Reciprocating Compressor Performance and Safety

Prediction for Control Panels

By Dwayne A. Hickman

SECTION 1: Introduction
There are three general methods of modeling reciprocating compressor performance and safety
prediction commonly used in unit control panel logic:

 Fast and Easy

 Covers 90% of Concerns
 Covers Everything according to the Compressor Manufacturer (OEM)

When overseeing a compressor or compressor station, one needs to know which method is being
implemented in each of your control panels (UCPs). If it’s not the OEM method, then there had better
be a good reason why something inferior is being used, especially if the unit is damaged due to
mechanical stresses.

Before discussing various modeling methods, it is important to note that real-time modeling of
performance and safety is different than real-time variable measurement, such as actual discharge
temperatures and internal cylinder pressures. Real-time modeling of performance and safety allows the
PLC to predict unit performance and to identify ahead of time within its defined operating map where it
is safe to run that unit, but it does assume that the unit is configured properly and that it is in reasonably
good health. Real-time monitoring reviews unit performance based on actual unit configuration (healthy
or damaged) and operating conditions, but provides no insight on how to get out of unsafe operations
other than flag alarms and shut down.

OEM Performance and Safety Prediction is required as it provides useful predictions for achieving user
goals, keeping units from overloading and out of unsafe areas. Adding optional real-time monitoring can
help prevent continued operations in high mechanical stress (e.g. like high rod load forces due to valve
breakages from slugs of liquids). Nevertheless, even if you choose to add real-time monitoring, you still
must make sure that the real-time monitoring devices are modeling your equipment in accordance with
OEM modeling specifications. Without this step, inappropriate shut downs or hardware damage can still
occur.

1
www.compressortech2.com
SECTION 2: Evaluating The Methods
For compressor stations with UCPs based on the Fast and Easy method, self-preservation suggests
only visiting those stations when their units are shut down. This method is often implemented by system
integrators (SI), whose programmers curvefit a handful of calculated points, and then apply a few
generic rules (such as maximum pressure differentials and maximum compression ratios) to keep the
units running and “safe”. This simple method typically does not pre-estimate interstage pressures
before load steps are changed or look ahead to determine if changing unit speed creates pin reversal
issues. For single-stage units that operate over narrow ranges, use only double-acting cylinders and do
not apply too much clearance to either head ends or crank ends. This approach is not completely
without some merit. It is commonly implemented in UCPs for many slow-speed, transmission and gas
gathering applications. The real problems begin when this Fast and Easy method is applied to more
generic units and/or wider operating ranges with the belief that this simple method is actually good to
use for all applications. It’s not.

The Covers 90% of Concerns approach involves more calculations

regarding unit performance and safety, and is the most frequently used
method by experienced end-users and their elected and experienced
“The industry does
system integrators. The calculations and methods for determining safe
operating limits are more robust, better understood, and are based on
offer a couple
actual modeling equations that calculate performance in real-time
versus a static curvefit approach. Here, thermodynamics are
approaches.”
considered, interstage pressures are reasonably predicted, rod loads
are better predicted, and the ability to model both double-acting and single-acting compressor modes is
standard. This approach works well for gas gathering, gas boosting, transmission, injection, withdrawal,
and even some process services. However, this method typically only works well for slow-speed units.
For those units, the legacy compressor manufacturer typically reviewed the operating ranges (and often
indicated the reviewed, and thus allowed, safety ranges on the unit nameplating) for complex issues
like internal rod loads, pin loading forces, degrees of pin reversal, and crank pin forces. The control
panel could effectively ignore those issues and concentrate on overload, high pressures, static (gas
force) rod loads, low volumetric efficiencies, and high discharge temperatures. The problem with this
method begins when it is applied to high-speed units under the belief that a method appropriate for
slow-speed units is generically valid for high-speed units. It’s not.

The need for a third option, an OEM-based method is evident. It is a method that covers unit
performance and safety according to all OEM limits and methods. While this may seem like the obvious
choice for all units to use, keep in mind that a higher level of modeling complexity brings about a higher
dollar cost. Additionally, how does one model unit performance and safety for each OEM’s line of
compressors if that knowledge is not readily available? How can one expect a low-level (and
computationally slower) PLC to handle the complex mathematical methods that take even a high-end
desktop PC a few seconds to compute? Fortunately, the industry does offer a couple approaches to
satisfy this option. The first approach is via unit-specific PLC algorithms based on OEM performance
and safety limits programmed directly into the UCP. The second approach is via use of an add-on
micro-controller that computes complex performance in real-time using the actual OEM-specified
equations and methods.

2
www.compressortech2.com
SECTION 3: The Importance of Using the OEM Method
Why is using correct OEM methods for performance and safety so important? The answer is safety,
warranty and optimization. OEM-X does not warranty their units based on OEM-Y’s method of modeling
mechanical issues like rod loads. The unit must run within the limits defined by OEM-X. OEM limits are
in part based on how the OEM calculates unit forces and stresses. Some use static pressures, some
use adjusted static pressures, some use dynamic internal gas pressures, and some use dynamic
internal gas pressures along with inertia forces. Additionally, while often reported as “rod loads”, some
OEMs really calculate pin loading forces since that’s the weakest area for their piston-rod-crosshead
assemblies.

