Approaching Compressor Control Retrofits

Rick McLin

Rockwell Automation

Houston, Texas

Modification of compressor control installations may cause fear and uncertainty in engineers

assigned to the task. After all multivariable, non-linear control systems are not for the faint of

heart. However, compressors are major pieces of capital equipment with long effective

lifespans. This length of service may necessitate one or more control system retrofits due to

control system obsolescence over the life of the compressor. Plant considerations may require a

compressor to be moved to another application or re-rated due to changing process requirements,

either of which may drive the need for a control system retrofit. Whether the retrofit is required

due to changes in the process requirements, or due to control system obsolescence, a compressor

retrofit is an opportunity to improve compressor operation and efficiency.

If a compressor control retrofit becomes necessary the compressor and the process should be

analyzed for opportunities to improve efficiency and provide smoother operations. The major

areas where compressor control improvements can be made are:

The compressor control algorithms

Control system interactions (capacity control)

Field instrumentation

Valve sizing and selection

Improving the Compressor Control Algorithms

The first area to investigate is the compressor control algorithm itself. If the compressor

installation is over 10 years old it is likely that the control system is based on outdated anti-surge

techniques that may not be as efficient as those that are in common use today. The incredible

improvements in control hardware capabilities allow rigorous models to be used to optimize

compressor performance together with tighter integration of the compressor controls into the

overall process itself. It’s no longer necessary to sacrifice process stability to protect the

compressor.

Compressor control algorithms are based on a performance map supplied by the compressor

vendor. These maps always have flow represented along the X-Axis while discharge pressure,

ressure ratio or head are located along the Y-Axis. Engineering units used on the X and Y-Axis

can be (and often are) nearly anything. About the only flow measurement I have not seen is

cubic furlongs per fortnight. Anything else seems to be fair game.

The figure below shows a typical compressor performance map.

Figure 1: Typical Compressor Performance Map

This example is a variable speed machine with its performance shown as polytropic head on the

Y-Axis versus flow on the X-Axis in thousands of cubic feet per minute. Compressor flow and

pressure follow a speed line until a surge point is reached. As the speed of the compressor

changes the flows and pressures change.

The bottom set of curves show the compressor efficiency at the various speeds and flows.

Compressor impellers are normally designed to achieve maximum aerodynamic efficiency near

the center of each speed line. If the compressor is not normally operated near its area of

maximum efficiency then a compressor re-rating should be considered.

The phenomenon of surge

Surge occurs when the kinetic energy imparted into the gas by the impeller is less than the

potential energy in the discharge. When this occurs the flow of gas will reverse direction.

When a compressor approaches the surge point along a speed line flow through the compressor

will reverse direction. This flow reversal happens at the speed of sound, far too fast for

instrumentation to detect, and once started cannot be stopped. A surge cycle will repeat unless

the surge control system intervenes. Repeated surge cycles can seriously damage or even

destroy a compressor. For this reason predicting the onset of surge is essential in modern surge

control algorithms.

What happens during a surge event

Figure 2 below is a representation of a surge event. Only a single speed line is shown for clarity.

Assume initially the compressor is operating at point (1) in the diagram below.

Figure 2: What Happens During Compressor Surge

The compressor is operating at its maximum flow capabilities at point (1). As the discharge

pressure increases the work the compressor must accomplish increases pushing the compressor

operating point along the speed line to point (2). If the discharge pressure continues to increase

the compressor operating point will move to (3) in figure 2. If the control system cannot reduce

the discharge pressure the compressor the operating point will cross the Surge Line and flow will

reverse through the compressor. When the flow is reversed through the compressor the

compressor operating point will rapidly move to point (4) on figure 2. The surge event reduces

the discharge pressure and increases the suction pressure of the compressor. The compressor will

then re-establish forward flow and the operating point will move from point (4) back to point (2).

This cycle will repeat unless the control system can intervene and break the cycle. Total time for

a surge cycle is 1 to 3 seconds, but the flow reverses through the compressor in under a

millisecond. This cycle will repeat until the compressor controls change the operating conditions

to stabilize compressor operation.

