/9

TURBULENCE MODELING NEEDS OF COMMERCIAL CFD CODES: COMPLEX FLOWS

IN THE AEROSPACE AND AUTOMOTIVE INDUSTRIES

Bizhan A. Befrui

adapcoMelville, New York

N95- 27896

CONTENT OF PRESENTATION

• STAR-CD: COMPUTATIONAL FEATURES

• STAR-CD: TURBULENCE MODELS

• COMMON FEATURES OF INDUSTRIALCOMPLEX FLOWS

• INDUSTRY-SPECIFIC CFD DEVELOPMENTREQUIREMENTS

• INDUSTRIAL COMPLEX FLOWS:APPLICATIONS & EXPERIENCES

- FLOW IN ROTATING DISC CAVITIES

DIFFUSION HOLE FILM COOLING

- INTERNAL BLADE COOLING

EXTERNAL CAR AERODYNAMICS

• CONCLUSION: TURBULENCE MODELINGNEEDS

I STAR-CD: COMPUTATIONAL FEATURES 1

BODY-FITTED NON-ORTHOGONALCOORDINATE SYSTEM

• UNSTRUCTURED COMPUTATIONAL MESH,DIFFERENT CELL TOPOLOGIES, IMBEDDEDMESH REFINEMENT, DISCONTINUOUSMESH INTERFACE, MOVING BOUNDARYAND INTERNAL INTERFACES

• PRIMITIVE VARIABLE, SELF-ADAPTIVEELLIPTIC-HYPERBOLIC PRESSURECORRECTION METHOD

• COLLOCATED-VARIABLE ARRANGEMENT

• EULER-IMPLICIT TEMPORAL INTEGRATION

• UD, CD, LUD, SFCD SPATIALDISCRETIZATION, WITH BLENDINGCAPABILITY

171

https://ntrs.nasa.gov/search.jsp?R=19950021475 2018-04-18T01:06:59+00:00Z

l STAR-CD: TURBULENCE MODELS I

° TWO-EQUATION MODEL

- STANDARD k-e WITH CORRECTIONS FORBULK DILATATION AND BUOYANCY

- HIGH REYNOLDS NO. RNG BASED k-¢MODEL

• TWO-ZONE (TWO-LAYER} MODEL

- HIGH REYNOLDS NO.: k-s VARIANTS

- LOW REYNOLDS NO.: k-t VARIANTS,PRANDTL MIXINGLENGTH MODEL

STAR-CD: TURBULENCE MODELS I

• REYNOLDS STRESS TRANSPORT MODEL"

- TRANSPORT EQUATIONS FORCARTESIAN STRESS TENSOR IN NON-ORTHOGONAL COORDINATE SYSTEM,ON NON-STRUCTURED MESH

- LAUNDER, RODI, REECE (1975|FORMULATION WITH LAUNDER (1989]MODEL CONSTANTS

- GIBSON & LAUNDER (1978} WALLREFLECTION MODEL

172

Sep94VIEW

.0oo

.0D01.000

ANGLE.000

DISTANCE.340

CENTER.076.06*7

.003EHIODEN PLOT

Oev,,, & Seege=W eadera_ Fac_g StapRow Oema_. -20"H Io _'H

Ues_. ;05 (Aml) x 4S (Radial)

L

-J • J ,4 • • 1i I.,1 1.,4 1i 1,4 ._Z *.1 J .1 J • A • • .T •

100uuAJfld'2 -100uvAJrM'2

tL'J

. f _,,..-'"

GRAPH PLOTFRAMES

• F..m. dam

RS modJ

KE mod_

Dav_ & SeeOmeJef8emmrd F_O StepData Inlet B.C.

LocaM_l X/H - 1.5

173

sJ

Jii

u" -__ !

• i• i

_J

th

u

: I

u

u

H

u

j j ,4 • • tJ 1Jr t,s 1.4 lal J.J _t J .! J • .4 • • 3

u

_k u" k ,

t H " •

I1 _ M t.a-.... Jr. - _." ......

7seps4

PLOT

FRAMES

Le_na

_ RS moClel

KE model

Driver & See(jmner Back'war¢l Facing Step

Data InM! B.C.; No waa Darap_ng Fur, or.

