentials, and nonconductive

coatings mar literally open the measurement circuit.
Two types of reference electrodes are available-ftowing and nonftowing. The ftowing version (Figure 6-10) is usually selected when the
highest possible accuracy or repeatability is needed. To prevent electrolyte contamination and to minimize coatings or deposits on the tip,
there is provision for a ftow of KCl electrolyte out through the paTOllS
electrode tip. Instrument air is applied inside the electrode. It maintains
a pressure on the electrolyte slightly higher than thai of the process
liquido A 3 psi (20 kPa) differential pressure forces a trickle of 1 or 2 ml
per dar of electrolyte out into the process solution.
The nonftowing version (Figure 6-11)of a reference electrode is often
considered standard. No external pressure is applied to ibis type of
electrode. The electrolyte is a paste of KCl and water thai actually does
ftow, or diffuse, into the process solution. The ftow rate mar be as low
as 0.01 ml per dar; ibis is a function of diffusion properties. After

Fig. 6-10. Flowing referenceelectrode.

168 FEEDBACK PROCESSCONTROL

Plastlc

Body

Internal
Electrode
Element
AgAgCI

Fig. 6 -11. Construction details of nonfiowing reference electrode.

depletion of the electrolyte, the nonflowing electrode is usually either

recharged or replaced.
Tem peratu re Com pensati on

The major potential difference exists in the reference electrode, between the metallic silver and the silver ions in the AgCI solution. It
follows from the Nemst equation:
(6-7)
Because R, n, and F are constants, and aAgoand AAg+ are fixed,
this potential (E) will vary only with the absolute temperature of the
electrode. Since there is a definite chemical relationship between the
ionic silver and the activity of chloride in the KCl electrolyte, the above
expression mar also be written as:
E = Eo -kT

log aCl-

(6-8)

To compensate for ibis temperature sensitivity, another silversilver chloride electrode is inserted into the top of the glass measuring
electrode, the internal conductor element mentioned previously. As the
temperature of the electrode changes, the potentials of the reference
electrode and the conductor element will vary, but will effectively cancel each other, assuming similar values for aAg+ (or aCI-) in each electrode as is usually the case.

ANAl..YT!CAL MEASUREMENTS 169

A remaining temperature effect influences the potential across the

membrane of the glass measuring electrode. This will vary with the
absolute temperature and is greatest at high- and 10w-pH values. At
values around 7, the variation with temperature is zero. This paint is
called the isopotential paint. Figure 6-12 shows the magnitude of the
temperature error, away from the isopotential paint, with glass electractes.
In general, some type of temperature compensation is essential for
accurate, repeatable pH measurement. If the process temperature is
constant, or ifthe measured pH is close to avalue of7, manual temperature compensation mar be used. Otherwise, for good results, automatic temperature compensation will be required.
Reading the Output

ot the pH Electrodes

The high resistance of the giass measuring electrode results in the need
for a millivolt meter with very high internal resistance or sensitivity to
measure the cell output. The pH electrodes mar be compared to a pair
of ftashlight cells in series, but unlike the ftashlight cell, the electrodes

Fig. 6-12. Graph oCpH measurement errors versus pH values at various solution
temperatures due to temperature differences across the measurementelectrode tip.

170 FEEDBACK PROCESSCONTROL

are characterized by xtremely high internal resistance (as high as a

billion ohms). Unless the millivolt meter employed to measure the
voltage created by the electrodes has almost infinite resistance, the
available voltage drop will occur across the internal resistance and
result in no reading, plus possible damage to the electrodes.
In addition to high impedance or sensitivity, the instrument must
bave some provision for temperature compensation. It must also bave
available adjustments that provide for calibration and standardization.
The Foxboro pH transmitter shown in Figures 6-13 and 6-14 is typical
of an instrument designed to make this measurement. Its output is
typically 4 to 20 mA dc which can be fed to a control system. When pH
is measured in the process plant, it is generally for the purpose of
control.
pH Control

Experience has shown that it is virtually impossible to control pH

without considering the process composition as well as more obvious
parameters offtow rate, pressure, temperature, and so on. This consideration is frequently even more critical where plant waste neutralization is being controlled; in such applications, there are usually wide,
relatively uncontrolled variations in composition as opposed to the
normally better defined, and controlled, process stream.
A pH control system can be a relatively simple on/off controlloop
in some batch processes. At the other extreme, it can be a complex,
feedforward/feedback system with multiple sequenced valves for continuous neutralization over a wide, dynamic pH range.

Fig. 6.13. Block diagrarnoftypical transmitter.

ANALYTICAL MEASUREMENTS 171

Fig. 6-14. Actualtransmitter.

The rate of change of inlet pH and fiow is a major consideration in

determining whether feedforward control is necessary. Since changes
in pH refiect logarithmic changes in composition, they can atrect reagent demands tremendously and rapidly. Since reagent addition varies
directly and linearly with fiow, doubling the fiow requires twice as
much reagent at a given pH. But a change of one pH unit at a given fiow
requires a tenfold change in reagent. Flow rate compensation is often
required in complex pH control loops, but usuany only when fiow
changes are greater than 3 : 1.
The amount of reagent addition for butrered and unbutrered solutions, as seen on conventional acid-base titration curves (Figure 6-15),
is also an extremely important consideration in the design of a pH
control system. Reaction time of the various reagents musi also be

,;,

172 FEEDBACK PROCESSCONTROL

005
BASE

001

REAGENT

001

005
AtiO

OI

REAGENT

MOLS REAGENTREQUIRED/LlTER INFLUENT

Fig. 6-15. Typical neutralization curves for unbuffered solutions (strong acid or strong
base) and buffered solutions. Examples of buffered solutions are (I) weak acid and its
sale with the addition of a strong base, and (2) strong acid or base, concentrated, 0.01

M.

given serious considerationin designingboth the processand the control system.

A pH measurementand control system, in conjunction with a
well-designedprocess,canbe a mostsuccessful,reliable systemwhen
proper considerationhas beengivento the various designparameters.
Summary
Continuous pH measurement is avaluable tool for industry and gives
information thai cannot be obtained economically in any other fashion.
It is generally more complex than, say, temperature or pressure measurements, but with constant use, it becomes a routine analytical tool.

Oxidation-Reduction

Potential

Oxidation-reduction potential or redox measurements determine the

oxidizing or reducing properties of a chemical reaction. In this application, the term oxidation is used in its electrochemical senseand applies
to any material which loses electrons in a chemical reaction. By definition, a reduction is the opposite of oxidation, or the gaining of external
electrons. There can be no oxidation without an attending reduction.

ANAL YTICAL MEASUREMENTS 173

For example, a ferrous iODmar lose an electron and become a ferric iOD
(gaining increased positive charge) if a reduction, say, of stannic to
stannous ions (which is the reverse ofthis operation) occurs at the same
time.
This measurement uses electrodes similar to those in pH measurement (except that metal is used instead of glass), but the two types of
measurement should DOtbe confused. The measurement depends on
the oxidizing and reducing chemical properties of reactants (not necessarily oxygen). The inert metal electrode versus the reference electrode
will produce a voltage that is related to the ratio of oxidized to reduced
ions in solution.
This measurement is similar to that of pH in the requirements that
it places on the voltmeter used with it. It is useful for determinations in
waste treatment, bleach production, pulp and paper bleaching, and

others.
The application of ORP measurement or control depends on a
knowledge of what goes on within the particular process reaction. The
ORP measurement can be applied to a reaction only under the following
conditions:
1. There are present in the solution two reacting substances: one that
is being oxidized and one that is being reduced.
2. The speed of reaction, following an addition of one of the substances above, is sufficiently fast for good measurement or control.
3. Contaminating substances are held to a minimum, especially those
capable of causing gide reactions of oxidation or reduction.
4. The pH ofthe solution is controlled in those areas ofthe applicable
curve where variations in pH can affect the ORP measurement.

lon-Selective

Measurement

Certain applications require that the activity of a particular iODin solution be measured. This can be accomplisehd with an electrode designed
to be sensitive to the particular iODwhose concentration is being measured. These electrodes are similar in appearanceto those employed to

measure pH, but are constructed as glass-membraneelectrodes,

solid-membrane electrodes, liquid-ion-exchange-membrane electrodes,
and silicone-rubber-impregnated electrodes.
A common example ofion-selective measurementis found in many
municipal water treatment plants. In this application, a measurement of

174 FEEDBACK PROCESSCONTROL

ftuoride iOD activity can be related to concentration. The measuring

electrode is made of plastic and holds a crystal of lanthanum ftuoride
through which ftuoride ions are conducted. The reference electrode is
the same as thai used for pH. The electrode output is read on a high
impedance voltmeter very similar to thai used with pH electrodes.
Many applications are possible using the ion-selective technique.
At present, measurements in water treatment include hardness,
chlorine content, waste, and pollution. Many other industrial applications bave been suggested and many bave proven effective.

Chromatography

The original application of chromatography consisted of studying the

migration of liquid chemicals through porous material, usually paper.
The word chromatography means color writing, which occurs when
certain extracts or dyes applied to paper produced colored bands.
Modem gas chromatography dates from 1952,when James and Martin
first used the principIe to separate a mixture of volatile fatty acids
having nothing to do with color.
The modem chromatograph is generally used for gas analysis. It
consists of a column, a tube or pipe packed with materials that will
absorb the gases being analyzed at different rates. The gas to be analyzed is carried through the columns by an inert carrier gas, usually
helium. As the gas mixture passes through the column, different components are delayed for varying increments of time. Thus, as the gas
stream leaves the column and is passed through a gas detector, a
chromatogram, or "fingerprint," is formed which mar be used to determine the components in the gas as well as the quantity present. The
packing material for the column must be properly selected to provide
the separation desired for the sample under analysis. The operating
parameters, such as temperature, carrier gas flow and pressure, sample
valve timing and detector sensitivity, all influence output. Therefore, in
a working chromatograph, these factors must be carefully controlled.
Chromatographs are widely used in process work; in fact, they are
among the most popular analytical tools. They are used for both liquids
and vapors, and adaptations are available to provi de an output in conventional analog form such as a 3 to 15 psi (20 to 100 kPa) pneumatic
signal. The details ofthis instrument' s operations are beyond the scope
of this book. Several references suitable for further study are listed at
the end of this chapter.

cable

ANALYTICAL MEASUREMENTS 175

Table6-4. ComparisonofCapaitance,Conductivity,pH, andORPTechniques

Specificity
Sensitivity
Conducting
fluids
Nonconducting
fluids

Maintenance
InstalIation
problems
Cost

Capacitance

Conductivity

pH or
ion-selective

ORP

Poor
Pair
Not
applicable
Good

Poor
Good
Good

Excellent
Excellent
Good

Poor
Excellent
Good

Not

Not

applil

appIiI

Low
Low

Low
Low

High
High

Not
applicable
Medium
Low

Low

Low

Medium

Low

Capacitance

Capacitance is Dot, as applied, an electrochemical measurement. However, a measurement of some characteristic in a nonconducting liquid is
frequently required, and in these applications, capacitance mar provide
the answer.
Electrical capacitance exists between any two conductors separated by an insulator (dielectric). The amount of capacitance depends
on the physical dimensions of the conductors and the dielectric constant of the insulating material. The dielectric constant (K) for a vacllum is 1; all other dielectric materials bave a K greater than 1. For
example, for air, K = 1.00588; for dry paper, K = 2 to 3; for pure
water, K = 80. The Table of Dielectric Constants of Pure Liquids
(NBS Circular 514), available Crom the U.S. Govemment Printing
Office, lists the dielectric constants of nearly all common liquids.
Mixtures of materials bave a composite value of K that can be
directly related to composition. This approach is readily applied to the
determination of water content in materials such as paper and crude oil.
Capacitance is also applicable to level, interface, octane, and other
measurement problems.
In general, the capacitance technique has provided a solution to
many measurement problems that cannot be solved easily by more
conventional means. Table 6-4 compares the application of pH, ORP,
ion-selective, conductivity, and capacitance measurements to process
situations.

