SURGE PREVENTION IN CENTRIFUGAL COMPRESSOR SYSTEMS Rainer Kurz and Robert C.

White Solar Turbines Incorporated San Diego, California Abstract Centrifugal compressor surge and its prevention have drawn significant attention in the literature. An important aspect of surge avoidance lies in the design of the compressor station and, in particular, the piping upstream and downstream of the compressor. Most anti-surge systems are perfectly capable of avoiding surge during normal operating conditions. However, unplanned emergency shutdowns present a significant challenge, and surge avoidance in these cases depends to a large degree on the station layout, in particular the volume of the piping system downstream of the compressor. Furthermore, the concepts used in the anti surge system (valves, piping, coolers) also impact the start-up of the station, or of individual units of the station. Start-up considerations for stations with and without cooled recycle loops are discussed.

Nomenclature A cv C Fp h Hcooler J k k Kv L N p Q SG SM T t V Y Z ,, Flow area Flow coefficient (cv=Q/SG/p) Compressible valve coefficient Piping geometry factor Head Gas cooler heat transfer Inertia Isentropic exponent Constant Valve coefficient Pipe length Speed (1/s) Pressure Volumetric flow Specific gravity Surge margin (%) Temperature Time Volume Expansion Factor Compressibility factor Constants Density

Subscripts avail Available compr Compressor op Operating point surge At surge std At standard conditions ss Steady state v Valve 1 Compressor inlet 2 Compressor discharge Introduction Recently, there have been attempts to increase the stability margin of a compressor by active (Epstein et. al. [1], Blanchini et. al. [2]) or passive means (Arnulfi et. al. [3]), or by increasing the accuracy of determining the surge margin (McKee and Deffenbaugh [4]). It is often overlooked that meaningful gains can be made by better understanding the interaction between the compressor, the anti-surge devices (control system, valves) and the station piping layout (coolers, scrubbers, check valves). This study focuses on centrifugal compressors driven by two-shaft gas turbines (Figure 1). The possible operating points of a centrifugal gas compressor are limited by maximum and minimum operating speed, maximum available power, choke flow, and stability (surge) limit (Figure 2). Surge, which is the flow reversal within the compressor, accompanied by high fluctuating load on the compressor bearings, has to be avoided to protect the compressor. The usual method for surge avoidance (anti-surge-control) consists of a recycle loop that can be activated by a fast acting valve (anti-surge valve) when the control system detects that the compressor approaches its surge limit. Typical control systems use suction and discharge pressures and temperatures, together with the flow through the compressor to calculate the relative distance (surge margin) of the present operating point to the predicted or measured surge line of the compressor (Figure 2). The surge margin is defined by:
SM = Qop Q surge Qop
N = const

(1)

If the surge margin reaches a preset value (often 10%), the anti-surge valve starts to open, thereby reducing the pressure ratio of the compressor and increasing the flow through the compressor. The situation is complicated by the fact that the surge valve also has to be capable of precisely controlling flow. Additionally, some manufacturers place limits on how far into choke (or overload) they allow their compressors to operate.

Figure 1: Compressor station

Figure 2: Typical compressor map A very critical situation arises upon emergency shutdown (ESD). Here, the fuel supply to the gas turbine driver is cut off instantly, thus eliminating the power to the driven compressor 1. The inertia of compressor, coupling and power turbine have to balance the compressor absorbed power, causing a rapid deceleration. Because the head-making capability of the compressor is reduced by the square of its running speed, while the pressure ratio across the machine is imposed by the upstream and downstream piping system, the compressor will surge if the surge valve cannot provide fast relief of the pressure. The deceleration of the compressor as a result of inertia and dissipation are decisive factors. The speed at which the pressure can be relieved not only de1

Some installations maintain fuel flow to the turbine for 1 to 2 seconds while the recycle valve opens. However, this can generate a safety hazard.

pends on the reaction time of the valve, but also on the time constants imposed by the piping system. The transient behavior of the piping system depends largely on the volumes of gas enclosed by the various components of the piping system, which may include, besides the piping itself, various scrubbers, knockout drums, and coolers. The system boundaries for this study are the first downstream check valve, while the upstream boundary may be either a check valve or an infinite plenum (at constant pressure, Figure 3). The requirements of the anti-surge system for such situations as ESD or other massive system disturbances are distinctly different from the usual process control case.

