AD-A233 526 N 9 ,NASA AVSCOMTechnical Memorandum 101416 Technical Report 88-C-037

Development of a Thermal andStructural Analysis Procedurefor Cooled Radial Turbines

Ganesh N. KumarSverdrup Technology, Inc.NASA Lewis Research Center GroupCleveland, Ohio

and

Russell G. DeAnnaPropulsion DirectorateU.S. Army Aviation Research and Technology Activity-A VSCOMLewis Research CenterCleveland, Ohio

Prepared for the33rd International Gas Turbine and Aeroengine Congress and Expositionsponsored by the American Society of Mechanical EngineersAmsterdam, The Netherlands, June 6-9, 1988

US ARMYAVIATIONSYSTEMS COMMANDAVIATION R&T ACTIVITY

Development of a Thermal and Structural AnalysisProcedure for Cooled Radial Turbines

GANESH N. KUMAR RUSSELL G. DEANNASverdrup Technology, Inc. Propulsion DirectorateNASA Lewis Research Center Group and US. Army Aviation Research andCleveland, OH 44135 Technology Activity-AVSCOM

NASA Lewis Research CenterCleveland, OH 44135

zfACff.iiecy level of two axial stages can beachieved by a single radial stage. This leads to

A procedure for computing the rotor a reduction in overall engine complexity, length,temperature and stress distributions in a cooled and weight. These attributes have resulted in theradial turbine is considered. Existing codes for widespread use of uncooled radial turbines inmodeling the external mainstream flow and the automotive turbochargers and auxiliary powerinternal cooling flow are used to ccp~te boundary units. The radial turbine has not been used as aconditions for the heat transfer pnd stress primary pawer source - where high tenperatures andanalyses. An inviscid, quasi thte*dimensional pressures are desired - because of thecode computes the external free stream velocity, difficulties associated with fabrication of theThe external velocity is then used in a boundary internal cooling passages. Most of theselayer analysis to compute the external heat fabrication barriers have been resolved to thetransfer coefficients. Coolant temperatures are point where it is feasible to design an ergnecomputed by a viscous one-dimensional internal with a cooled radial turbine.flow code for the momentum and energy equation. Several programs have looked at theThese boundary conditions are input to a three- fabrication and performance of small cooled radialdimensional heat conduction code for calculation turbines. The first major attempt at anof rotor temperatures. The rotor stress internally cooled radial turbine rotor wasdistribution may be determined for the given initiated under the U.S. Army in 1968 [2,3]. Thethermal, pressure and centrifugal loading. The turtine was designed for a 2300 F turbine inletprocedure is applied to a cooled radial turbine temperature a 2300 ft/sec tip speed, a pressurewhich will be tested at the NASA Lewis Research ratio of 5:1, and a corrected specific work of 40Center. Representative results from this case are Btu/lbm. The design required 3% of engine airflowincluded, for rotor blade cooling and achieved a total

efficiency of 87.5%. Although this design clearlyfNT JCTION displayed the radial turbines advantage, the

fabrication of test hardware was met withIncreasing the turbine-inlet temperature is problems. In 1977 the Army sponsored a radial

the most lucrative approach to improving thermal turbine program with the objective of providingefficiency and specific thrust of gas turbines, and demonstrating the technology required toHowever, this practice is limited by the economically manufacture a cooled high temnperatureavailability of materials which can withstand radial turbine (4,5]. One approach used phototbpee high +emperatures. Current materials cannot etched laminates bonded together to form abe exposed to terperatures above 1250 K without a complete wheel. The other approach used a castdramatic decrease in life [1]. Blades must be airfoil shell diffusion bonded to a powdered metalcooled if significantly higher turbine inlet disk. Both programs showed promising resultstemperatures are desired. Mainstream air diverted An important step in the design of a cooledfrom the compressor exit is used to cool the turbine is thc knowledge of the temperature andturbine. This reduces the thermal efficiency of stress distribution throughout the rotor. Asthe system. Therefore, minimizing the cooling prototype experiments are usually extremelyair stream is an important task in gas turbine expensive, there is a great need for theoreticaldesign. Cooled axial flow turbines have prediction methods [1]. A crmputational proceduretraditionally been used in these engines. has been developed by the authors to assist in the

