Effects of Asymmetric Tip Clearance on Compressor Performance and Stability by Thomas S. Wong B.S. Aerospace Engineering Illinois Institute of Technology, Chicago, Illinois, 1994 SUBMITTED TO THE DEPARTMENT OF AERONAUTICS AND ASTRONAUTICS IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN AERONAUTICS AND ASTRONAUTICS AT THE MASSACHUSETTS INSTITUTE OF TECHNOLOGY JUNE 1996 © Massachusetts Institute of Technology, 1996. All rights reserved. - Z?~ Signature of Author Depa eof Aeronautics and Astronautics /7 May 17, 1996 Certified by f-I Professor Edrd M. Greitzer, Thesis Advisor H.N. Slater Professor of Aeronautics and Astronautics Certified b Dr. Choon S. Tan, Thesis Advisor Principal Research Engineer Accepted by Professor Harold Y. Wachman Professor of Aeronautics and Astronautics Chairman Departmental Graduate Committee OF TECHNOLOGY JUN 111996 AerO Effects of Asymmetric Tip Clearance on Compressor Performance and Stability by Thomas S. Wong Submitted to the Department of Aeronautics and Astronautics on May 17, 1996 in Partial Fulfillment of the requirements for the degree of Master of Science in Aeronautics and Astronautics ABSTRACT An experimental study has been carried out to assess effects of asymmetric tip clearance on compressor performance and stability. The experiments were designed using a theoretical model of compressor behavior with asymmetric tip clearance and were conducted in a four stage, low speed compressor at the General Electric Aerodynamic Research Laboratory. It was found that the reduction in stability margin with asymmetric clearance is significantly more severe than that estimated based on an average clearance. Specifically, the stall point can be closer to that obtained with axisymmetric flow at the largest clearance of the asymmetry than to that obtained with the average clearance. Further, the circumferential harmonic content (length scale) of the asymmetry is an important factor in determining the amount of the stability margin reduction. For the same average clearance, an asymmetry with wavelength equal to the compressor circumference showed a reduction in peak pressure rise at stall one-and-a-half times as large as that for an asymmetry with half the wavelength, both having the same average clearance. The theoretical model was shown to capture the loss in steady-state pressure rise and stable operating range associated with the clearance asymmetry. The model also reproduced the trend of increasing velocity asymmetry as compressor axial velocity parameter is reduced. Finally, the time resolved data showed that the stability limiting process can be linked to the unsteady structure of the calculated disturbance modes in the compressor. Thesis Supervisor: Edward M. Greitzer Title: H.N. Slater Professor of Aeronautics and Astronautics Thesis Supervisor: Dr. Choon S. Tan Title: Principal Research Engineer Acknowledgments I would like to thank everyone involved in this research project. In particular, I wish to thank: Professor Edward Greitzer and Dr. Choon Tan for their support, guidance, experience, and encouragement that were greatly needed in the completion of the project and this thesis. Dr. Martin Graf whose theoretical contribution lead to this experiment. Dr. David Wisler for his support that made the experiment at GE Aerodynamic Research Laboratory possible. Dr. Hyoun Shin for his assistance and guidance at GE. His expertise in flow measurements was invaluable to the success of this experiment. Mr. Don Menner for his assistance in preparing the Low Speed Research Compressor for testing and his support on data reduction. Dr. Brent Beacher, Dr. Dave Halstead, and Mr. Bill Groll for their knowledge and experience on the LSRC. Professor Frank Marble and Professor Alan Epstein for their insightful discussions and suggestions. Roommate and pal John Brookfield, all my friends, and the staff at the MIT Gas Turbine Laboratory for their help and friendship. I would especially like to thank my parents and sister for their unconditional love and constant encouragement which kept me going at the worse of times. This research was supported by the Air Force Office of Scientific Research, under Grant Number F49620-93-1-0015, Dr. J. M. McMichael, Program Manager, and General Electric Aircraft Engine Company, D.C.Wisler, Program Manager. Table of Contents Chapter 1 Introduction 1.1 Background ............................................................. 1.2 Overall Background.................................................... 1.2.1 Tip Clearance Effects on Compressor Performance........ 1.2.2 Asymmetric Clearance........................................ 1.3 Thesis 1.4 Scope of the Thesis.................................................... Chapter 2 Objective....................................................... Formulation of the Compressor Model with Asymmetric Tip Clearance 13 2.1 A ssum ptions............................................................ 13 2.2 Brief Description of the Compressor Model........................ 14 2.2.1 14 Steady Flow Field............................................. 2 .2 .2 S tab ility ......................................................... 17 2.3 Overall Performance Calculations..................................... 17 2.4 S um m ary ................................................................ 18 Experimental Setup 22 3.1 Low Speed Research Compressor Rig Description................ 22 3.2 Experimental Configurations.......................................... 23 3.3 Instrum entation......................................................... 24 Chapter 3 3.4 Chapter 4 4.1 Error Analysis....................................................... Experimental Results and Assessment of Model 29 Steady State Performance.............................................. 