2. an engine research program focused on low pressure turbine

8/14/2019 2. an Engine Research Program Focused on Low Pressure Turbine

1/6

Proceedings of IGTI:ASME TURBO EXPO 2002

3-6 June 2002Amsterdam RAI International Exhibition & Congress Center

Amsterdam, The Netherlands

GT-2002-30004

AN ENGINE RESEARCH PROGRAM FOCUSED ON LOW PRESSURE TURBINEAERODYNAMIC PERFORMANCE

Raymond CastnerNASA Glenn Research CenterSanto Chiappetta

Pratt & Whitney Canada

John WyzykowskiPratt & Whitney CanadaDr. John Adamczyk

NASA Glenn Research Center

ABSTRACTA comprehensive test program was performed in thePropulsion Systems Laboratory at the NASA GlennResearch Center, Cleveland Ohio using a highlyinstrumented Pratt and Whitney Canada PW 545 turbofanengine. A key objective of this program was thedevelopment of a high-altitude database on small, high-bypass ratio engine performance and operability. Inparticular, the program documents the impact of altitude(Reynolds Number) on the aero-performance of the low-pressure turbine (fan turbine). A second objective was toassess the ability of a state-of-the-art CFD code to predictthe effect of Reynolds number on the efficiency of the low-pressure turbine. CFD simulation performed prior andafter the engine tests will be presented and discussed.Key findings are the ability of a state-of-the art CFD codeto accurately predict the impact of Reynolds Number onthe efficiency and flow capacity of the low-pressureturbine. In addition the CFD simulations showed theturbulent intensity exiting the low-pressure turbine to behigh (9%). The level is consistent with measurementstaken within an engine.INTRODUCTIONIt is well known that the aero-performance (adiabaticefficiency) of the low-pressure turbine (LPT) of a turbofanengine decreases with altitude or Reynolds number. Thisreduction in aero performance is commonly referred to asReynolds number lapse.

This is a preprint or reprint of a paper intended for presentation at aconference. Because changes may be made before formalpublication this is made available with the understanding that it willnot be cited or reproduced without the permission of the author.

A number of studies (Halstead, et al. (1997); Hodson,(1990), LaGraff and Ashpis, (1997) suggest that theboundary layers on LPT blading transitions towards alaminar flow state as Reynolds number is reduced. Thusfor a fixed level of aerodynamic loading a reduction inReynolds number can result in flow separation. If theseparated flow regions are large the efficiency of the LPTwill be compromised. Having a flow model, which canaccurately predict the Reynolds number lapse of a LPT iskey to the execution of successful designs. This is ofparticular importance today because of the emphasis onreducing design time and reducing LPT blade countwithout sacrificing LPTefficiency. In addition the recentinterests in Uninhabited Aerial Vehicles (UAV) for highaltitude surveillance has added even more emphasis onthe need for models that can accurately predict the LPTReynolds number lapse in efficiency.The work in this paper outlines a test program in whichaero-performance data is acquired for an LPT turbineoperating in an engine environment over a range ofReynolds number typical of a UAV application. Theengine used in this test is a Pratt & Whitney Canada(PWC) PW 545 jet engine. The PW 545 engine is a high-bypass engine with a thrust rating of 3000 Ibs. The LPTturbine in the PW 545 engine has three stages. Thisengine was highly instrumented in order to determine theReynolds number lapse in efficiency of the LPT. CFDsimulations were preformed using the CFD codeAPNASA, Adamczyk et.al. (1990) prior to the tests. These

Copyright _ 2002 by ASME


2/6

simulations established the ability of a state-of-the-artCFD code to predict the Reynolds number lapseof the LPT efficiency. Results from simulations executedafter the tests will also be presented. These post-testsimulations further support the ability of the CFD code toaccurately estimate the Reynolds number lapse of theLPT efficiency.Engine Test Set-upThe PW 545 High Altitude Test was performed in theNASA Glenn Research Center Propulsion SystemsLaboratory, Cell 4. An overall installation sketch of the testengine installation is shown in Figure (1).

InletPlenum PW545 Exhaust

Engine Collector

Figure 1 - PW 545 in PSL4

The test engine, a PW 545 turbofan, was hard-mountedvia engine stand to a single-axis thrust stand. The metricbed is designed to accommodate the PWC engine stand,and also support the inlet ducting. One measurement andone calibration load cell are included on the stand. An on-board hydraulic cylinder and hand pump is used tocalibrate the stand prior to all testing. The engine wasbuilt, instrumented, and acceptance tested in a sea leveltest facility in Mississauga, Ontario, Canada and flighttested on the PWC Boeing B720 Flying Test Bed (FTB).