176 FEEDBACK PROCESSCONTROL

References
Conductivity cells. Technical Information Sheet 43-10a. Foxboro, MA., The
Foxboro Company.
Fluoride measuring systems-potable water. Technical Information Sheet
43-21a. Foxboro, MA., The Foxboro Company.
Fundamentals of pH measurement. Technical Infor~{ltion Sheet 1-90a.
Foxboro, MA., The Foxboro Company.
pH e1ectrodesand ho1ders. Technical Information Sheet43-11a. Foxboro,
MA., The Foxboro Company.
Theory and app1icationof oxidation-reduction potentials. Technical
Information Sheet 1-61a. Foxboro, MA., The Foxboro Company.
Pneumatic composition transmitter. Technical Information Sheet 37-1.30a.
Foxboro, MA., The Foxboro Company.
Liptk, B. G. Instrument Engineer's Handbook, Sec. 81, Volume I. Radnor,
PA: Chilton.
Shinskey, F. G. pH and pION Control in Process and Waste Streams.
New York: John E. Wiley and Sons, 1973.

Questions

6-1. Conductivityand pH measurements

are:
a. Two differenttechniques
b. Similar in operation
c. ldentical but given differentnames
d. Two techniquesthat usethe sameequipment
6.2.
a.
b.
c.
d.

The three factors that control the conductivityof an electrolyteare:

Specificgravity, density,and volume
Concentration,material in solution,and temperature
Color index, turbidity, and temperature
Hydrogenion concentration,temperature,and pressure

6-3. pH is a measureor:
a. Effective acidity or alkalinity of a liquid
b. The oxidationor reductionpropertiesof a solution
c. Specificconductanceof an electrolyteor total ionic activity
d. Purity in an aqueoussolution
6.4.
a.
b.
c.
d.

A buffer solutionis usedwith pH-measuringinstrumentsto:

Standardizethe equipment
Protectthe equipment
Cleanthe electrodes
Platinizethe referenceelectrode

6-5. Oxidation-reductionpotential (ORP)is a measurement

or:

ANALYTICAL MEASUREMENTS 177

8.
b.
c.
d.

The oxidizing or reducingchemicalpropertiesof a solution

The oxygenpresentin any quantityof a givengasmixture
The hydrogenion concentrationin a given solution
The degreeof ionizationfor a particularsolution

6-6. Capacitancemeasurements
are usuallyappliedto:
8. Conductingliquids
c. Gasmeasurements
b. Nonconductingliquids
d. lonized gases
6.7.
8.
b.
c.
d.

Which of the following are electrochemicalmeasurements:

Humidity and density
Turbidity and differentialvaporpressure
pH and ORP
Dew point and boiling point rise

6-8. A salt (NaCl)solutionmustbe controlled at a concentrationof 12

percent.The bestchoiceof measurement
for the control systemwould be:
8. Conductivity
c. ORP
b. pH
do Capacitance
6-9. An industrial effluentstreamis to be neutralizedby addinga sodium
hydroxide solution.The bestchoiceof analyticalmeasurement
for the control
systemwould be:
8. Conductivity
c. ORP
b. pH
d. Capacitance
6-10. lon-selectivemeasurements:
8. Are similar to conductivity in operationbut usea differentcell
b. Are similarto capacitancemeasurements
but usea differentinstrument
c. Are similarto pH measurements
but use differentelectrodes
d. Are similarto densitybut usemore exacttechniques
6-11. A pH control systemis usedto neutralizea chemicalwastestream
beingdumpedinto a municipalsewagesystem.The desiredpH is 7 or
completeneutrality. An unfortunateaccidentshortcircuits the cable
connectingthe electrodesto the measuringtransmitter.The result will be:
8.
b.
c.
d.

The control valve admittingthe neutralizingagentwill fully open

The control valve will close
The control valve will remainapproximatelyat its hal openposition
The systemwill cycle

6-12. The concentrationof salt in a liquid usedto carry a slurry niust be

monitored.The bestchoice of measurementwill be:
8. Electrodelessconductivity
c. ORP
b. pH
d. An ion-selectivesystem
6-13. A pH systemis to be selected.It is requiredthat the systemfunction
with little or no maintenance.The referenceelectrodeselectedwould be:

178 FEEDBACK PROCESSCONTROL

8. A ftowingtype
b. A pressurizedftowing type

c. A nonftowingtype
d. A pressurizednonftowing
type
6-14. The pH of 8 streamis to be monitoredaccurately.It is discoveredthat
the temperatureof this streamvariesfrom 40 to 60F:
8. A measuringsystemwith automatictemperaturecompensation
is
indicated
b. Temperaturecompensationis not necessary
c. A manualtemperaturecompensationwill be adequate
d. The pH rangewill determinethe needfor temperaturecompensation
6-15. The most popularcarriergasusedin gaschromatographs
is:
8. Helium
c. Hydrogen
b. Air
d. Oxygen

SECTION

II PNEUMATIC

AND ELECTRONIC

CONTROL

SYSTEMS

The feedback controlloop, introduced in Chapter 1, requires further

discussion. This chapter is devoted to a more detailed description of
loop operation and its implicit mathematical relationships.
ln the feedback loop, the variable to be controlled is measured and
compared with the desired value-the set point. The difference between the desired and actual values is called error, and the automatic
controller is designed to make the correction required to reduce or
eliminate this error.
WitQin the controller, an algebraic "summing point" accepts two
inputs-one from measurement, the other from set. The resulting output
represents the difference between these two pieces of information. The
algebraic summing point is shown symbolically in Figure 7-1, where c
represents measurement; r represents set point; and e represents error.
Both rand c must be gi ven their proper signs (+ or -) if the value of e is
to be correct.
A typical closed-loop feedback control system is shown in Figure
7-2. Assume that the algebraic summing point is contained within the
feedback controller (FBC). ln order to select the proper type of feedback controller for a s~cific process application, two factors-time
and gain-must be considered. Time consists of dead time, capacity,
179

180 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

e~r
e=r-c
Fig. 7-1. Algebraic summing point.

and resistance. These factors cause phase changes that will be described in this chapter.
Gain appears in two forms-static and dynamic. Gain is a number
that equals the change in a unit's output divided by the change in input
which caused il.
Gain = AAoutput
input
The static gain of an amplifier is easily computed if, for a given step
change in input, the resulting change in output can be monitored. Case
A in Figure 7-3 shows an amplifier with a static gain of one. In Case B,
the output is magnified when compared with the input by a factor
greater than one. In Case C, the output change is less than the input.
Rere, the input has been multiplied by a number less than one.
The dynamic gain of an amplifier mar be computed by inducing a
giDewave on the input and observing the resulting output. Figure 7-4
illustrates this procedure with an amplifier that has no time lag between
its input and output. If the amplitude of the output is only hali as high as

Fig. 7-2. Closed-loopcontrolsystem.

TUE FEEDBACK CONTROLLOOP 181

Case

20%
5%
10% [ 10%

Output change

10%
10%

Static gain

!Q=1
10

22
10=2

Inputchange

1.--1
10-2

Fig. 7-3. Static gain.

the amplitude of the input giDe wave, the amplifier is said to bave a
dynamic gain of 0.5 for the particular frequency of the input wave, for
example, 1 Hz.
By monitoring the output amplitude for many different input frequencies, a series of dynamic gain numbers mar be found. A plot of the
amplitude ratio (gain) as a function of frequency of the input giDewave
is the gain portion of a Bode diagram, or a frequency-response curve
(Figure 7-4). Note that the higher the frequency, the lower the gain.

Fig. 7-4. Dynamicgain.

)(

182 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

This is true for nearly all processes and instruments. The frequency
scale is normally a logarithmic scale. Gain or amplitude ratio is also
normally expressed in decibels, where l decibel equals 20 times the
logarithm to the base 10 of the amplitude ratio.
As the frequency becomes lower and lower, and finally approaches
zero, a measure of the amplifier's static gain can be obtained.
Just as the static or dynamic gain of an individual amplifier can be
computed, so can the static or dynamic gain of a process controlloop.
Figure 7-5 shows a temperature controlloop. The static loop gain has
been computed by multiplying the static gains of each of the individual
components of the loop. Likewise, the dynamic loop gain could be
calculated by multiplying the dynamic gains of each element at a particular frequency. Pure dead time has a gain of 1.
Each element in the controlloop contributes gain to the totalloop.
Increasing the size of a control valve or narrowing the span of a transmitter has exactly the same effect as increasing the gain of a controller.

Fig. 7-5. ResponsecharacteristicsoCeachelementin a Ceedback

controlloop-heat,
exchangerprocesso
TT Transmittergain = psif'F
TC Controllergain = psi/psi
Diaphragmactuatorgain = inches/psi
Valve gain = Btu/inches(lift)
Reat exchangergain = F/Btu
F
Si ~'
"
.,
-.,
.
~
.~
Static loop gain = l' ~

)(

PS! ,

PS!

)(

mches

)( ~

Btu

The

THE FEEDBACK CONTROLLOOP 183

Proper selection of valve size and transmitter span is as important as

selecting the gain range in the controller.
In practice, the gain of the components in the loop mar nat be
uniform throughout their operating range. For example, control valves
having several characteristics will exhibit linear, equal-percentage
(logarithmic), or other types of gain curves. Unless special precautions
are taken, the gain of many control loops will change if the operating
level shifts. The gain most then be readjusted at the controller to provide satisfactory control.
Let us investigate how a feedback control system functions under
closed-loop conditions.
The Closed-Loop

Control

System

process consists of multiple capacitances (to store energy) andresistances

(to energy flow), as shown in Figure 7-6. Similarly, the
measuring system generally contains multiple capacitances and resis-tances.
The controller also consists of capacitance and resistance net-works.
The final operator, or valve motor, likewise contains capacitance and resistance, and the control valve is essentially a "variable
resistance." Thus, the analysis of the system is simply an analysis of
how signals change as they pass through resistance-capacitance and
dead time networks of various types.

Fig. 7.6. Closed-loopprocesscontrol systemillustratestheRC combinationsthat could

exist arounda feedbackloop. Eachatfectsprocessstability.

184 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Phase Shift Through

RC Networks

Figure 7-7 illustrates the phase shifts that occur when a sine wave is
applied to a network ofresistances and capacitances. Each succeeding
RC combination contributes its own phase shift and attenuation. In
Figure 7-7, bottom, a phase shift of 180degrees (a full half cycle) takes
place. That is, the output (eRa)is 180degrees out of phase with the input
(e applied). The signal has also been attenuated because each resistor
causes a loss in the energy level in the system. This phase change can
lead to instability in a closed-loop system.
Oscillation
In a physical system employing feedback, instability will occur if
energy is fed back in such direction (phase) as to sustain the instability
or oscillation.
Oscillators are often divided into two major categories, those thai
utilize a resonant device and those thai do noto The resonant oscillator
uses a device thai requires a minimum of added energy each cycle to
maintain oscillation. On the other hand, the nonresonant oscillator is
simply an amplifier with some type of phase-shifting network between
output and input. A public address system with the gain so high thai it
howls is a nonresonating type.

I
-"
I
SIMPLIFIEO PHASE SHIFTER

Fig. 7-7. Phase ofsigna! acrossR, R2' andRa is different. At one frequency, the output
signa! can be exact!y 180degrees out of phase with the input signa!, as shown in the

graph.