In small engines, the radial turbine has analysis and design of a cooled radial turbine.several inherent advantages over the axial This procedure predicts the three-dimensionalturbine. Normally, the work extracted and rotor temperature and stress distribution for a

given set of operating conditions. The method

Presr nteri al the Gas Turbine and Aeroenqine Conq-,Amsterdam The Netherlands -- Pine 6 9 1988

uses a quasi three-dimensional inviscid code in procedure is based on the method developed byconjunction with a boumndary layer code to compute Christie, et al, (12].the external blade surface heat transfercoefficients. Coolant passage heat transfer MDDEL GENERATICNcoefficients a,-4 coolant teperatures are computedby a viscous one-dimensional internal flow code. NAS ModelA three-dimensional heat conduction code uses Based on roal points along the streamlinesthese results to compute the rotor temperature of the blade suction and pressure surfaces, thedistribution. A detailed stress analysis can also disk geometry and the coolant passages geometrybe requested. input data, a NASTRAN model of the blade and the

To minimize the time and effort required associated disk is created by the code RITICNfor complete analysis, the entire procedure is [13). The data generated by this code can be usedauto-ated. Once the procedure is started, it will to:

go through each step of the procedure in theappropriare scqueance, iterating between solutions 1. Perform NASTRAN structural analysis on theuntil ccrnvergencx, is achieved. blade and the disk.

MALYSTS MrC RE 2. Provide two of the six input files requiredby the code SMAIN4 [14]. The code SMAIN4

Fcir main codes are linked into a procedure converts the NASTRAN model of RITICN to

to compute the steady-state rotor temperature and SINDA model for theral analysis.stress distributicns. The four codes are:M7l:DIjSqONIC/STkN5, CPFTHA, Srh'D3 and NASTrRAN. 3. Provide plate and solid-element NASTRAN

The three code set, MERID4PI/CrfC/STAN5, is modeling, which when combined with RELACTto compute :..c exLna; :.:!at transfer temperature output provides input to the

coefficients along streamlines on both the suction PATRAN 3-D color graphics plottingand pressure surfaces [6,7,83. MERIDL is used to routines.compute the mid-channel streamline positions fromhub to shroud. TSONIC uses each of these This preprocessing program generates thestreamlines to determine the flow conditions on a grid point and element connectivity data for asurface between two blades. Stacking these CCSMIC/MSC NASTRAN finite element grid of a radialsurfaces from hub to shroud produces a quasi inflow turbine rotor with internally cooledthree-dimensional solution. M=IDL and TSONIC are blades. The grid generated by this programsteady state, inviscid, two-dimensional codes consists of one blade and the associated sector ofapplicable to a perfect gas. the rotor disk. The NASTRAN cyclic symmetry

The inviscid blade surface velocities from feature is employed for structural analysis, thusthe MERIDLISONIC combination form the input to rendering the above as sufficient to model theSTAN5. These are used as external free stream entire rotor.velocities in the t.&urdary layer analysis. STAN5 The basic turbine model generated employs auses these velocities to compute a two-dimensional hollow blade which is modeled with triangularboundary layer along each streamline. The plate element. and a disk sector which is modeledcalculated heat transfer ooefficients are used as with brick elements. Blade halves are attached toboundary conditions in the conduction analysis. the disk using rigid elements. Along the bladeThe free stream velocities mist be "smoothed" if tip, along the leading edge, and in the scallopSTAN5 predicts regions of sepairated flow. region, the blade halves are tied together with

CPFRIHA (Coolant Passage Flow with Rotation quadrilateral plate elements . dhe exhdaust end ofand Heat Addition) is used for computing the the blade halves is open. The sides of the rotorinternal coolant temperature, pressure and heat hub are "tied" together (CYJOIN) for utilizationtransfer coefficients along the coolant flow path of the NASIRAN cyclic symmetry feature. All of[9). It integrates the one-dimensional mouentum the above is automatically generated by theand energy equations along a path. The equations program. The user has great flexibility inaccount for area change, rotation, friction and generating a wide range of models due to theheat addition. Flow splits cannot be determined number of user specified input variables. Awith this code. The internal cooling passages typical model is depicted in Figure 1. The userthat can be acoanTudated in the procedure include may generate additional elements manually for theplain passages, finned passages, trip strips, pin NASTRAN deck to model features which may be uniquefins, tip cap impingement, etc. The coolant to a given concept/design (e.g. pins, Laffles,passage wall temperature distribution and coolant ribs, etc. internal to the blade halves (skins)).inlet conditions are a required input. Thus, aniterative process between CPFRHA and the heat SINDA Modeconduction code must be used since the wall The SINDA model consists of nodes andterperature distribution is not known apriori. conductors which provide a lnnoed parameter