The flow capacity of the test cell ranges from 250 pps to400 pps at a typical inlet pressure of 4 psia. For thesetests the inlet flow was reduced to 12 pps at an inletpressure of 1.13 psia. The low flow rates required for thistest program, relative to what is typical, necessitatedadditional inlet instrumentation. The station 1.0instrumentation duct (Figure 2.) provided mounting forfour boundary layer rakes, one cross-duct rake, and wallstatic instrumentation ports. The cross-duct rake wasused to verify the total pressure profile at the duct inlet as

measured by the boundary layer rakes. The forward edgeof the duct was the metric break location for the entireinlet assembly.

I CrossDuctRake_- ,_

_t_= \

t3oundary ,,Layer Rakes'_,

.>'_Duct Static................Pressures

Figure 2 - PW 545 Station 1.0 InstrumentationSix thermocouple rakes with 12 elements each wereinstalled near station 1.0 to measure the incoming totaltemperature profile. The station 2.0 duct mounts directlyto the engine compressor inlet case. Four boundary layerrakes were installed at this location to record the totalpressure profile and airflow at this location. Themeasurement accuracy of the incoming engine mass flowwas estimated to be better than 1%.Turbine MeasurementsFor this test program the LPT was heavily instrumented torecord the lapse in turbine performance with Reynoldsnumber. Over 150 pressure and temperature sensorswere installed in the LPT module. The LPTinstrumentation plus the instrumentation in the fan andcore compressor were used to infer the incoming flowconditions to the LPT (corrected flow) and its efficiency.The accuracy of the efficiency estimates derived from thedata was +/- 1 point.Test and CFD Simulation ResultsPrior to the engine tests CFD simulations were performedat four Reynolds numbers. The rotation speed of the LPTwas fixed at a corrected speed parameter of 280. Thisspeed parameter is defined as the physical rotationalspeed (rpm) of the LPT divided by the square root of thetotal temperature (measured in Degs. Rankine) of thegas stream entering the LPT. The inlet flow conditionsspecified in the simulations (total temperature, totalpressure, flow angles, and turbulence level) wereprovided by Pratt and Whitney Canada.

2 Copyright @)2002 by ASME


3/6

The CFD simulations were performed using the APNASAcode, Adamczyk et al. (1990). The APNASA code solvesthe average passage equation system as formulated byAdamczyk (1985). This equation system governs the timeaverage flow field within a typical passage of a blade rowembedded in a multistage axial flow turbomachine.The turbulence model used by APNASA in this study is anenhanced k-E model (CMOTT turbulence model)developed by Shih et al. (1995) modified to be consistentwith the average passage equation system. The uses ofthis turbulence model in CFD simulations ofturbomachinery is reported upon by Shabbir et al. (1996),Adamczyk et al. (1998). The CMOTT turbulence modeldoes not explicitly account for transition. As reported byAdamczyk et al. (1998) the model appears to providereasonable estimates of the aero performance of axialflow compressors in flow regimes where the flow is knownto be in transit ion.The CFD simulations performed prior to the engine testsincorporated models for purge flows and leakage aboutthe rotor tip knife-edge seals (all rotors are tip shrouded).The predicted normalized efficiency lapse with Reynoldsnumber is shown on Figure (3). The Reynolds number isdefined in terms of the chord of the first LPT vane, and theflow conditions at midspan at the exit of the vane. Thesimulations span a large range of Reynolds numberranging from 30,000 to 295,000. The efficiency estimatesderived from the CFD simulations used mass averagedinlet and exit flow values of total temperature and totalpressure. In addition the efficiency estimates explicitlyaccounted for leakage and purge flows.

Normalized Efficiency vs ReynoldsNumber

1.02 ]101

oo94__og_'} 0 97 l

i0.94 4093

0

APNASA_ ---C-'-- NASA Data

J .....::_-. FTB Data5O 100 150 2OO 25O

ReynoLds Number (xl000)Figure 3

300 350

The corresponding experimental results derived from theengine tests at NASA Glenn are shown on the figure.This data ranges from a Reynolds number of 50,000 to165,000. In addition efficiency estimates derived from afight test bed (FTB) program have also been included.The Reynolds numbers for the fight test program rangefrom 110,000 to 235,000. All efficiency estimates (engineresults are not a direct measurement) have beennormalized by their respective value at a Reynoldsnumber of 165,000. Thus all results have a value of one aReynolds number of 165,000.Figure (3) shows that the Reynolds number efficiencylapse as predicted by the CFD simulations is in very goodagreement with that deduced from the engine data. Froma Reynolds number of nearly 300,000 to 30,000 the CFDsimulation predicts nearly a 7 percent reduction inefficiency. The predicted lapse is well within the accuracyof the efficiency estimates derived from engine data.The next figure shows the lapse in LPT inlet correctedflow with Reynolds number. Once again CFD simulationresults as well as results derived from both engine tests(Tests at NASA Glenn, and fight test bed) are shown.All inlet corrected flow estimates have been normalizedwith respect to their value at a Reynolds number of165,000. All estimates show a well defined trend, andare self consistent with each other. Once again the CFDresults are well within the accuracy of the flow rateestimates derived from the engine test data.