ENGINE

COMPRESSOR VV

SV

AFTERCOOLER

DV

TT LV SCRUBBER

FT

PT

PT

TT

ANTI-SURGE CONTROLLER LIMIT SWITCH 4 - 20mA SOLENOID ENABLE 24VDC POSITION TRANSMITTER 4 - 20mA

SV LV VV DV

= = = =

SUCTION VALVE LOADING VALVE VENT VALVE DISCHARGE VALVE

TT = FT = PT =

TEMPERATURE TRANSMITTER FLOW TRANSMITTER PRESSURE TRANSMITTER

FAIL OPEN ANTI-SURGE CONTROL VALVE

Figure 3: Anti-surge and recycle system. The former is a massive, fast change in conditions that first and foremost requires fast system reaction. This requires, among others, extremely fast opening valves with sufficiently large flow areas. The latter requires the capability to precisely control slow changes in the process, such that no oscillations occur, which can be accomplished with precisely positioned valves. Surge control valves are primarily sized to fit the compressor. During steady-state recycling, the required capacity of the recycle valve can be directly derived from the compressor map. To handle transient conditions, the required capacity must be greater to allow for the volumes on either side of the compressor. With the initiation of a shutdown, the compressor can be expected to decelerate approximately 30% in the first second. With a 30% loss in speed, the head the compressor can develop at its surge limit will drop by approximately 50%. The recycle control valve

must, therefore, reduce the pressure ratio across the compressor by one-half in that first second to avoid surge. The following guidelines pertain to a typical one valve, one compressor arrangement. More complex systems of cascaded valves or valves around multiple compressors require a more detailed analysis. To facilitate both precise throttling at partial recycle and the need to reduce the pressure difference across the compressor quickly during a shutdown, surge control valves with an equal percentage characteristic are advantageous (Figure 4). The equal percentage characteristic spreads the first half of the valve's fully open capacity over the first 2/3 of the valves travel for a globe valve, and about one third of the valve's fully open capacity over the first 2/3 of the valves travel for a ball valve. This greatly improves controllability at partial recycle throttling. In order to avoid surge during a shutdown, the valve must open to the required capacity in significantly less than one second.

Figure 4: Typical valve characteristics for globe and ball valves. Surge control systems must be sized to meet two diverse objectives: During steady-state recycling, the required capacity of the recycle valve can be directly derived from the compressor map: the smaller the valve, the smoother the control. During transient conditions, the required capacity increases due to the volumes on either side of the compressor. Therefore, to avoid surge

during a shutdown, the bigger the valve, the better. To facilitate both smooth throttling at partial recycle and the need to reduce the pressure differential (DP) across the compressor quickly during a shutdown, control valves with an equal percentage characteristic are recommended. With an equal percentage characteristic, the more the valve is opened, the greater the increase in flow for the same travel. We recommend two types of valves for surge control: globe valves and noise-attenuating ball valves. The globe valves capacity (Cv) varies with the square of the percentage travel. The noiseattenuating ball valves capacity (Cv) varies with the cube of the percentage travel. Both valves are sized to be throttling at about two-thirds open at surge conditions. As such, the noiseattenuating ball valve will have 50% more capacity to depressurize the discharge volume than the globe valve. This additional capacity makes the noise-attenuating ball valve the better choice in installations where there is a single surge control valve (i.e. there is no hot bypass) and the discharge volumes are large (e.g., the discharge system includes an aftercooler). The valve is matched to the compressor (Figure 5).

Figure 5: Matching of Valve and Compressor. The valve characteristic for a number of opening positions (60%,70% and 100%) is superimposed to the compressor performance map.

