You are on page 1of 30

Compressor Performance Analysis

Presented By

Hans Mathews Hoerbiger Corp of America


At the

2000 GMC

Table of Contents Introduction ......................................................................................................................... ii Terms and Definitions What is Horsepower ................................................................................................ 1 Mean Effective Pressure.......................................................................................... 2 Theoretical Compressor Horsepower (Mathematical Integration).......................... 3 Measured Compressor Horsepower (Numerical Integration) ................................ 5 Mechanical Efficiency............................................................................................. 7 Volumetric Efficiency ............................................................................................. 7 Clearance................................................................................................................. 8 Capacity................................................................................................................. 10 Polytropic Compression and Expansion Exponents (log-log) .............................. 12 Valve Losses ......................................................................................................... 13 Evaluating Performance The Single Channel P-V Card............................................................................... 14 The Multi-Channel P-V Card................................................................................ 15 Valve Losses (Part II)............................................................................................ 15 Pulsation Loading.................................................................................................. 17 Compressor Efficiency and Thermodynamics ...................................................... 19 Calculating Isentropic Efficiency.......................................................................... 21 Making Decisions.................................................................................................. 23 Valve Losses ............................................................................................. 23 Thermodynamic Performance ................................................................... 24 Cost of Poor Performance ......................................................................... 24 Selling the Benefits ................................................................................... 25

Introduction Compressor performance, it is about measuring and predicting how a unit will behave under certain set of pipeline conditions. Compressor performance is also about identifying poor performing compressors, and identifying the areas that should be addressed to improve the overall compressor efficiency. In the first half of this course you learn the terms and definitions that are used to measure compressor performance. You will learn what terms like Volumetric Efficiency, Clearance and Capacity really mean. You will learn how they are measured, and how they are calculated. Along the way, you will also learn of the errors that can occur when analyzing measured data. The second half of the course will deal with evaluating compressor performance. We will identify the measured losses in the compressor cylinder and theorize about others. This segment will conclude with a discussion of isentropic (often called adiabatic) efficiency. Along with the basic theory, the measurement requirements and use of isentropic efficiency calculations will be discussed.

ii

Compressor Performance Analysis Terms and Definitions


The Fundamentals

Terms and Definitions This section contains brief descriptions of the most commonly used compressor performance terms and definitions

What is Horsepower? Horsepower is a rate of work, defined as the ability to perform 33000 ft-lbs of work per minute. The term horsepower was coined by James Watt, who determined that an average horse could lift a 550 lb bale of hay, 1 ft per second. Brake Horsepower Brake Horsepower (BHP) is the power required to drive the compressor and must include compressor mechanical and compression losses.

Where: ICHP = EFFm = indicated compressor horsepower compressor mechanical efficiency, %

Indicated Compressor Horsepower Indicated Compressor Horsepower (ICHP) is the compressor cylinder horsepower determined from data obtained from measured compressor P-V diagrams. The indicator diagram is integrated to obtain the cycle work (W) in inch-pounds. From this, cylinder indicated horsepower is determined from:

Where: W N Constant 396,000 = cycle work in inch-pounds = crankshaft speed in revolutions per minute = converts in-lbs/min to HP

How is Horsepower REALLY Determined from a P-V Card? The governing equation for determining horsepower is: Hp = MEP SV N 396,000 MEP = Mean Effective Pressure in Psi. SV = Swept Volume in cubic inches. N = Speed in RPM 396000= Constant for conversion to horsepower. Mean Effective Pressure The key term in the horsepower equation is Mean Effective Pressure (MEP), but just what is it. By definition MEP is the pressure, that if acted on the piston for the entire power stroke would do an amount of work equal to the work actually done by (or on) the piston. Confused? Look at Figure 1 below. From points 1 to 2 and 2 to 3, the gas is pushing the piston back down the bore. That is work done ON the piston by the gas. From point 3 to 4 and 4 to 1, the piston is compressing (pushing) on the gas. That is work done BY the piston on the gas.

Where:

Figure 1

The difference between the work done by the piston and the work done on the piston is the total work for the cycle. MEP is simply a way of stating this as the net pressure for the work done by the piston part of the cycle, it sets the work done on the piston part of the cycle equal to zero (0). The important thing about MEP, is that we must find it to use in our horsepower equation. How to determine MEP is demonstrated in the sections that follow.