3 Copyright _ 2002 by ASME


4/6

Normalized Inlet Corrected Flow vsReynolds Number

1 ' Omu. 0.99

i

.,,., 0.98 --e--APNASA_ NASA DataP 0.97" _ FTB DataO0.96 ].95

0 100000 200000 300000 400000Reynolds Number

Figure 4At the start of the CFD simulation exercise questionsarose as to the dependency of the simulation results onthe values specified for purge and leakage flow rates. Itwas hypothesized that the value for efficiency is highlydependent upon the values set for these flow ratesbut that the Reynolds number efficiency lapse is a weakfunction of these flow rates. This was the caseirrespective of whether the efficiency estimates werederived from CFD results or engine tests data. Toestablish the validity of this hypothesis a series of CFDsimulations were performed without purge and leakageflows. These CFD simulation results are shown in Figure(5) along with the results from Figure (3). Once again theefficiency results are normalized as outlined above. Withthe exception of the results at the lowest Reynoldsnumber, the Reynolds number efficiency lapse for thisengine is a weak function of the purge and leakage flowrates.

Normalized Efficiency vs Reynolds NumberFixed Corrected Flow

1.02o 1=.2_.0.98

ILl-00.96N_0.94E_0.92z 0.9

"__Flow, APNASA

V + Leakage and Purge Flow, APNASA

0 100000 200000 300000 400000Reynolds Number

Figure 5

Figure (6) shows lapse in inlet corrected flow withReynolds number, both with and without purge andleakage flows. The results show that the dependence ofthe lapse of inlet corrected flow on Reynolds number is aweak function of the leakage and purge flows.

Normalized Inlet Corrected Flow vsReynolds Number Fixed Corrected

Speed Parameter

1.011OE O.99

"O 0.98.m O.97Oo 0.96

0.95

_, APNASA

Purge and Leakage Flows, APNASAb

0 100000 200000 300000 400000Reynolds Number

Figure 6Based on the results presented in Figures (5) and (6) alladditional CFD simulation results to be presented will notinclude the effect of purge and leakage flows.A second set of CFD simulations were generated(executed after the engine tests was completed) at aspeed parameter of 250. These simulations wereexecuted in order to compare with engine data forReynolds numbers below 50,000. Four simulations weredone ranging from a Reynolds number of 25,000 to135,000. Results from this second set of simulations areshown in Figure (7). This time the efficiency has beennormalized with respect to the efficiency at a Reynoldsnumber of 135,000. Efficiency estimates derived fromengine data are also shown. These results have alsobeen normalized with respect to their value at a Reynoldsnumber of 135,000. The range of the engine results isfrom 30,000 to 135,000.

4 Copyright/; 2002 by ASME


5/6

Normalized Efficiency vs Reynolds Number atCorrected Speed of 250

1.02

OC6 0.98ELUo 0.98._.=_ 0.94Oz 0.92

0.910

atata

50000 100000Reynolds Number

150000

Figure 7This time the correlation between engine data and CFDresults is not as clear as that on Figure (3).There is even a lack of correlation between the estimatesderived from the two sets of engine test data.However, if one neglects the estimate at a Reynoldsnumber of 40,000 derived from the tests at NASA Glennthere is good agreement between estimates derived fromthe CFD simulations and those derived from the tests atNASA Glenn. Given that the estimates derived from theengine tests data have an error of plus or minus one pointone could also state that the CFD based estimates are inreasonable agreement with those derived from the FTBstudy.The CFD simulations provided a creditable estimate of theefficiency lapse of the LPT with Reynolds number. Basedon these simulations an estimate of the dependency ofentropy rise across the LPT on Reynolds number wasattempted. The results are shown in figure (8) in whichthe log of the entropy rise across the LPT is plotted as afunction of the log of the Reynolds number. A reasonablefit to the simulation results is a linear curve whose slope is-0.274. This dependence of entropy rise across the LPTon Reynolds number lies between that for a laminar flow(-0.5) and that for a fully turbulent flow (-0.2).

Log of Entropy Rise vs Log of Reynolds Number

-0.5.e

oI_-1.5C...I ;ol

-,: ]

2_.5

Ln Reynolds Number2 4 6 8 10 12 14

O Corr Spd=270, APNASAD Corr Spd=250, APNASA

Figure 8

Key to the ability of APNASA to predict the efficiencylapse of the LPT with Reynolds number is capturing theturbulence level through the LPT. The turbulence level isdefined in terms of turbulent intensity as:

I= _/q (1)

where k is the axi-symmetric average of the turbulentkinetic energy, and q is the axi-symmetric average of theabsolute velocity. A plot of the turbulent intensity at mid-span at various axial locations within the LPT is shown inFigure (9).