Surge Phenomenon Day [5] has provided very detailed measurements of compressor surge cycles, pointing out the dominating influence of the discharge volume. Figure 6 shows the head-versus-flow characteristic of a typical centrifugal compressor at constant speed, including the areas of unstable operation. At flows lower than the stability limit

(surge line), the compressor initially shows a reduced capability to generate head with reduced flow, until it experiences reverse flow, that is, the gas now flows from the discharge to the suction side of the compressor. Once flow reversal occurs, the amount of flow depends on the pressure ratio across the compressor, since in this situation the compressors acts more or less like an orifice. The flow reversal means that the pressure downstream of the compressor is gradually reduced. The speed of pressure reduction depends largely on the size of the volume downstream of the compressor. Once the pressure is reduced sufficiently, the compressor will recover and flow gas again from the suction to the discharge side. Unless action is taken, the events repeat again. Ongoing surge can damage thrust bearings (due to the massive change of thrust loads), seals, and eventually overheat the compressor.

Figure 6: Simplified surge cycle Modeling the Piping Surge Control Interaction Design of the piping and valves, together with the selection and the placement of instruments will significantly affect the performance of an anti-surge control system. This should be a major issue during the planning stage because the correction of design flaws can be very costly once the equipment is in operation. Typical configurations for recycle systems are outlined in Figure 3. In its simplest form, the system includes a flow-measuring element in the compressor suction, instruments to measure pressures and temperatures at suction and discharge, the compressor, an aftercooler and a discharge check-valve, as well as a recycle line with a control valve, connected upstream of the discharge check valve and compressor flow-measuring device. The control system monitors the compressor operating parameters, compares them to the surge limit, and opens the recycle valve as necessary to maintain the flow through the compressor at a desired margin from surge. In the event of an ESD, where the fuel to the gas turbine is shut off

instantly, the surge valve opens immediately, essentially at the same time the fuel valve is closing. In a simple system, the boundaries for the gas volume (V) on the discharge side are established by the discharge check valve, compressor, and recycle valve (Figure 3). The volume on the suction side is usually orders of magnitude larger than the discharge volume and, therefore, can be considered infinite. Thus, for the following considerations, the suction pressure remains constant. This is not a general rule, but is used to simplify the following considerations. This yields the simplified system, consisting of a volume filled by a compressor and emptied through a valve (Figure 7).

Figure 7: Simplified system and transient characteristic. The basic dynamic behavior of the system in Figure 7 is that of a fixed volume where the flow through the valve is a function of the pressure differential over the valve. In a surge avoidance system, a certain amount of the valves flow capacity will be consumed to recycle the flow through the compressor. Only the remaining capacity is available for de-pressurizing the discharge volume. The worst case scenario for a surge control system is an ESD, particularly if the compressor is already operating close to surge when the engine shutdown occurs2. With the initiation of shutdown, the compressor will decelerate rapidly under the influence of the fluid forces counteracted by the inertia of the rotor system. A 30% loss in speed equates to approximately a loss in head of about 50%. The valve must, therefore, reduce the pressure across the compressor by about half in the same time as the compressor loses 30% of its speed. This speed loss is very rapid. The larger the volumes are in the system, the longer it will take to equalize the pressures. Obviously, the larger the valve, the better its potential to avoid surge. However, the larger the valve, the poorer its controllability at partial recycle. The faster the valve can be opened, the more flow can pass through it. There are, however, limits to the valve opening speed, dictated by the need to control intermediate positions of the valve, as well as by practical limits to the power of the actuator. The situation may be improved by using a valve that is only boosted to open, thus combining high opening speed for surge avoidance with the capability to avoid oscillations by slow closing.
2

Similar considerations are to be made for the trip of an electric motor driver. The main difference is the different inertia of the motor (and the gearbox).