The heat conduction code SINDA tSystems representation of the turbine rotor and itsImproved Numerical Differencing Analyzer) (10] is thermal boundary conditions. Nodes and conductorsused in the analysis procedure to perform detailed are generated along the streamlines, nodal lines,three-dimensional heat transfer analysis of the and in the tangential direction following an orderrotor. A wedge consisting of one blade and the similar to that used in RrICN. Most nodes havecorresponding disk region is analyzed. three conductor numbers associated with their

Based on the converged temperature nodal number. One conductor is oriented in thedistribution in the rotor , the MERIDL/TSONIC streamline direction, one in the nodal lineexternal pressure distribution, and the CPFRHA direction, and one in the tangential direction.internal coolant passage pressure distribution, a The code SMAIN4 generates a SINDA heatNASIRAN (11] structural analysis of the rotor is transfer model of a radial inflow turbine withperformed. From this analysis, nodal internallyL uled blades. The program obtains thedisplacementq -*-ess di ributi. and modal geometric data of the model from the program:;haec7 Ll 4-'e iuLor are obtained. The above RMTI.

2

Figure 1: NASTRAN Modl Generated by RITIC.

A SIN node is generated for each NASTRAN Within the disk, a SINDA node is generatedgrid point within the blade. Adjacent to these, for each NASTRAN grid point. Additional nodes areSINDA nodes are generated at the metal surfaces generated for the teoperature boundaries along theand at the points that represent the temperature rim, back face, bore, hot gas face, and exducerboundary. Conductors are generated ;hich end. Conductors are generated along streamlines,represent the metal conduction along streamlines, nodal lines, and in the tangential direction, andalong nodal lines, and through the thickness. to represent the external heat transferAdditional conductors are generated to represent coefficients. Additional options have beenthe heat transfer coefficients between the metal's incorporated which generate the thermal boundarysurface and the boundary nodes. A typical segmenL conditions for the coolant inlet channel, exit endof the blade model is shown in Figure 2. Similar face (exducer) and the disk's cooling passages.nodes and conductors are generated for the bladetips, leading edge, and root edge (Figures not Boundary Conditionsshown due to space limitations). The equation governing three-dimensionaI

steady state heat conduction is of the ellipticform. Therefore, heat transfer coefficients andrelative total terperature boundary conditions arerequired over the entire surface. The model under

Pr, .. - Side onsideLdtion consists of an entire blade with the.-_ .... corresponding disk sector. Figure 3. shows a

Bu~d. .. 11-rt1-' Lin..

Figre2 SI 1 Bld Model.d L

Fiue :Rto 'oudr 7-dtol adItra

T<-

Figure ~ ~ -1 2:SII Bad kdFBue3.Ctr.i~ir oiltcl ndItra

C2rerjNod. - 3

typical blade/disk combination. The numbers cn related to the SINDA relative node number which isthe figure refer to the numbers of the following provided by adding a few FORMAN statenents to theparagraphs- SINDA analysis output calls. The above is

performed for all grid points in the blade and

1. Suction and Pressure External Surface: Tne disk. Generating the plate element field cardsheat tran-fer coefficients and free stream requires calculating the plate element's material

total temperatures are obtained with the temperature, the bottom surface teperature, theMERIDL/TSONIC/STAN5 combination. top surface temperature, an the thermal gradient

through the thickness of the element. In addition

2. Internal Cooling Passage: The heat to the relation tables described above, the

transfer coefficients and coolant passage program requires the element connectivity andterperatures are computed by CPFHA. thickness data. All the relation tables and

element data files are generated by either RITIN,3. Disk Back Face: This region is connected SMAN4, or SINTM. Very little inprut is generated

to a shaft. A contact resistance heat by the user. RELACT output can be directly usedtransfer coefficient is used. by PATRAN [16) for obtaining the three'dimensional

temperature plots (in the form of temperature4. Blade Leading Edge: The heat transfer bands and contour lines) onthe blade and the disk.