29 4.1.1 Pressure R ise................................................... 29 4.1.1.1 Axisymmetric Tip Clearance....................... 29 4.1.1.2 Non-Axisymmetric Tip Clearance................. 30 4.1.2 E fficiency ....................................................... 31 4.1.2.1 4.2 Chapter 5 Reference Axisymmetric Tip Clearance....................... 31 4.1.2.2 Non-Axisymmetric Tip Clearance................. 31 4.1.2 Circumferential Flow Distribution in the Compressor..... 31 Unsteady Compressor Response 32 4.2.1 Unsteady flow in compressors............................... 32 4.2.2 Stall inception.................................................. 33 Summary and Conclusions 45 47 List of Figures Chapter 1 Page 1.1 Effects of tip clearance on peak pressure rise. [Smith (1958)] 11 1.2 Effects of tip clearance on stalling pressure rise coefficient [Koch (1981)]. 11 Effects of increasing tip clearance on overall compressor performance [Wisler (19840]. 12 1.3 Chapter 2 2.1 Pressure rise characteristics of a three-stage compressor with axisymmetric tip clearance of 1% to 5% of chord. 19 2.2 Local compressor operating points around the annulus. 19 2.3 Steady-state circumferential flow coefficient variation. 20 2.4 First mode wave envelope 20 2.5 Stall points for axisymmetric and non-axisymmetric compressor. 21 2.6 Stall points for clearance asymmetry of cos (0), cos (20) and cos (40). 21 Chapter 3 3.1 Schematic of the General Electric Low Speed Research Compressor. 27 3.2 Casing panel positions for a) the one-lobed configuration and b) the two-lobed configuration. 28 Axial location of the instrumentation 28 3.3 Chapter 4 4.1 Compressor pressure rise characteristics for axisymmetric tip clearance. 35 4.2 Compressor pressure rise characteristics with axisymmetric tip clearance. 35 4.3a Change in peak pressure rise relative to the baseline configuration. 36 4.3b Change in stalling flow coefficient relative to the baseline configuration. 36 4.4 Compressor efficiency for axisymmetric tip clearance. 37 4.5 Compressor efficiency with asymmetric tip clearance. 37 4.6 Circumferential steady flow coefficient variation for the one-lobed configuration. 38 Circumferential steady flow coefficient variation for the two-lobed configuration. 39 4.8 Harmonic amplitudes of time-mean flow coefficient distribution. 40 4.9 First mode wave envelope. 41 4.7 4.10 Mode shape harmonic content. 41 4.11 Normalized RMS static pressure fluctuation. 42 4.12 Pressure transducer traces for axisymmetric baseline configuration. 42 4.13 Pressure transducer traces for one-lobed configuration. 43 4.14 Pressure transducer traces for two-lobed configuration. 43 4.15 Histogram of stall inception locations. 44 CHAPTER 1 INTRODUCTION 1.1 Background Tip clearance is the gap between rotor blade tips and the casing of a turbomachine. There is a large body of experimental information which shows that increases in tip clearance have a significant adverse effect on axial compressor performance and stability. An operational problem for gas turbine engines connected with can be the development of a clearance asymmetry due to such effects as casing distortion or rotor eccentricity. Almost all the studies of tip clearance in the literature, however, have focused on the situation with axisymmetric clearance, and the impact of such asymmetry is much less understood. This thesis thus addresses the effect of non-axisymmetric tip clearance on compressor performance and stability. Clearance asymmetry can be separated into two categories. The first is stationary asymmetry which can be caused by relative displacement (off-center) between rotor and casing or by deformation of the casing. The second category is rotating asymmetry, which can be generated by non-uniform blade height in the rotor, by a whirling rotor, or by a socalled "bowed rotor." The focus of the present investigation will be the first of these stationary asymmetry. 1.2 Overall Background 1.2.1 Tip Clearance Effects on Compressor Performance As stated, a number of investigations have demonstrated the effect of tip clearance on axial compressors performance and stability. Examples are the data shown in the studies by Smith (1958) and Koch (1981) which are given in Figure (1.1) and Figure (1.2). The curves shown in the figures illustrate the decrease in stalling pressure rise associated with the increase in tip clearance. Increasing the mean level of clearance causes a decrease in the range of compressor stable operation. Tip clearance also affects the peak efficiency performance of a compressor as seen in Figure (1.3) [Wisler (1984)] which shows pressure rise and efficiency characteristics for a multistage compressor run at two different clearances. The data demonstrate the decrease in both pressure rise capability and efficiency due to increased clearance. 1.2.2 Asymmetric Clearance This thesis is complementary to a previous study by Graf (1996) which described the formulation of a model to calculate the steady and unsteady behavior of a compressor with asymmetric tip clearance. The results of that study indicated that asymmetric tip clearance could have a significant impact on compressor pressure rise capability and stall margin. Also, changes in behavior due to asymmetric tip clearance were linked to the sensitivity of the compressor to axisymmetric clearances. Parametric studies showed that the severity of the effect depends strongly on the wavelength of the asymmetry. Data for an asymmetric clearance are given by Freeman (1985) but no information is provided about the test details. 