THE FEEDBACK CONTROLLOOP 185

Academically, both the resonant and nonresonant oscillators follow the same mathematicallaw, but the amplifier gain requirements are
quite different for the two categories. With a nonresonant circuit, oscillation will occur (1) if the feedback is positive (inphase) and (2) if the
gain is unity or greater. These two conditions are necessary for sustained oscillation.
Figure 7-8 shows a simple mechanical oscillator. ln ibis system,
energy stored in the spring is transferred to the weight, back to the
spring, to the weight again, and so on. The oscillations soon die Dut
because some ofthe energy is dissipated in each cycle. lfthe oscillation
is to be continuous, energy musi be added to make up the losses.
ln the bottom part ofthe figure, ibis energy is added by depressing
and releasing the weight. If the energy is to enforce the oscillation, it
musi be in the proper phase with the motion. That is, enforcing energy
musi be added at a time when both the energy and the oscillation are
moving in the same direction.
Figure 7-9 shows a free-running oscillator in which the required
energy is introduced through feedback. This circuit is called a phase
shift oscillator. lt consists of a transistor amplifier and a feedback path
comprised of three resistance-capacitance combinations. This circuit is
important because it is a simple illustration of the manner in which an
ordinary amplifier can be made to oscillate simply by use of feedback.

l@~
Fig. 7-8. At top, the magson the spring received a single downward displacement and
the oscillation is delayed. Below, arter the initial displacement, the magsreceives
periodic small displacements and the oscillation is continuous.

11

186

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

The transistor amplifier Dot only makes up for circuit losses, but
also contributes a phase shift of 180degrees; that is, the collector is 180
degrees out of phase with the base signal. EachRC element contributes
additional phase shift, depending on the values of the components and
the frequency involved. If each RC combination contribute s a phase
shift of 60 degrees, the three together will result in a 180-degreephase
shift. This, added to the 180-degreephase shift through the amplifier,
results in 360 degrees of phase shift. The feedback signal is now in
phase with the input signal. This inphase feedback produces continuous oscillation.
Since only one frequency will be shifted in phase exactly 360 degrees by the RC networks, the oscillator output is of this one fre-

quency.
The closed-loop control system (Figure 7-6) also has phase-shifting
networks, made up of all process and controller <:omponents,together
with an amplifier capable of contributing sufficient gain to overcome the
system's losses. This closed-loop system contains the same basic ingredients as the phase-shift oscillator. Each resistance-capacitance
combination will shift the phase of the energy flowing around the controlloop in the same manner as in Figure 7-9.
Oscillation (instability) will occur whenever (1) the phase relationships through the vario us resistance-capacitance combinations provide
feedback in proper phase, and (2) the system gain is unity, or greater at
the frequency at which the phase shift is 360 degrees.
Figure 7-10 illustrates a plot of system gain (output/input signalBode diagram) and output/input phase shift versus frequency for a
three-mode controller. The phase shift (expressed in degrees of lead or

Fig. 7-9. Shift of phasein threeRC sectionsmaintainsoscillationin phase-shift

oscillator.

THE FEEDBACK CONTROLLOOP 187

o
~

'"o
~
Z
~
c
2

10
"
100
1000
LOGARITHM FREOUENCY-~AOIANS PER SECONO

10000

Fig. 7-10. Bode diagram shows gain of system (top) and phase change (bottom) .at
various frequencies. The gain (output signal/input signal) and phase (output/input) are
measured with the system in open-loop operation.

lag) is plotted linearly against the logarithm of frequency. Note that the
integral is phase lagging and the derivative is phase leading.
The proportional adjustment in the controller adjusts its overall
gain and hence the loop gain. The integral adjustment governs the
low-frequency response, and the derivative adjustment governs the
high-frequency gain of the control mechanism.
In addition to controlling gain, these adjustments also affect the
phase shifts. Thus, an improper setting of any one of the three control
modes can cause the control system to satisfy the two criteria for
oscillation and become unstable,
Stability in the Closed-Loop System
Under normal operating conditions, the control system should produce
stable operation. That is, the controller should retum the system to set
point, in the event o an upset, with mnimum overshoot and oscilla-

tion.
Too much overall gain(too narrow a proportionalband)can cause

188 PNEUMAT1CAND ELECTRON1CCONTROL SYSTEMS

the system to oscillate if the feedback reinforces the oscillation. Too

little gain can cause the system to deviate too far Cromthe desired set
point.
The damped oscillation shown in Figure 7-11 is characteristic ofthe
curve a closed-loop control system will produce when it is subjected to
a step change. If the phase and gain relationships are proper, energy is
Dot fed back to sustain the oscillation, and the cycles die out. The
pattem of decay generally can be controlled by the adjustments available in the control mechanism. Engineering judgment and skill are
required to select the recovery curve best suited to the process being

controlled.
A high-gain setting is desired because this gives the fastest and
most accurate control action. However, too much gain produces oscillation. The best compromise is to use enough gain to produce a damped
oscillation, as shown in the middle of Figure 7-11 (0.25 damping ratio).
Nonlinearities

Thus far it has been assumed that the capacitances and resistances
found in the process control loop bave a fixed value that does not
change with process conditions. This is not always true in practice.
Frequently, process conditions vary the value of the resistances and
capacitances involved and, as a result, the phase and gain relationships
are in constant transition Crom one value to another. At other times,
these values change, limiting or restricting the natural behavior of the
'-PERIOO~
GAI!

MEASURED DR
CDNTRDLLED
VARIABLE
(CLDSED LODP
SYSTEM)

GAIN',

TIMEFig. 7-11. Closed-loopresponseto an upsetdependson gainofloop.

THE FEEDBACK CONTROLLOOP 189

system. Such changes are often considered as a group and are called

nonlinearities.
A thorough understanding of the operation of the linear system is a
prerequisite to an understanding of the nonlinear type. It is common
practice to assume a system is linear for the purpose of basic control,
and then deal with the nonlinear characteristics as problems anse.

Controllers

and Control

Modes

A wide variety of responsecharacteristics involving gain and time

create the different modes of process control. The controller mode
selectedfor a givenprocesswill dependon:
Economics
Precisionof control required
Time responseof the process
Processgaincharacteristics
Safety
Now let us investigatethe most commoncontrol modes.
Two-Position Control
Within this classification there exist several specific methods-on/off,
differential gap, and time-cycle control.
On/off control is the most common. As soon as the measured variable differs from the desired control point, the final operator is driven to
one extreme or the other.
In the usual sequence (Figure 7-12), as soon as the measured variable exceeds the control point, the final operator is closed. It will
remain closed until the measured variable drops below the control
point, at which time the operator will open fully. The measured variable will oscillate about the control point with an amplitude and frequency that depend on the capacity and time response of the processo
As the process lag approaches zero, the curve will tend to become a
straight line. Then the frequency of the final operator open/close cycle
will become high. The response curve will remain constant (amplitude
and frequency) as long as the load on the system does Dot change.
On/off control is found in many household applications, such as the
refrigerator, heating system, and air-conditioning system. A simple

190 PNEUMATIC AND ELECTRONICCONTROLSYSTEMS

MEASURED
VARIA BLE
FINAL
OPERATOR
POSITION

riME'"
FINAL
OPERATOR
POSIT/ON

MEASURED

VARIABLE

TIME -+

Fig. 7-12. Two-positioncontrolacts on operator.

thermostaticallycontrolledelectric heateris anotherfamiliar example.

The room in which the heateris placeddeterminesthe capacityof the
process,and, hence,the responsecurve.
The following requirementsare necessaryfor on/off control to produce satisfactoryresults:
1. Precisecontrol must Dotbe needed.
2. Processmustbave sufficientcapacityto allow the final operatorto
keep up with the measurementcycle.
3. Energy inflow is smallrelative to the energyalreadyexistingin the
processo

Differential-gapcontrol is similar to on/off except that a band, or

gap, is createdaroundthe controlpoint. In Figure7-12,notethat, when

THE FEEDBACK C~TROL LOOP 191

the measured variable exceeds the upper boundary of the gap, the final
operator is closed and remains closed until the measured variable drops
below the lower boundary. It now opens and remains open until the
measured variable again exceeds the upper boundary. In a process
plant, differential-gap control might be used for controlling noncritical
levels or temperatures.
A time-cycle controller is normally set up so thai when the measured variable equals the desired control point, the final operator will
be open for hal the time cycle, and closed the other hal. As the
measured variable drops below the control point, the final operator will
remain open longer than it is closed.
Two-position control is nearly always the simplest and least expensive form of automatic control. Any of the forms discussed can be
implemented with commercially available mechanical, pneumatic, or
electronic instrumentation. On the other hand, two-position control
mar Dot meet the requirements often demanded by today's sophisticated processes. Now let us investigate a typical on/otI controlloop.
Assume we bave a liquid-Ievel process as shown in Figure 7-13.
Throttling

Control

Proportional or throttling control was developed to meet the demand

for a more precise regulation ofthe controlled variable. Throttling control indicates any type of control system in which the final operator is
purposely positioned to achieve a balance between supply to and demand from the processo
The basic type of throttling control is proportional-only control.
This term applied to a control action wherein the position of the final
operator is determined by the relationship between the measured variable and a reference or set point. The basic equation for a proportional
only controller is:

m = Gain (Error) + Bias

(7-1)

where:
m = controller output or valve position
Error = difference between set (r) and measurement (c)

or

e=r-c

(7-2)

v=

Fig. 7-13. PROBLEM: DIFFERENTIAL

GAP CONTROL, CALCULATE PERIOD.

Question I: What is the period of oscillation ofthe controlloop?

Given:
Controller: on/otI with 7 percent differentia! gap.
lnstrumentation response: assume instantaneous.
Load = Q = 60 gpm.
Manipulated Variable = M = O gpm or 90 gpm.
Tank: 6 feet diameter, 12 feet high.
Level transmitter: Oto 8 feet.
Set Point = 50 percent
Question 2: What is the period of oscillation ifthe load is 50 gpm?
Question 3: At which load will the period be shortest?
So/ution:
1. Volume oftransmitted signa! = area x height
1TD2

xH=

2. Volume within 7 percent differential

(0.07)(1692) = 118.4gal.
3. Level cycle:
3a. Rate of rise = (90 -60)

TlIne
" to nse
.118.4gal.
= 30

gpm = 30 gpm

gpm

3b. Rate offall = (O-60)

=.

395

mm.

gpm = -60 gpm

Tirne to fall =

-118.4 gal.
= 1.973min.
-60 gpm
3c. Total tirne = (3.95 + 1.973)min. = 5.92 min.
Solution to no. 2:

3.lD1e
T ' to nse
, = (90118,4
,
-40) gal,
gpm = 2.96 mm.
TlD1e
'
-118.4gal.
'
to "la11 = (O
-50) gpm = 2. 37 mm.
Period oCoscillation = 5.33 min,

Solutionto no. 3:
When Q = Mon + Moll = 90 gpm + O gpm = 45 gpm
2
2

m=
Application

THE FEEDBACK CONTROLLOOP 193

Bias = usually adjusted to p-lacethe valve in its 50 percent open position with zero error.
The term proportional band is simply another way o expressing

gain.
% ProportionalBand =

100

Controller Gain

or
Controller Gain =

100

% Proportional Band

Substituting the error formula (e = r -c),

proportionalband formula ( % PB = ~),

Equation 7-2, and the

Equation7-3, into Equa-

tion 7-1, it mar be expressedas:

100

%PB (r -c)

+ 50%

Another approach to visualizing the effect of varying the proportional band is shown in Figure 7-14. Each position in the proportional
band dictates a controller output. The wider the band, the greater the
input signal (set point -measurement) musi change in order to cause the
output to swing from O to 100percent. Manual adjustment of the bias
shifts the proportional band so thai a given input signat will cause a
different output level.
ot Proportional

Control

Proportional control attempts to retum a measurement to the set point

arter a load upset has occurred. However, it is impossible for a proportional controller to retum the measurement exactly to the set point,
since, by definition (Equation 7-4), the output must equal the bias setting (normally, 50 percent) when measurement equals the set point. If
the loading conditions require a difIerent output, a difIerence between
measurement and set point must exist for this output level. Proportional control mar reduce the efIect of a load change, but it cannot
eliminate it.
The resulting difIerence between measurement and set point, arter
a new equilibrium level has be en reached, is called ofIset. Equation 7-5

194 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

SET POINT
MINUS
MEASUREMENT

GAIN. 1

OUTPUT

TIME-..
OPEN-LOOP

GAIN RESPONSES

0%

1001

CONTROLLER

OUTPUT

MEA5UREMENT,I

"I.