Theoretical Compressor Horsepower (Mathematical Integration) To determine compressor horsepower, you first determine the work represented by a single cycle and then multiply by the rate of work (rpm) and convert to horsepower. We have already discussed the base horsepower equation and MEP, now we need to find a way to calculate the horsepower. Hp = MEP SV N 396,000

You can see by looking at equation, that we know everything but the MEP. Technically speaking, when we integrate the equation: W = Pdv or W = Vdp
Note: Either approach works and provides the same answer, but Vdp is easier.

We get the area of the P-V card, which represents the work in the cycle. The swept volume (SV) is a constant and is pulled out of the integral, so what we really get out of the integral is MEP. This is apparent in the solution to the integral presented below: The key relationship is that PVn = C(onstant), so that: PsVnbdc = Ccomp = PiVni where i is any point in the compression process. Similarly: PdVntdc = Cexp = PiVni where i is any point in the expansion process. And one final piece of algebra: P1V 1 = P2V 2, V2
n n
1 1 1 P Vn n n n 1 n 1 1 = =C =C Pn P2 P2 1 1

n P Area = Vdp = Ccomp Ps


1

Pd

n dp

nP Cexp Ps

Pd

n dp

Area =

n Ccomp

n 1 n 1 n 1 n 1 1 n n n n n n Pd Ps n Pd Ps Cexp n 1 n 1

n 1 n 1 1 1 n n n n Pd Ps n = Ccomp Cexp n 1

n 1 n 1 1 1 n nV nV ( RPs ) n Ps n P RP = ( ) s bdc s tdc n 1


n 1 n 1 1 1 n Ps n R n 1 Ps n Vbdc R nVtdc n 1

1 nn 1 n = Ps R 1 Vbdc R nVtdc n 1

) )

n 1 1 n Ps R n 1 LA + c L LA R n c L LA n 1 n 1 1 n Ps R n 1 LA 1 + c L R n c L n 1

n 1 n 1 1 nn 1 1 n n = Ps LA R + R c L R n R n c L 1 c L + R n c L n 1

1 n 1 nn 1 n n Ps LA R 1 + c L R + R n R 1 = n 1

The area of the card represents work (torque), as can be seen by the unit analysis.
Lbf 2 ( in)(in 2 ) in Lbf in

Note that the swept volume of the cylinder (L and A) has gone through the process intact. If you look at the rest of the term, it is the theoretical MEP for the cylinder.
1 n1 nn 1 n n MEP = Ps R 1 + c L R + R n R 1 n 1

Complete the equation to a include rate of work and correct to horsepower. Note that this is Indicated horsepower, not Brake: IHP =
1 n 1 1 n Ps LAN nn R 1 + c L R n + R n R 1 n 1 396,000

To determine brake Hp, we will need to account for the mechanical efficiency. BHP =
1 n 1 nn 1 1 n Ps LAN n R + R R 1 R n 1 + c L n 1 396,000 eff

Where: eff = mechanical efficiency of the unit. Measured Compressor Horsepower (Numerical Integration): For our example we will analyze the ideal card in Figure 2 (same as Figure 1). It has no valve losses and a very clearly defined valve event. We will do our integration in 4 easy steps. This will result in some errors that would be corrected if we took smaller steps. The technique used is also applicable to real cards, but requires small (we use 512) steps for an accurate analysis.

Figure 2

Necessary Information: Swept Volume (SV) = 14in(bore) x 17in(stroke) = 2617in2 Clearance = 168.83% Suction Valve Open (SVO) = 35% Discharge Valve Open (DVO) = 54% Ps = 609.78 +14.73 = 624.51 Pd = 779.76 + 14.73 = 794.49 794.49 R= = 1.272 624.51 N = 300.6 Processing the expansion (from 1 to 2) and suction (from 2 to 3) events first.
MEPexp = 794.49 + 624.51 [0.35 0] + 624.51 [1 0.35] 2

MEPexp = [709.5 0.35] + [624.51 .65] MEPexp = [248.325] + [405.93] = 654.257 Processing the compression (from 3 to 4) and discharge (from 4 to 1) events.
MEPcomp = 794.49 + 624.51 [1 0.54] + 794.49 [0.54 0] 2

MEPcomp = [709.5 0.46] + [794.49 .54] MEPcomp = [326.37] + [429.026] = 755.395 Finding the cycle MEP: MEP = MEPcomp - MEPexp MEP = 755.395 - 654.257 MEP = 101.14 Finding the Indicated Horsepower:
IHp =

2617 300.6 101.14 = 201 396000

Finding the Brake Horsepower (corrected for frictional losses in the unit to represent true flywheel horsepower, the industry standard for integrals is 95%):
BHp = IHp = 212 .95