Turbulence Level Through LPT

12

t.

._=

U==4"5 2I-

Inlet R1 exit R2 Exit R3 ExitAxial Location

Figure 9

5 Copyright G: 2002 by ASME


6/6

These values are derived from the CFD simulations for aReynolds number of 295,000, which corresponds to SeaLevel Take-Off conditions. The results show the turbulentintensity increases through the first two stages of the LPTbut decreases slightly across the last stage. Thedecrease is the result of a reduction in the aerodynamicloading of the last stage relative to the second stage.Figure (9) shows that the turbulence intensity is more thandoubled between the inlet and the exit of the LPT.Sharma (1998) reported measurements of turbulentintensity downstream of a LPT at Sea Level Take-Offconditions ranging from 12% to 16%. The turbulentintensity derived from the CFD simulation is not out of linewith the measurements reported by Sharma (1998). Thehigh level of turbulence intensity aft of the first stageraises issues as to the nature of the transition process inthe inner stages of a LPT.

ConclusionsA comprehensive engine research program wasconducted to establish the Reynolds number efficiencylapse of an LPT under engine operating conditions. Insupport of this engine research program a series of CFDsimulations were preformed to establish the ability of aCFD code (APNASA) to predict the Reynolds numberefficiency lapse as well as the lapse in LPT inlet correctedflow with Reynolds number. Both the CFD simulationsand the engine tests points spanned a wide range ofReynolds numbers, which makes the current studyimportant.The CFD simulation results presented in this papercapture the lapse in aerodynamic efficiency with Reynoldsnumber quite well. It also appears the CFD simulationsaccurately capture the lapse in LPT inlet correct flow withReynolds number. CFD simulations show that the effect ofleakage and purge flow on the efficiency lapse and inletcorrected flow lapse is small.The CFD simulations were executed using a turbulencemodel that does not explicitly account for flow transition.The model does however account for the production ofturbulence due to the straining of wakes as they convectthrough a blade row. It is the straining of wakes that leadsto an increase in turbulence intensity through the LPT.The model also accounts for the damping of turbulentkinetic energy near solid surfaces, the extent of whichincreases as the Reynolds number is reduced. The highturbulence level of the free stream penetrates the outerregion of the blade boundary layers to a depth establishedby the wall-damping model. The resulting state of theboundary layer thus determines its response to animposed pressure gradient. These key elements of the

turbulence model play a key role in generating the resultspresented in this paper.ReferencesAdamczyk, J.J., 1985, "Model Equation for SimulatingFlows in Multistage Turbomachines," ASME paper 85-GT-226.Adamczyk, J.J.,Celestina, M.I., Beach, T.A., and Barnett,M., 1990,"Simulation of Three-Dimensional Viscous Flowwithin a Multistage Turbine", Trans. ASME, 112,370-376.Adamczyk, J.J., Hathaway, M.D., Shabbir, A., andWellborn, S.R.,1998,"Numericat Simulationof MultistageTurbomachinery Flows," Presented at the VehicleTechnology Symposium onDesign Principles andMethods for Aircraft Gas Turbine Engines, hosted byAGARD in Toulouse, France May 11-15, 1998.Halstead, D.E., Wisler, D.C., Okiishi, T.H., Walker, G.J.,Hodson, H.P., and Shin, H., 1997,"Boundary LayerDevelopment in Axial Compressors and Turbines Part 1 o4: Composite Picture,"Journal of Turbomachinery, Vol.119, p.114.Hodson, H.P., 1990, "Modeling Unsteady Transition andIts Effects on Profile Loss," ASME Journal ofTurbomachinery, Vol. 112, No. 4.LaGraff, J.E., and Aships, D.E., 1998, "Minnowbrook II1997 Workshop on Boundary Layer Transition inTurbomachines," NASA/CP-1998-206958Shabbir, A., Zhu, J. and Celestina, M.L.,1996,"Assessment of Three Turbulence Modelsin a Compressor Rotor",ASME Paper No. 96-GT-198Sharma, O.P., 1998,"Impact of Reynolds Number on LPTurbine Performance", NASA/CP-1998-206958Shih, T.H., Liou, W.W., Shabbir, A., Zhu, J. and Yang, Z.1995, "A New k-e Eddy Viscosity Model for High ReynoldsNumber Turbulent Flows", Computers Fluids, 24, 3, 227-238.

6 Copyright 2002 by ASME

2. an engine research program focused on low pressure turbine

Documents