If the discharge volume is too large and the recycle valve cannot be designed to avoid surge, a short recycle loop (hot recycle valve) may be considered, where the recycle loop does not include the aftercooler. While the behavior of the piping system can thus be predicted quite accurately, the question about the rate of deceleration for the compressor remains. It is possible to calculate the power consumption for a number of potential steady-state operating points. The operating points are imposed by the pressure in the discharge volume, which dictates the head of the compressor. For a given speed, this determines the flow that the compressor feeds into the discharge. In a simple system as described above, mass and momentum balance have to be maintained (Sentz [6], Kurz and White [8]). From this complete model, some simplifications can be derived, based on the type of questions that need to be answered. Obviously, for relatively short pipes, with limited volume (such as the systems desired for recycle lines), the pressure at the valve and the pressure at compressor discharge will not be considerably different. For situations like this, the heat transfer can also be neglected. The set of equations then is reduced to:
dp 2 k p 2 [Q Qv ] = dt V

(2)

The rate of flow through the valve is calculated with the standard ISA method [7]3:
Qstd p p1 1 = 1360 F p c v Y 2 SG T2 Z 2 p2
0 .5

(3)

and
Qv = Qstd

std ( p 2 , T2 , Z 2 )

(4)

The compressibility Z2 is calculated with the Redlich-Kwong equation of state. Equations 2 , 3 and 4 mean that the discharge pressure change depends on the capability of the valve to release flow at a higher rate than the flow coming from the compressor. It also shows that the pressure reduction for a given valve will be slower for larger pipe volumes (V). Kurz and White [8] have shown the validity of the simplified model. The discharge pressure p2 in Eq. 2 is a function of the compressor operating point, expressed by:
p2 k 1 h (Q , N ) SG k 1 = 1 + p1 k 287 ZT1
k

h Q Q = + + N2 N N

(5 a, b)

Alternatively, a lookup table, showing the head-flow relationship for the compressor, can be used.
3

Qstd is the standard flow. Fp is the piping geometry factor. It is usually not known and can be assumed to be 1. The pressure is assumed to be constant in the entire pipe volume. It is thus the same just upstream of the valve and at the discharge pressure of the compressor.

The above relationship can be used for any positive flow. If the compressor exhibits reverse flow, it can be modeled as an increasing flow backwards through the compressor with time across the surge limit, and a 1% hysteresis in recovery head (head must decrease 1% below the surge limit before the compressor will recover from surge). When recovery head is reached flow through the compressor begins decreasing, through zero and then flow begins increasing in the forward direction. This is purely for esthetics. No attempt is made to accurately represent what happens after the compressor stalls. It should be noted that this somewhat crude formulation suffices for the present study because we want to determine whether the compressor will go into surge at ESD or not. The post-surge behavior is, thus, not important and is only introduced to keep the numerical model stable. The behavior of the compressor during ESD is governed by two effects. The inertia of the system consisting of the compressor, coupling and power turbine (and gearbox where applicable) is counteracted by the torque (T) transferred into the fluid by the compressor (mechanical losses are neglected). The balance of forces thus yields:
T = 2 J dN dt

(6)

Knowing the inertia (J) of the system and measuring the speed variation with time during rundown yields the torque and, thus, the power transferred to the gas:
P = T N 2 = (2 ) J N
2

dN dt

(7)

If the rundown would follow through similar operating points, then P~N3, which would lead to a rundown behavior of:
dN k N2 k = N2dN = dt + c N(t) = 2 2 dt J (2 ) J(2 ) 1 (8) 1 k t 2 Nt=0 J(2 )

Regarding the proportionality factor (k) for power and speed, this factor is fairly constant, no matter where on the operating map the rundown event starts. Thus, the rate of deceleration, which is approximately determined by the inertia and the proportionality factor, is fairly independent of the operating point of the compressor when the shutdown occurred; i.e., the time constant (dN/dt(t=0)) for the rundown event is proportional to k/J. However, the higher the surge margin is at the moment of the trip, the more head increase can be achieved by the compressor at constant speed. The model described above, which contains and accounts for the primary physical features of the discharge system, can be used to determine whether the combination of discharge volume and valve size can prevent the compressor from surge during an ESD. It allows the two important design parameters to be easily varied to avoid surge during ESD. The surge valve size and opening speed can be increased for a given discharge volume or the maximum allowable discharge volume for a given configuration of valves and compressor characteristic can be limited. The second method, which has the advantage of being more transparent for the station design, is used here.