If you plug the information through the ideal horsepower equation, you will find that the actual brake horsepower represented by this card is 210. So our approximation missed by about 1%. The error would have been much greater if we had a real card with varying pressures during the suction and discharge events. We can get the approximation closer and deal with the more complex real cards by reducing the step size for the numerical integration process. Of course instead of 4 easy steps, we would require hundreds of little steps to work our way around the card. Mechanical Efficiency The mechanical efficiency is nothing more than the ratio of work done by the driven component (indicated Hp) divided by the work done by the driver (brake Hp). Although it is not exactly correct, the industry standard for integral reciprocating compressors is 95%. That says that for every 1000 horsepower available at the crankshaft we will get 950 horsepower at the compressors. It is interesting to note that there is no difference noted for piston scavenged units, which often develop 10 to 15% more power-side horsepower than compressor horsepower. This is because the scavengers require horsepower also. It does not change the unit efficiency though, because the 5% loss is taken between the driver and the driven component. In other words, the scavenger horsepower is already factored out of the available brake horsepower. If we could remove the scavengers, then the unit would have more horsepower available for driving the compressors. Volumetric Efficiency Volumetric efficiency is a measure of absolute cylinder capacity at a given condition. The condition most commonly referenced is Suction Volumetric Efficiency (EVs). When most folks talk about volumetric efficiency (and dont specify suction or discharge), they are referencing the EVs. The EVs is referenced to the volume of the suction event, and the Discharge Volumetric Efficiency (EVd) is referenced to the volume of the discharge event. In either case, the event swept volume (capacity) is expressed as a percent of the cylinder end swept volume (capacity). The volumetric efficiency is always less than 100%. Volumetric efficiency can be calculated (theoretical), or measured. The theoretical, efficiency is derived from the gas properties and the physical cylinder parameters. It is useful for predicting the compressor cylinder flow for a given condition.

Theoretical Suction Volumetric Efficiency


1 n EV s = 100 CL% R 1

Theoretical Discharge Volumetric Efficiency


1 1 1 n 1 n EV d = CL% 1 + 100 R R

Where: PD PS R CL% n = = = = = discharge pressure (psia) suction pressure (psia) absolute pressure ratio (PD/PS) clearance from manufacture in % effective cylinder isentropic exponent

Measured Volumetric Efficiency Referring back to our P-V card in Figure 2, the volumetric efficiency is defined by the valve opening events. Keep in mind that cylinder filling and expulsion (un-filling?) can only occur when the valves are open. The suction valve open event was at 35% of stroke, from that point to 100%, the valve was open. The discharge valve opened at 54% of stroke, and remained open until piston moved back to the zero position. EVs = 100-35=65% EVd = 54% Volumetric efficiency is the critical measured term in determining both the cylinder clearance and capacity. Other terms in the equations are easily measured, or are can be looked up in a book, but the volumetric efficiency is a bit more tricky to determine reliably and repeatedly. Those of you familiar with analysis are very aware of the subjective nature of this measure. When we move on to a real card, we will discuss the interpretive nature of this measured parameter. Clearance Clearance (CL) refers to the actual volume of gas trapped in the cylinder when the piston is at top dead center. Clearance includes the fixed clearance of the cylinder and any

added volumes (pockets) that are in use. Clearance is typically expressed as percent of cylinder swept volume. Clearance is determined by rearranging the theoretical volumetric efficiency equations and solving for the clearance term. All of that math and theory pays off here, since you will be using the measured volumetric efficiency plugged into the theoretical equation to determine a measured clearance. This works because the theoretical equation defines the relationship between volumetric efficiency and clearance. Clearance from Suction Volumetric Efficiency CL% = 100 EV s 1 R n 1

Clearance from Discharge Volumetric Efficiency


1 n EVd 100 R CL% = 1 1 n 1 R
1

Gathering the necessary information from Figure 2, and the volumetric efficiencies we calculated above, we have the following data available to determine the clearances for our card. EVs = 65% EVd = 54% Ps = 609.78 +14.73 = 624.51 Pd = 779.76 + 14.73 = 794.49 794.49 R= = 1.272 624.51 n = 1.28 Clearance from Suction Volumetric Efficiency CL% = 100 EV s R 1
1 n