Common compressor control algorithms

Minimum Flow Recycle

There are several control approaches to prevent surge in compressors. The oldest and least

efficient is minimum flow recycling. This approach simply picks a flow at which the compressor

can be guaranteed not to surge. If the flow drops below this point the recycle or blow off valve

opens and maintains a redefined minimum flow through the compressor.

Figure 3: Minimum Flow Recycle

This approach can be effective but it is not efficient. When the compressor operates at lower

speeds a large flow is still required to protect the compressor. This approach also does not take

into account changes in gas properties which may alter the compressor surge line

While this approach is not the most efficient it can be useful as a fallback algorithm in an

advanced surge control application. Fallback algorithms are used when field instrumentation

faults prevent an accurate calculation of the compressor operating point. With degraded field

instrumentation minimum flow fallback may be the only practicable control algorithm.

Maximum Discharge Pressure

This approach to surge control relies on the relationship between the maximum achievable

discharge pressures a compressor is able to produce at various temperatures. Discharge pressure

control is commonly used on constant speed packaged air compressors (typically integrally

geared machines) where the suction pressure does not vary. This approach has the advantage of

being extremely inexpensive (read that as cheap) to implement since minimal instrumentation is

required. A discharge pressure transmitter and an ambient temperature measurement are all that

is required. Flow through the compressor is not even measured.

Several compressor maps are supplied by the manufacturer that relates the maximum pressure

the compressor can produce at summer and winter conditions. During the winter when the air is

colder and the air density is higher the compressor can produce a higher discharge pressure

before a surge occurs. In the summer when the air density is lower the compressor cannot

produce as high a discharge pressure. Control is very simple with a discharge pressure PID

having a variable pressure setpoint that is adjusted for ambient temperature.

Unfortunately simplicity often wastes energy and does not provide adequate protection for the

compressor. This approach is often used on packaged air compressors where the surge data is

generic, not specific to a particular machine. Variations in manufacturing require a conservative

approach which results in further inefficiencies.

In addition as the compressor impellers experience wear or intercoolers become fouled, the

maximum discharge pressure the compressor can achieve, decreases. This requires a lower

pressure setpoint to protect the compressor. If ignored compressor damage will occur. Older

packaged compressors often used pneumatic controls. Older pneumatic temperature

measurement was difficult to calibrate and consequently was often ignored by maintenance

personnel. This can result in a pressure setpoint that does not change with temperature which

will result in damage to the compressor. More modern controls, even if ignored, allow for

automatic adjustment of the surge line so once a surge occurs the margin can be adjusted to limit

the number of surge cycles a compressor would experience. Multiple surge cycles can also be

set to trip the compressor to prevent damage.

Delta P vs. h

The Delta P vs. h algorithm, also known as Pressure Rise, was originally developed in the 1970s

when it was observed that the pressure ratio across the compressor closely followed the

measured differential pressure across a flow measurement device. Still widely used, Delta P vs.

h has the advantage of simplicity and low cost. A flow and a differential pressure measurement

across the compressor are all that is required for compressor control. Delta P vs. h is a major

improvement on minimum flow recycle but it still has significant problems. It cannot account

for changes in gas properties and it requires a suction pressure that does not significantly change

during operation.

As control systems have become more advanced it has become possible to implement more

elaborate thermodynamic models for compressor control. Delta P vs. h has largely been replaced

by some form of a compressor head model.

Compressor Head vs. Flow

Compressor head vs. flow is an algorithm that is based on calculating the head produced by the

compressor and plotting it versus the temperature and pressure compensated flow being

produced. Whether the algorithm is based on polytropic or adiabatic head, this approach has the

advantage of being able to accurately predict the compressor operating point at various

temperatures and pressures and has an added benefit of not being affected by changes in the

molecular weight of the gas. The basic equations for this algorithm are shown below.