Here lies the real problem. If the wrong OEM method is applied, then there are four potential outcomes:

1. The results (almost) perfectly match the OEM’s results. That doesn’t mean they are right.

2. The results create a conservative operating map. That is, the calculations prevent the
unit from running in areas that it can safely operate. While this is better than the next two
outcomes, it also means that the unit is not always being optimized and thus not likely
delivering maximum flows and revenues.

3. The results allow the unit to operate in unsafe areas. That is, the calculations indicate
that certain areas in the operating map are safe, when in fact, according to the OEM
running the unit in those areas results in specific safety limits being exceeded. This is
the worst case scenario for safety.

4. The results are conservative in some areas of the operating map while allowing unsafe
operations in other areas. This is likely the common case for many current PLC
implementations of compressor performance and safety in UCPs – limiting the available
operating map on one hand and unknowingly exposing the unit to cumulative and
potentially catastrophic damage on the other hand.

The proper control and performance prediction of reciprocating compressors presents a serious issue
for those who operate them. Most large reciprocating compressors available today are high-speed,
separable units. In almost all cases, the OEM for the compressor does not offer an engine or motor,
and may have limited engineering connections with the driver OEM. While some exceptions do exist,
the majority of today’s units are the result of the end-user specifying a preferred driver and a preferred
compressor.

3
www.compressortech2.com
SECTION 4: Adding “Smarts” via PLCs
Most of today’s modern high-power reciprocating engines and electric motors include OEM designed,
developed and approved control panels with the purchase of the driver. This makes sense as the driver
OEM knows more about the abilities of their equipment than anyone else; It makes sense that they
would develop a robust control panel that both maximizes the unit’s ability while also maintaining proper
safety based on current operating conditions as determined by the engineering limits and designs of the
driver. Moreover, the driver OEM panel can act independently and change driver settings (speed, fuel
rate, turbo speed, current draw, etc.) directly to keep the driver safe, and it can also independently act if
necessary to shut down the unit if serious issues are identified.

Unfortunately, the equivalent of a driver OEM control panel does not readily exist for reciprocating
compressors. This leaves control, performance prediction, safety prediction, and dynamic compressor
reconfiguration left to the end-user, the packager, or a third party automation company. Since two very
expensive pieces of equipment must be coupled together (a driver with a compressor), along with other
numerous auxiliary items (yard valves, bypass valves, coolers, etc.) the control concepts (start up,
idling, shut down, unloading device actuation, liquids removal, speed control, etc.) necessary for the
wide variety of reciprocating compressor applications makes it difficult for a single set of control logic to
address all situations. However, it is quite possible for two complimentary and interactive control panels
(one for the driver and one for the compressor) to coexist with a third panel that manages the balance
of package control requirements. With today’s open distributed control architectures, this is a very
reliable approach.

“It is quite For some situations, such as gas gathering, only a few system elements
such as speed governors, suction pressure regulators, and recycle
possible for two valves are automated and thus controllable by a control panel. Some
standardized control panels have typical controls for these three basic
complimentary devices encoded into them , allowing the control panel to adjust one,
two, or all three types of devices to keep the unit running within the end-
and interactive user’s requirements. This type of solution works well when the only
issue of concern is overload (the standard panel can estimate load
control panels based on the driver’s fuel usage or electricity usage), low suction
pressures, high discharge pressures, or to maintain a desired flow rate.
to coexist with a Nevertheless, since operators can still manually adjust manual Variable
Volume Clearance Pockets (VVCPs) to set head end clearances,
third panel.” add/remove valve spacers, or pull valves to deactivate ends, all
combinations of pressures vs. speed vs. suction gas temperatures vs.
end clearance settings vs. ends deactivated must be reviewed beforehand. This is required to make
sure that running the unit in one of the potential hardware configurations cannot lead to mechanical
failure. Areas that lead to violations of the OEM’s Safety Limits must then be “locked out” from
allowable operations.

For example, on a typical two-stage, two-throw high-speed unit running from 50 to 150 psig (345 to
1035 kpag) suction pressure, 800 to 1100 psig (5516 to 7584 kpag) discharge pressure, 1100 to 1500
rpm, suction gas temperatures ranging from 40° to 90°F (4.4° to 32.2 °C), VVCPs on both stages, and
a need to pull suction valves from the first stage’s head end at times, a review should be made that
covers 2 to 16 million operating points before any control panel is allowed to arbitrarily adjust speed or
suction pressure. Unless the panel on a compressor has been reviewed for all possible combinations
4
www.compressortech2.com
before it was allowed to control the unit, how can one be confident that there are no areas where the
panel may turn down the driver speed in an effort to control low suction pressure only to put the unit in
pin non-reversal conditions that burn up bushings and possibly destroy crossheads, rods, pistons, or
even crankshafts? If those fail, then the possibility of a fire, an explosion, loss of station operations, and
even injury or loss of life exists.