100 65
1 1.272 1.28 1

= 169.3%

Clearance from Discharge Volumetric Efficiency

1 n 1 1.28 EVd 100 54 100 R 1.272 = CL% = = 168.4% 1 1 1 n 1 1.28 1 1 R 1.272 The exactly correct answer is 168.8%, so why the error and why the discrepancy between the two equations? Round off error. In the material presented here, the valve events are rounded to the nearest whole number. The pressure ratio is truncated at 3 decimals as well. If you care to set this up in a spreadsheet, you will find that the answers match exactly. For the record, the actual EVs = 65.065% and the EVd = 53.910%, you can work it from there.
Capacity

Capacity is traditionally calculated from the volumetric efficiency equations. If you take the volumetric efficiency and multiply by cylinder swept volume, you will have an actual cubic feet of gas moved by the cylinder in one stroke at the particular condition evaluated (suction or discharge). Multiply throughput per stroke times the unit speed (RPM) and you have Actual Cubic Feet per Minute (ACFM), for the condition of interest. Note that the only time the cylinder moves gas, is during the discharge or suction events. That is why these are the only two points of interest.
Capacity from the Suction Event

CQ * SV N EVs Ps Z STD Qs = Ts Z s
Capacity from the Discharge Event

Qd = *

CQ * SV N EVd Pd Z STD Td Z d Use CQM for capacity in MMSCFD. Use CQS for capacity in SCFM.

Where: Qs Qd SV = = = suction capacity in MMSCFD or SCFM discharge capacity in MMSCFD or SCFM cylinder swept volume in cubic inches

10

N EVs EVd Ps Pd Ts Td Zs Zd ZSTD CQM

= = = = = = = = = = =

unit rpm suction volumetric efficiency in % discharge volumetric efficiency in % suction pressure in psia discharge pressure in psia absolute suction temperature in 0R (0 F + 459.67) absolute discharge temperature in 0R (0 F + 459.67) suction gas compressibility discharge gas compressibility compressibility at 14.7 psia and 60 0F 0.2314x10-6 for flow in MMSCFD = CQs x 60rnin x 24hr x 1 SCF hr day 1,000,000 MMSCFD

Going back to our P-V card in Figure 2, we can calculate the throughput for the card. We will calculate at both the suction and discharge events. They should be equal, if not then we have a leak. If the suction capacity is greater than the discharge capacity it is a suction leak. If the discharge capacity is greater than the suction capacity, then we have a discharge leak. You can not determine anything about ring leaks by comparing measured capacities for the suction and discharge events.
Capacity from the Suction Event

CQ * SV N EVs Ps Z STD 0.2314 10 6 2617 300.6 65 624.51 .99 Qs = = Ts Z s 519.6 .90 = 15.64MMSCFD
Capacity from the Discharge Event

Qd =

CQ * SV N EVd Pd Z STD 0.2314 10 6 2617 300.6 54 794.49 .99 = Td Z d 551.19 .90

= 15.58MMSCFD

11

As you can see, the answers are nearly the same for both events. The small errors are again round off errors that were incurred when truncating the numbers for the presentation. An alternate method for calculating capacity is by measuring the energy rise of the gas across the compressor (reported as BTU/MMSCF). With some conversion you can calculate Hp/MMSCF. Since you measure the horsepower it is simple math to get the throughput from there.
Polytropic Compression and Expansion Exponents (log-log)

The polytropic compression and expansion exponents are critical to all of the previous analysis. The term n, which is sometimes referred to as the isentropic compression exponent is the polytropic exponent. A polytropic process is one that is repeatable and governed by the equation:
PV n = C

Where C is a constant defined by the actual process. It is from this simple governing equation that all of the other parameters for the process are determined. Since any corresponding P and V for the process provides the same C, you can restate the equation as: n n P1V1 = P2V 2 This is a very simple, yet powerful way of restating the equation. By fixing state 1, (say as the suction pressure and volume at BDC), you can determine the pressure at any point during the stroke (think of the remaining volume). With some substitution, the equation can be manipulated to yield the volumetric efficiency equations, and the clearance equations. It is the fundamental equation of reciprocating compression, but it leaves one question. What is n? n is the polytropic exponent and it is determined from the measured data. When you plot log P .vs. log V, n is the slope of the line. The equations below show how to determine the polytropic exponents using just 2 data points and analyze the data from Figure 1. In practice, the exponents are usually determined by a least squares curve fit of all the data collected during the compression and expansion events. Polytropic Compression Exponent
Pdi log P BDC nc = V BDC log V di 794.49 log 624.51 = 1.283 = 100 + 168.83 log 168.83 + 54

12

Polytropic Expansion Exponent


PTDC log P si ne = V si log V TDC 794.49 log 624.51 = 1.278 = 100 - 65 + 168.83 log 168.83