1.3 Thesis Objective The objective of the present study is to provide steady and unsteady performance data which can not only serve as a guide to the magnitude of the effects expected with asymmetric tip clearance but can also help evaluate the analytical model. The experiment was carried out at the Low Speed Research Compressor (LSRC) at the Aerodynamic Research Laboratory (ARL) of General Electric Aircraft Engine Division. There were two phases to the experiment. The first was aimed at obtaining the performance and stability characteristics of the compressor at different axisymmetric clearance levels. Those were needed for input to the model. The second set of tests then examined the performance of the compressor with two different casing asymmetries. 1.4 Scope of the Thesis In Chapter 2, the asymmetric tip clearance model is briefly described and the trends of Grafs results (Graf, 1996), which served as a guide to the design of the experiment, are illustrated and discussed. The experimental facility and setup are described in Chapter 3. In Chapter 4, the results of the experiment are presented and discussed. The summary and conclusions are given in Chapter 5. 130 120 110 100 90 80 70 0 2 1 5 6 Clearance / Chord, /Ct, % 3 4 7 8 9 Figure 1.1. Effects of tip clearance on peak pressure rise. [Smith (1958)] 1.25 1.20 1.15 1.10 1.05 1.00 0.95 0.90 0.85 0.05 0.10 Tip Clearance: dg 0.15 Figure 1.2. Effects of tip clearance on stalling pressure rise coefficient. [Koch (1981) / \ 90 \ 89 87 / \ / 1.5 pt l II 88 Nominal Rotor Tip Clearance e/h = 1.38%/ Increased Rotor Tip Clearance 85 /h = 2.80% Rotor B/Stator B Four-Stage Configurations DA 0.30 0.40 0.50 Flow Coefficient, po Figure 1.3. Effects of increasing tip clearance on overall compressor efficiency. [Wisler (1984)] CHAPTER 2 FORMULATION OF COMPRESSOR MODEL WITH ASYMMETRIC TIP CLEARANCE 2.1 Assumptions The asymmetric tip clearance model (Graf, 1996) described in this Chapter is an extension of the approach used by Hynes and Greitzer (1987) to assess effects of circumferentially non-uniform inlet flow on compressor performance and stability. It is assumed that the compressor has a high hub-to-tip ratio so that the flow through it can be considered as two dimensional (the variations are in the axial and circumferential directions). The relevant Mach numbers and the compressor pressure rise are assumed to be low enough that compressibility effects can be neglected. A simple accounting for blade row viscous losses, similar to that described by Haynes et al. (1993), is also included. From the outset the model is not designed to capture the detailed fluid mechanics of the clearance flow but is aimed at giving useful information about overall compressor and compression system behavior, including stability. Steady state performance characteristics of the compressor with difference axisymmetric clearances are input to the model. To illustrate the manner in which these are used, consider Figure (2.1) which shows three steady state total-to-static pressure rise characteristics of a compressor with axisymmetric clearance values ranging from "tight" (say 1% of rotor chord length) to "loose" (say 5% of rotor chord length). The characteristic with the highest pressure rise capability corresponds to the tightest clearance, and the lowest curve belongs to the largest clearance. The curve in the middle would correspond to performance of the compressor with nominal clearance of 3% chord. The view taken is to regard a non-axisymmetric situation as composed of a distribution of different "compressors in parallel" around the annulus, each with a characteristic that corresponds to the local tip clearance level as shown in Figure (2.2). In the figure, the locus of operating points around the annulus for a compressor with a sinusoidal clearance asymmetry is shown as the heavy dashed lines along with the three characteristics previously shown in Figure (2.1). The clearance asymmetry implies that the compressor operates over a range of flow coefficients. The local slope of the compressor pressure rise characteristic is also varying around the circumference. It is, therefore, possible for a portion of annulus to operate on a positively sloped region of the local characteristic, a situation which is associated with a tendency toward instability. 2.2 Brief Description of the Compressor Model 2.2.1 Compressor Steady Flow Field A full description of the model has been given by Graf (1996). In this section, we present only a summary of some of the features. An implicit equation for asymmetric steady flow through a compressor has been given by Hynes and Greitzer (1987) as Pexit - Ptinlet d (2.1) pU 2 dO (2.1) In equation (2.1) j~(O) is the pressure rise characteristic with axisymmetric flow and 4 is the steady state flow coefficient. X is an unsteady blade inertia parameter defined as follows bx IUCOS2,) ucos2 , U: U With a non-axisymmetric tip clearance, the compressor pressure rise characteristics around the annulus are functions of both local flow coefficient and local clearance (the local pressure rise characteristics depend on the tip clearance distribution around the circumference). The steady state performance of the compressor is thus given by Graf (1996) as d= 4~ dO Pexit - Ptinlet pU 2 (2.2) where E is the tip clearance normalized here by rotor chord length. The asymmetric steady flow is obtained by the solving equation (2.2) along with an equation to determine the circumferential average flow through the compression system. This is a non-linear description of the asymmetric steady flow around the annulus. Although it is to be emphasized that Grafs analysis for the background flow is non-linear, we present here