SECTION 5: Failure is Not an Option

If such a failure were to occur, who would accept responsibility? The end-user for not requiring the pre-
checks across their entire operating range? The packager for not requiring the checks from the UCP
provider? The control panel provider? The compressor OEM? It is not in the best interest for any party
associated with the design, installation and operation of the compressor to have that unit fail. OEMs
build hardware to meet the challenges of end-user applications. However, all mechanical systems have
limits. Reciprocating compressors often run under conditions of high stress on the components. OEMs
design their units to handle those high stresses. But if the unit is operated in conditions that exceed
allowed limits, then damage can (and will likely) occur. Stressed metal does not heal itself.
Consequently, operating the unit under excessive forces weakens the components and thus the unit
may fail later even when it is running within OEM specifications effectively making the unit’s safety
unpredictable. That is not a good situation for anyone, especially the operators.

End users certainly expect that any installed UCP would act to protect their compressor units, the
station, and personnel, and not arbitrarily adjust control valves, unloading hardware, or speed in such a
manner that the mechanical forces acting on rods and pins exceeded allowed OEM limits damaging or
potentially destroying compressors. End users must make sure that their expectations are clearly
specified in bid requests – that automation control panels must protect the unit according to all of the
OEM’s specifications for safe operations.

Packagers certainly expect that any installed UCP would act to

protect their sold/rented unit and personnel, and not just
“It is not in the best
arbitrarily adjust hardware and potentially destroy the
compressor. Therefore, packagers must make sure that their
interest ... to have
expectations are clearly specified in requirements to the
automation group (whether intra-company or a third party
that unit fail.”
company) that the supplied control panel must protect the unit
according to all of the OEM’s specifications for safe
operations.

System integrators want an installed UCP to properly start up, safely operate the unit, and properly shut
down the compressor according to the OEM’s specifications. However, while most compressor OEMs
provide high-end Windows-based performance modeling software for determining if a specific condition
is safe or not, they do not generally provide the actual equations and modeling methods for others to
duplicate the OEM’s calculations and methods. Adding a Windows-based co-processing unit to run the
OEM software is certainly an option, but most automation experts shy away from using a high-level,
complex operating system (OS) for critical applications due to concerns with security, crashing, stability,
computational speed, driver conflicts, and software update issues.

5
www.compressortech2.com
Even on a fast Windows-based PC, some OEM software can take as long as half a minute to calculate
the performance required for all potential load steps. By the time the required performance is
calculated, the high-speed unit would already have completed 100 to 900 revolutions, potentially
damaging itself. System integrators need to be advised by the end-user, the packager, the OEM, or by
an end-user/industry accepted compressor expert on how best to safely model the unit while staying
within the computing ability of the control panel.

SECTION 6: Modeling OEM Methods

Many OEMs have openly released sanitized versions of their modeling equations. This is a good start,
but SI programmers still do not have all of the data and methods to be 100% compliant with that OEM’s
safety requirements. Typical (and critical) items not released by OEMs are equations or methods
related to thermodynamic calculations; determination of degrees of pin reversal and pin loading;
determination of expected interstage pressures; determination of cylinder end internal pressures; and
determination of internal gas force plus inertia rod loads. Nevertheless, SI programmers are being
asked to fully incorporate OEM safety limits for each OEM based on that OEM’s developed methods to
determine unit safety.
Consequently, a SI must:

1. Go to the OEM for guidance

2. Ask the end-user what methods they want to implement
3. Work with an expert recognized in compressor performance for all OEMs.

The first option makes the most sense as the OEM knows more about their equipment and safety limits
than anyone else. However, asking for OEM guidance can be awkward. OEMs want end-users to
control their compressors safely, but also need to protect their proprietary methods (developed via
costly R&D) that may give them certain advantages over their competitors in select areas (e.g.
advanced adjustments to volumetric efficiencies to handle challenging process gases). Furthermore,
some items such as gas thermodynamic routines, used in OEM software cannot run efficiently on
PLCs, and the OEM may not even have access to the complex source code for those routines to
provide the automation programmers.

The second option, asking the end-user for guidance, works

“Many OEMs have better with slow-speed units (less than 450 rpm) as end-
users have had many years of experience with slow-speed
openly released units to develop and perfect how they want those units
modeled. But reapplication of even slow-speed units to
sanitized versions different operating ranges from which they were originally

of their modeling
applied creates concern. Things as simple as changes to
the lubrication fluid (to new more environmentally friendly

equations.”
oils, etc.) can cause units to fail where they seemed to run
fine before (i.e. the new lubrication fluid does not handle the
forces as well and is
wiped clean more readily leading to increased metal to metal friction). Moreover, many of the slow-
speed end-user performance gurus have retired, leaving a void in many end-users’ knowledgebase of
compressor performance and safety prediction.

6
www.compressortech2.com
The third option, consulting with compressor experts, is what most end-users are opting for in today’s
high-speed compressor era. As such, end-users often direct packagers and/or automation specialists to
consult with specific experts to create appropriate models for the control panel. A compression expert
will know the calculation differences between the various OEMs’ rod load and pin load calculation
methods. They will be able to review thousands to millions of operating points to see where the
compressor may experience problems. They will be able to calculate valve losses similar to each
OEM’s method, and they will know how to predict interstage pressures and safety issues for all
hardware unloading configurations (load steps) before any unit hardware is changed (including speed
changes).