coefficients are obtained by -odeling theleading edge as a cylinder in cross flow. NASTRAN Structural Analysis

The blade and disk modeling for NASTRAN5. Blade Trailing Edge: The average of the finite element analysis is created by RITICN as

suction and pressure_ =rface heat transfer explained before. The internal geometry of onecoefficients at the trailing edge are used. blade (Coolant passages) is modeled as follows:

6. Disk Exducer End Face: A contact Baffles -CQUAD4 quadrilateral plate

resistance heat transfer coefficient is elementsused. Pin Fins - CBAR rod-type elements

Colling Ribs - CONM2 extra mass elements7. Blade Platform: The heat transfer

coefficicnts are taken to be the mean of The pressure loads applied to the modelthe suction and pressure surface values at arise from two areas:the blade hub. Hub platform film cooling 1. Internal coolant Pressure (Datais accounted for Ly adjusting the external Supplied by the Code CPFRHA)total temperatures. 2. Working Fluid Pressure differential

across the blade (Data Supplied by8. Blade Scallop: Heat transfer coefficients the MERIDL/TSONIC/STAN5

are taken to be the average of the external combination).suction and pressure values at the hubstreamline in the scallop region. The thermal load applied to the model is

generated by the code RELACT using the output from

9. Disk Rim: This is the region where the SINDA.

coolant cames on board the rotor. The heat A static stress and deflection analysis astransfer coefficient is found by modeling well as a modal analysis can be performed on thethe flow as an enclosed rotating disk next cooled radial turbine using NAgmm.to a stationary wall. ILLUSTRATIVE APPLICATION

10. Blade Tip: The heat transfer coefficientsare the average of the suction and pressure A cooled radial turbine will be tested atvalues along the streamline. the NASA Lewis Reasearch Center. The conditionsfor the test and sample application are given in

Table 1. The turbine internal geometry is similarUsing the boundary conditions mentioned to that shown in Figure 3. Cooling air entering

above, SMAIN4 generates the CONDUCIOR Data and the the rotor gets directed to the leading edge.NODE Data required as input for analysis by SINDA. From there it flows down to the hub and branchesThe user is to supply zhe metal's thermal into three paths exiting at the trailing edge.conductivity as a function of temperature in the The flow splits indicated in the figure wereARRAY data as described in the SINDA user's obtained with a flow network analysis code whichmanual. The SIV cards zequired to simulate includes the momentum equation but not the energybaffles, pin fins, etc. are to be hand generated equation.and added to the SINDA input deck created bySMAIN4.

The output from SINDA will be a detailedthree dimensional temperature distribution in the Table 1: Turbine Inrut Conditions

blade and disk.The computer program RELACT [15] then Rotor Inlet Total Temperature 316 C

converts the output of a SINDA heat transfer Rotor Inlet Total Pressure 2.52 Baranalysis into the thermal 1-ad input for a NASTRAN Mass Flow Rate 2.08 Kg/sec

structurai analysis. RELACT has two major Speed 21570 rpnfunctions. The first is to generate the grid Total Pressure Ratio 4.05xrint" trinrture fi-ld 'rdzL; LhC sconu is to Work uocff-cdtiui ah/ 1.0

generate pldt.e element field cards. The first is Total Enthalpy Change 170 KJ/Kgeasily accomplished by using relation tables that Coolant Flow Rate 4.3% Mainflowrelate SINDA relative node numbers to SINDA actual Coolant Onboard Total -30 Cnode numbers, and SINDA actual nodes numbers to TemperatureNASTRAN grid point numbers. Using thesp tables Coolant Onboard Total 1.58 Barthe NASTRAN grid point temperature is easily Pressure