10

The simplified model calculates the maximum discharge volume where the head across the compressor can be reduced by half in one second, based on the assumption that this reflects the speed decay during an ESD as outlined above. Therefore, the calculation of the instant compressor speed is replaced by a fixed, presumed to be known, deceleration rate. The assumption is made that the power turbine and compressor will lose about 20 to 30 % speed in the first second of deceleration. This is, for example, confirmed by data from Kurz and White [8] showing a 30% speed reduction of a gas turbine driven compressor set, and Bakken et. al. [9], where the gas turbine driven configurations lost about 20-to-25% speed in the first second, while the electric motor driven configuration lost 30% speed in the first second. As a result of the loss of 25% speed, the head the compressor can produce at the surge line is about 56% lower than at the initial speed, if the fan law is applied. A further assumption is made about the operating point to be the design point at the instant of the ESD. Any ESD is initiated by the control system. Various delays in the system are caused by the time for the fuel valve to shut completely, the time until the hot pressurized gas supply to the power turbine seizes, and the opening time of the recycle valve. ESD data show it is a valid assumption that the surge control valve reaches full open simultaneously with the beginning of deceleration of the power turbine / compressor. This is the starting time (T0) for the model. Usually, the suction volume (no check valve) is more than three orders of magnitude greater than the discharge volume and is therefore considered at a constant pressure. The general idea is now to consider only the mass flow into the piping volume (from the compressor) and the mass flow leaving this volume through the recycle valve. Since the gas mass in the piping volume determines the density and, thus, the pressure in the gas, we can for any instant see whether the head required to deliver gas at the pressure in the pipe volume exceeds the maximum head that the compressor can produce at this instant. Only if the compressor is always capable of making more head than required can surge be avoided. A further conceptual simplification can be made by splitting the flow coefficient of the recycle valve (cv) into a part that is necessary to release the flow at the steady-state operating point of the compressor (cv,ss) and the part that is actually available to reduce the pressure in the piping volume (cv,avail). The first stream and, thus, cv,ss of the valve necessary to cover it are known. Also known is the cv rating of the valve. Thus, the flow portion that can effectively reduce the backpressure is the determined by the difference:

c v ,avail = c v c v , ss

(9)

The model is run at constant temperature. Most of the compressor systems modeled contain aftercoolers. The thermal capacity of the cooler and the piping are much larger than the thermal capacity of the gas; thus, the gas temperature changes are negligible within the first second. The flow calculated above in each step of the iteration is then subtracted from the gas contained in the discharge and a new pressure in the pipe volume is calculated. The calculation yields the maximum allowable piping volume for the set parameters that will not cause surge at ESD. Application

11

The simplified model lends itself to iteratively determine the maximum allowable discharge volume for a given valve configuration. This is important, because the valve size can be determined early in the project phase, based on the considerations outlined in Figure 5. With a known valve configuration, the station designer can be provided with the maximum volume of piping and coolers between the compressor and the check valve, that allows to avoid surge during ESDs. The calculation requires to specify the head-flow-speed relationship of the compressor, and the definition of the surge line as a function of either compressor speed, compressor head or compressor flow. Further, the valve needs to be described by its maximum capacity (cv), as well as by its capacity as a function of valve travel (cv~(travel)3 for a ball valve, cv~(travel)2 for a globe valve), and the opening behavior, including the delay. The train deceleration is modeled as shown in Figure 8, based on test data and the calculations outlined by eq. 7 and 8. The discharge check valve is assumed to be closed as soon as the recycle flow exceeds the compressor flow, i.e. the depressurization begins. The calculation procedure is started by initiating the deceleration of the train and the valve opening. For each time step, the compressor head and flow (based on speed and system pressures), and the flow through the valve (based on system pressures and valve opening) are calculated. The mass of the gas trapped between surge valve and compressor discharge is subsequently determined, yielding a new discharge pressure. If surge occurs (i.e if the flow drops below the flow at the surge line), the backwards flow through the compressor is assumed to increase with time in surge, with a recovery once the required head drops 1% below the head at surge. The modeling of the backwards flow is not critical (and it is only made to avoid numerical instabilities), because the only information that is expected from the model is whether the compressor will surge for the given configuration, or not. Figures 9 and 10 show typical results of these simulations. In Figure 11, the discharge volume is small enough , and while the actual flow of the compressor approaches the minimum allowable flow (surge flow) at about 500ms after the initiation of the ESD , surge can be avoided.