System integrators for a reciprocating compressor UCP are expected to be experts in automation,
control logic, data logging, HMI screens and SCADA, but not experts in compressor performance
predictions. Nevertheless, they do need to know with whom to work to incorporate the correct modeling
approaches and correct safety implementations for the select OEM’s equipment. Ultimately, the end-all
expert for a compressor’s performance and safety is the OEM of that compressor, as the OEM
determines the limits, the methods for predicting forces and stresses, and the modeling equations used.
Any selected compressor expert should have an excellent working relationship with the compressor
OEMs.

SECTION 7: OEM Solutions to Modeling Requests

Where does the compressor OEM’s responsibility begin and end? In short, they cannot reasonably
expect anyone to be able to model and safely control their compressors without full disclosure of all
limits involved in protecting the unit, and all methods for calculating those limits. However, these
disclosures do not necessarily have to be made public. The OEMs can work with a few trusted
companies with whom they maintain secrecy agreements.

“Having an These select companies can then employ the correct methods and

approved third
limits and develop simplified equations and methods for the PLC
programmers on a per job basis. This provides the desired safety for

party SI handle
that specific unit without fully disclosing actual OEM calculation
methods. This approach works well with many PLC programmers as it
the controls of reduces the number of calculations needed to model the unit. The
main drawback is that this approach can reduce the size of the
the allowed safe operating map. That is, in favor of simpler modeling
equations and limiting constraint equations, the potential operating
compressor... map where the compressor can safely run may be reduced by 5%,
10%, or even 50%. Why? To keep the modeling equations as simple
makes sense.” as possible, often the worst cases are considered when creating the
array of constraint equations needed to keep the unit out of pin non-
reversal and rod load issues. For example, for a given discharge pressure, if the unit can run safely at
1000 rpm, but is unsafe at 890 rpm, but again safe at 750 rpm, then the 890 rpm cutoff may be used
when creating the constraint equations that limit the minimum allowed suction pressure at that
particular discharge pressure. Overly conservative constraint equations can effectively reduce the
allowed operating map, as well as the operational flexibility needed by the end-user to fully achieve
their goals.

7
www.compressortech2.com
That’s the way it has been up until about 2012. Now, a growing number of OEMs (currently
Cameron/Ajax, GE Oil & Gas, and Knox Western, with LeROI Gas Compressors and Arrow Engine
being added in 2014) have incorporated their full set of compressor performance and safety limits into
an actual micro-controller add-on for UCPs. The UCP can now achieve the exact same performance
and safety checks that the OEM’s Windows-based performance modeling software provides.

Most other major OEMs have provided many of their modeling equations and safety check methods so
that the micro-controller add-on provides similar results when calculating performance and safety limits
for those OEMs too. While some restrictions still apply, this compressor performance and safety
advisory UCP add-on is expected to fully support hydrocarbon liquid dropout calculations and other
advanced features in future versions. The full power and modeling abilities of OEM software is making
its way down to the control panel level, providing cost effective, powerful options to UCP system
integrators, compressor packagers, and end-users.

Having an approved third party SI handle the controls of the compressor (when to start it, what speed to
run it at, when to load and unload it, when to open and close bypass valves, when to open and close
recycle valves, how best to cool interstage gas, when to throttle gas going into a specific stage, etc.)
makes sense. Customizing the actual UCP to best meet the needs of the station can lead to optimal
use of the equipment, and likely to maximum revenues. Unit Control Panel logic should not be confused
with compressor performance and safety. The latter is a basic necessity for the former. UCP logic deals
with “what do I want to achieve”, and “what sequence of events and time delays are required to try to
achieve those goals”. Compressor performance and safety predictions by a real-time micro-controller
add-on provides answers to the UCP’s questions: If the compressor were to run under the proposed
operating conditions, what hardware configurations would be safe? And for the ones that are safe, how
much load is required? How much flow is predicted? How efficient is the unit at those conditions? And,
does the compressor need to be shut down for safety reasons?

8
www.compressortech2.com
SECTION 8: Concerns for PLC Programmers
Programmers that program control panels for controlling reciprocating compressors really do like their
jobs (Forbes Magazine, #1 job in America for 2013 is Software Developer). However, they are often at
a major disadvantage from their counterparts in other disciplines. Namely, someone puts together a $2
million packaged unit and hands control of that unit over to the PLC programmers, and casually
mentions, “Make sure this thing runs wherever and whenever I need it, but be careful, as failure of the
unit can cause a lot of damage, including injury and death to personnel. And by the way, I’m not really
sure how the compressor’s safety and performance is determined, but I trust you can figure that out.”

“A simple action of
System integrators are challenged to design, code and test
the control logic for properly starting compressors, safely

a control panel
operating them across defined operating maps, and properly
shutting them down. Part of most control logic involves

could actually lead

adjustments to suction pressure regulators, actuation of
unloading hardware, control of recycle valves, and
adjustments to operating speed. With newer separable units,
the last item is usually not done directly. to catastrophic
Most engine and motor OEMs provide panels with their damage to the
products. The end user’s control panel can only control the
engine/motor indirectly via requests sent to the driver OEM’s unit.”
panel. Any requested change that creates a safety issue
typically results in the engine’s/motor’s OEM panel not
permitting the action to be taken, thus protecting the driver. The same is not generally true for the
compressor.

For the compressor, an action as simple as increasing the speed of the compressor from 800 rpm to
875 rpm could wreck the compressor, even though ironically running it at 1000 rpm might be perfectly
safe. A simple action of a control panel (perhaps lowering speed to reduce flow rates) could actually
lead to catastrophic damage to the unit.