4

Figure 4 shows the freestream critical FREE STREAM VELOCITV MOOTkWO

velocity ratio distribution along the meanstreamline. These are the MERIDL/TS(NIC computedvelocities which are used as input to STN5. Ten 0-10S

streamlines between the hub and tip were chosen.The free stream velocity was "smoothed" on thesuction surface of streamlines 1,2,5,6, and 7 to 10-16

eliminate separation. Figure 5 shows the location U

and amount of "smoothing." Generally, STAN5predicts separation along the suction surface in 95-35

regions of high diffusion. "Smoothing" the freestream velocity, or reducing the velocitygradient, will eliminate the separation. Figures6-8 show some of the heat transfer resultspredicted by STAN5 for the mean streamline. STAN5input assumptions include turbulent flow, arecovery factor of one, and a flat inlettemperature profile. Assuming a small boundarylaver thickness ca%3ared to the surface's radiusof curvature allows the curved blade to be modeledas a flat surface in STAN5.

The CPFRHA predicted coolant temperatureand heat transfer coefficient distribution isgiven in Figures 9 & 10. High heat transfercoefficients are predicted near the tip. Thisleads to a cool pressure side metal temperature Figure 5: Freestream Velocity Smoothing.since the external heat transfer coefficients onthis side are low. The suction side temperaturesare higher due to the higher external heattransfer coefficients. HEAT TRANSFER COEFFICIENT

Several interlinking codes were used in the _ _ _ _ __,_ _

procedure to generate input files required bySMAIN4. For the above sample application, it tookonly two iterations between SINDA and CPFRHA toobtain a converged temperature distribution (wherethe maximum change in temperature was less than S ,b2%).

s.,fo, S.,0c

PATRAN plot routines are used to display t0both SINDA and NASTRN cout.ts. SINDA torperaturedistribution in the form of oontours is displayed

in Figure 11. The stress contours predicted by 3W -

NASTRAN are shown in Figure 12 and the nodaldisplacements are shown in Figure 13. These 20 .... __O___figures are a representative sample of the typical ,0graphical output generated by the above __.procedure.0..

0 02 04 06 Oa

FIACTOON Of STREAMUN OISTANCC

Figure 6: External Heat TransferF, iTPLAM v2ELCC TY Coefficients

EXTERNAL STANTON NUMBER/ 00030 S2CT040 SUR.ACE 2(00 S1TEMUN

/ .0033S.. 00032

003 / 031i / 003

0025

0N o 00027

021

0 00 0000;00018, ,,

FPACTI OF STRAO JNE DISTMCE

Figure 4: Blade Velocity. Figure 7: Stanton Number Along Suction Surface.

5

EXTERNAL STANTON NUMBERPRESSURPE SURFACE 440fl SYAEMSL44

Io. PRESSURE SURFACE

0 2, o6

Figure 8 : Stanton Nube Along Pressure Surface.

CONTOURTIimP-LEUEL (K)

lU~ a 380.

D 3SB.E 335.F 320.

129 7 G 305.H 290

1 275.

4 32 896 34e

50 1.3%

6 - 1.6%

SUCTIONO SURFACE

Figure : ntrna eat~ TlansfPesreSra.

coefficients (W(K)K).

2 6 12

25 395.

COO 335.

1295T 97 4%85FL298.

Figure 10. Coolant Temprture (K). Figure 11: Teperature cotor (SINDA).

6 268.

PRESSURE SURFACE CONTOUR STRESSLEVEL (l1'A)SUTOSRFC

A 368.B 325.C 298.0 255.E 220.

./ F 185.'/'J "' 150.,".f-- __ [4 5" /iH 115.

J 845.

G

G ~ 6

Figure 12: Stress Contours (NAsTRAN).

The NASTRAN model generation and the NASTRAN toSINDA conversion codes must be modified beforeapplication to axial turbines.

There are several prominent differencesbetween axial and radial turbines. The axialturbine does not have the blade scallop found inradial madines; thus, one boundary condition iseliminated. The relative total tevperature alongthe blade external surface is strictly a functionof the radial position within the blade. A largeradius change occurs in a radial turbine as thegas passes between inlet and exit. This causesthe relative total temperature to drop. Thistemperature change does not exist in an axialturbine since the radius change is small. Theprocess of introducing the coolant to the bladediffers between axial and radial machines. Thisprocess in an axial design may be quite similar toor different from the radial machine under study.There are numerous other minor differences betweenthe axial and radial turbine which the procedureshould readily aroomodate.