In a good cylinder, the compression and expansion exponents are equal. In this case we are plagued by the same round off errors that we have seen from the beginning. These exponents show that this is a good card (it should be, its an ideal card). If the compression exponent is larger than the expansion exponent, then you have a discharge leak. If the expansion exponent is larger than the compression exponent, then you have a suction leak. If you plot log(P) .vs. log(V) for your measured data, you should have straight lines, it helps to overlay the ideal log-log line on the graph. If the measured data bows, then you may have a ring leak. The log-log plot is the best P-V tool for determining ring leaks. One critical point, is that you need to have your clearances correct in order to accurately calculate the polytropic exponents. If you use analysis software that has this feature, you are already aware of this.
Valve Losses

Valve losses, which are sometimes referred to as Differential Indicated Power (DIP), represent the difference between the indicated power for some portion or zone of the card bounded by a reference pressure line. The reference pressure may be constant or varying, and depends on amount of data available for analysis. All of this can be confusing, but to demonstrate properly, we need to look at a real card, not an ideal card. We will work with real cards in the next section of the class and will discuss the calculation of valve losses there.

13

Compressor Performance Analysis Evaluating Performance

Evaluating Compressor Performance The Single Channel P-V Card

Now that we have established the terms and measures we will use to determine performance, we can analyze a real P-V card. Figure 3 below is a diagram of the typical single channel P-V card with all of the parts labeled.

Figure 3

With single channel data, you have to make some assumptions about what is going on in the cylinder. The biggest assumptions are that the pressure at BDC and TDC (the toes of the card) are equal to the line pressures. The single channel analysis uses these projected line pressures to establish the valve events, which are used to determine valve losses, volumetric efficiencies, clearances, capacity and the polytropic exponents. It is important to remember that except for gross horsepower, everything you are interested in depends on correctly picking the valve crossing points.

14

The Multi-Channel P-V Card

Figure 4 below is a diagram of the typical multi-channel P-V card. You can see that we have introduced an ideal card and pulsation loading.

Figure 4

There is much more information available with the multi-channel card. The ideal card is drawn using the unit parameters and assuming ideal compression between the average nozzle pressures. The valve opening events are clearly defined as they occur where the PV card crosses the nozzle trace. Pulsation loading is identified and will be discussed in a later section.
Valve Losses (Part II)

Figure 5

15

Valve losses are derived by performing an MEP calculation for the appropriate nozzle trace and the portion of the card bounded by that trace. The valve losses are the shaded areas shown in Figure 5. Remember that the single channel approach simply assumes the line pressure to be equal to the toe pressures. The valve loss horsepower is found by multiplying the MEP by the affected stroke increment, the piston area, the rpm and appropriate horsepower correcting factor. Valve losses are not constant over a range of pressure ratios, so we must have an equation that models that changing loss. As the pressure ratio goes up, the valve losses go down. At Tennessee Gas Pipeline, we express valve losses as a percent of ideal compression horsepower and define the term waste K as:
K = %ValveLoss( R 1)

R = Pressure Ratio

Rearrange to find that:


%ValveLoss =

K ( R 1)

During the valve events, you have both valve losses and pulsation loading. Sometimes that loading is a combined loading, with pulsation being negative and valve losses being positive, and some times the loadings are distinctly separate. Let the perimeter of the real card be your guide, and keep in mind that valve losses can never be negative.

Figure 6

When the perimeter of the card falls inside a nozzle trace, then it becomes the controlling part in the analysis. In Figure 6, the card is indicating that the valves actually closed during that part of the discharge event. The area between real card trace and the ideal card trace in the highlighted region is negative pulsation loading. Valve losses can never be negative. One final note on valve losses. Lost work, is work that is done beyond the minimum required to meet the pressure and throughput demands. All of the losses in the compression process are caused by throttling at either; the valves, the passages, or even the pocket throats. In a normal P-V card, valve losses are the only measured loss.