Automation specialist and end-users often assume that because there is a plethora of pressure,
temperature and vibration sensors installed throughout the system, that any issues will be caught by
those sensors – another bad and potentially dangerous assumption. Very few reciprocating
compressors actually have real-time internal pressure sensors. Standard pressure sensors only
measure the suction and discharge pressures nearby the cylinder flanges. These are useful for
checking against low and high pressure limits, but not useful for OEM models that base rod loads on
dynamic internal pressures.

Discharge temperature sensors can catch high discharge temperatures, but only if the discharge valves
open and release the hot gas. Deactivated ends and ends operating in very low suction volumetric
efficiencies do not lead to hot gases exiting out the discharge valves. As such, internal temperatures
can climb to 350°F, 400°F, 500°F, (175°/204°/260°C) and hotter. Eventually valves, packing, and rings
will fail leading to consequential and potentially catastrophic damage. A few sensors cannot make up
for not implementing the correct performance and safety models in the PLC. Any expectation otherwise
can be a dangerous choice.

9
www.compressortech2.com
SECTION 9: The Dangers of Applying the Wrong Modeling

Figure 1A. This chart shows safety based on rod loads determined by dynamic internal gas pressures and inertia, and crank pin forces (as
specified by OEM).

Figure 1B. Specifications are given for three-stage compressor tested.

10
www.compressortech2.com
Figure 1A shows the safe operating map of an OEM’s 3-stage compressor (Figure 1B) across a suction
pressure range of 0 to 100 psig (0 to 690 kpag), a discharge pressure range of 100 to 1000 psig (690 to
6895 kpag), a speed range of 700 to 1000 rpm, six load steps, and suction temperatures for the three
stages varying from 100° to 140°F (38° to 60°C). The safe operating map is generated after reviewing
over 8.5 million distinct operating points using the OEM’s exact methods for calculating safety issues.
Red areas are where the unit cannot safely operate regardless of what load step is used and
regardless of what speed the unit is running. Yellow areas reflect where some combinations of load
step and speed can be used to safely operate the unit while other combinations are unsafe. Green
areas reflect when all load steps are safe at all speeds. The key issues noted on the graphic are “C:
Crank Pin Forces Exceeded”, “R: Rod Load Forces Exceeded”, “U: Gas is Throttling through one of the
stages”, and “O: Unit is Overloaded”. Excessive crank pin forces can damage the crankshaft, and
excessive rod loads can damage the piston, rod, crosshead, and even crankshaft. Either one can lead
to catastrophic damage of the unit.

Figure 2A. This table shows safety based on rod loads based on static gas pressures (using old, inappropriate rod load methods). Green area
represents all load steps predicted to be safe at all speeds, and all suction gas temperatures.

Figure 2A shows the exact same compressor, but this time it is being modeled by standard compressor
performance and safety calculations one would find by searching the Internet, or in older compression
theory textbooks. Notice that Figure 2A shows no areas of excessive crank pin forces. Areas marked as
rod load concern (“R”) are quite different than those indicated in Figure 1A.

11
www.compressortech2.com
 Figure 2B. This table shows safety based on rod loads based on static gas pressures (using old, inappropriate rod load methods). When
overlaid on actual OEM’s safety map, the majority of the “predicted safe” area is not always safe. In this case, the unit can fail due to crank pin
forces being exceeded at various speeds for various load steps.

Figure 2B overlays Figure 2A over Figure 1A with some translucency. The blue boundary line helps to
highlight where the non-OEM, generic method would indicate to the end-user and the control panel that
the unit is safe even though the unit may very well be under excessive stresses.

12
www.compressortech2.com
 Figure 3A. This table shows safety based on rod loads based on adjusted static gas pressures (using another OEM’s method, which is
inappropriate since the calculation method is not appropriate for the limits used). The green area represents all load steps predicted to be safe
at all speeds, and all suction gas temperatures.

Figure 3A shows the exact same compressor, but this time it is being modeled by another OEM’s
equations and safety limits. Again, notice that Figure 3A shows no areas of excessive crank pin forces.
Areas marked as rod load concern (“R”) are quite different than those indicated in Figure 1A. Figure 3B
overlays Figure 3A over Figure 1A with some translucency. The blue boundary line helps to highlight
where the alternate OEM method would indicate to the end user and the control panel that the unit is
safe even though the unit may very well be under excessive stresses. For some combinations of load
step and speed, the unit is safe, while for other combinations of load step and speed, the unit is unsafe.

Since the earliest control panels, end-users held high hopes that control panels could keep the unit
safe, but knew due to limitations in computing power that some safety items were not being checked.
However, since at least 2004, new control panels are expected to correctly know when it is safe and
unsafe to operate the unit being controlled. Anything less is totally unacceptable.

For more clarity, let’s look at just the calculations for a single operating point as modeled for a large
injection and withdrawal unit (Figure 4A) running in its single stage mode, compressing natural gas
from 640 to 1600 psig (4413 to 11032 kpag) at a suction gas temperature of 75°F (24°C).

When modeled using the OEM’s performance software, the calculated rod loads are 102.9%
compression and 88.8% tension on throws #1/#3/#5 (see Figure 4B). Any control panel should be
predicting these rod load values at the same operating point by at least within ±1 percentage point.

