4.
Process Control Preliminaries
The final control element, process and sensor/transmitter all have
their own gain, time constant and dead time
controller
error
Set Point
controller
output
Controller
manipulated
process
variable
measured
process
variable
Final
Control
Element
Process
Disturbance
feedback
signal
Measurement
Sensor/Transmitter
Copyright 2002
Douglas J. Cooper
All Rights Reserved
The Process Designers Process
For new installations, the dynamic behaviors of valves and sensors
should be carefully considered prior to purchase
If undesirable behavior in an existing loop can be traced to certain
equipment, then consider relocation or replacement
A final control element and sensor should start to respond quickly
(add little dead time) and complete the response quickly (have a
small time constant)
The qualifier quickly is relative to the overall time constant of the
process (a P of 9 min is large relative to a P of 10 min but small
relative to a P of 1000 min)
Copyright 2002
Douglas J. Cooper
All Rights Reserved
Redefining Process for Controller Design/Tuning
The controller sends a signal out on one wire and receives the result as a
change in measurement on another wire
From the controllers view, the individual gains, time constants and dead
times all lump into one overall dynamic behavior
Here, process dynamics refers to these combined behaviors
controller
error, e(t)
Set Point
+-
manipulated variable is the
controller output, u(t)
Controller
controllers
process
Final Control Element
Process
Sensor/Transmitter
Disturbance
measured process
variable, y(t)
Copyright 2002
Douglas J. Cooper
All Rights Reserved
On/Off Control The Simplest Controller
For on/off control, the final control element is either completely
open/on/maximum or closed/off/minimum
To protect the final control element from wear, a dead band or upper and a
lower set point is used
As long as the measured variable remains between these limits, no changes
in control action are made
V a r ia b le
C o n t r o l l e r O u Pt pr uo tc e s s
P ro c e s s : C u s t o m P ro c e s s
56
C o n t ro l l e r: M a n u a l M o d e
upper dead band limit
54
52
50
48
46
44
lower dead band limit
55
ON
50
45
OFF
250
Copyright 2002
Douglas J. Cooper
All Rights Reserved
300
350
400
450
500
T im e ( tim e u n its )
550
600
650
700
Usefulness of On/Off Control
On/off with dead band is useful for home appliances such as furnaces,
air conditioners, ovens and refrigerators
For most industrial applications, on/off is too limiting (think about
riding in a car that has on/off cruise control)
Copyright 2002
Douglas J. Cooper
All Rights Reserved
Intermediate Value Control and PID
Industry requires controllers that permit tighter control with less
oscillation in the measured process variable
These algorithms:
compute a complete range of actions between full on/off
require a final control element that can assume intermediate
positions between full on/off
Example final control elements include process valves, variable speed
pumps and compressors, and electronic heating elements
Copyright 2002
Douglas J. Cooper
All Rights Reserved
Intermediate Value Control and PID
Popular intermediate value controller is PID (proportional-integralderivative) controller
PID computes a controller output signal based on control error:
u (t ) ubias KC e(t ) +
Proportional
term
KC
e(t )dt
Integral
term
de(t )
dt
Derivative
term
KC D
where:
y(t)
u(t)
ubias
e(t)
KC
D
Copyright 2002
Douglas J. Cooper
All Rights Reserved
= measured process variable
= controller output signal
= controller bias or null value
= controller error = ysetpoint y(t)
= controller gain (a tuning parameter)
= controller reset time (a tuning parameter)
= controller derivative time (a tuning parameter)
5. P-Only => The Simplest PID Controller
The Proportional Controller
The simplest PID controller is proportional or P-Only control
It can compute intermediate control values between 0 100%
C o n tr o l S ta tio n : C a se S tu d ie s
C o n t r o l l e r O u t p uL t e v e l / S e t p o i n t
P ro c e s s : G ra v i t y D ra i n e d T a n k
2 .8
2 .6
2 .4
2 .2
2 .0
1 .8
60
58
56
54
52
50
10
Copyright 2002
Douglas J. Cooper
All Rights Reserved
C o n t ro l l e r: P ID ( P = R A , I= o ff, D = o ff )
15
20
25
30
35
T im e ( m in s )
T u n i n g : B i a s = 5 5 .2 , G a i n = 1 1 .5 , S a m p l e T i m e = 1 .0 0
40
The Gravity Drained Tanks Control Loop
Measurement, computation and control action repeat every loop
sample time:
a sensor measures the liquid level in the lower tank
this measurement is subtracted from the set point level to
determine a control error; e(t) = ysetpoint y(t)
the controller computes an output based on this error and it is
transmitted to the valve, causing it to move
this causes the liquid flow rate into the top tank to change, which
ultimately changes the level in the lower tank
The goal is to eliminate the controller error by making the measured
level in the lower tank equal the set point level
Copyright 2002
Douglas J. Cooper
All Rights Reserved
P-Only Control
The controller computes an output signal every sample time:
u(t) = ubias + KC e(t)
where
u(t)
ubias
e(t)
KC
Copyright 2002
Douglas J. Cooper
All Rights Reserved
= controller output
= the controller bias
= controller error = set point measurement
= ysetpoint y(t)
= controller gain (a tuning parameter)
controller gain, KC steady state process gain, KP
a larger KC means a more active controller
like KP , controller gain has a size, sign and units
Design Level of Operation
Copyright 2002
Douglas J. Cooper
All Rights Reserved
A controller is designed for a particular process behavior (or
particular values of the FOPDT parameters KP, P and P)
Real processes are nonlinear, so their behavior changes as
operating level changes
Thus, a controller should be designed for a specific level of
operation
Collect Process Data at the Design Level
Copyright 2002
Douglas J. Cooper
All Rights Reserved
the design value for the measured process variable is where the