Lo(:a_oct X/H : 1.5

174

Z

175

2s,po4PRESSURE

N/M"2

L.O_M. MX= 329.9

L.OCAL MNs ,-29.3"/

4t0.0

3e0.7

351.3

322.O

25_7

2e_5

234.O

204.7

173..5

144..0

116.7

87"_

SS.00

28.67

-.e887

-30.00

0.5 LUO OIFFERENCINGTURBULENCE MOOEL _ - UPPER, RSM • LOWER

COMMON FEATURES OF INDUSTRIALCOMPLEX FLOWS

• THREE DIMENSIONAL WITH MULTIPLE FLOW"COMPLEXITIES"

BODY-FORCE FIELDS

- STREAM SURFACE CURVATURE

- STRONG PRESSURE GRADIENTS

- COMPRESSIBILITY EFFECTS

- LAMINAR-TURBULENT TRANSmON

- COMBUSTION, SHOCK, MULTIPHASE, NON-NEW'rONIAN

.

* LARGE SCALE DOMAIN AND COMPLEXGEOMETRIC CONRGURATION

• IRREGULAR, UNSTRUCTUREDCOMPUTATIONAL MESH

- SPATIAL RESOLUTION DIFFICULT TO ACHIEVEON O(10 s - 10 8) MESH CELLS

• INSUFFICIENT AND UNCERTAINEXPERIMENTAL DATA FOR TURBULENCE

MODEL VAUDATION/IDENTIFICATION OFDEFICIENCIES

176

INDUSTRY-SPECIFiC CFD DEVELOPMENT

REQUIREMENTS

• AUTOMOTIVE INDUSTRY

- EFFICIENT COMPLEX-GEOMETRY. MOVING-BOUNDARYCAPABILITIES

- MEMORY/SOLUTION PERFORMANCE FOR LARGESCALE DOMAIN CFO SIMULATION

- DIAGNOSTIC/COMPARATIVE EVALUATIONOBJECTIVES

- GEOMETRIC FIDEUTY AND SPATIAL RESOLUTION AREPRIMARY ACCURACY FACTORS

• AEROSPACE INDUSTRY

- REGULAR AND SMALL-SCALE FLOW DOMAIN (BENCH-MARK EXPERIMENTAL MODELS)

- DESIGN/PERFORMANCE OPTIMIZATION OBJECTIVES

- NUMERICAL AND TURBULENCE MODEL ACCURACYIMPORTANT

- REQUIREMENTS

• HEAT TRANSFER

• LOW REYNOLDS NO. FLOW

• BODY FORCE FIELDS

177

EXTERIOR TUBEREMOVED)

ORIGINAL PAGE IS

OF POOR0UALrrv

OUTLET

SYMMETRY PLANE

MOVING WALLSIMULATION OF GROUNDMOTION RELATNE TOVF.HX:LE (40 _)

MOVING WALLSIMUt.ATION OF ROTATING TIRES(371 RPM)

INLETSIMULATION OF VEHICLETRAVELING AT 40 Imh('T = 30C, P. 100 kPa)

FIGURE 1: EXTERIOR BOUNDARY CONDmONS FOR W202 40 _ ANALYSIS

178

oF eooleOu/ ry

11 De¢93

V1EI.OC r'l"Y MAGNT'RJDE

M/S

ITER= 140

LOCAL MX= 50,.81

LOCAL MN= .0000E+00

15.00

1425

13,50

12.75

12.00

11.25

loJ0

9./S0

II.o00

8.28oi.

7.5C0

e,.'/_o

6.000

S.2_

4,800i

3.700

3.o00

2_50

1.500

.7500

.0G_OE.=,00

W202 UNDERHOOD FLOW AN_J.YSIS

C/_.qE 3:40 _ SIMULATION

VeloCily near the surlace of the vehlde.