13
www.compressortech2.com
So, what values of “predicted” rod loads are many field control panels likely to calculate for this unit?
And how close are they to the OEM’s predicted forces?

Classic Rod Load Compression Formula:

= [Area Head End (in.2)] x [Discharge Pressure (psig)] – [Area Crank End (in.2)] x [Suction Pressure
(psig)]

Classic Rod Load Tension Formula:

= [Area Crank End (in.2)] x [Discharge Pressure (psig)] – [Area Head End (in.2)] x [Suction Pressure
(psig)]

SECTION 10: Case #1 – Applying Simple Static Gas Rod Load Calculations
Case #1: Rod loads based on flange pressures. This approach is standard for legacy, slow-speed units.
Unfortunately, many PLC programs have implemented this approach for high-speed units.

• Compression forces based on flange pressures (imperial units shown here): o Compression = (1600
psig) * 𝜋/4*(10.5 in.)2 – (640 psig) * 𝜋/4*[ (10.5 in.)2 – (3.125 in.)2 ]

• Compression = 138544.24 lbf – 50508.96 lbf (psig = lbf/in.2, gage)

• Compression = 88035.28 lbf out of 100,000 lbf rated, or 88.0%. § That’s 15 percentage points from
the OEM’s 102.9% value.

This method indicated the unit is safe when according to the OEM, it is not.

• Tension forces based on flange pressures: o Tension = (1600 psig) * 𝜋/4*(10.52 – 3.1252) – (640
psig) * 𝜋/4*(10.52)

• Tension = 126272.39 lbf – 55417.69 lbf

• Tension = 70854.70 lbf out of 95,000 lbf rated, or 74.6% § That’s 14 percentage points from OEM’s
88.8% value.

This method indicated the unit is safe when, according to the OEM, it is safe, but off by 14 percentage
points.

14
www.compressortech2.com
SECTION 11: Case #2 – Applying Adjusted Static Gas Rod Load Calculations
Case #2: Rod loads based on 10% pressure drop adjustment to flange pressures. This approach was
sometimes implemented on the basis that by increasing the pressure ratio, higher rod loads would be
predicted and this method would be more conservative (and safer).

• Compression forces based on 10% adjusted flange pressures: o Compression = (1600 + 160 psig) *
𝜋/4*(10.52) – (640 – 64 psig) * 𝜋/4*(10.52 – 3.1252)

o Compression = 152398.66 lbf – 45458.06 lbf

o Compression = 106940.60 lbf out of 100,000 lbf rated, or 106.9% § It is off by four percentage points
from OEM’s 102.9% value.

§ This method would have predicted unsafe conditions a lot sooner than the OEM’s method did, making
the operating map useless. For high ratio applications with low valve losses, this method could indicate
a significant portion of the user’s operating map as unsafe. Moreover, for high valve loss, low ratio
applications, this method would likely indicate unsafe areas as being safe. The belief that it is always
being conservative in safety calculation is incorrect.

• Tension forces based on 10% adjusted flange pressures: o Tension = (1600+160 psig) * 𝜋/4*(10.52 –
3.1252) - (640 – 64 psig) * 𝜋/4*(10.52)

o Tension = 138899.63 lbf – 49875.92 lbf

o Tension = 89023.71 lbf out of 95,000 lbf rated, or 93.7% § It is off by five percentage points from
OEM’s 88.8% value.

SECTION 12: Case #3 – Applying Correct OEM’s Rod Load Method – Lots of Equations!
Case #3: Appropriate OEM’s rod load calculation method for end-users to use (imperial units used):

• OEM’s formula for double-acting rod load compression:

= [area head end] x [internal discharge pressure (psia)] – [area crank end] x [internal suction pressure
(psia)]

• OEM’s formula for double-acting rod load tension:

= [area crank end] x [internal discharge pressure (psia)] – [area head end] x [internal suction pressure
(psia)]

• Internal suction pressure = [suction flange pressure (psia)] * ( 100 – PLSUCT ) / 100

• Internal discharge pressure = [discharge flange pressure (psia)] * ( 100 + PLDISCH ) / 100

15
www.compressortech2.com
• Suction flange pressure (psia) = [suction line pressure (psig) + AtmPress ] * (1 –
SuctionPressureDrop%)

• Discharge flange pressure (psia) = [discharge line pressure (psig) + AtmPress ] * (1 +

DischargePressureDrop%)

• PLSUCT = (GVS2/6550000)*(SpecificGravity / ZSUCT) * (520 / TSUCT_INT)

• PLDISCH = (GVS2/4550000)*(SpecificGravity / ZDISCH) * (520 / TDISCH_INT)

• GVS = (PistonArea * AveragePistonSpeed) / AVC

• AveragePistonSpeed = 2 * Stroke * Speed / 12

• PistonAreaHEAD_END = π/4 * CylinderBoreDiam2

• PistonAreaCRANK_END = π/4 * (CylinderBoreDiam2– RodDiam2)