16

Results from the PV Card

Single and multi-channel results from Figure 5 are presented in Table 1. Both cards are identical, the only difference is the inclusion of nozzle traces on the mult-channel card. Looking at the single channel results, you could be misled into thinking that you had a problem, especially during the discharge event. What appears to be a significant loss, is really nothing unusual. The more information you have, the better equipped you will be to make the correct call. Single Channel Result 612.01 754.06 5.09% 11.62 HP 15.82% 36.1 HP Table 1 Multi-Channel Result 610.58 779.78 3.57% 8.15 HP 4.56% 10.41 HP

Suction Pressure Discharge Pressure Suction Valve Loss Discharge Valve Loss

Pulsation Loading

Pulsation loading is probably the term that you are least familiar with in compressor performance analysis. This term is used by to describe the effects of undulating nozzle pressures on compressor performance (Figure 4). In a single channel analysis, there is no pulsation loading, because the line pressures are assumed to be the toe pressures. Expansion and compression always begin right at the toes. The problem is that an operator can not see inside the compressor to read the toe pressures. The operator reads a gauge and operates the unit based on this reading. Look back at the start of re-expansion in Figure 4 (TDC). You can see that the incylinder pressure at TDC dropped significantly below the average discharge pressure. This caused re-expansion to start at a lower pressure than expected (the average line pressure). Re-expansion now occurs outside of, but parallel to, the ideal re-expansion line. This creates more area in the P-V card and represents more work. This is work that must be accounted for in the predictive horsepower equation, otherwise the unit will be overloaded. Calculating pulsation loading for a multi-channel analysis is done in much the same manner as calculating valve losses, only in this case we focus on the difference between the real trace and the ideal trace. Loading during the compression or expansion event is easiest to calculate and typically accounts for the bulk of pulsation loading (see Figure 4). Pulsation loading during the valve event is defined by the inner trace (typically the nozzle trace) as it moves above and below the ideal card. This loading can be either positive (outside the ideal card) or negative (inside the ideal card). See Figure 7.

17

Figure 7

Looking back, Figure 6 shows the area of the card that is most troublesome During the valve events, you have both valve losses and pulsation loading. Sometimes that loading is a combined loading, with pulsation being negative and valve losses being positive, and some times the loadings are distinctly separate. Let the perimeter of the real card be you guide, and remember, Valve losses CAN NOT be negative. All predictive horsepower equations have factors to account for non-ideal work. With a single channel analysis, this extra work is usually lumped together with the valve losses. In a multi-channel analysis, pulsation loading and valve losses can have separate terms.
Modeling Pulsation Loading

Now that we have identified pulsation loading, we need a way to predict it. At TGP, we express pulsation induced horsepower as a percent of ideal horsepower. We then roll each individual cylinder loading up to the unit level for developing unit operating curves. Once this is done, we plot the data as a function of pressure ratio and curve fit the equation with either a first or second order polynomial. See Figure 8.

PULSATION LOADING CURVE

0.150 0.140 0.130 0.120 0.110 0.100 0.090 0.080 0.070 0.060 0.050 0.040 0.030 0.020 0.010 0.000 -0.010 -0.020 1.20

LOADING FACTOR

Ao = -6.21782 A1 = 8.76755 A2 = -3.02448

1.23

1.25

1.28

1.30

1.33

1.35

1.38

1.40

1.43

1.45

1.48

PRESSURE RATIO (R)

Figure 8

18

Compressor Efficiency and Thermodynamics

Thermodynamic compressor efficiency calculations are not foreign to compressor performance. They are commonly used for evaluating centrifugal compressor performance. By applying the principles of isentropic compression to reciprocating compressor analysis, it is possible to evaluate the units independent of the pressure ratio, clearance volume or some other physical constraints. By contrast, volumetric efficiency is dependent on all of these variables. Isentropic efficiency is the ratio of ideal energy rise across the compressor compared to the actual energy rise across the compressor. This energy rise is measured for the working fluid (gas) in the process. Every substance has a state energy. For a gas that is the energy of pressure and temperature for a given condition (state). It is reported in BTU, just like the heating value of the gas. Dont get confused, it is not related to the heating value, even though it is reported in the same units. The equation for calculating isentropic efficiency is:

hideal where h is the enthalpy of the fluid. hreal

Before we can calculate an isentropic efficiency for a compressor, we must determine whether or not the process fits the definition of isentropic. To be isentropic, a process must be both reversible and adiabatic (no heat transfer to the surroundings).
Is the process reversible?

The answer is mostly yes. Controlled compression and expansion is considered reversible. In our process, the re-expanding gas pushes the piston down the bore, against the rod, with the same forces that the rod exerted on the piston to move it up the bore during the compression part of the stroke. The balanced forces, over opposite parts of the stroke yield equal work, and we can conclude that the process is reversible. The irreversible part is the friction encountered while the piston moves in the cylinder. For the typical pipeline unit the parasitic horsepower is very small compared to its developed horsepower. In fact, it is small enough to be neglected and we can assume that the process is reversible.
Is the process adiabatic?