PROPORTIONAL
BANO

OUTPUT
50%

50

100%

0% CONTROLLER

PROPORTIONAL

BANO

-SET

POINT

OUTPUT

.!~I:,_C2~~R2!;~E~
OUTPUT
BIAS (MANUALLY
EFFECT
-

OF

CHANGES

IN

PROPORTIONAL

SET)

= 50%
BAND

Fig. 7-14. Outputresponsereactsto changein gain.

indicatesproportionalresponseto a load upset,and the resultingoffset.

The amount of offset may be calculatedfrom:
Ae =% Proportional Band AM
100

(7-5)

where ~e is changein offset, and ~M is changein measurementrequired by load upset.

From Equation 7-5, it is obvious that, as the proportional band
approacheszero (gainapproachesinfinity), offset will approachzero.
This seemslogical, becausea controller with infinite gain is by definition an on-otI controller that cannot permit a sustainedoffset. Conversely, as the proportional band increases(gain decreases),propor-

___!_'22-:_C2~~~~~!~

THE FEEDBACK CONTROLLOOP 195

tionately more and more offset will exist. Offset mar be eliminated by
manually adjusting the bias until the measured variable equals the set
point.
Narrow-band proportional-only controllers are often used in noncritical, simple temperature loops, such as in maintaining a temperature
in a tank to prevent boiling or freezing. A low, dynamic gain allows a
narrow band to be used. Controllers of ibis nature are typically fieldmounted and pneumatically operated. Many noncritical, level-control
applications having long time constants also use proportional-only control. Now let us apply proportional-only control to the level problem
shown in Figure 7-15.
Proportional-Plus-lntegral

Control

If offset cannot be tolerated, another control mode must be added.

Integral action will integrate any difference between measurement and
set point, and cause the controller's output to change until the difference between measurement and set point is zero.
The response of a pure integral controller to a step change in either
the measurement or the set point is shown in Figure 7-16. The controller output changes until it reaches O or 100 percent of scale, or the
measurement is retumed to the set point. Figure 7-16 assumesan openloop condition where the controller's output is dead-ended in a measuring device, and is Dot connected to the processo
Figure 7-16 also shows an open-loop proportional-plus-integral response to a step change. Integral time is the amount oftime required to
repeat the amount of change caused by the error or proportional action.
In Figure 7-16, integral time equals t r, or the amount of time required to
repeat the amount of output change (xJ. (Some instrument manufacturers define integral as the inverse of the above, or the number of
times per minute the amount of change caused by proportional action is
repeated.)
Another way of visualizing integral action is shown in Figure 7-17.
In this example, a 50 percent proportional band is centered about the
set point. If all elements in the controlloop bave been properly chosen,
the controller's output should be approximately 50 percent. If a load
upset is introduced into the system, the measurement will deviate from
the set point. Proportional response will be immediately seen in the
output, followed by integral action.
Integral action mar be thought of as forcing the proportional band
to shift, and, hence, causing a new controller output for a given rela-

196 PNEUMATIC AND ELECTRON1CCONTROL SYSTEMS

Fig. 7-15. Sample problem: PROPORTIONAL CONTROL, CALCULATE

MEASUREMENT.
Question: What is the level under steady-state control?
Given:
Range of level transmitter: Oto 70 inches
Level controller: Proportional only, PB = 75 percent, bias = 50 percent, Set
Point = 40 inches
Load, q: Fixed at 3.5 gpm
Valve: Pressure drop, ~P, is constant, Linear characteristic, delivers 6 gpm at 100
percent stroke.
Solution:
I. Under steady state, inftow must equal outftow.
6
- 0 gpm = 58.3%
2. Inftow = 3.5
.gpm

3. Inftow % = Controller output

m = 58.3%
.40
4. Set pomt = r= 70 inches
. h = 57.1%

mc es

5. Equation (7-4) for proportional control:

100
m = -(r

%PB

-c)

+ 50%

100(57.1 -c) + 50
58.3 = 75
Solving for c: c = 50.9%
Converting to inches: 0.509 x 70 = 35.6 inches

tionship between rneasurernentand set point. Integral action will continue to shift the proportional band as long as a difference exists between rneasurernentand set point. Integral action has the sarne function
as an operator adjusting the bias in the proportional-only controller.
The width of the proportional band rernains constant, and is shifted in a

THE FEEDBACK CONTROLLOOP 197

MEASUREMENT

CONTROLLER
OUTPUT

TIME-'

Fig. 7-16. Outputrespondsto stepchangein input.

direction opposite to the measurement change. Thus, an increasing

measurement signal results in a decreasing output, and vice versa.
If the controller is lloable to retum the measurement to the set
point, integral action will drive the proportional band until its lower
edge coincides with the set point (see Figure 7-17); hence, a 100 percent
output. Ifthe measurement should then start to corne back within control range, it must cross the set point to enter the proportioning range,
and cause the output to begin throttling. If the system has any substantial capacity, the measurement will overshoot the set point, as shown
on the right gide of Figure 7-17.
This basic limitation is called integral windup. It must be seriously
considered on discontinuous or batch processes where it is common for
the controller output to become saturated and overshoot. A pneumatic
mechanism called a batch switch designed to prevent this overshoot is
described in Chapter 8.

~
.=

LOAD

198 PNEUMATIC AND ELECTRONICCONTROLSYSTEMS

MEASUREMENT
% OF SPAN

50

2~

-~"--""

~;;;;",

'

','~
,

SET paiNT

'~

CHANGE

Fig. 7.17. Integral actionshiftsproportionalband in referenceto differencebetween

input signa!and setpoint.

Integral setting is a function of the dead time associated with the

process and other elements in the controlloop. Integral time should Dot
be set faster than the process dead time. If it is set too fast, the controller's output will be capable of changing faster than the rate at which the
process can respond. Overshoot and cycling will resuIt. An altemate
definition of integral action is the fastest rate at which the process can
respond in astable manner.
Proportional-plus-integral controllers are by far the most common
type used in industrial process control. Pneumatic and electronic
analog, and digital hardware are available with proportional-plusintegral action.
The classical equation or expression for a proportional-plusintegral controller is:

(7-6)

%PB

= gam

e = error (deviation)R
= integraltime
If rou compare this equation with Equation 7-4, the proportionalonly expression, rou will find the bias term replaced by the integral

term.
Fig.

rHE FEEDBACK CONTROLLOOP 199

In operation, then, the output of this controller will continuously

change in the presence of an error signat. Once the error has been
reduced to zero, the output will stop changing and star fixed until an
error redevelops. At this time, the output again will change in such
direction as to eliminate the error once again.
Adding

Derivative

ln controlling

(Rate)

multiple-capacity

processes, a third mode of controls is

often desirable. .
By definition, derivative is the time interval by which derivative
action will advance the effect of proportional action on the final
operator. lt is the time difference required to get to a specified level of
output with proportional-only
action as compared to proportionalplus-derivative action. Figure 7-18 illustrates derivative response to a
camp change; where td equals the controller's derivative time.
Derivative
action occurs whenever the measurement si~nal
changes. On a measurement change, derivative action differentiates the
change and maintains a level as long as the measurement continues to
change at the given rate. Under steady-state conditions, the derivative
acts as a 1 to 1 repeater. lt has no inftuence on a controller's output..By
reacting to a rate of input change, derivative action allows the controller to inject more corrective action than is initially necessary in order to
overcome system inertia. Temperature presents the most common application for derivative. Derivative action should Dot be used on processes that are characterized by predominant dead times, or processes
that bave a high noise content, that is, high-frequenY extraneous signaIs such as are present in the typical ftow application.

MEASUREMENT

,
PROPORTIONAL

-ONLY

ACTION

PROPORTIONAL PLUS
DERIVATIVE ACTION

TIME-

7.18. -Derivative

action increases rate oCcorrection.

DERIVATIVE
TIME, td

Selecting

200 PNEUMATIC AND ELECTRONICCONTROLSYSTEMS

The equationor expressionfor the proportional-plus-integral-plus

derivative controller is:
(7-7)
where:
e = error
R = integral time
D = derivative time
Question: How long will it take for the output to change 5 percent if
the measurement remains constant?

Given:
Proportionalband = 50 percent
Integral time = 3 minutes
Derivative time = 3 minutes
Set point = 45 percent
Measurement= 35 percent
Control action: increase-decrease
So/ution:
1. Error staysconstant,thereforeno output changedue to derivative.
2. No output changedueto proportionalaction. However, the proportional action affectsthe integralresponse.
3. Error = r -c = 45% -35% = 10%
4. R = 3 minutes = time to changean amountequalto the error
5. Becausethe proportionalband = 50%,the output in Step4 is multiplied by

= 2.

Therefore 2(10%)= 20%changein 3 minutes

or

5% changein 3/4minute.
the Controller

Now that the variety of available control modes has been described,the
next logical question is: which should be selected to control a
particular process? Table 7-1 relates process characteristics to the
common control modes. Let us apply the chart to the beat exchanger
processo
The beat exchanger acts as a small-capacity process; that is, a
small change in steam can cause a large change in temperature. Accu-

THE FEEDBACK CONTROLLOOP 201

Table 7-1. Process Characteristics

Control
Mode
On/OlI
Floating
Prop.

Transfer
Lag
Min

Small
Small

Dead
Time

Capacitance

Reaction Load
Rate
Changes

SelfRegulation

Min
Min

High
Low
Moderate

Slow
High
Slow

Any

Slow
Slow

Must bave

SmaIl

Sma11

Prop. +
Integral

Moderate

Moderate

Moderate

Any

Any

Any

Any

Any

Any

Any

Prop. +
Integral +

Deriv.

rate regulation of processes such as this calls for proportional rather

than onloff control.
Variations in water rate cause load changes that produce offset, as
described previously. Thus, the integral mode should also be used.
Whether to include the derivative mode requires additional investigation ofthe process characteristic. Referring to the reaction curve (Figure 7-19), notice that the straight line tangent to the curve at the point
of inftection is continued back to the 150F (starting) level. The time
interval between the start of the upset and the intersection of the tangentialline is marked TA(in Figure 7-12, this was dead time); the time
interval Crom this point to the point of inftection is TB' If time T B
exceeds time TA,some derivative action will prove advantageous. 1fT B
is less than TA, derivative action mar lead to instability because of the
lags involved.

Fig. 7-19. The process reaction curve is obtained by imposing a step change at input.

7-1.
7.3.
7-4.

202

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

Table 7-2.

OFFSET
CA. BEy

~'ES

USE

TO'ERATEO?

P-O."

.s

-'+1

USE

The reaction curve of Figure 7-19 clearly indicates that some derivative action will improve control action. Thus a three-mode controller with proportional, integral, and derivative modes satisfies the needs
of the beat exchanger processoTable 7-2, as applied to this problem,
will yield the same resulto
Ouestions
With adequate gain and inphase feedback. any system will:
8. Drift
c. Increase amplitude
b. Oscillate
d. Degenerate
7-2. The natural frequency at which a closed-loop system will cycle depends

upon:
3. The amplifier gain
b. The attenuation provided by the process
c. The phase shift provided by the resistance-capacitanceand dead time
networks that exist in the system

d. Resonance
8.
b.
c.
d.