a simple linear description to illustrate some of the basic parameter dependence. Linearizing equation (2.2) yields the following expression for the steady flow variations 8Pexit - pU 8Ptinlet - I8 = o 2 p U2 + 8 - ,d 88 O dO (2.3) (2.3 (E We can expand the axial velocity parameter variations and the clearance variations in a Fourier series 80 = I Ane ino (2.4) 8E = _ Bne ine (2.5) There are no losses upstream of the compressor, so the total pressure is circumferentially uniform at the inlet. Further because the flow is assumed to exit the last stator, exit static pressure of the compressor is also uniform (Lonley and Greitzer, 1992). This exit condition is verified experimentally from the results of this investigation. Using these inlet and exit conditions along with the equation (2.4) and (2.5), the variation in axial velocity is described by 54 = Bne in0 8, inx -v (2.6) The extent of the velocity variations around the annulus thus depends on 1) the harmonic content of the clearance variation, 2) the flow inertia parameter X, 3) the compressor pressure rise sensitivity to axisymmetric clearance 8/8E, and 4) the local slope of the pressure rise characteristics 8V/50. In general, a more negative slope of the compressor characteristic, a larger inertia parameter, a higher harmonic content of the clearance variation, and a decreased sensitivity to clearance in the asymmetric situation will all tend to promote smaller variations in axial velocity parameter, and thus, presumably, less of an adverse effect on overall performance and stability. Figure (2.3) gives an example of the steady flow at the compressor face obtained by solving equation (2.2) and the mean flow equation of the compression system. The figure illustrates the non-uniformity in steady state inlet flow to the compressor, due to asymmetric tip clearance, which has developed from uniform flow far upstream of the compressor. 2.2.2 Compressor Stability* Once the distorted steady flow through the compressor is obtained, the stability of the compression system is assessed by adding a small unsteady perturbation to the steady flow and determining whether it grows or decays. This involves solving the linearized equations of the compression system for small unsteady perturbations. The details of their development can be found in (Graf, 1996). The loss model employed in the asymmetric clearance model and a summary of the compression system equations can also be found in Appendix A of Haynes et al. (1993). Stability of the compression system is defined by the eigenvalues of the linearized equations. The mode shape of the first eigenmode (least stable) for the three-stage compressor example is shown in Figure (2.4). The wave envelope of this particular mode represents the axial velocity perturbation wave at neutral stability, at different times. While traveling around the annulus, the fluctuations grow in certain regions and are damped in other regions as indicated. The growth or decay in a particular region depends on the location of the local operating point on its pressure rise characteristic (i.e. positively sloping or negatively sloping). The magnitude of the flow fluctuations is thus not constant around the annulus. 2.3 Overall Performance Calculations To assess effects of asymmetric tip clearance on compressor stall margin, the stall point of the example compressor with axisymmetric clearance can be compared to that of the compressor with asymmetric clearance. Both compressors have the same average tip * Note that the results of this and subsequent sections are obtained using the non-linear model of Graf (1996) clearance. Figure (2.5) shows the stall points for the two configurations; the clearance asymmetry causes a 3.7% reduction in peak pressure rise. The case illustrated is for a sinusoidal tip clearance distribution with wavelength equals to the circumference of the compressor (cos 0). Two other cases with wavelength of a half (cos 20) and a quarter (cos 40) of the compressor circumference are also shown. Their stall points are given in Figure (2.6) along with those of the baseline and the cos 0 clearance distribution. As the wavelength of the asymmetry is reduced, the reduction in stall margin becomes smaller because of the increased importance of unsteadiness which causes the circumferential flow variation to decrease in magnitude. The locus of operating points around the annulus [shown in Figure (2.2)] thus reduces in width. As a result, the performance of the compressor with "short wavelength" clearance asymmetry approaches that of the axisymmetric baseline compressor. 2.4 Summary The effect of asymmetric clearance is modeled using steady-state compressor design parameters and system dynamic parameters. Non-axisymmetric clearance distribution causes a circumferential distortion of the steady flow through the compressor and a reduction in stability. The predicted decrease in stability of a compressor with asymmetric clearance is more severe than based on average clearance. 0.9 0.8 0.7 0.6 Loose Clearance 0.5 0.4 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 4, Figure 2.1. Pressure rise characteristics of a three-stage compressor with axisymmetric tip clearance of 1% to 5% of chord. 10.95 0.9 0.85 r 0.8 0.75 0.7 0.65 0.6 - 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0 Figure 2.2. Local compressor operating points around the annulus. 0.7 0.7 0.65 S0.6 Z 0 0.55 0 o 0.45 0 . 0.45 . 0.4 0.35 .... 0.3 0.5 0 1 1.5 2 Circumferential Location (PI) Figure 2.3. Steady-state circumferential flow coefficient variation. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 (Theta. - Pi) Figure 2.4. First mode wave envelope 1.8 2 0.84 Stability Point with Axisymmetric 0.82 0.8 0.78 v / 0.76 Stability Point with Non-Axisymmetric Clearance 0.74 0.72 0.7 - I 0.35 I 0.45 0.65 0.55 Figure 2.5. Stall points for axisymmetric and non-axisymmetric compressor. 0.84 - 0.82 0.8 0.78 v 0.76 0.74 cos(e) 0.72 0.7 - 0.35 0.45 0.55 0.65 Figure 2.6. Stall points for clearance asymmetry of cos 0, cos 20 and cos 40. CHAPTER 3 DESCRIPTION OF EXPERIMENT 3.1 Low Speed Research Compressor (LSRC) The experiment was performed at the General Electric Low Speed Research Compressor (LSRC). The facility is shown schematically in Figure (3.1). Flow enters from the top through the cylindrical filter. A ring of eleven wall static pressure taps located at the flow measurement plane shown in the figure serve to meter the flow rate. The air passes through the four stages and exits to a plenum below the test floor and through a circular throttle plate to atmosphere. The throttle plate moves vertically to change the exit area, regulating the flow through the machine. In the configuration tested, the compressor had a constant casing diameter of 1.524 meter (60 inches) and a hub-to-tip ratio of 0.85. There were four identical blade rows. The casing is comprised of forty-eight Plexiglas windows mounted on steel frames. They were shimmed to provide the non-uniform clearance variation. There is extensive instrumentation associated with the rig which will be referred to as needed. A detailed description of the LSRC can be found in Wisler (1984). The compressor consisted of fifty-three inlet-guide-vanes. For each stage, there are fiftyfour rotor blades and seventy-four stator blades. The blading used in this experiment was representative of modem designs. Chord lengths and stagger angles for the compressor are summarized in Table (3.1). The Reynolds number was 3.5x10 5 based on rotor chord length and mid-span blade velocity. The tip Mach number was 0.2, so that the flow can be considered incompressible. Blade type Chord (inches) Stagger angle (degree) IGV 3.300" -3040' Rotor 3.590" 46050 ' Stator 3.114" 13040 ' Table 3.1. LSRC Blading geometry. 3.2 Experimental Configurations The tip clearance of the compressor was altered to assess performance sensitivity with axisymmetric and non-axisymmetric tip clearance. To alter the clearance, the twelve Plexiglas windows over each stage were shimmed out to the desired clearance values. The shimming was done for all four stages to assess the performance of the entire compressor. In the first set of experiments, the first characteristics of the compressor with different axisymmetric clearance levels were attained; these serve as input to the analytical model. Three clearances were tested having 2%, 4%, and 6% of rotor chord. The performance of the 4% chord axisymmetric configuration served as the baseline case to which the performance of the other configurations will be compared. The second set of experiments involved two different asymmetric clearance distributions. The circumferential harmonic content of the clearance variation was predicted to have a significant influence on the severity of the effects of asymmetric clearance and the tests thus included one configuration with a single lobed clearance variation around the circumference and one with two lobes. The two configurations are illustrated schematically in Figure (3.2). Both the one-lobed and two-lobed configurations had an average tip clearance of 4%, the same as the baseline configuration. A summary of the axisymmetric and non-axisymmetric test configurations examined is shown in Table (3.2). Test Average Clearance/chord Configuration Clearance Asymmetry/chord J Wavelength of Asymmetry Axisymmetric: 1 2% 2% N/A N/A 2 4% (baseline) 4% N/A N/A 3 6% 6% N/A N/A Non-axisymmetric: 4 One-lobed 4% ±2% circumference 5 Two-lobed 4% ±2% one-half circumference Table 3.2. Summary of test cases. 3.3 Instrumentation The overall compressor parameters of interest include compressor pressure rise and efficiency. The pressure rise characteristic is presented as the overall total-to-static pressure rise of the four stages [Wisler (1984) gives details of the instrumentation]. The efficiency, which is obtained from overall torque and pressure rise, is also a value for the four stages. In addition to overall performance parameters, the time mean flow field associated with asymmetric clearance is also of interest. At the inlet of the compressor, steady flow measurements were obtained by eleven Kiel probes and eleven static pressure taps upstream of the IGV and evenly spaced around the circumference. Data were acquired at three points on the compressor characteristic corresponding to the wide open, design, and near stall. The exit static pressure was measured at the fourth stage exit by eleven static taps. It was circumferentially uniform as expected (uniform to within 0.6% of the mean dynamic pressure). Unsteady compressor pressure measurement were also obtained using eight evenly spaced Kulite high-response pressure transducers mounted on the casing over the first-stage rotor. The transducers had a response up to 20,000 Hz. The sampling frequency used is 5000 Hz with anti-aliasing filter settings at 1000 Hz. In comparison the compressor has a rotor frequency of less than 15 Hz. Table (3.3) is a summary of all the instrumentation used. Station# Location Instrumentation Parameter 1 1/2 chord in front of IGV 11 Kiel Probes (steady state) Pt0 1 1/2 chord in front of IGV 11 casing static pressure taps (steady state) P1 2 1/2 chord of 1st stage rotor 8 Kulite high-response pressure transducer P2 3 1/2 chord behind 4th stage rotor 11 casing static pressure taps (steady state) P3 Rig standard Pressure rise, flow shaft speed, power i 0, '1 Table 3.3. Summary of instrumentation. 