CONCIUDI G R KS

An automated analysis procedure has beenFigure 13: Nodal Displacements (NASTRAN). developed which can actxoplish a rigorous thermal

and structural analysis of a cooled radialturbine. This procedure after proper validationwith the experimental data should result in a

Experiments leading to validation of the versatile analysis and design tool for cooledanalysis procedure are planned for late 1988. The radial turbines.turbine will be tested at the conditions in Table1. Rotor aerodynamic and heat transfermeasurements are planned. Blade surface pressure, REFENCEStemperature and heat flux measurements will beobtained. A subsequent paper will compare the 1. Rodi, W. and Scheuerer, G. "Calculation ofanalytical and experimental results. Heat Transfer to Convection Cooled Gas

Turbine Blades," Transactions of the ASME,AXIAI/RIAL TURBINE ANALYSIS C PARISON Vol. 107, July 1985, pp 620-627

fhe analysis presented applies to radial 2. Calvert, G. S. and Okapuu, U., "Design andturbines; although, the procedure may be extended Evaluation of a High Temperature Radialto include axial turbines. The internal and Turbine Phase I," USAAVLABS Technicalexternal aerodynamic analysis to predict heat Report 68-69, January 1969.transfer coefficients and the internal coolantpassage temperature prediction procedures can be 3. Calvert, G. S., Beck, S. C. and Okapuu, U.,applied to an axial geometry without difficulty. "Design and Experimental Evaluation of a

7

High-Temperature Radial Turbine," USAAMRDLTecLnimal kirrt 71-20, May 1971

4. Vershure Jr., R. W., Large, G. D., Meyer,L. J., and Lane, J. M., "A Cooled LaminatedRadial Turbine Technology Demonstration,"AIAA-80-0300, January 1980.

5. Ewing, D. A., Monson, D. S., and Lane, J.M., "U.S. Army/Detroit Diesel Allison High-Temperature Radial Turbine Demrnstration,"AIAA-80-0301 January 1980.

6. Katsanis, T. and McNally, W. D., "RevisedFORTRAN Program for Calculating Velocitiesand Streamlines on the Hub-ShroudMid channel Stream Surface of an Axial,Radial, or Mixed Flow Turbomachine orAnnular Duct-I User's Manual," NASA TN D-8430, 1977.

7. Katsanis, T., "FORTRAN Program forCalculating Transonic Velocities on aBlade-to-Blade Stream Surface of aTurbocdhine," NASA TN D-5427, 1969.

8. Crawford, M. E. and Kays, W. M., "STAN5 - AProgram for Numerical Couputation of Two-Dimensional Internal and External BoundaryLayer Flows," NASA CR-2742, 1976.

9. Meitner, P. L. , "Computer Code forPredicting Coolant Flow and Heat Transferin Turbomachinery," NASA TP. (To bepublished)

10. Smith, James P.; "SINDA User's Manual"Cosmic Program MSC-18597, March 1983.

11. MSC/NASTRAN, User's Manual,MacNeal Schwendler Corporation, LosAngeles, 1984.

12. Christie, R., Tornabene, R., and Stevens,M., "Thermal Analysis of Cooled MetalTurbine Rotor", Final Report, NASA T.O.#856, March 1984.

13. Christie, R.T., and Stevens, M.A., "RITICN-Mesh Generation Program for Radical InflowTurbines with Internal Cooling forCOSMIC/MSC NASTRAN", Final Report, NASAT.D. #578.

14. Christie, R.J., and Tornabene, R.T., "SINMYModel Generating Program for Radial InflowTurbines with Internal Cooling", FinalReport NASA T.O. #823.

15. Christie, R. J., "REIACT - SINDA Output toNASTRAN Input Data Conversion Program forRadial Inflow Turbines with InternalCooling", Final Report, NASA T.O. #750.