The Bode diagram describes:

Gain and phase shift through the usable frequency range
The system's linearity
The reaction to a step change
The recovery curve that will result from a load change

All systems may be assumed to be:

a. Linear
b. Nonlinear

THE FEEDBACK CONTROLLOOP 203

c. Linear for the purpose ofinitial consideration but with full knowledge
that this may not be the case
d. Nonlinear for purposes of analysis with rhe exception that the system
may prove to be linear
7-5. A closed-loop control system that employs a three-mode controller:
a. Can oscillate or cycle at only one frequency
b. Can oscillate or cycle at several frequencies. depending on controller
adjustment
c. Will not oscillate because of the stability provided by derivative
d. Will produce only damped oscillations with a 0.25 damping ratio
7-6. We have a closed-loop control system which is cycling. We should:
a. Increase the proportional band
c. Check and adjust both
b. Increase integral action
d. Immediately shut it down
7-7. A proportional controller is being used to control a process. and the
offset between set point and control point must be held to a minimum. This
would dictate that rhe proportional band:
3. Be as narrow as possible
b. Be as wide as possible
c. Be of moderate value
d. Does not relate to the problem
7-H. The system gain ofthe closed-loop control system:
3. Refers to the process gain
b. Refers to the gain of the measurement and control devices
c. Refers to total gain of all components. including measurement, controller, val ve operator, valve, and process
d. Relates only to the gain ofthe controller
7-9. If a closed-loop control system employs a straight proportional
controller and is under good control. offset:
3. Will vary in magnitude
b. Will not exceed one-half of the proportional band width
c. Will exceed the deviation
d. Is repeated with each reset7-10.
Any closed-loop system with inphase feedback and a gain of one or
more will:
a. Degenerate
c. Exhibit a 0.25 damping ratio
b. Cycle or oscillate
d. Produce square waves7-11.
How long will it take for the Otltput to change 5 percent if the
measurement remains constant?
Gil 'l'li :

Proportional band '" 50 percent

Integral time '" 3 minutes
Derivative time '" 3 minutes

204 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Set paint = 45 percent

Measurement = 35 percent
Control action = increase/decrease7.12.
If a closed-loop control system is adjustd to produce aO.25 damping
ratio when subjected to a step change, the system gain is:
3. 0.1
c. 0.5
b. 0.25
d. 1.07-13.
A straight proportional controller is employed to control a processo
Narrowing the proportional band will:
3. Not change the offset
c. Always cause cycling
b. Decrease the offset
d. Never cause cycling7-14.
The type of process that most often can benefit from derivative is:
3. Flow
c. Temperature
b. Level
d. Pressure
7-15. Pure dead time ina process contributes a gain or:
3. Zero
c. Depends upon dynamics
d. Infinite
d. One
7-16. RefeITing to Figure 7-5, the transmitter span is 200F; the controller
proportional band is adjusted to 150 percent; the equal percentage valve
delivers saturated steam containing 40,000 Btu per minute at full open; and
water enters the exchanger at 50F and is heated to 2000Fat a 40 gpm
maxirnum flow rate. The static gain of this control loop is approximately:
3. 0.5
c. 1.5
b. 1.0
d. 2.0
7-17. Ifthe flow rate ofheated water in Problem 7-16 is reduced to 5 gpm,
you would expect the gain to:
3. Increase
c. Remain the same
b. Decrease
7-18. The integral control mode is:
3. Phase-Ieading
b. Phase-Iagging

c. Inphase
d. Phase-reversing

7-19. The derivative control mode is:

8. Phase-Ieading
b. Phase-Iagging

c. Inphase
d. Phase-reversing

7-20. The most common combination of control modes found in the typical
process plant is:
8. Proportional-only
b. Proportional, integral, and derivative
c. Proportional-plus-integral
d. On/off

Pneumatic Control Mechanisms

The basic pneumatic control mechanism is the ftapper-nozzle unit. This

unit, with amplifying relay and feedback bellows, is a simple, rapidacting control mechanism.
The basic pneumatic mechanism converts a small motion (position)
or force into an equivalent (proportional) pneumatic signal. Since
pneumatic systems mar use a signal of 3 to 15 psi (20 to 100kPa) (3 psi
or 20 kPa at O percent and 15 psi or 100kPa at 100 percent scale), the
instrument must bave the ability to convert a position or small force
into a proportional pneumatic span of 12 psi (3 to 15 psi or 20 to 100kPa).

The Flapper-Nozzle

Unit

Figure 8-1 shows the principIe of the ftapper-nozzIe device. Input air
(reguIated at 20 psi or 138 kPa) is fed to the nozzIe through a reducing
tube. The opening of the nozzIe is Iarger than the tube constriction.
Hence, when the ftapper is moved away from the nozzIe, the pressure
at the nozzIe falls to a Iow vaIue (typicalIy 2 or 3 psi, or 10 or 20 kPa);
when the ftapper is moved cIose to the nozzle, the pressure at the
205

206 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Fig. 8-1. Flapper-nozzleis the basicpneumaticcontrolelement.Flappercan be

positionedby temperature(as shown),pressureelement,or any sensor.

nozzle rises to the supply pressure (20 psi or 138 kPa). Flapper movement of only a few thousandths of an inch (Figure 8-2) produces a
proportional pneumatic signal that mar vary from near zero to the
supply pressure. Some pneumatic control mechanisms use the air at the
nozzle (nozzle pressure) to operate a control valve.
The simple ftapper-nozzle unit shown in Figure 8-1 has several
basic limitations. The output air must all corne through the input constriction if it is used directly to operate a control valve. Hence, the
output pressure can change only slowly, causing sluggish action. Just
as the rate of increase in pressure is limited by the input constriction,
the rate of decrease in pressure is similarly limited by the slow rate of
air passage through the nozzle to the atmosphere.

Fig. 8-2. Flapperneedmove only a fewthousandthsof an inch for full rangeof output
(nozzle)pressure.

The
Fig.

PNEUMATICCONTROLMECHANISMS 207

AIso, the measuring element that positions the ftapper must be

strong enough to overcome the blast of air leaving the nozzle. Since
many measuring sensors are relatively low-force elements, they could
Dot be used to position the ftapper accurately and positively if the
nozzle were made large enough to pass sufficient air to overcome this
limitation.
A second limitation is that the ftapper moves only a few thousandths of an inch for complete action from minimum to maximum nozzle pressure. This small trave I for a change in output from zero to full
value makes the entire element susceptible to vibration and instability.
All these limitations are overcome by employing (1) a relay, or
other amplifier, which amplifies the nozzle pressure; and (2) a feedback
system which repositions the ftapper or nozzle.
Valve Relay or Pneumatic

Amplifier

Figure 8-3 shows a celar connected to a nozzle. The nozzle pressure is

applied to the pneumatic relay, which contains a diaphragm operating a

small ball valve. Because the diaphragm has a large area, small
pressure changes on its surface res uit in a significant force to move the
ball valve. The ball valve, when open, permits the full air supply to
reach the output; when closed, it permits the nozzle pressure to bleed
to atmosphere.
If the ftapper position is changed with respect to the nozzle, the air
pressure acting on the celar diaphragm changes and either opens or
closes the celar ball valve, thus either increasing or decreasingthe ftow
of supply air, which now can ftow directly from the supply to the
output, overcoming the first deficiency (slow action) of the ftapper noz-

zle.
relay is often called a pneumatic amplifier becausea small

8-3. Relay is an arnplifier. That is, small change in nozzle pressure (the input to the
relay) causes a large change in output pressure (to the control valve).

The

208 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

change in nozzle pressure causes a large change in output pressure and

flow. The pneumatic relay shown in Figure 8-3 (used in the Foxboro
Type 12A temperature transmitter and the Model 130 controller)
amplifies the nozzle pressure by a factor of 16; that is, a change in
flapper position of 0.0006 inch, which produces a change in nozzle
pressure of3/4psi, and results in a change arrelar output of 12 psi (from
3 to 15 psi).
The relay shown in Figure 8-3 increases output pressure as nozzle

pressureincreases.Relays can also be constructedto decreasethe

output pressure as nozzle pressure increases.
Linear Aspirating

Relay or Pneumatic

Amplifier

Flapper-nozzle detectors employed in the set-point transmitter and derivative sections of the Model 130 controller use a different type of
pneumatic amplifier. If the flow or volume output requirements are
small, an aspirating relay mar be employed. The pneumatic amplifier,
shown in Figure 8-4, makes use of the Venturi tube principIe and resembles a small Venturi tube. With a 20 psi supply, the throat pressure
of the Venturi can vary 3 to 15 psi or 20 to 100 kPa for a change in
flapper position with respect to the nozzle of less than 0.001 inch. The
aspirating relay accomplishes ibis with almost perfect linearity and
does Dot require parts thai are subject to wear. This recent develop-

.,.SUPP"
,

TO

FLAPPERtNO'

Fig. 8-4. Aspirating cone, showingVenturitube principIe.

Fig.

PNEUMAT1CCONTROLMECHANISMS 209

ment offers advantages over conventional pneumatic amplifiers, provided that a high volume output is Dot required.
Proportional Action
If a controller had only the units described thus"far, ~apper-nozzle and
relay, it would only bave on/otI action. On/otI control is satisfactory for
many applications, such as large-capacity processes. However, if
on/otI action does Dot meet the control requirements, as in low-capacity
systems, the ftapper-nozzle unit can easily be converted into astable,
wide-band proportional device by a feedback positioning system that
repositions the moving ftapper.
An example of a proportional action mechanism is the Foxboro
12A pneumatic temperature transmitter (Figure 8-5). This device,
called a "force-balance pneumatic transmitter," develops a 3 to 15 psi
(20 to 100kPa) output signal proportional to the measured temperature.
Thus, it is functionally a proportional controller.

8-5. Pneumatic temperature transmitter. with tubing and sensor.

210

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

Figure 8-6 shows .the principIe of the Model 12A transmitter. The
forces created by the bellows are automatically balanced as follows.
When the temperature sensor is subjected to an increase in temperatufe:
1. The increased temperature expands the gas contained in the sensor
and increases the force exerted by its bellows. This increases the
moment of force, tending to rotate the force bar clockwise.
2. The flapper now approaches the nozzle and the nozzle pressure
increases. This pressure is applied to the pneumatic relay (Figure
8-3), increasing its output and the pressure applied to the feedback
bellows, thus increasing the counterclockwise moment of force
sufficiently to restore the force bar to equilibrium.
The force bar (flapper) is now repositioned slightly closer (less than
0.001 inch) to the nozzle, and the output pressure has reached a new
levellinearly related to the measured temperature.
In the actual unit, two additional forces act on the force bar. One is
thai applied by the zero elevation spring. Adjustment of tbis spring
determines the constant force it adds to thai supplied by the feedback
bellows. This allows a given span of temperature measurement to
be raised or lowered. The other force is applied by the ambienttemperature- and barometric-pressure-compensating bellows. This
force compensates for ambient temperature or atmospheric pressure
changes acting on the gas-filled thermal system, thus minimizing errors
caJlsedby these changes.

Fig. 8-6. PrincipIeoCpneurnatictemperaturetransmitter.Forcescreatedin bellowsare

balanced.

PNEUMAnc CONTROLMECHANISMS 211

Fig. 8.7. Generalpurposecontroller.

The Model 12A temperature transmitter produces an output proportional to measurement and is a proportional control mechanism.
However, a useful general purpose controller should incorporate additional modes, and also bave the ability to adjust the parameters of the
mechanism along with other convenient features. For these reasons the
unit described, although a proportional controller, is used primarily as a
transmitter to send an input signal to a generalpurpose controller (such
as the 130 Series), as shown in Figure 8-7.

Control

Mechanism

Requirements

A control mechanismshouldbe ableto controlthe processandeasethe

job of the operator. To achievethesegoals, it must meetthe following
requirements:
1. The required control modesmustbe availableand easilyadjustable
throughthe required range.