The basic equation for polytropic head is defined as:

MWn

n

ZRTs

Ps

PdHp

n

n

*1

*1

1

Figure 4: Polytropic Head Equation

Where:

Hp = polytropic head

Pd = Discharge Pressure

Ps = Suction Pressure

Td = Discharge Temperature

Ts = Suction Temperature

n = number of moles of gas in a given pressure/volume

Z = gas compressibility

R = universal gas law constant

MW = molecular weight of the gas

The difficulty with using this equation for surge control is that not all the variables can be

measured directly. Gas compressibility and molecular weight cannot be determined except by

offline analysis.

To eliminate these variables from equation it is necessary to utilize the flow relationships of

differential pressure flow measurement devices. Orifice, venturi, annubar and other head type

measurement devices have flow equations that include terms for molecular weight and

compressibility. The classis orifice equation:

PsMW

ZRTshKQ

*

**

Figure 5: Head Type Flow Measurement Equation

Where:

Q Flow, in appropriate units

h differential pressure across flow measurement device (head)

K Orifice coefficient, dependent on flow units and geometry

The MW

ZRTsterm is present in both the head and flow equations which are used to generate the

plotting coordinates used on the polytropic head versus flow compressor control map. The

effects of changes in molecular weight, suction temperature and compressibility affect the X and

Y coordinates by the same amount, allowing the generation of compressor maps that are valid for

variable composition gas streams. This approach generates a compressor control map that is

often referred to as a Universal Surge Curve. This approach is valid for all temperatures and

pressures as defined by the manufacture’s compressor map and accounts for changes in

molecular weight. Changes in gas properties such as the heat capacity ratio can also be

accommodated by incorporating the thermodynamic relationships derived from Charles and

Boyle’s law.

Control System Interactions (Capacity Control)

An operating compressor is an integral part of the process in which it is installed. Control of the

volume of gas delivered by the compressor is necessary to match process requirements. Capacity

control, while often done by other controllers or a plant wide distributed control system (DCS) is

best handled by the compressor controller.

Modern compressor controllers have the capability to incorporate capacity control. This allows

compressor surge conditions to be factored into the capacity control. The plantwide DCS then

sends a setpoint for the capacity controller which allows compressor protection while meeting

plant requirements. This is accomplished by decoupling. Decoupling (described in more detail

later) prevents a process or capacity controller from pushing a compressor into surge by forcing

it out of its operating envelope. .

Methods of Compressor Capacity Control

Variable speed

The flow rate and discharge pressure a compressor can produce is related to the speed the

compressor is being driven at. The figure below illustrates the movement of the operating point

when the speed is reduced. Reducing the compressor driver speed from point 1 to point 2

reduces the flow produced by the compressor. In this example the pressure ratio across the

compressor does not change, allowing the compressor to supply a lower flow rate at a reduced

speed. Speed control also allows the compressor to stay in its most efficient operating range.

Flow

Pre

ss

ure

Ra

tio

Surge Line

Speed Lines

Compressor Performance Map

Q1Q2

Speed 1Speed 2Pd/Ps

Min Speed

Max Speed

Figure 6: Variable Speed Control

If the speed were lowered rapidly it is possible the compressor would pass the surge line and the

compressor would surge. Having the speed controlled by the compressor controller allows

decoupling to be implemented to protect the compressor from surge.

Inlet guide vanes

Inlet guide vanes are stationary blades with variable pitch that provide a mechanism to alter the

swirl pattern on the inlet flow to a compressor. They are commonly used on fixed speed

compressors to increase the operating range of the compressor. Inlet guide vanes are connected

together with a mechanical linkage so all the guide vanes move together.

Inlet guide vanes can significantly increase the efficiency of the compressor and improve the

turndown ratio of the machine. A compressor performance map for a constant speed compressor

with inlet guide vanes is shown in Figure 7 below. While this map looks similar to a variable

speed map there are significant differences in the control methodology used. The slope of each

guide vane angle line is usually much steeper than a speed line.

Flow

Pre

ss

ure

Ra

tio

Surge Line

Guide Vane

Angles

Compressor Performance Map

-30 deg

-5 deg

-15 deg

0 deg

5 deg

Pd/Ps

Q1Q2

Figure 7: Inlet Guide Vane Compressor Map

Guide vanes dramatically alter a compressor performance and special control techniques are used

which take guide vane position into account. Maximum compressor discharge pressure

capabilities vary greatly with guide vane position. For this reason guide vane position feedback

is strongly recommended. If the guide vane position is not accurately reported to the surge

controller due to mechanical problems or incorrect calibration severe damage to the compressor

can result.