Overall, a process can be considered adiabatic if it meats the following criteria: 1. No external cooling. 2. Minimal temperature difference between the process and the surroundings. Low ratio pipeline cylinders typically meet these criteria. Helping us stretch this assumption is the small amount of time available for any heat transfer to take place. At 300 rpm, there is only 1/5 of a second available to affect any heat transfer. You can also

19

argue that any heat transfer would balance out on each side of the cycle, with equal heat transferred to the cool suction gas from the relatively warm(er) cylinder and from the hot discharge gas to the relatively cool(er) cylinder, netting to zero.
Applicability

Using the thermodynamic (adiabatic) efficiency model it is possible to compare the performance of similar units under vastly different operating conditions. Even different units can be qualitatively compared. Figure 9 shows some of the results of the PCRC Compressor Efficiency Survey. The results are presented as the average efficiency on a company by company basis. Note that company (1) averages 86% efficiencies. While company (7) manages only 74%. We dont know who company (1) or (7) is, what type of units they are operating, or even the operating conditions that prevailed during the test, yet we are able to fairly evaluate their performance.
PCRC COMPRESSOR EFFICIENCY SURVEY - Company Averages
Average Isentropic (Adiabatic) Efficiency % 90 88 86 84 82 80 78 76 74 72 70 1 2 3 4 5 6 7 Company I.D. Number

Figure 9 To calculate the compressor efficiency, you need to measure: 1. Suction Temperature. 2. Suction Pressure. 3. Discharge Temperature. 4. Discharge Pressure. 5. A good, 10 part Gas Analysis. 6. Horsepower (Optional, needed for flow calculations).

20

Make all of the measurements as close to the cylinder as possible. The further away from the cylinder you get, the more time and surface area you allow for heat transfer to take place. Heat transfer violates the number two requirement for a process to be considered isentropic. Pressure measurements are less sensitive to location, as long as there are no significant restrictions imposed in the system. Orifices, and choke tubes create pressure drops in the system that will affect the efficiency calculations and violate the requirement for reversibility. An accurate gas analysis is needed to perform the BWR calculations that are necessary for calculating the state enthalpies of gas. The gas analysis should be taken as close to the testing location as possible. In general, this is more critical the closer you get to gathering locations. If possible take a gas sample at the test site and send it to a lab for an accurate gas analysis. Often times a continuous analysis will be performed on a chromatograph located near a sales point. If this is available it can be useful for providing continuously updated data throughout the test.

Figure 10
Calculating Isentropic Efficiency

Now that we know what data to take and how the isentropic efficiency is calculated, its time to work an example. The example refers to the Thermo routine, which is a small program that will be passed out during the class. The routine was first distributed at the PCRC Gas Machinery Conference in October of 1996. It is useful for making the necessary calculations, but neither Tennessee Gas Pipeline, El Paso Energy, or the Gas

21

Machinery Research Council accept any responsibility or liability for the use and accuracy of the program. Use the program at your own risk. Assume the following conditions reported for 2 units in the same service: Suction Pressure = 650.45 Psig. Suction Temperature = 60 oF. Discharge Pressure = 850 Psig. Unit 1 discharge temp = 99.2 oF Unit 2 discharge temp = 106.0 oF Solve for Efficiency Unit 1 = 84% efficiency Unit 2 = 68% efficiency Solve for HP/MMSCF.
Unit 1 ActualHP / MMSCF = 13.73HP / MMSCF Unit 2 ActualHP / MMSCF = 16.96 HP / MMSCF

Whats the cost of the efficiency difference? Assume a total operating cost $0.03/HP-HR.

HP $ 24 HR $ = HP HR Day MMSCF MMSCFD


The 84% compressor costs $9.88/MMSCFD delivered. The 68% compressor costs $12.21/MMSCFD delivered.
It costs 23.6% more to deliver gas with the 68% compressor than the 84% compressor.

Critical operating information that would be difficult to accurately derive from the typical P-V analysis.