3.4 Error Analysis Pressure Coefficient, 1.: The instrumentation for pressure coefficient measurements is accurate to within 0.15 percent. Day-to-day variation was observed to have a precision of 0.16 percent. The overall error, calculated using Root-Sum-Square Method, is 0.22 percent. Flow Coefficient, (: The instrumentation for flow coefficient measurements are accurate to within 0.15 percent with day-to-day variation of 0.36 percent. The overall error is 0.39 percent. Flow Coefficient Variation, &: The instrumentation error of variation in flow coefficient is 0.7 percent with day-to-day variation of 0.5 percent. The overall error in flow coefficient variation is 0.73 percent of mean. Efficiency, 1: Efficiency measurements are accurate to within 0.15 point. Flow Direedoc Flow Straightening Vanes Bellmont Inlet Screen Figure 3.1. Schematic of the General Electric Low Speed Research Compressor. . . ,- . 6% chord \ . 4% chord .. 2% chord a) One-Lobed 360 degree 0 degree - -6% \ /\-4% - chord chord - 2% chord b) Two-Lobed Figure 3.2. Casing panel positions for a) the one-lobed configuration and b) the two-lobed configuration *1 *I 1-1. 6 6 C C S Y- - - Y'I T f '1 Y ? X A A 6 Inlet e ' IGV A A R1 A x x x x S1i I R2 S2 I IR31 IS3 I IR4 I IS4 x x x x A A A 0 Kiel Probes x Exit x x Static Taps A Kulite Transducers Figure 3.3. Axial location of the instrumentation. CHAPTER 4 EXPERIMENTAL RESULTS AND ASSESSMENT OF MODEL 4.1 Steady State Performance Steady state compressor pressure rise characteristics were obtained with both axisymmetric and non-axisymmetric tip clearance. The baseline configuration had a 4% clearance/chord and this was also the average clearance of the two non-axisymmetric clearance configurations. The other two axisymmetric configurations had clearance level equal to the maximum and the minimum clearance of the non-axisymmetric configurations. 4.1.1 Pressure Rise 4.1.1.1 Axisymmetric Tip Clearance Figure (4.1) shows the total-to-static pressure rise characteristics of the compressor with different axisymmetric tip clearance values. The increase in tip clearance decreases the pressure rise and increases the stall mass flow. When the tip clearance was increased from 2% to 4% of chord, the compressor was less sensitive to clearance change than for the increase from 4% to 6%. The implication drawn was that the machine was hub critical at tight clearance (Wisler, 1996). 4.1.1.2 Non-Axisymmetric Tip Clearance The compressor measured pressure rise characteristics for non-axisymmetric clearance are shown as the data points in Figure (4.2). The axisymmetric clearance data of Figure (4.1) are given as dashed lines. The figure demonstrates the effects of non-axisymmetric tip clearance. If we compare the 4% axisymmetric data (the middle dash line) with the nonaxisymmetric data (points), we see a degradation in pressure rise and in flow range of the latter. At high flow coefficients (design point or higher, say), the characteristics for asymmetric clearance are close to the baseline characteristics and the largest deviations between the configurations occur at low flow. Also the degradation in compressor performance is reduced when the dominant wavelength of the clearance halved with the two-lobed clearance variation where the behavior more closely approaches that for axisymmetric clearance. The theory is shown as solid lines in the Figure. The model appears to capture both the effect of asymmetry and of reduced frequency in the overall curves. The experimental and calculated stall points also show good agreement. To assess the change in stall margin for the different configurations in more detail, the change in peak pressure coefficient and in stalling flow coefficient, relative to the baseline configuration's stall point, are shown in Figure (4.3a) and Figure (4.3b). As a reference, the left hand side of the figure shows the results for axisymmetric clearance. The alterations in stall conditions for the two asymmetric configurations can be seen to be significant compared to the alterations for the same clearance change in an axisymmetric configuration. The one-lobed experiment shows a decrease in peak pressure rise of more than eight percent compared to that of the baseline configuration. The stall point for this asymmetric configuration is in fact closer to the stall point with the loosest axisymmetric clearance than the average. 4.1.2 Efficiency 4.1.2.1 Axisymmetric Tip Clearance Efficiency characteristics for the three different levels of clearance are shown in Figure (4.4). The efficiency has been normalized by dividing by the peak value at the tightest clearance. The data indicate a roughly two percent change in peak efficiency for 2% chord change in tip clearance. 4.1.2.2 Non-Axisymmetric Tip Clearance Efficiency measurements for the one-lobed and two-lobed non-axisymmetric configurations are shown as the data points in Figure (4.5). The axisymmetric data of Figure (4.4) are given as dashed lines. The peak efficiency for the two asymmetric configurations are close to the that of the axisymmetric configuration so asymmetric clearance has only a small impact on peak efficiency of the compressor. This trend is consistent with that of pressure rise characteristics for asymmetric clearance which show only a small deviation from the baseline at design mass flow. This point will be addressed subsequently. 