Suction or discharge throttling

The least efficient method of compressor capacity control is throttling. Throttling can be used to

lower compressor flow by increasing the pressure ratio the compressor has to achieve which

increases the amount of work the compressor has to do.

Suction throttling is more efficient than discharge throttling at lowering flow across a

compressor since the gas has not been compressed first. Suction throttling reduces the suction

pressure which increases the pressure ratio and hence reduces the flow through the compressor.

Discharge throttling increases discharge pressure which also increases the pressure ratio.

Flow

Pre

ss

ure

Ra

tio

Surge Line

Speed Lines

Compressor Performance Map

Q1Q2

Speed 1 Speed 2Pd/Ps 1

Min Speed

Max Speed

Pd/Ps 2

Figure 8: Changing Pressure Ratio Using Throttling

If the compressor installation has a suction throttle valve which is used to decrease suction

pressure, rapid changes in the valve position can push the compressor into surge by increasing

the pressure ratio above the surge line of the compressor. Similarly if the compressor has a

discharge throttle valve which increases discharge pressure, changes in valve position can

increase the pressure ratio above the surge line of the compressor resulting in surge.

Decoupling

In the instances above; Reducing Speed, Changing Inlet Guide Vane Angle or Increasing

Pressure Ratio by Throttling, are actions taken by capacity controllers responding to some

process demand. Decoupling is the technique utilized to protect the compressor against being

forced into surge by capacity controllers. Decoupling temporarily suspends the control action

that is driving the compressor towards an unstable operating area while it establishes recycle

flow to stabilize the compressor. Once the compressor move toward the unstable region is halted

the capacity control action is once again enabled to satisfy the process demand.

Decoupling can only be effectively implemented in the surge controller since the surge controller

determines the compressor operating point and can determine when and how quickly a

compressor is approaching the surge line. Capacity demand from a DCS can be accomplished by

a remote setpoint to the compressor controller which can implement decoupling to protect the

compressor from surge.

Field Instrumentation

Transmitter Speed of Response

The types of transmitters selected can have an enormous impact on the success or failure of a

control retrofit. For compressor applications the speed of response is important for the pressure

and flow measurements. Some smart transmitters on the market today have a very slow response

time and are unsuitable for compressor applications. Response time for transmitters is defined as

the time required for the sensor output to reach 63.2 percent of its final steady state value.

Figure 9: Transmitter Speed of Response

Every effort should be made to obtain flow transmitters with a response time of under 100

milliseconds. Some widely available smart transmitters on the market have response times under

50 milliseconds and a few special transmitters can be found that advertise a response time under

10 milliseconds. There are however some smart transmitters with response times of 350

milliseconds or greater which make them unsuitable for compressor control applications.

Transmitter Location

Compressor installations are particularly susceptible to incorrectly located or installed

instruments. Special care should be taken to insure short impulse lines to transmitters and to

avoid low spots that would allow condensation to collect in the lines. Long impulse lines also

reduce the speed of response of the signal and can seriously impair control functionality.

Temperature and pressures transmitters should be mounted near the compressor suction and

discharge to accurately reflect actual operating conditions.

Flow transmitters can be located in the compressor suction or discharge; however if mounted in

the discharge downstream of any after coolers, temperature and pressure transmitters must be

added and located near the flow measurement. This will insure that proper temperature and

pressure compensation can be implemented.

In general, flow measurement in the compressor suction is more desirable than a discharge

location since suction flow measurement is not dependent on pressure-temperature

compensation. Flow measurement in the compressor discharge also requires more elaborate

fallback strategies upon failure of these other transmitters.

Flow Measurement

Flow measurement for anti-surge control represents a particular challenge for compressor

applications and the correct flow measurement technology must be selected.