W202 UNDERHOOD FLOW ANALYSIS

CASE 3:40 k.ph SIMULATION

179

APPLICATIONS & EXPERIENCES

APPLICATION RNDINGS T.M. NEEDS(DATA}

ROTATING

DISC CAVITY 1

DIFFUSION

HOLE FILM

COOLING 2

FLOW TURBULENCE

COMPLEXITY MODEL

• FORCE FIELD • k-s •

• WALL • 2 LAYER k-t

EFFECT •

• JET-CROSS

FLOW

• WALL

ANISOTROPY

• k-E •

• RNG, k-_

• 2 LAYER k-t

EKMAN LAYER

RESOLVED

FAIRPRESSURE

DROP

EXCESSIVE

E.V.

JET

SEPARATION

SENSITIVE TO

MESH

TOPOLOGY/

RESOLUTION

POOR

SPANWISE

SPREAD

• RSTM +

SUITABLE

2 LAYER

• LOW ReRSTM

• RSTM +

SUITABLE

2 LAYER

• LOW Re

RSTM

1GRABER et al (1987)

2GOLDSTEIN et al (1968), LIGRANI et al (1992)

COMPRESSOR DRUM TEST RIG STAR-CD CONJUGATE HEAT TRANSFER MODEL

AxisOf Rotatlo_"_

180

HALF-CAVI3"_ TEST MODEL: MESH rS TYPICAL FOR MULT-CAVtTY MODEL

¢fs_ tim (r= 11.01n.)

coolant inject=on

Iocabo,n (inlet)

(1.250in.)

last near wan cell ( 0.00026918 In. 0-,¢_)

dl_ bo_ (r - 4.0 In+)

exhaust (ouOm)

lxxe tube O.D. (t = 3.0 In.)

Compressor Drum Test Rig Cold Flow Benchmark Ar;alysis

Secondary F_ow

14-Jun-93

VELOCITY

COMPONENTS U W

FTFoEC

PSYS,= 2

LOCAL MX= 36.44

LOCAL MN= 0.2792E-01

• PRESENTATION GRID"

36.44

33,84

28.64

2O`04

2_44

2O`M

I lL,24

15.6;3

13+00

I0.43

7.831

5.23O

2.6211=

0.27'9_E-01

181

i

Disk 3 Face (z - 0.25 In.)

um m eJ_

_ Drum T_t Rig Cold Row _ _P/IdS

Sec_n_y R0w In Cavity 2

Velocity Ve¢lo_ At r = 7A0 lilies

ii}111

pllllJIIIIII

!!!!!

IIIII

ll_ t1111

_ !!!!!

I

III;IJill!Hi!iII1:!PII III;

I

14-Jun-93

VELOCITY

_ENTS U W

FT/SEC

PSYS= 2

LOCAL MX= 22.30

LOCAL MN_ 0.4491

2230

21,21

20.12

19.03

17.11_

16.84

15.75

14.(B

13.56

12.47

11.3_

Io.28

8_

7.OO6

SJ)13

4.820

3727

2.83S

1.$42

0.4491

CAVITY 2: PRESSURE DROP

u

DI

• . ;

n.ira i --

o .== i in i i

i

$ " I

P

re _m /"

, • //s

", = V./o

¢

D . • ._

o -

p • , /

I • -

/Z

/I i

p ! I i

Me= o_= ¢lw

DImer-=>loNess Radius Ratio

i ; i

Dbnem,_",less Pressure Drop

STAR-CO Model

• Test Data

Forced Vortex

Free Vortex

182

mQ

n

o

_ -lJm

a$

P

r

o

$

u

r _e

D -Lm

r

o

p _

CAVITY 4: PRESSURE DROP

f

A

/

/1

/

/

/

//!

/i

i:"

4J_ z I I I! F m r m

<u> aim ¢l¢e _ ¢Jm elm iJ_ eJ_

Dlmemskxdess Radius Rs_o

i i

oJm

Pms_re D_o

STAR-CD Model

• Test Data

Fomed Vortex

Free Vortex

1.(_0

1.(00

1.000

_NGLE

.000

DISTANCE

84234

nefk3ed mesh. M = O,S, pipe gdd 8butting plato gMd.

Mesh = 330000 f_ Ceils

Two-Layer mesh

Y

183

HH" .................ItSt', ...............Its,,. ..............Tlls.. .............

,lIII is"#1/1111 • • " .........

x./d - 8

lJtt*N,, ...................llsol.*. ..................If/i/ms. ...............