12

Figure 8: Valve opening characteristic and power turbine deceleration

Figure 9: Actual Flow and Flow at the surge line during ESD. Recycle valve sizing and discharge volume allow for ESD without surge

13

Figure 10: Actual Flow and Flow at the surge line during ESD. Compressor is in surge at 0.6 s after trip. The simulation results for another configuration (Figure 10) show a surge event at about 700ms after the initiation of the ESD. For this configuration, either the valve size has to be increased, or the discharge volume has to be reduced, to avoid compressor surge during an emergency shutdown. Start-up Considerations The design of the anti surge and recycle system also impacts the start-up of the station. Particular attention has to be given to the capability to start up the station without having to abort the start due to conditions where allowable operating conditions are exceeded. Problems may arise from the fact that the compressor may spend a certain amount of time recycling gas, until sufficient discharge pressure is produced to open the discharge check valve (Figure 3), and gas is flowing into the pipeline. Virtually all of the mechanical energy absorbed by the compressor is converted into heat in the discharged gas. In an un-cooled recycle system, this heat is recycled into the compressor suction and then more energy added to it. A cubic foot of natural gas at 600 psi weighs about 2 lb (depending on composition). The specific heat of natural gas is about 0.5 Btu/lb (again depending on composition). 1 Btu/sec equals 1.416 hp. If the recycle system contains 1000 cubic feet, there is a ton of gas in it. 1416 hp will raise the temperature of the gas about 1 degree per second. This approximates what happens with 100% recycle. At 100% recycle, eventually this will lead to overheating at the compressor discharge. The problem usually occurs when there is a long period between the initial rotation of the compressor and overcoming the pressure downstream of the check valve. Low pressure ratio compressors often do not require aftercoolers. There are three primary strategies that can be employed to avoid overheating the un-cooled compressor during start-up:

14

-accelerate quickly -delay hot gas re-entering the compressor -throttled recycle. Compressors without cooling must be accelerated and placed on line quickly to avoid overheating. Un-cooled compressor sets cannot be started and accelerated to idle. They must be accelerated quickly through the point where the discharge check valve opens and the recycle valve closes. If acceleration slows when the discharge pressure is met, and recycle valve closes slowly, a shutdown may still occur. Often standard start sequences are very conservative and can be shortened to reduce the time it takes to get a compressor on line. Extending the length of the recycle line downstream of the recycle valve increases the total volume of gas in the recycle system. This reduces the heat buildup rate by delaying when the hot gas from the compressor discharge reaches the suction. Some heat will be radiated through the pipe walls. If the outlet is far upstream into a flowing suction header, dilution will occur. Figure 11 outlines a solution to a rather difficult starting problem for a compressor station without after cooling capacity: To start the first unit is relatively easy, because there is virtually no pressure differential across the main line check valve, and therefore the unit check valve will open almost immediately, allowing the flow of compressed gas into the pipeline. However, if one additional unit is to be started, the station already operates at a considerable pressure ratio, and therefore the unit check valve will not open until the pressure ratio of the starting unit exceeds the station pressure ratio. Ordinarily the unit would invariably shut down on high temperature before this can be achieved. By routing the recycle line into the common station header, the heat from the unit coming on line is mixed with the station suction flow. This equalizes the inlet temperature of all compressors; higher for the compressors already on line, lower for the compressor coming on line. With this arrangement overheating of a compressor coming on line is nearly always avoided.

15

Figure 11 a and b: Original (a) and improved (b) station layout. The original layout features individual recycle lines, while the improve layout allows to feed the recycled flow into the common station header.