• AVC = ( ValveADJ_EQUIV_AREA-2 + CylinderADJ_EQUIV_AREA -2 )-0.5

• Data from unit model (data is based on each application): o SuctionPressureDrop% = 1%,

o DischargePressureDrop% = 1%,

o Specific gravity = 0.65,

o GasK = 1.3042, (NOTE #1: This value can change with operating conditions.) § A verified curvefit
equation over the specified operating ranges, or an acceptable equation of state can be used to predict
this value at various operating conditions. For many applications, a constant GasK value can be used
without producing any significant effect on unit performance and safety predictions. However, this must
be reviewed across the operating ranges and not just assumed.

o Stroke (in) = 8.50,

o Speed (rpm) = 750

o Gas suction temperature TS (degR) = 534.67 (= 75 degF + 459.67)

• Fixed data from unit model (data is based on frame and cylinders being modeled): o Cylinder bore
diameter (in.) = 10.5

o Rod diameter (in.) = 3.125

o Valve adjusted equivalent area for suction head end (in.2) = 13.268,

o Valve adjusted equivalent area for suction crank end (in.2) = 13.268,

o Valve adjusted equivalent area for discharge head end (in.2) = 13.268,

16
www.compressortech2.com
o Valve adjusted equivalent area for discharge crank end (in.2) = 13.268,

o Cylinder adjusted equivalent area for suction head end (in.2) = 18.4,

o Cylinder adjusted equivalent area for suction crank end (in.2) = 18.4,

o Cylinder adjusted equivalent area for discharge head end (in.2) = 28.1, and

o Cylinder adjusted equivalent area for discharge crank end (in.2) = 28.1.

• Compression ratio = (flange discharge pressure) / (flange suction pressure)

• Internal compression ratio = (internal discharge pressure) / (internal suction pressure)

• Gas discharge temperature TD (degR) = (gas suction temperature) * ( compression ratio )1-1/GASK

• NOTE #2: The following two equations are similar to what this OEM uses, but not exactly the same as
used in their Windows-based performance software. However, the OEM certifies these simplified
versions for PLC modeling: o TSUCT_INTERNAL = (gas suction temperature) + [0.02 + 0.002 *
(cylinder bore diameter) ] * (TD – TS)

o TDISCH_INTERNAL = TSUCT_INTERNAL * ( compression ratio )1-1/GASK

The sequence of calculations taken to determine acceptable OEM rod loads on this 10.5 in. bore throw
is indicated below. They are shown here for readability and are not optimized for PLC encoding. Some
items need only be calculated once as they are fixed thereafter, while others need to be calculated
dynamically as pressures, speed and temperatures change. The calculations are a bit lengthy, but once
coded into the PLC, they execute swiftly. More importantly, as the final results will show, the
determination of rod loads is very close to what the OEM expects.

• Calculate AVC for the four states:

o Head end suction = (13.268-2 + 18.4-2 )-0.5 = 10.761888

o Head end discharge = (13.268-2 + 28.1-2 )-0.5 = 11.997809

o Crank end suction = (13.268-2 + 18.4-2 )-0.5 = 10.761888

o Crank end discharge = (13.268-2 + 28.1-2 )-0.5 = 11.997809

• Calculate head end and crank end piston areas:

o Head end piston area = π/4 * 10.52 = 86.590148

o Crank end piston area = π/4 * (10.52 – 3.1252) = 78.920244

• Average Piston Speed = 2 * 8.50 * 750 / 12 = 1062.5

17
www.compressortech2.com
• Calculate GVS for the four states:

o Head end suction = (86.590148 * 1062.5) / 10.761888 = 8548.874719

o Head end discharge = (86.590148 * 1062.5) / 11.997809 = 7668.236113

o Crank end suction = (78.920244 * 1062.5) / 10.761888 = 7791.640208

o Crank end discharge = (78.920244 * 1062.5) / 11.997809 = 6989.006014

• Calculate suction flange pressure (psia) = [640 + 14.7 ] * (1 – 0.01) = 648.153

• Calculate discharge flange pressure (psia) = [1600 + 14.7 ] * (1 + 0.01) = 1630.847

• Calculate compression ratio = 1630.847 / 648.153 = 2.516145

• Calculate TD (degR) = (534.67) * (2.516145 )1-1/1.3042 = 663.064813

• Calculate TS_INTERNAL = 534.67 + [0.02 + 0.002 * (10.50) ] * (663.064813 – 534.67) = 539.934187

• Calculate TD_INTERNAL = (539.934187) * (2.516145 )1-1/1.3042 = 669.593134

• NOTE #3: Use a thermodynamic equation of state (e.g. Redlich-Kwong, Peng–Robinson, BWR, etc.)
or use a full thermodynamic application (e.g. AGA-8, NIST, VMG, etc.) to determine compressibility
factors: o ZS = EoS(suction pressure flange, suction temperature, gas) = AGA8(648.153, 534.67, Nat.
Gas) = 0.9096

o ZD = EoS(discharge pressure flange, discharge temperature, gas) = AGA8(1630.847,

663.06, Nat. Gas) = 0.9270

o NOTE #4: For exotic gases, or gases near dew points, or gases in super critical fluid areas,
make sure your EoS option matches the predictions from the OEM reasonably well.