212

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

2. The controller shouldhave an easilyadjustedset point.

3. The mechanismshould clearly display what is going on for the
benefitof the operator.
4. It shouldbe convenientfor the operatorto switch over to manual
control either for the purpose of operatingthe processmanually
during startupor in otherunusualsituations,or to performmaintellance or adjustmentson the automaticcontrol unit.
5. After manualoperation,it shouldbe a simplematterto switch back
to automatic control smoothly without shockingor bumping the
processo

6. If for any reasonthe unit requires maintenanceor adjustment,it

shouldbe simpleto remove it for this purposewithout interfering
with the processo
7. In manysituations,it mar alsobe convenientand practicalto bave
a continuousrecord of the controllervariableand an alarm system
to signalthe operatorwhensomepredeterminedlmit is exceeded.
Figure 8-8showsthe completeModel 130controller in block diagramformo
The total control mechanismconsistsof the following functionalparts:
1. A set-point unit that produces a 3 to 15 psi or 20 to 100kPa
pneumaticsignalto be fed into the controllermechanism.This unit
is generallylocated within the controller, but it mar be remote.
2. Thederivative unit that acts directly on the measurementsignal.
3. The automatic controller itself-headquarters for the control action.
4. The manualcontrol unit, whichperformstwo functions-switching
Cromautomatic to manual and Crommanual to automatic, and
manualadjustmentof the control valve (regulatedvariable).
5. The automatic balancingunit, consistingof a simple floating controller through which the operator can switch Cromautomaticto
manualand back without any balancingadjustmentsand ret without causinga bump or suddenvalve change.
6. A measurementindicator that displays with pointersand scalethe
measurementbeingfed into the controller, and the set point.
7. An output indicator-a display of controller output that mar be
interpreted in terms of valve position.
The Automatic

Controller

Figure 8-9 develops the mechanism of a proportional controller in

schematic formo This controller, like the Model 12A shown in Figure
8-7, has measurement and feedback bellows and a flapper-nozzle relay

~
~
unit.
Fig.

PNEUMATICCONTROLMECHANISMS 213

Fig. 8-8. Block diagramoCFoxboropneumaticConsotrol130Mcontroller.

The basic added features are a set bellows, which permits adjustment of the set point; and an integral (R) bellows.
ln the actual Model 130 controller (Figure 8-10), a ftoating disk actsas
the ftapper of the ftapper-nozzle system. The resultant moments offorce
due to the four bellows determine the position of the disk with

0f?5!)@
OUTPUT

b=ooa

obOR
b
B
P

b.40

OUTPUT

: o.
o bM
P
b.lo
4

OUTPUT

8.9. Schematicdevelopmentofproportionalcontroller.

PROPORTIONAL
~
=1

214 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

('=I--~ --.

REDUCING
TUBE-

~
~

BAND
DIAL
REVERSING
SWITCH

~.-

ADJUSTING ARM AND

FULCRUM BAR

NU~~LI:

SET (B) OR MEASUREMENT

(A) BELLOWS
--

FORCE BALANCING
FLOATING DISC

-INTEGRAL
BELLDWS
PROPORTIONING BELLOWS --'I

-MEASUREMENT
(B) OR
SET (A) BELLOWS
CONTROLLER ACTION
REVERSING SWITCH

/1

\:

r=

=!
,
r~1
'=--,~_f=

INTEGRAL
RESTRICTOR

Fig. 8-10. Control unit with floating disk flapper nozzle.

respect to the nozzle. Therefore, the relay output pressure varies with
changes in pressure in any ofthe bellows. The mechanism is aligned to
produce a midrange output of9 psi (60 kPa) when the error signal is O.
This is called bias.
The width of the proportional band is adjusted by the position of
the proportional-band-adjusting lever. (The term gain is also used in
place of proportional band. Gain is the reciprocal of proportional
band.) This lever positions the fulcrum abolli which the moments of
force created by the pressure in the four bellows act. These moments of
force tilt the floating disk upward or downward, thereby causing a
change in nozzle pressure. The reguli is an increased or decreased
output pressure Crom the controller. This output pressure, which is
proportional to changes in measurement or set pressure, acts to reposition the control valve, thus influencing the process and bringing abolli a
change in the measurement bellows to establish a new equilibrium in

the system.

PNEUMATICCONTROLMECHANISMS 215

Derivative and Integral

The general purpose controUer needs derivative and integral action, as
described in Chapter 1. These are added to the pneumatic controUer.
The derivative function is added to the measurement signal before it
reaches the controUer (block 4, Figure 8-8). This eliminates derivative
response to set-point change and practicaUy eliminates any interaction
between derivative and the other control modes.
During steady-state conditions, the unit acts as al: 1 repeater, but
upon a change in the measurement the unit adds a derivative inftuence
to the measurement change.
In the derivative unit (Figure 8-11), the force moment (beUows area
times distance from fulcrum) ofbeUows A is 16 times that ofbeUows B,
and the force moment of beUows B plus beUows C equals that of bellows A. As the measurement signal increases, the immediate change in
feedback pressure in B is 16 times the change in pressure in beUows A.
Simultaneously, air starts to ftow through the restrictor to beUows C,
graduaUy reducing the pressure needed in beUows B to restore equilibrium. Thus, the output ofthe derivative unit, which is the signal to the
automatic control unit measurement beUows, reftects the change in

measurementplus a derivative responseadded to that change. The

graph at the bottom of Figure 8-11 shows the signal that the measure-

TIME

Fig. 8-11. Derivative unit, schematicdiagramand responsecurve.

216

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

ment bellows in the automatic control unit receives from the derivative
unit at the measurement change shown.
The integral function takes place within the automatic control unit
itself. The integral bellows opposes the feedback bellows; thus, if the
feedback bellows introduces negative feedback, the force created by
the feedback bellows will act like positive feedback; that is, it tends to
move the flapper in the same direction as error. This would lead to
instability if the integral restriction were Dot used. anly very slow
signals (minutes in duration) can affect the integral bellows because of
the very small integral restriction. The integral action occurs only arter
the proportional and rate actions h~ve affected the processoIf an error
remains (such as offset due to load change), the integral action takes
place slowly. Because integral action will continue in the presence of a
continuous error, such as we mar find in a batch process, it can cause
the output to go to an extreme. This is called integral windup.
Integral time is adjusted by setting an adjustable restrictor, or needle valve. The dial on this pneumatic val ve is calibrated in minutes of
integral time.
Figure 8-12 illustrates the Foxboro Model 130controller. The control rnodes-proportional, integral, and derivative-are in the controller shown. The signal level with respect to time mar be changed by
appropriate adjustments of proportional, integral, and derivative
modes. The proportional band adjustment is calibrated from 5 to 500
percent, while the integral and derivative dials are calibrated from 0.01
to 50 minutes. These adjustments are calibrated in "normal" times.
The effective times are somewhat different.
In any controller, some interaction between control actions mar
exist because change in one action (proportional, integral, or derivative) can create change in the others. This unavoidable interaction,
even if small, should be kept in mind when adjusting or tuning a control
ler.
Manual Control

Unit

Manual operation is achieved by using the manual control unit shown

schematically in Figure 8-13 and depicted in Figure 8-14. With the
transfer switch in the manual position, the thumbwheel is engaged, and
by its action the flapper-nozzle relationship governs the pneumatic output. The operation of the manual control unit is similar to that of any
pneumatic transmitter in that it employs a feedback bellows. Full-scale
manipulation of the thumbwheel corresponds to a 3 to 15 psi (20 to 100
kPa) pneumatic output.

PNEUMATICCONTROLMECHANISMS 217

Fig. 8-12. Pneumatic Consotrol130M controller. (Top) right sicle cover removed and
manual controls. (Battam) the leCtsicle cover is removed to show pneumatic printed
circuit board.

Transfer

Automatic to Manual
With the transfer switch in the automatic position, the thumbwheel in
the manual control unit is disengaged and the automatic controller's
pneumatic output is fed into the manual control unit bellows. The manual control unit remains balanced at this output. At the time of transfer
to manual, the manual unitinstantfy starts transmitting this same output. Thus, the response record of the transfer from automatic to manual (Figure 8-15) is smooth and bumpless.
Manual to Automatic
The transfer Crom manual to automatic requires the use of an
automatic-balancing unit, which consists of a singie-pivoted diaphragm
with four air pressure compartments. This is shown in schematic Corm

AIR
SWITCH
SUPPLY SIGNAL

OUTPUT

Fig. 8-13. Schematicof manualcontrolunit.

Fig. 8-14. Manual controlunit;

TRANSFER

TIME-Fig. 8-15. Responserecord oftransfer Cromautomaticto manual.

218

PNEUMATICCONTROLMECHANISMS 219

in Figure 8-16. At the left, the unit is in the automatic position; at the
right, it is in manual. It will act as a simple proportional controller with
a fixed proportional band of approximately 30 percent. The basic probIem in transferring from manual to automatic lies in the fact that the
output of the automatic controller must equal that of the manual unit at
the moment of transfer, and then a change at a predetermined integral
rate is necessaryto bring the measurement to the set paint. The output
ofthe manual station is the control unit' s output. Full air supply is sent
to the three pneumatic switches, closing two and opening one, as
shown in Figure 8-16.
The proportional bellows of the automatic control unit is disconnected from the output of the controller. The integral restrictor is
bypassed, and the integral bellows is disconnected from the proportional bellows. The input signals to the automatic balancing unit shown
represent the automatic control unit proportional bellows pressure in
bellows D, and the manual control output in bellows A. Bellows B is
the balancing pressure and bellows C is the output to the integral bel-

lows.
If either the measurement or the set paint to the automatic control
unit changes, the pressure in the proportional bellows must also
change, because it is operating as a proportional-only control unit.
When the change in pressure in the proportional bellows is sensed by
the balancing unit, the unit' s output will change the pressure in the
controller's integral bellows. This, in turo, will cause the proportional
bellows pressure to change in the opposite direction until it once again
equals the output of the manual control relay, or until the supply
pressure limits are reached.
Any diiference between the set and measurement bellows is thus
balanced by the difference between the integral and proportional bel-

lows.
If the output of the manual control unit changes, a similar action
occurs, forcing the proportional bellows pressure to equal the manual
control unit output.
When transferring from manual to automatic (Figure 8-17), the
output will remain at the level determined by the operator when the
controller is in manual. If the measurement input is equal to the set
paint, the output remains constant until corrective action is required. If
the measurement does Dot equal the set paint at the moment of transfer ,
the output will camp from the level of manual operation to the level
necessary to make the measurement equal to set paint a function of the
controller's integral rate.

220

PNEUMATIC ANDELECTRONIC CONTROLSYSTEMS

Fig. 8-16. Schematic of auto-manual mechanism with transfer switch in automatic

position (above) and manual position (opposite page).

Set Po nt

The set-point knob is attachedto a pneumatictransmitter, which relates the transmitter's output positionor the set-pointpointer (Figures
8-18and 8-19).The output or the transmitteris applied to the set bellows or the controller. If the automaticcontroller is operating, normal

PNEUMATICCONTROLMECHANISMS 221

changes can be made by turning the set-paint knob. Proportional-plusintegral action will occur. Since the derivative amplifier exists only in
the measurementcircuit, derivative action will nat occur. If it is desired
to bring the process slowly to the new set paint with integral action
only, the controller is simply switched to manual, the set-paint change
is made, and the controller is switched back to automatic. Now the
measurement will approach the new set paint, with integral action only,
and no overshoot will occur.

222

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

SET paiNT

LTIME-i

TIME-

Fig. 8-17. TransCerCrommanual to automatic ([eft) when measurementequals set point

and (right) when measurement does not equal set point.

Fig. 8-18. FoxboropneumaticConsotrol130Mcontrolletpulled out to show

access-to-mode
adjustments.