set point will be set during normal operation
the design values for the important disturbance variables are
their typical or expected levels during normal operation
perform the dynamic test as near practical to the design level of
the measured process variable when the disturbances are quiet
and near their typical values
Controller Gain, KC, From Correlations
P-Only control has one adjustable or tuning parameter, KC
u(t) = ubias + KC e(t)
KC sets the activity of the controller to changes in error, e(t)
if KC is small, the controller is sluggish
If KC is large, the controller is aggressive
To determine KC, use this controller design procedure:
generate dynamic process data at the design level of operation
fit a FOPDT dynamic model to this data
use the FOPDT model parameters in a correlation to compute
initial estimates of KC
Copyright 2002
Douglas J. Cooper
All Rights Reserved
Controller Gain, KC, From Correlations
Integral of time-weighted absolute error (ITAE) tuning correlations:
if set point tracking (servo control) is the objective:
KC
Kp
( p / p ) -1.219
if disturbance rejection (regulatory control) is the objective:
KC
Copyright 2002
Douglas J. Cooper
All Rights Reserved
0.202
0.490
Kp
( p / p ) -1.084
Controller Gain, KC, From Correlations
Copyright 2002
Douglas J. Cooper
All Rights Reserved
Correlations provide an initial guess or starting point only
Final tuning requires online trial and error because:
the designer may desire performance different from that
provided by the correlation
the FOPDT model used for tuning may not match the actual
dynamic behavior of the plant
performance must be balanced over a range of nonlinear
operation
performance must be balanced for set point tracking and
disturbance rejection
Notes:
the designer defines best control performance
it is conservative to start with a small KC value
Understanding Controller Bias, ubias
Thought Experiment:
Consider P-Only cruise control where u(t) is the flow of gas
Suppose velocity set point = measured velocity = 70 kph
Since y(t) = ysetpoint then e(t) = 0 and P-Only controller is
u(t) = ubias + 0
If ubias were set to zero, then the flow of gas to the engine would be
zero even though the car is going 70 kph
If the car is going 70 kph, there clearly is a baseline flow of gas
This baseline controller output is the bias or null value
Copyright 2002
Douglas J. Cooper
All Rights Reserved
Understanding Controller Bias, ubias
Copyright 2002
Douglas J. Cooper
All Rights Reserved
In the thought experiment, the bias is the flow of gas which, in
open loop, causes the car to travel the design velocity of 70 kph
when the disturbances are at their normal or expected values
In general, ubias is the value of the controller output that, in open
loop, causes the measured process variable to maintain steady
state at the design level of operation when the process
disturbances are at their design or expected values
Controller bias is not normally adjusted once the controller is put
in automatic
Reverse Acting, Direct Acting and Control Action
If KP is positive and the process variable is too high, the controller
decreases the controller output to correct the error
=> The controller action is the reverse of the problem
When KP is positive, the controller is reverse acting
When KP is negative, the controller is direct acting
Since KC always has the same sign as KP , then
KP and KC positive
KP and KC negative
Copyright 2002
Douglas J. Cooper
All Rights Reserved
reverse acting
direct acting
Reverse Acting, Direct Acting and Control Action
Copyright 2002
Douglas J. Cooper
All Rights Reserved
Most commercial controllers require you to enter a positive KC
The sign (or action) of the controller is specified by entering the
controller as reverse or direct acting
If the wrong control action is entered, the controller will drive
the valve to full open or closed until the entry is corrected
Offset - The Big Disadvantage of P-Only Control
Copyright 2002
Douglas J. Cooper
All Rights Reserved
Big advantage of P-Only control:
=> only one tuning parameter so its easy to find best tuning
Big disadvantage:
=> the controller permits offset
Offset - The Big Disadvantage of P-Only Control
Offset occurs under P-Only control when the set point and/or
disturbances are at values other than that used as the design level
of operation (that used to determine ubias)
u(t) = ubias + KC e(t)
How can the P-Only controller compute a value for u(t) that is
different from ubias at steady state?
The only way is if e(t) 0
Copyright 2002
Douglas J. Cooper
All Rights Reserved
Offset and KC
As KC increases: Offset decreases Oscillatory behavior increases
Impact of KC on Offset and Oscillatory Behavior
C o n tr o lle r O u tp u t P V / S e t p o in t
P r o c e s s : G r a v ity D r a in e d T a n k
3 .0
2 .8
2 .6
2 .4
offset
offset
2 .2
KC = 40
KC = 11.5
75
70
65
60
55
50
45
10
Copyright 2002
Douglas J. Cooper
All Rights Reserved
C o n tr o lle r : P I D ( P = R A , I = o f f , D = o f f )
20
30
40
50
T im e ( m in s )
T u n in g : B ia s = 5 5 .2 , G a in = 4 0 .0 , S a m p le T im e = 1 . 0 0
60
70
Proportional Band
Manufacturers want you as a repeat customer so they confuse the
issue with different tuning parameter names and units
With units of percent that range from 0 to 100%, then:
100
PB
KC
Copyright 2002
Douglas J. Cooper
All Rights Reserved
So if KC is large, then PB will be small
Bumpless Transfer to Automatic
A bumpless transfer" achieves a smooth transition to closed
loop by automatically:
Copyright 2002
Douglas J. Cooper
All Rights Reserved
setting ubias equal to the current controller output value
setting the set point equal to the current measured process
variable value
So when put in automatic, there is no controller error and the
bias is properly set to produce no offset
As a result, no immediate control action is necessary that would
bump the measured process variable
Workshop 2
P-Only Control of Tank Level
Copyright 2002
Douglas J. Cooper
All Rights Reserved