1_1=4

IKiUH_.I.U/ / ;'/ / / H , , , _ , ,MIA_///.I/J/// / / / s t t,,,,,

/////1111 I" Aii.......P_'fs*l,* ,* s . . •.j,,s_.sss • _. •

CFOOknte HoleFire _ Ver_c_n Stuo'y_tj_ d eq)ettne_ o_Goldste_,or.al,[lg681;B_owlngraUoU-0.£veecltyvectononspanwisei_os; 2.1altarmodel.

,,,o ................

Illll., ..............

tti..................||1 ...................II_ .... ,,,oo . .

i_,_..,._ ...............

x/el. 20

TII. ................Ii. .................I, ..................'sOl.,_o, ............

• .o,,,#,., ...........

x/d- 16

_,,/sll.. .........

x/d,. 12

5 JulIP4

VEL COMP VW

M/S

LOCAL MX- 36.13

LOCAL MN- .O000E+OO

3O.OO

28.50

-- 27.00

ZS,SO

24.OO

22.5O

21.OO

Ill.50

1LOO

1S.OO "

1&SO

I_.OO

-- 10.S0

O.OOO

-- 7500

-- 8.OOO

-- 3.000

-- 1.500

Y

1_=0.

CFDDlcreteHoleFilmCooUngVed_a_onStudySlmu_atk:,nof experimentofGol_tetn,eL al,(19681;Blowingm,oM ,,0.5Temperature contours off _anwlse planes : 2.1syer model.

=tl,,llO

_1,, I$

x/d= 12

184

| mi

Im

c

o

I

I

n

g

f

e

I

v

e

s

s.im

o

\\ o111 •

\\\ o :\

i

expm'mlenml, z,'d

• 0.0

o 0.25

• 0.5

rl 0.75

• 1.0

," 1.5

0.0

-- 0..25

0.5

0.75

1.0

1.5

o g

0

! • A

x/d

EXPERIMENTS OF GOLDS'T'EIN El" AJ..,1 g_,8

COMPARISON OF FILM COC'UNG EFFECTIVENESS

M - 0.5 : Mesh II.

f .im

I

I

m._

co

I

ns¢g

f

e

Cj_t

I

vn4_o

e

s

s lm

.J_

o

\ •o

\o

\.

-.....• & _ "Jr--..-.-

e_o_nenmL z/d

o 0.0

• 0.5

n 0.75

• 1.0

/, 1.5

-- 0.0

0.25

0.5

0.75

1.0

1,5

0

• • J-

z n 0 _ -"7 n- --r-- ,i -(', _. • _ ro

.o zJee _ 73e0 Io10_ Itl_ iseoo 171oo :lOlem _,.sQ1 _sr_ox/d

EXPERIMENTS OF GOLDS'rEIN E'i" AL,1968

COMPARISON OF FILM COOLING EFFECTIVENESS

M = 0.5 : 2 Layer mesh.

185

APPLICATIONS & EXPERIENCES (cont'd)

APPLICATION

(DATA)

INTERNAL

BLADE

COOLING 3

EXTERNALCAR AERO-

DYNAMICS 4

FLOW

COMPLEXITY

• FORCE FIELD• B.L.

DISRUPTION

• B.L.

STRUCTURE

INTERACTION• COMPLEX

WAKE

TURBULENCE

MODEL

• k-E

• k-s

• RNG k-E

• 2 LAYER k-g

FINDINGS

• DEPENDENCE

ON MESH

RESOLUTION

• GOOD Z_P,h

• DEPENDENCE

ON MESH

RESOLUTION

• GOOD C D• POOR LIFT

T.M. NEEDS

• RSTM

• LOW Re

RSTM

3GE AIRCRAFT ENGINES [ABUAF & KERCHER (1991)]

410 FORD 1/4 SCALE MODELS IN WIND TUNNEL TEST [WILLIAMS et al (1994)]

186

29xugs4

PRESSURE

Nnd"2

LOCAL MX,..4427E_06

LOCN. MN=-.2719E,_06

.4427E+_

.3406E_

.23_E,,06

.1875E_

.13_E_06

.8541E_

• Jr/73E*¢5

-.11ME_

-.laME_

-.22WE_

-2"/lgE*_e

eS_

MAGNITUDE VELOCITY

M/SEC

PSYS,, 2

LOCAL MX- 314.6

LOCAL MN= 4.850

• PRESENTATION GRID"

315.0

294.3

273.7

2S3,0

232.3

211.7

1;)1.0

170.3

149.7

_29.0

"_ 87.6"/

(17.00

46.33

25.87

5.000

187

2_g4H. T. COEFFICIENTS

W I,._Q, METIER - K

LOCAL MX- .10_E_05

LOCAL MN,, .0(X_E+00

.10_E_05

.1014[+05

8314.