PNEUMATICCONTROLMECHANISMS 223

LlNEARIZING
ASPIRATOR
RELAY

iC

TO REVERSE
FROMREMOTE
SWITCH AND
SET POINT
AUTOMATIC
SIGNAL
L.A.R.
CONTROL U 1T
.L

'j

LOCAL
I ~.
SETPOINT1n-"1
TRANS

AlS \~~-.
r-rr

II

REMOTE
SET
POINT

RECEIVERI

SET KNOB

Fig. 8-19. Schematicoflocal and remote set-pointmechanisms.

The Closed-Loop

Pneumatic

Controlled

System

Now let us apply the pneumatic controller to a process controlloop.

The objective of a control system is to maintain a balance between
supply and demand over time. (See Chapter 1 for a discussion of supply
and demand.) The closed-loop control system achieves this balance by
measuring the demand and regulating the supply to maintain the desired balance.
Figure 8-20 illustrates a typical and familiar controlled system in
which the temperature of heated water (controlled variable) is regulated by control of i~put steam. The control system comprises a

IIPULATED

224

PNEUMATIC AND ELECTRONIC CONTROL SYSTEMS

VARIABLE 1

Fig. 8.20. The processto becontrolledoccursin a beatexchanger.All elementsofthe

pneumaticcontrol systemare shown-transmitter, controller, valve, input water,
output water, and steam.

pneumatic temperature transmitter (Foxboro Model 12A), a forcebalance pneumatic controller (Foxboro 130 Series), a pneumatic (diaphragm) valve actuator, and a control val ve (wide-range V-port).
Figure 8-21 shows an accepted manner of representing a system in
a block diagram. All the elements of the actual system are included.

CONTROLLER

,'

:.--J ERROR 1 fAMPLIFIER'

"

iDETECTORM (RELAY)t=l

L-

-INTERNALFEEDBACKLOOPjI

SET

II M~.A_S.u.~.!.NG
II
MEANS
~EXTERNAL

CONTROLLED

FEEDBACK

LOOP

MANIPULATEO

Fig. 8-21. Block diagramshowingall the elementsor Figure8-20.

PNEUMATICCONTROLMECHANISMS 225

The Process

The process to be controlled is a shell-and-tube beat exchanger. Lowpressure steam applied to the shell beats water flowing through the
tubes. The exchanger has 33.5 square feet ofheat-transfer surface, and
the time required for the beat exchange to take place across this surface causes the exchanger to bave a delayed response. The response
characteristic is that of a multiple-capacitance, multiple-resi&tance
circuit.

The response or reaction curve of the beat exchanger was 'obtained

by (1) allowing the beat exchanger to stabilize with a constant flow of
both steam and water, (2) holding the water flow constant, (3) increasing the steam flow suddenly by opening the control valve, and (4)
plotting the increase in water temperature. The resulting curve is the
reaction curve illustrated in Figure 8-22.
'
Since the water temperature was measured with the pneumatie
temperature transmitter, the transmitter output depends nat only on the
response of the beat exchanger, but also on the response of the transmitter, which is nat instantaneous.
The Controller

Set Point

The set point o the controller is established by setting an air pressure

in a "set" bellows by means o an air regulator. The setting dial (Figure
8-23) is spread over the span o the measured (controlled) variable.

Fig. 8-22. Responseof outlet temperatureto stepsin steamvalve positiontest results.

226 PNEUMATIC AND ELECTRONICCONTROLSYSTEMS

OUTPUT

SET KNOB

Fig. 8-23. Schematic of set-point mechanism.

Valve and Actuator

The controller output positions a control valve which, in turn, governs

the energy or material inflow to the processoThe speed of response of
the valve depends on the type of valve actuator, that is, the device that
transduces the controller output signal into a valve position.
The actuator can be pneumatic or electric. The system under study
uses a familiar pneumatic actuator, which consists of a large diaphragm
(110 square inches) to which the pneumatic signal is-applied. When 3
psi (20 kPa) is applied to an area of 110 square inches, a force of 330
pounds is developed; when 15 psi (100 kPa) is applied, the thrust will be
1,650 pounds. This downward diaphragm force is opposed by a large
coil spring that pushes upward with a force of 330 to 1,650 pounds.
Increasing or decreasing the pneumatic signat to the diaphragm will
cause the diaphragm to move until it reaches a position where diaphragm force and spring force are in equilibrium. Since the relationship
between spring thrust and excursion is a linear one (Hooke's law),
valve trave} is related linearly to controller output.
Valve actuators mar be arranged either to open the valve or close it
on increasing air pressure. The choice of action is usually dictated by
the process being controlled. For a beat exchanger process, the air-toopen valve usually is selected because the spring will close the val ve
and cause the temperature to drop in event of air failure. This is called
fail-safe action. If the process were one that became hazardous when
the valve closed, the action would be reversed to make the valve open
on air failure.
The valve actuator has a response time because it has the capacity

PNEUMATICCONTROLMECHANISMS 227

to hold air and the connectingtubing offers resistanceto air ftow. The
time constant for a pneumatic operator is typically several seconds
with normallengths of connectingtubing. Adding tubing lengthensthe
time constant.Thus, the valve actuatorcontributestime lag or phase
shift to the controlloop, dependingon the lengthof connectingtubing

used.
Final Control

Element

The function of the pneumaticvalve actuatoris to positionthe control

valve. The valve canbe one of manytypes, dependinguponthe process
to be controlled. For the beat exchangerprocessbeing discussed,a
single-seatwide-rangeequalpercentagevalve is used (Figure 8-24).
The single-seatequalpercentagevalve hasa contouredinner valve
that provides an exponentialrelationship betweenvalve stroke and
valve capacity.

Fig. 8-24. Cutawayview of single-seat,wide-range,equalpercentvalve.

228 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Fig. 8-25. Closed-loopcontrol system,showingall basiccomponents.

Dynamic Behavior ot Closed-Loop Control Systems

When all components are connected to form the closed-loop control
system (Figure 8-25), each component contributes to control system
operation. For example, the total gain (output signal/input signal) is
affected by the gain of every one ofthe loop components. However, the
only loop component that has an adjustable gain is the controller. Adjustment ofthe proportional band results in adjustment ofthe totalloop
or system gain.
Let us assume that the control system has stabilized with the outftow temperature at 1500F,66C and that the system is subjected to an
upset by raising the set paint suddenly to 1600F(71C).

~SE

PNEUMATICCONTROLMECHANISMS 229

".,3."CON.'.,.
'" ,. 6.

,.

The action around the controlloop (Figure 8-26) will be as follows:

1. Increasing the set point increases air pressure in the set bellows,
thereby causing the ftapper to approach the nozzle.
2. Increased nozzle pressure is amplified and becomes output to the
control valve. Because the valve actuator is a resistancecapacitance element, the actual change in valve position will be
delayed behind the change in output signal.

230 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

3. The addedbeat (Btu) will be nearly in stepor phasewith the valve

position.
4. The added beat causesthe temperatureof the beat exchangerto
rise, and as it Tises,the pressurein the measurement
bellows will
increase.
5. Any unbalanceof the momentsof force contributed by the measurementand set bellows will causethe integral bellows to continue to change through the integral restricter, and the output
pressureto increaseuntil the temperaturereachesthe set point.
The momentsof force are now balanced,and the ftapperis within
its throttling range(a band of travelless than0.001 inch wide).
6. After approximately2 minutesbaveelapsed,the forcesexertedby
the various bellows will be in equilibrium; the ftapperis within its
throttling range; and the processhas stabilized at 160F,71C.
Thesestepsare illustrated in Figure 8-26.
(Note: If the upsetwere causedby a load change,derivative action
would bave beenaddedto the functions described.)
Adjusting the Controller
To accomplish effective control and cause the process to feaci in an
optimum fashion, the controller behavior, with respect to time, musi be
matched to thai of the processoThis is done by adjusting the proportional band, integral time, and derivative time.
Thning or adjusting the controller for optimum performance can be
achieved either by trial-and-error, or by more exacting methods. The
usual trial-and-error procedure in adjusting the controller settings to
the process conditions is to set the integral restrictor at maximum and
the derivative restrictor at minimum, and then to adjust the proportional band to produce the minimum process stabilization time. Then
the derivative restrictor is increased gradually, and the proportional
band narrowed, until a combination of proportional and derivative action is obtained which produces a shorter stabilization time than normal
proportional, and with less upset to the processo
To eliminate offset, the integral restrictor is then set to avalue thai
will bring the control point to the set point in a minimal time without
upsetting the stability of the system.
Other methods of adjustment by mathematical and analog analysis
are discussed in Chapter 12.
Suppose a set point of 20 percent scale change is made in the
Foxboro 130 Series controller with the controller on automatic. Note

PNEUMATICCONTROLMECHANISMS 231

OUTFLOW

WATER
TEMPERATUREI

Fig. 8-27. Controlled set-point change made in automatic mode (top) and in manual mode,
then switched to automatic (bottom).

the damped oscillations that occur as the system Testabilizes (Figure

8-27, top). The instability lasts for several minutes. If the controller is
first switched to manual for the set-point change, a reaction will occur
when switched back to automatic. This is shown in the lower part of
Figure 8-27 and discussed previously. The controlled variable retums
to the new set point at the integral rate. Any adjustments tend to cause
a temporary upset; therefore, the technique of switching to manual,
making any required adjustments and switching back to automatic, can
be used to advantage with the Foxboro 130Series controller.

Batch Controli er

In Chapter 7, reference was made to integral saturation or integral

windup. This occurs when the integral mode is used on a batch or
discontinuous processo During the time the process is shut down, the
integral circuit of a conventional two- or three-mode controller will
saturate at the supply pressure.
After the process is restarted, measurement overshoot of the set
point can occur unless precautions are taken. The batch switch is a
special pneumatic mechanism designed to prevent this overshoot.
During the time a process is otI control, the measurement will be

-Fig.

232 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

riME

CuRVE

8-28. (Left) Recovery from sustained deviation; proportional plus integral controller
without batch feature. (Right) With batch feature.

below the set point and the controller's output will be at its maximum
value. Ifthe time oftbis deviation is long enough, the integral circuit of
a conventional controller will also reach ibis value. When process conditions retum to normal, no change in controller output can occur until
the measurementreaches the set point. Figure 8-28, curve A illustrates
ibis control action.
The batch switch eliminates tbis integral circuit saturation and
conditions the integral bellows to permit output to start to change
before the measurement reaches theset point. If the controller output
starts to change before the measurement reaches the set point, overshoot can be prevented and the measurement can retum to the set point
smoothly. Figure 8-28, curve B illustrates tbis control action.
Principie

ot Operation

A schematic diagram of a proportional-plus-integral batch controller is

shown in Figure 8-29. The only addition to a conventional two-mode
controller is a specially designed pressure switch between the relay
output and the integral circuit.
The switch is actuated by the output pressure of the controller
relay. Its trip point is adjusted by a spring, or an external pneumatic
signal, to trigger at 15 psi (100 kPa). When the controller's output is
below this pressure, the ball valve closes the vent port and permits
passage of the relay output to the integral circuit. This allows normal
integral response as long as the measurement remains within the proportional band (below 15 psi or 100 kPa). Should the measurement
reach the 15 psi or 100 kPa limit of the proportional band throttling
range, the force on the batch switch diaphragm will seat the ball valve
in the opposite direction. This cuts off the relay output pressure to the
integral circuit and simultaneously vents the circuit to the atmosphere.

PNEUMATICCONTROLMECHANISMS 233

Fig. 8-29. SchematicoCproportional-plus-integraJ

batchcontroller.