16

Figure 12:Temperature build up in a recycle loop, consisting of 3000 ft of 24in pipe. Compressor shutdown levels would be reached after about 20 minutes (assuming a shut down setpoint of 350 F).

Figure 13a,b,c: Temperature rise in the recycle loop during startup at a function of (a)Power turbine and compressor speed, (b) gas producer speed, (c) time (in minutes). The power turbine starts to turn at about 75% gas producer speed, at which point the temperature starts to rise. After the discharge check valve opens (at 0.2 minutes after the compressor starts to rotate), 95% gas producer speed and 70% power turbine speed), the temperature drops rapidly.

17

Figure 12 shows the problem of a conventional system without after-cooling: The temperature in the recycle line starts to rise, eventually leading to a shut down of the compressor once the set temperature limits are reached. Figure 13 outlines the start-up event with the revised system. The power turbine and the compressor start to rotate once the gas producer provides sufficient power. Subsequently, the gas temperature rises, but, because the discharge pressure required to open the check valve is reached fast enough, overheating can be avoided. Further analysis of the start-up problem indicates the advantage of throttling the recycle valve, rather than starting the unit with the recycle valve fully open. Figure 5 illustrates this: At 70% open setting, the startup of the compressor is relatively closer to the surge line than at 100% open setting. For any given speed, the power requirement of the compressor is lower when it is closer to surge than when it is farther in choke. Therefore, for a given amount of available power, the start is quicker if the compressor operates closer to surge. If the rate of acceleration is quicker, the heat input into the system is lower. Actively modulating the surge during start-up is virtually impossible as the parameters defining the surge limit of the compressor are too low to be practically measured. Returning to Figure 5 the surge limit of a compressor matches well with a fixed travel (constant Cv ) line for a recycle valve. As such, a compressor can be started with a fixed recycle valve position

Conclusions and Recommendations A model to simulate shutdown events was developed and used to define simpler rules that help with proper sizing of upstream and downstream piping systems, as well as the necessary control elements. The model coincides well with the data, particularly with regards to proper prediction of surge events. The inaccuracies and limitations inherent in the current model are only problematic if the entire rundown process needs to be described. The key variables -compressor characteristic -valve size and characteristic -discharge volume are included into a simple, yet sufficiently accurate, modeling procedure. Furthermore, the related issue of recycle loop sizing and design considering the requirement to bring units on-line without overheating the system are discussed. Acknowledgements The authors would like to acknowledge the contributions of Daryl D. Legrand and Roland Kaiser, both with Solar Turbines Incorporated. References [1] Epstein, A.H., Ffowcs Williams, J.E., and Greitzer, E.M., Active Suppression of Compressor Instabilities, AIAA-86-1994, 1994. [2] Blanchini, F., Giannatasio, P., Micheli, D., and Pinamonti, P., Experimental Evaluation of a High-Gain Control for Compressor Surge Suppression, ASME 2001-GT-0570, 2001. [3] Arnulfi, G.L., Giannatasio, P., Micheli, D., and Pinamonti, P., An Innovative Control of Surge in Industrial Compression Systems, ASME 2000-GT-352, 2000.
18

[4] McKee, R.J., and Deffenbaugh, D., Factors that Affect Surge Precursors in Centrifugal Compressors, Proc. GMRC Gas Machinery Conference, Salt Lake City, Utah, 2003. [5] Day, I.J., Axial Compressor Performance During Surge, Proc.10th International Symposium on Air Breathing Engines, Nottingham, UK,1991, pp. 927-934. [6] Sentz, R.H., The Analysis of Surge, Texas A&M Turbomachinery Symposium, 1980. [7] ANSI/ISA S75.01, Flow Equations for Sizing Control Valves, 1995. [8] Kurz, R., White, R.C., 2004, Surge Avoidance in Gas Compression Systems, TransASME JTurbo Vol.126, pp.501-506. [9] Bakken, L.E., Bjorge, T., Bradley, T.M., and Smith, N., Validation of Compressor Transient Behav ior, ASME GT-2002-30279, 2002.

19