4.1.3 Circumferential Flow Distribution in the Compressor Figure (4.6) shows the time-mean inlet axial velocity around the circumference of the compressor with the one-lobed clearance variation. As described in Chapter 2, the velocity distortion increases as the flow and the average compressor pressure rise characteristic slope decreases. This can be explained if one notes that the axisymmetric clearance characteristics are far apart near stall and closer together near design and high-flow conditions. Flow field variations are thus greater at low flow than at high flow. The model captures this trend and the shape of the axial flow variations. The circumferential flow distributions for the two-lobed clearance configuration are shown in Figure (4.7). There are two essentially identical lobes around the annulus and the amplitude of the axial velocity variation has decreased. The two-lobed configuration is a more demanding test of the assumptions on which the model is based but there is still good agreement between theory and data. The calculated and experimentally measured amplitudes of the flow variation at design point and stall point are shown in more detail in Figure (4.8) which gives the first and second harmonics of the velocity variation. For the one-lobed clearance distribution, the amplitude of the first and second harmonics of the variation are shown. For the two lobed distribution, only the second harmonic is given. The figures show an increase in amplitude as the overall flow is throttled from design conditions to near stall. They also show the effect of reduced frequency on decreasing the flow variation amplitude from the one-lobed clearance distribution to the two-lobed. 4.2 Unsteady Compressor Response 4.2.1 Unsteady Flow in Compressors Now thus turn to examination of the unsteady flow processes that limit stability. For a compressor with asymmetric clearance, the steady flow field entering the compressor is non-axisymmetric. As the rotor moves through this spatially non-uniform flow, it perceives the distorted flow as an unsteady inlet flow, similar to the situation with circumferential inlet distortion. Also similar to inlet distortion, both experiment and calculation indicate that lower spatial harmonics produce the largest loss in stability. The wave shape of the calculated least stable eigenmode is illustrated in Figure (4.9), which shows the circumferentially traveling wave at different times. Because of clearance asymmetry, the local damping of this traveling wave is different at different circumferential location, so the wave grows at certain locations and decays at others. Figure (4.10) shows the harmonic content of the eigenmode. For this configuration the largest harmonic is the first but others harmonics are also strong. The calculated wave envelope Figure (4.9) indicates that the local amplitude of the velocity fluctuation varies circumferentially. The level of unsteadiness will thus be a strong function of annulus location. Eight equally-spaced, high-response pressure transducers were used to measure the unsteadiness in the compressor. The normalized root-mean-square values of the static pressure fluctuation are shown in the Figure (4.11). The data indicate a widely varying level of fluctuation for the eight locations. The highest level of measured fluctuation coincides with the calculation of the location of the peak level of unsteadiness. 4.2.2 Stall Inception The onset of stall is characterized by a rapid increase in fluctuation amplitude from the growth of an underdamped circumferentially traveling wave. Figure (4.12) shows traces of the normalized unsteady data from the eight pressure transducers from a test of the baseline axisymmetric configuration. The data are taken at times just before and during the development of a stall cell. The horizontal axis is time in units of rotor revolutions. The vertical axis is the circumferential locations of the eight transducer. The time traces are normalized static pressure fluctuation. The origin (defined here as have pressure fluctuation, 8W, of 0.1) of the stall cell can be traced roughly to the location at 250 degrees but the traveling wave shows similar growth over all the annulus. With asymmetric clearance, the unsteady flow is quite different. Figure (4.13) shows normalized pressure transducer data with the one-lobe configuration. The measured level of fluctuation are now much higher at locations near 150 degrees and the stall cell can be traced back to this same location. The cell begins to grow at roughly 150 degrees mark and its amplitude increases until it reaches roughly 200 degrees. From 250 to the 50 degrees however, the amplitude does not increase, and there is growth at some locations and decay at others. This phenomenon is also shown by the model. Data from the two-lobe configuration indicate the same behavior as shown in Figure (4.14). For this configuration, the traveling wave experiences two growth and two decay regions as it travels once around the annulus. Examination of the stall cell inception location from different test runs of the baseline axisymmetric configuration and from the one-lobe asymmetric configuration are also useful. To determine these locations, the inception points are defined by the onset of a prescribed value of the RMS pressure fluctuation amplitude (6 w=0.1). A histogram of the stall cell inception locations is shown in figure (4.15). The compressor with axisymmetric clearance does not appear to have one specific location from where the stall cells emerge, but stall cells in the asymmetric compressor seem to always appear at the location with highest unsteadiness. The model shows that the locations with the highest unsteadiness offers the most favorable environment for the growth of small traveling disturbances. 1.5 1.4 1.3 I 1.2 1.1 -- ----h 0.9 2% chord 4% chord 6% chord 0.8 0.4 0.45 0.5 0.55 0.6 0.65 0.7 Figure 4.1. Compressor pressure rise characteristics for axisymmetric tip clearance. 