16. PATRAN Plus User Manual, PDA Engineering,2975 Redhill Avenue, Costa Mesa,California, 92626, July 1987.

.,, .,. ,nmm m innitliaI I II8

NASA Report Documentation Page

1 Report No NASA IM-101416 - 2 Government Accession No 3. Recipient's Catalog No.

AVSC()M TR-88-C-0- 17

4 Title anc Sublitle 5. Report Date

l),.clopmcnt of a Thermal and Structural Analysis Procedure for

Cooled R ,dial Turbhines 6. Performing Organization Code

7 Ajtho,'sl-- 8. Performing Organization Report No.

(ianch N. Kutmar and Russell G. DeAnna E-4515

10. Work Unit No.

9 Performing Organization Name and Address

NASA le sk Research Center 535-05-01

Clceclard. Ohiii 44135-319) 11. Contract or Grant No

Propull i n Director:.itcti.S Arm. A iation Research and Technology Activity-AVSCOMCleo.cland, Ohio 44135-3127 13. Type of Report and Period Covered

12 Sponsoring Agency Name and Address Technical Memorandum

National Aeronautics and Space AdministrationWashinton. D.C. 20546-0001 14. Sponsoring Agency Code

andU.S..Arini Asiation So stems CommandSt. Louis. Mo. 63120-1798

15 Supplementary Notes

Prescntcd at the 33rd International Gas Turbine and Aeroengine Congress and Exposition sponsored by theAmerican Society of Mechanical Engineers, Amsterdam, The Netherlands, June 6-9, 1988. Ganesh N. Kumar,Sscrdrtlp Technolog\. Inc.. NASA Lewis Research Center Group, Cleveland, Ohio 44135; Russell G. DeAnna,PropOusiit Directorate.

16 Abstract

A procedure for computing the rotor temperature and stress distributions in a cooled radial turbine is considered.Existing cotdcs for modeling the external mainstream flow and the internal cooling flow are used to compute

boundary conditions for the heat transfer and stress analyses. An inviscid, quasi three-dimensional code computesthe external free stream velocity. The external velocity is then used in a boundary layer analysis to compute theexternal heat transfer coefficients. Coolant temperatures are computed by a viscous one-dimensional internal flowcode for the momentum and energy equation. These boundary conditions are input to a three-dimensional heatconduction code for calculation of rotor temperatures. The rotor stress distribution may be determined for thegicn thermal, pressure and centrifugal loading. The procedure is applied to a cooled radial turbine which will betested at the NASA Lewis Research Center. Representative results from this case are included.

17 Key Words (Suggested by Author(s)) 18. Distribution Statement

Turbomachinery heat transfer Unclassified - UnlimitedRadial turbines Subject Category 07Structural analysisComputational methods

19. Security Classif (of this report) 20. Security Classif. (of this page) 21. No of pages 22. Price

Unclassified Unclassified 10 A02

NASA PORM 1626 OCT 86 For sale by the National Technical Information Service, Springfield, Virginia 22161

0

z

00

0 0:

0 0 H

10 0 <f

0z H) 1-02~ U) cam

- Ul

oo z Z M

o 4) 0 W C3E -4 >w -4 H

o4 u m4 W E-CJC4

>. 0 QA p (

2) m

- z C4mm~I

0 w - 4

() O 'l r- -4Z [

- cc 04

0 D

0 C14 0- (1 0 ti m

4 zi u U) N

T C H

o

4

Cc -'A 'A4 zx

- e 0 Ef~ 03

0 cc a- z- 0ECA 41U UA 0<

0-..-0

C u U -4 CcH44~ ~~~ N J

Wc- K:C 0) cc0V)Cci 'A 0AI 00 4JU zz

Uo w''. r- < 0 0a'ci o cc 0, (1) C n

Cx. D Ncl4 i Ez 0W w u0 w W ~iC 0 .

Z~ 2w >I o~ I~ j o z z 0 -

Nd>N -WW4 U N CIC O c ., c o ) wm w u 2

-- ccl z 50

LJ N Z* 0

ii W Ui-r pW N w z ww~d m CC cc C

0O 0 00- c

ational Aeronautics and FOURTH CLASS MAIL!11111

)ace Administration

PwIs Research Center ADDRESS CORRECTI(,N REQUESTEDlevland, Ohio 44135 USAIL

aly for PrIvele U"e 5300

UASA