22

Making Decisions

All of the data in the world doesnt do us any good if we dont analyze it, and use it to make our decisions. While this section is brief, it does provide a guide on where to look for support in your decisions. The discussion in this section assumes that the compressors are in good mechanical condition. However, some of the techniques can be applied to calculating the costs associated with running a unit which has a problem.
Valve Losses

The easy part about valve losses is that less is better, kind of like a golf score. While that concept is simple, the practice of accurately determining valve losses is a bit more touchy. When you include a nozzle trace in your analysis, and use it in your valve loss calculation, you will get a number that represents the total passageway and valve loss. Usually, the valve loss is the dominate loss in the system. Many arguments have been made to discredit the nozzle trace as a poor measure of valve performance. Yet, it is the best practical measure we can make, and it beats the heck out of a single channel analysis (which can be affected by pulsation). Dont get caught up in these arguments, use your information to make a comparative analysis. If the only change you make is the valve assemblies and the losses go down, the new valve is better than the old one. When evaluating valve and passage losses you should try to confirm your conclusions with the thermodynamic results. By quantifying your losses as a percent of ideal horsepower, you can predict your best possible efficiency. 100%(Ideal) - Valve /Passage Loss(%of Ideal) = Efficiency (Best Achievable) The ideal card represents perfect compression, any losses will detract from that ideal performance. If your valve losses are 10% of the ideal horsepower, then the best compressor efficiency your unit will attain is 90%. You can not have an efficiency better than you will predict with the equation above. If you do, there is an error. You probably wont measure the best achievable performance, since there are some losses that we dont capture. Generally, if you can predict the efficiency with 2% you are doing very well. When you see that the actual efficiency is significantly less than you would expect, given your valve losses, you need to look elsewhere in the cylinder. Poor valve or piston ring sealing will cause the isentropic efficiency to drop. Another thing to watch is the number of pockets you have open. If the pockets have a restrictive inlet, then they will also cause your efficiency to decrease.

23

Thermodynamic Performance

The beauty of the thermo analysis, is its ability to fairly compare different units across a wide range of operating conditions. That ideal standard is the same regardless of what or where you measure. 100% is as good as it gets. Use your thermo analysis as a comparative tool. It is especially useful for comparing similar compressor cylinders. If you have one set of Clark VBAC cylinder operating at 90% and another set at 80%, then youre on to something. You can determine the additional cost per MMSCFD for running the poor compressor, and then justify the expenditures necessary for improving its performance.
Cost of Poor Performance

Assigning costs to poor performance is actually pretty easy. If you have specific unit, or location level data, then you can use the figures to determine the cost of poor performance. You can roll up to company wide, or industry wide data if necessary. An industry survey that was done by Southwest Research Institute in the early 90s found that the average industry cost for reciprocating compression was $0.75/HP-Day. This is the number that was used to determine costs for the Compressor Diagnostic Software (CDS) that was published by the PCRC. That works out to about 3 cents per HP-HR. In order to determine the cost of poor performance, you need a good ideal of what is realistically achievable. DO NOT USE 100%. Getting a feel for what is achievable will take some time and test results, but once you develop a database with the correct information, you can readily identify the opportunities for improvements. Example: Clark HBA-6T, rated at 1550 HP with 6 valve per end VBAC compressors. Station A : Station B : compressor efficiency = 90%. compressor efficiency = 80%

Cost of poor performance at Station B =

(90% 80% ) 1550 $0.75 = $145


80

Day

Day

Pretty easy stuff, thermo is your friend.

24

Selling the Benefits

This is the cool part, but it can also be the toughest part of the analysis to figure out. As engineers and analysts, we are usually pretty good at quantifying failure, or improvement, but we need data. The challenging part of predicting the effect of increased throughput on the bottom line is coming up with that data. When determining increases in throughput, you get to legitimately use the larger number. If you increase your efficiency from 80% to 90%, you have made a 10 out of 80 improvement, or 12.5%. When our 1550 HP unit operated at 80%, it effectively delivered 1240 HP worth of compression. When we added 10 points (not %) to the efficiency, our addition was 100% effective. We added 155 HP worth of compression. 155/1240 =12.5% easy as pie.
How do you sell that 12.5%?

Reduced Transportation Costs On pipelines, (mainline transmission companies) use the reduced transportation cost approach. Show that while the machine will still cost the same to operate, it will pump more gas and allow you to shut a unit down somewhere else on the pipeline. This is an easily understood approach. When you are modifying several units at a big station, you may be able show that you can shut a unit down at that location and still meet demand. This has pay back written all over it. Increase Revenues At a gathering site, use increased revenues. The producer is paid for each cubic foot delivered, and its easy to calculate the increased capacity. Show that each additional cubic foot pumped is money in the pocket without any increase in compression costs. In this case it becomes very easy to show payback.
On a pipeline, one units performance cant be projected across the whole pipeline, unless you are making improvements at a bottleneck location. Even then, you have to ensure the bottleneck is the plant, and not the surrounding pipeline. For a pipeline, it is nearly impossible to estimate the revenue impact without sophisticated pipeline modeling software.

25

You might also like