1.6 1.5 --- 1.4 Two Lobe 1.3 One Lobe 1.2 ------- 2% chord 1.1 ------- 4% chord ----- 6% chord 1 O Experiment, one lobe Theory, one lobe 0.9 A 0.8 0.4 .. 0.45 ',, Experiment, two lobe Theory, two lobe 0.5 0.55 0.6 0.65 0.7 Figure 4.2. Compressor pressure rise characteristics with asymmetric tip clearance. 10 Axisymmetric Clearance Non-axisymmetric Clearance 5 0- -5 , - -2% chord from Nominal % rd m N inal E LSRC Data -10 8 Calculation -15 Figure 4.3a. Change in peak pressure rise relative to the baseline configuration. 15 Axisymmetric Clearance Non-axisymmetric Clearance 10 5 = 0% &rd -5 m inal -2% chord from Nominal Two Lobe One Lobe at 0 LSRC Data E3 Calculation -10 Figure 4.3b. Change in stalling flow coefficient relative to the baseline configuration. 100 99 98 92 --- 4% chord 91 - 6% chord 90 .. 0.4 0.45 0.5 0.55 0.6 0.65 C Figure 4.4. Compressor efficiency for axisymmetric tip clearance. 100 . - 99 1 98 II 97 • ,i,' A 0 A- 10• 0 96 0 k 0 95 94 - 93 ------- 2% chord ' ------- 4% chord 92 ------- 6% chord 91 - 90 ! 0.4 0 Experiment, one lobe A Experiment, two lobe 0.45 0.5 0.55 0.6 0.65 Figure 4.5. Compressor efficiency with asymmetric tip clearance.. 0.7 A Theory - 0.65 -A A Experiment LI A ......iiillll " tX ZX 0.6 0.55 0lillilllllIIIIIII A0 12 , 24 00 3 0 A8 0 0 Ui 0.5 0 8 Tight LIClearance 0j 8 Loose Clearance 0.45 Casing Geometry 0 60 120 180 240 300 360 Circumferential Location (degrees) Figure 4.6. Circumferential steady flow coefficient variation for the one-lobed configuration. 0.7 A Experiment Theory 0.65 0.6 4 0.55 o o M 0.5 0 0 -J ..a Loose Clearance 0.45 Tight Clearance Casing Geometry 0 60 120 180 240 300 360 Circumferential Location (degrees) Figure 4.7. Circumferential steady flow coefficient variation for the two-lobed configuration. 15 One Lobe Two Lobe Near Stall 0 Cu U Experiment d.O E Calculation Design • = St St CPoint 5 Design NPoint Near Stall o Design Point 7 0 I~ 1st Harmonic L 2nd Harmonic 2nd Harmonic Figure 4.8. Harmonic amplitudes of time-mean flow coefficient distribution. 0.5 0 -0.5 -1 0 50 100 150 200 250 300 350 Figure 4.9. First mode wave envelope. 1 0.5 - I.. in I- [jL N 0 0 1 2 3 4 5 Harmonic Number Figure 4.10. Mode shape harmonic content. 41 6 7 8 1 0.5 0 0 60 120 180 240 300 360 Circumferential Location (degrees) Figure 4.11. Normalized RMS static pressure fluctuation. -10 -5 0 # of Revolutions to Stall Figure 4.12. Pressure transducer traces for axisymmetric baseline configuration. 5 250 0 200 0 150 00 C, E 0 .'100 # of Revolutions to Stall Figure 4.13. Pressure transducer traces for one-lobed configuration. 350 300 0 250 0 O a 100 # of Revolutions to Stall Figure 4.14. Pressure transducer traces for two lobe configuration. 43 O Axisymmetric 4 SOne-Lobed E3 0 0 20 65 110 155 200 245 290 Circumferential Location (degrees) Figure 4.15. Histogram of stall inception locations. 335 CHAPTER 5 SUMMARY AND CONCLUSIONS 5.1 Summary and Conclusions 1) Experiments have been performed at the General Electric Low Speed Research Compressor to investigate effects of asymmetric tip clearance on compressor performance and stability. 2) It was found that asymmetric tip clearance resulting from casing distortion can significantly reduce compressor peak pressure rise and stable flow range. The decrease in stability is more severe than based on average clearance. 3) The harmonic content of the asymmetry is an important factor in determining the severity of the impact on compressor performance. Asymmetry with lower harmonics is more detrimental than asymmetry with higher harmonics. Clearance asymmetry with wavelength of its dominant spatial harmonic equal the circumference of the compressor produces the greatest reduction in stability margin. 4) There is a reduction in peak efficiency due to asymmetric clearance, but the percentage change will generally be smaller in magnitude than for the stall pressure rise because of the generally steeper speedline slopes at peak efficiency flows. As example, the change in peak efficiency of the compressor with asymmetric clearance was less than 0.5 % compared to a change of 8% for peak pressure rise. 5) Sensitivity to asymmetric clearance is a function not only of steady-state compressor design parameters but also of the parameters that reflect the unsteady response of the compressor. 6) The theoretical model (Graf, 1996) gave a good description of both the steady state (overall as well as local) and the unsteady behavior with asymmetric clearance. References Freeman, C., Effect of Tip ClearanceFlow on CompressorStability and Engine Performance. von Karman Institute for Fluid Dynamics, Lecture Series 1985-05, 1985. Graf, M.B., Effects of Asymmetric Tip Clearanceon CompressorStability, Master Thesis, MIT Department of Aeronautics and Astronautics, June 1996. Longley, J.P. and Greitzer, E.M., Inlet DistortionEffects in Aircraft PropulsionSystem Integration,AGARD Lecture Series -183, May 1992. Haynes, J.M., Hendrick, G.J., and Epstein, A.H., Active Stabilizationof Rotating Stall in a Three-Stage Axial Compressor,ASME Paper No. 93-GT-346. 1993. Hynes, T.P. and Greitzer, E.M., A Methodfor Assessing Effects of CircumferentialFlow Distortionon CompressorStability, ASME Journal of Turbomachinery, 1987, Vol. 109, 371-379. Koch, C.C., Stalling PressureRise Capabilityof Axial Flow CompressorStages, ASME Paper No. 81-GT-3. 1981. Smith, L.H., Jr., The Effect of Tip Clearanceon the Peak PressureRise of Axial-Flow Fans and Compressors,ASME Symposium on Stall, 1958, 149-152. Wisler, D.C., Loss Reduction in Axial-Flow Compressors Through Low-Speed Model Testing, ASME Paper No. 84-GT-184. 1984. Wisler, D.C., personal communication, 1996.