• Calculate PL% for the four states:

o Head end suction = (8548.8747192/6550000) * (0.65 / 0.9096) * (520 / 539.934187) =

7.678953

o Head end discharge = (7668.2361132/4550000) * (0.65 / 0.9270) * (520 / 669.593134) =

7.037291

o Crank end suction = (7791.6402082/6550000) * (0.65 / 0.9096) * (520 / 539.934187) =

6.378843

o Crank end discharge = (6989.0060142/4550000) * (0.65 / 0.9270) * (520 / 669.593134) =

5.845820

18
www.compressortech2.com
• Calculate internal cylinder pressures for the four states:

o Head end suction = 648.153 * ( 100 – 7.678953 ) / 100 = 598.381636

o Head end discharge = 1630.847 * ( 100 + 7.037291 ) / 100 = 1745.614449

o Crank end suction = 648.153 * ( 100 – 6.378843 ) / 100 = 606.808338

o Crank end discharge = 1630.847 * ( 100 + 5.845820 ) / 100 = 1726.18338

• Compression forces: (Internal pressures converted from absolute to gage pressures for these
calculations)

o Compression = (1726.18338 – 14.7) * 86.590148 – (598.381636 – 14.7) * 78.920244 =

102133.302

o Relative to 100,000 lbf rated rod load compression limit, this becomes 102.1%.

o OEM’s Software reports 102.9% using their internal calculations.

o Very close, only off by -0.8 percentage point. Acceptable.

• Tension forces: (Internal pressures converted from absolute to gage pressures for these calculations)

o Tension = (1726.18338 – 14.7) * 78.920244 – (598.381636 – 14.7) * 86.590148 =

84529.60671

o Relative to 95,000 lbf rated rod load tension limit, this becomes 90.0%.

o OEM’s software reports 88.8% using their internal calculations.

o Very close, only off by +0.2 percentage point. Acceptable.

SECTION 13: Case #4 – Applying a Different OEM’s Rod Load Method

Case #4: A different OEM’s rod load calculation method:

• Many OEMs use dynamic gas forces and inertia forces when calculating rod loads (and/or pin loading) forces. If
these calculations were applied to the previous unit model, then the calculated compression force = 98% (five
percentage points off the actual OEM’s method) while the calculated tension force = 82% (seven percentage
points off the actual OEM’s method), based on weights for the actual piston+rod assemblies for the cylinders, and
on the other OEM’s rod load method based on using dynamic, internal cylinder pressures.

o This particular method indicates that the unit is safe when, according to the actual OEM’s method, it is
not. This could cause a catastrophic error.

o Most analyzers tend to calculate rod load forces based on dynamic internal gas pressures plus
calculated inertia forces. These results are then shown on the analyzer reports. This situation creates

19
www.compressortech2.com
 confusion if the OEM did not calculate their forces in that exact manner, or used different weights than
what the analyzer is set to use. § The good news is that recent changes in analyzer software now allow
analysts to specify which OEM compressor is being modeled, and based upon that setting, the analyzer
will calculate the unit’s rod loads according to the appropriate OEM methods.

Some items that end-user, packagers, and automation specialists should consider when deciding upon what
methods to implement in unit control panels, in regards to performance and safety prediction are:

• If a non-OEM method is implemented, does that implementation potentially void the unit’s warranty?

o If there is a failure, failure analysis testing will likely show that the metal (rods, bolts, pistons,
crankshafts, etc.) experienced prolonged stresses before breakage – a telltale sign that the unit was run
in unsafe areas of high stress for extended periods.

• If a non-OEM method is implemented, does this implementation potentially create legal consequences?

o Consult your legal specialists here, but if a case goes to court, the defense is most likely going to
attack using the classic legal argument that the defendant implemented “services involved that deviated
from the best practices of the control system industry”.

• Have sufficient operating points across the specified operating map been checked for safety?

o If so, then the automation specialists have a better idea of what problems to expect, and where they
are likely to occur.

o If not, then you are likely allowing a control panel to blindly adjust speed and/or suction pressures
without knowing ahead of time if those adjustments are in fact safe.

20
www.compressortech2.com
SECTION 14: Closing Comments
The worst thing you can do is knowingly operate a compressor in unsafe areas. The next worst
situation is to unknowingly operate it in unsafe areas. Next comes operating a compressor by using
overly conservative methods and thus restricting its operating map (lost revenues, potentially higher
emissions, more shut downs, more wasted energies from pressure throttling and recycling, etc.).

As an end user, you expect to run the compressor to its maximum potential so you can safely cover the
widest operating map, run the unit most efficiently, and accurately predict load, flow, and unit safety. If
the unit is going to be controlled by a control panel, then you need to require that the automation
specialists integrate into their control logic the proper OEM-approved methods for determining unit
safety. Anything short of full implementation of the OEM-approved performance and safety methods is
not really an acceptable best practice for today’s control panel.

Finally, many optimization and/or monitoring service companies provide valuable information and
insight into the fleet of compressors being modeled. These service-based companies take operating
data from the field, run the data through a series of calculations, trend the data, compare to historic
values, and generate reports. Just like automation companies, however, inappropriately applied
equations are used, potentially rendering the information from those reports anywhere from
meaningless often to misleading and possibly dangerous.

The concluding rule to use for end users is simple: When automating a unit or adding it to a monitoring
service, make sure that your selected vendor applies the correct OEM safety and performance methods
when modeling the unit. If the bidding vendor cannot do that, then select a vendor who can.

21
www.compressortech2.com