Most suppliers of compressor anti-surge control products today utilize a compressor head versus

flow model for compressor anti-surge. This non-linear multivariable approach provides much

more accurate and efficient control than older minimum flow bypass or delta P vs. h techniques

but it requires more process measurements be made to perform the necessary calculations. This

thermodynamic modeling approach promises much more efficient compressor operation but to

be effective it must be carefully applied.

While the various compressor control system manufacturer’s implementations differ somewhat

they all require a head type flow measurement device. This means that flow measurement

technology based on velocity, such as vortex shedding flow meters, are not appropriate for

compressor control applications.

The most common head type flow measurement devices are:

Orifice plate

o Advantages

Low cost

High measurement pressure differential

Easily flange mounted in pipeline

o Disadvantages

Higher process pressure loss

Venturi and Flow Nozzles

o Advantages

Low process pressure loss

Non-plugging

o Disadvantages

Higher cost

Requires the most space for installation

Annubar

o Advantages

Easy to install

Low process pressure loss

o Disadvantages

Low measurement pressure differential

Each of these flow measurement devices has advantages and disadvantages but all can be used

successfully in compressor applications as long as their particular characteristics are taken into

account. With annubar flow meters care should be taken to insure a large enough differential

pressure can be obtained. Low measurement differential pressures may require special low range

transmitters and the signal to noise ratio of the measurement may become an issue resulting in an

unstable flow signal. In addition low differential pressure range transmitters generally have

slower response speeds due to internal mechanical considerations than transmitters designed to

operate at higher differential pressures.

Valve Sizing and Selection

Valve sizing

Valves should be sized to handle 1-1/2 to 2 times the surge flow. A common error in surge valve

sizing is to select the normal operating pressure ratio and size the valve to handle surge flow

conditions at that point. This allows no margin for error and changing gas compositions may

alter flow across the valve, resulting in inadequate flow even at full open recycle valve position.

Check the valve flow at multiple locations along the surge line. At lower rotational speeds the

pressure ratio across the compressor decreases. The recycle valve should be sized to handle flow

at 1-1/2 to 2 times the surge flow at all the pressure ratios the compressor is expected to see

during operation. The attached figure shows an example.

Figure 10: Valve Sizing

Select a valve with linear characteristics

Equal percentage valves are good at throttling applications but their flow characteristics mean

that when the valve is open 50% only 25% of its maximum flow is achieved. During a surge

event the valve will have to open farther to prevent surge, delaying the increase in flow through

the compressor.

It is quite common on air compressors to use a butterfly valve as a blow off valve to protect the

compressor. Butterfly valves are inexpensive but they have an inherently equal percentage

characteristic. If an equal percentage valve is unavoidable it is possible to modify the valve

positioner (usually with a special cam available from the valve manufacturer) to give the valve

an approximate linear characteristic. If it is not possible to obtain a linear valve or to linearize

the flow using the valve positioner, then a non-linear output to effectively provide a linear

analog output from the controller should be implemented the desired linear flow response.

Figure 11: Valve Characteristics

Quick opening valves establish flow rapidly but their flow characteristics usually result in very

poor throttling abilities. A linear valve trim generally provides the best compromise for surge

control applications.

Valve Full Stroke Speed

A valve stroke speed from full closed to full open should be accomplished in 2 seconds or less.

When approaching surge the most important thing is to provide adequate flow through the

compressor. Speed of response of the entire control loop is essential to prevent surge. All items

should be considered, transmitter response time, controller loop cycle time, valve stroke speed,

and the recycle flow loop volume. Valves with 2 second stroke times are available and should be

used. Boosters can be added to positioners to increase the valve stroke speed.

Managing a Successful Compressor Retrofit

Compressor control system retrofits can be intimidating but they are not impossible to

accomplish if the basic design principles detailed here are followed. The current generation of

PLCs offers amazing performance at a very effective price point. Features that previously were

unavailable together with algorithm execution speeds that are better than made for purpose black

box controllers are available from multiple suppliers.

Compressor control itself is slowly coming out of the realm of a black art to take its place

alongside other complex control applications. Multi-variable non-linear applications will never

be easy, but systems can be designed that improve operations and energy efficiency. These

control applications can be made much easier if fundamental mistakes are not made during the

design phase.