This causes the pressure in the circuit to dropo As long as the relay
output pressure is above the trip point, the integral circuit pressure will
drop until there is no longer any pressure in the integral bellows. At this
point, the proportional band throttling range will have shifted completely below the set point as indicated in Figure 8-28B, curve B. During this operation, the controller's output pressure is at 15 psi or 100
kPa (or more) and the valve is wide open. When the batch process is
restarted, the control valve will begin to throttle as soon as the measurement reaches the 15 psi or 100kPa lmit of the proportional band
throttling range. Note that as soon asthe controller output drops below
15 psi or 100kPa, normal integral response is restored. This overcomes
the tendency to overshoot the set point.
Load Bias (Preload)

Due to the time characteristics of certain processes, shifting the proportional band completely below the set point will cause an intolerable
delay in bringing the batch to the set point. In such cases, an adjustable
back pressure mar be applied to the vent port to "bias" the batch
switch. Thus, the pressure in the integral circuit mar be prevented from
dropping below a preselected amount and the proportional band throttling range will shift only partially below the set point. Although this
will allow faster recovery, a slight overshoot will occur if the "bias" is

234 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

increased too much. Increasing the bias to 15 psi or 100 kPa would
obviously completely eliminate batch action.
One limitation thai applies to all pneumatic control systems is the
distance thai mar be accommodated between components. The maximum distance depends on two major factors: the pneumatic signal
travels at the speed of sound, and the loop components along with the
tubing all bave capacity and resistance. Thus, an RC time constant
musi exist. Volume boosters often are applied to lengthen ibis working
distance. Unfortunately, a volume booster will do nothing to speed up
signal velocity. A booster will help if a large volume, such as thai found
in a pneumatic valve actuator, is involved.
The distance lmit for pneumatic control systems is approximately
200 feet (60 meters). Iftbis distance factor is ignored, the dynamics ofthe
control system will suffer and reguli in poor control operation. If it
becomes necessary to lengthen these distances substantially, the only
practical solution is to utilize electrical signals thai mar work over an
almost unlimited distance.
Questions

8.1. The integraldial in a pneumaticcontrolleris calibratedin:

8. Minutes or repeats
d. Percentage
b. Integralunits
e. OtIset

c. Gain
8-2. True or false: In a proportional-onlycontroller,ifthe measurement
equals
the set point the output will equalthe bias.
8-3. True or false: In an integralcontroller,the rate of changeof the output
is proportionalto the error.
8.4. True or false: The largerthe numberon the integraldial the greaterthe
effectof the integralaction.
8-5. True or false: In a batchoperation,if a controllerhaswound up, it is
quite possiblethat the valve may stay in an extremepositionuntil the
measurementactuallygoesbeyondthe set point beforethe valve beginsto
changeits position.
8-6. True or false: What somemanufacturerscall rate others call derivative.
8-7. Indicate all the correctstatements:
8. Gainis the reciprocatof the proportionalband.
b. The proportionalbandis the reciprocatof gain.

PNEUMATICCONTROLMECHANISMS 235

c.
d.
e.
f.

The
The
The
The

proportional band times the gain equals 1.

gain divided by the proportional band equals 1.
narrower the proportional band, the higher the gain.
wider the proportional band, the lower the gain.

8-8. In a process controlled by a proportional-plus-integral controller, the

measurement was at the set point and the output was 9 psi (60 kPa). The
measurement then quickly decreased to a certain value below the set point
and leveled out there. The output responded by changing to 8 psi (55 kPa). As
time progresses, what would you expect of the output pressure?
a. Increase to 9 psi (60 kPa) to bring the measurement back to the set

point.
b. Remain at 8 psi (55 kPa) as long as the measurement stays where it is.
c. Decrease and continue to decrease to 3 psi (20 kPa).
d. Continue to decrease until the measurementreaches the set point, or if
it does not retum to the set point, decrease to Opsi.
8-9. The range of the temperature measuring system used in conjunction
with a Model 130proportional-only contro1ler is O to 150oP(66C). The output
is 9 psi (60 kPa) when the set point and indicator are both at 75P (24C). If
the proportional band is 200 percent, what is the output when the
measurement is 150oP(66C)?
a. 9 psi (60 kPa)
c. 12 psi (83 kPa)
b. 3 psi (20 kPa)
d. 15 psi (100 kPa)8-10.
If the span of a measuring transmitter in a control system is made
one-half of its value, the proportional-band adjustment in the controller must
be -to
maintain the same quality of control.
a. Cut in half
d. Narrowed
b. Doubled
e. None of the above.

c. Squared
8-11. With a proportional-plus-integral controller, a sustained error will result

in:
a.
b.
c.
d.
e.

Windup
A fixed offset
A temporarynarrowingof the proportionalband
A delay in the process
None of the above

8-12. By locating the derivative function in the input measurement

circuit,
which of the following advantagescanbe realized?
a. Smoothbumplesstransfer
b. No derivative bump with a set-pointchange
c. No proportionalresponseto a set-pointchange
d. Integral adjustmentis isolated from response

8-13.

236 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

a.
b.
c.
d.

If proportional-plus-integral
controlis good,the additionof derivative:
Will anticipatechangesand speedup corrections
Will alwaysimprove control
Will makethe controlleradjustmentseasierto accomplish
May create stability problemsin somesystems

8-14. The advantageof addingderivativeto a controlleris always:

a. Increasedstability
b. The ability to overcomea big pure deadtime lag
c. The ability to reactmore quickly to measurement
change
d. A decreasein the pure deadtime of the process
8-15. For fail-safeactionthe control valve should,uponenergy(air)failure:
a. Open
b. Close
c. Move in suchdirectionas to makethe processnonhazardous
d. Stay in its previous position
8-16. If the closed-loopcontrol systemhastoo muchgain, it will cycle. The
only loop componentthat hasconvenientlyadjustablegainis the:
a. Measuringtransmitter
c. Process
b. Valve operator
d. Controller
8-17. Adjusting the controller for optimumperformance:
a. Is not required,becauseit adjustsitself
b. Requiresa specialtool
c. Is usuallydone by trial and error
d. Always requiresa very involved mathematicalanalysisof the process
8-18. A processis to be controlledusingan all pneumaticsystem.The
maximumdistancebetweenloop componentswill be:
a. 1,000feet
c. 200feet
b. 500feet
d. 20 feet
8-19. If the distancebetweenloop componentsin the all pneumaticcontrol
systemmustbe increasedit will require:
a. A pneumaticvolume booster
b. Largersizedtubing
c. Smallersizedtubing
d. Conversionto an electricalsignal
8-20. PneumaticsignalstraveI throughthe signaltubing al:
a. 100feet per second
b. Approximatelythe speedof sound
c. A rate that dependson tubing size
d. Approximatelythe speedof light

237

Electronic control systems,already widely accepted,are gainingrapidly in popularity for a numberof reasons:
1. Electrical signalsoperateover greatdistanceswithout contributing
time lags.
2. Electrical signals can easily be made compatible with a digital
computer.
3. Electronic units can easily handle multiple-signalinputs.
4. Electronic devices can be designedto be essentiallymaintenance
free.
s. Intrinsic safety techniques bave virtually eliminated electrical
hazards.
6. Generally, electricalsystemsare lessexpensiveto install, take up
less space,and can handle almost all processmeasurements.
7. Electronic devices are more energy efficient than comparable
pneumatic equipment.

8. Specialpurposedevices suchas nonlinearcontrollers and analogcomputingunits are simplified.

This chapter will describethe electronic instrumentcomponents
that make up a typical multiloop system. The processis one found
within the boiler room of most processplants-a feedwater control
systemfor a boiler drum.

238 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

Feedwater

Control

Systems

A sample electronic control system is shown in Figure 9-1. The water

level in the drum is to be controlled. This is accomplished by measuring
Dot only the water level, but the steam ftow out of the boiler and the
water ftow into the drum. The three measured variables are fed to an
electronic control system, which controls the water to the drum.
Transm itters

Two ditIerential pressure transmitters (Fig. 9-2) are used, one to measure water flow; the other to measure drum water level.
Referring to Fig. 9-3, in operation the ditIerence in pressure between the high and low gide of the transmitter body is sensed by a twin
diaphragm capsule (1) which transforms the ditIerential pressure into a
force equal to the ditIerential pressure times the etIective area of the
diaphragm. The resultant force is transferred through the C-flexure (2)
to the lower end of the force bar (3). Attached to the force bar is a
cobalt-nickel alloy diaphragm which serves as a fulcrum point for the
force bar and also as a seal to the process in the low-pressure cavity
gide of the transmitter body. As a result of the force generated, the

Fig. 9-1. Peedwater control system. (I) Drum water level (2) Steam ftow from boiler,
via turbine steam pressure. and (3) Water ftow into drum.

Fig. 9-2. Differentialpressuretransmitter.

Fig. 9-3. Operationof electronicforce balancedifIerentialpressuretransmitter.

239

force

240 PNEUMATIC AND ELECTRONICCONTROL SYSTEMS

bar pivots abolli the CoNi alloyseal, transferring a force to the

vector mechanism (5).
The force transmitted by the vector mechanism to the lever system
(11) is dependent on the adjustable angle. Changing the angle adjuststhe
span of the instrument. At point (6), the lever system pivots and
moves a femte disk, part ora differential transformer (7) which serves
as a detector. Any position change of the ferrite disk changes the output of the differential transformer determining the amplitude output of
an oscillator (8). The oscillator output is rectified to a d-c signat and
amplified, resulting in a,4-20 mA d-c transmitter output signal. A feedback motor (9) in series with the output signal, exerts a force proportional to the error signal generated by the differential transformer. This
force re balances the lever system. Accordingly, the output signal ofthe
transmitter is directly proportional to the applied differential pressure
at the capsule.
Any given applied differential pressure within the calibrated measurement range will result in the positioning of the detector' s femte
disk, which, in turn, will maintain an output signal from the amplifier
proportional to measurement, thus keeping the force balance in equilibrium. A simplified schematic of the electronic circuitry is shown in Fig.

9-4.
Steam ftow out ofthe drum is measured in terms of steam pressure,
which is measured with a pressure transmitter. The operation of the
pressure transmitter is similar to the operation of the differential

Fig. 9-4. Working schematic oCelectronic transmitter.

ELECTRONIC(ANALOG) CONTROLSYSTEMS 241

pressure transmitter_,The same feedback technique used in the differential pressure transmitter is employed in the pressure transmitter. The
major difference lies in the size and construction of the sensor.
The pressure transmitter measures the pressure in the first stage of
the power turbine. Since there is a linear relationship between first
stage turbine pressure and steam flow, the output signal from the
pressure transmitter is linear with steam flow out of the boiler,
A variety of electronic differential pressure transmitters are available from a number of manufacturers. Some of these make use of strain
gauge detectors, capacitive detectors, resonant wire detectors, and inductive detectors. Many are motion or open loop devices which are still
capable of accuracies within a fraction of a percent. In the application
of any electronic transmitter, the user must guard against subjecting the
electronics to temperatures which might result in damage. This probIem generally can be avoided by observing the precautions recommended by the manufacturer.

The Controllers

The controllers described befe are the Foxboro SPEC 200 type. The
SPEC 200 is generally a split-architecture system. In this system two
areas may be used, a display area and a oest area. Field equipment,
such as measuring transmitters, electrical valve actuators, and the like,
generally operates on 4 to 20 mA dc. Within the oest, SPEC 200 operates on O to 10 volts.
The display area contains control stations, manual stations, recorders, and indicators to provide the necessary operator displays and
controls. These units are shelf-mounted and contain only the electronic
circuitry required to communicate the display and adjustments necessary for an operator to control and monitor a processo
The nest area contains the analog control, computing, alarm, signal
conditioning, and input and output signal converter units. These units
are in the form of "modules" and "circuit cards." The oest itself is
basically an enclosure provided for the mounting of "modules."
System power is supplied to the oest. This power supply must
deli ver + 15 and -15 V dc for operation of the display and nestmounted instruments. Recorder chart drives and alarm lights require 24
V ac.
When a single location is required, the oest and display areas may