?804.

124,1.

S4_

3121.

2341.

780.4

.00_E+00

2._

20¢

lS0

100

LEADING EDGE C_INEL

1! °

. C/D Coa_ A_

0 Ci=O FI_e/¢qi -

FLOWIO° TURN

50 N n n n . n n n n

x/D

]ROOTINLET

CFD BLADE AND LEADING EDGE MODELS

Marinaccio (1989,1990a,1991)

Hp=r¢ 4 a Lcad_g odgc ¢hanndhe._tmm.d'er_t.ion _th d_umccfrom the _¢L Com-par/=o,',of ,,,odd n._-bu]atodcorn,c( mrfac_mad,m=_ =d=/=_= anducrag¢ mca.mrement_w/thbl_," CFD avengepr_

188

THIRD GRID REDUCTION (3:1)

SECOND GRID REDUCTION (2:1)

WAKE REGION

FIRST GRID REDUG'RON (2:1)

NOTCHBACK WIND TUNNEL AERODYNAMIC STUDY MODEL

COMPLETE MODEL DOMAIN

DGRID

02-Jul.93

MAGNITUDE VELOCFr'Y

M/S

LOCAL M.X= 53.46

LOCAL MN= O.0000E+O0

• PRESENTATION GRID"

35.00

3,4.00

33.00

32.00

31.00

30.00

29.00

28.00

27.00

26.00

25.00

24.00

23,00

22.00

21.00

2000

'10.00

15,00

'17.00

'_8.00

1500

WIND TUNNEL AERODYNAMICS STUDY OF NOTCHBACK TEST SHAPE

KE RESULTS - KE TURBULENCE MODEL WITH LUD

VELOCITY MAGNITUDE NEAR THE VEHICLE VIEW FROM REAR

189

PRESSURE COEFFICIENTS AT THE CENTI_:IUN E OF VEHICLE

PRESSI.IRECOEFF1CIEN_AT11'1EClB_TERLINEOFVEI'UCt.E

,.=_'

_,,::. . ;_

o

•e,,il0O4.moo4,_o0 -Lm_ u=__aso _0O.LO000_ _um l.t_. .1

'" .... _.v_ ._-_: ............_i_-_._SS_ c,*_..'.,l._,_'_",':l,'"I

• --I -I¥ I..- I-.'_l_,_b_-,_'-'_____ _ _"_'_ ' I' ]

190

r,N t ,,

. ./_/

o /..too

AL "'= •

o //L /, / TT _

0N ..Ibm

.._ _ n i I E r i J ; a r

EXPERIMENTRESULTS

COMPARISON OF EXPERIMENTAL AND COMFWJlqONAL UFT COEFFICIENTS

K EPSILON llJRBULENCE MOOEL - "" INITIAL RESULTS ""

• SrAncm W,_ON

• N(_HL_X el

• _rC>e._x Cl

O _OTCNL_:X

• FMSTL,_X1

O FLLTL_X Z

• FkSlTL_CX3

O F_.57_A_X4

m _MSI1L_XS_*_£S

, /D .m

c //A

L

C

L , ///_/ /

A

T

I0

. //A I

EXPERIMENTAL RESULTS

COMPARISON OF EXPERIMENTAL AND COMPUTIONAL DRAG COEFFICIENTS

K EPSILON TURBULENCE MODEL - "' INITIAL RESULTS "'°

• _TCHL_X F1

0 _TOL_C m

• Norcm,_<_ c_

0 NO_O,eK_ <_

• _'_I,,KXI

• ;_'rll.,<x

. .

.. o

191

CONCLUSIONS: TURBULENCE MODELINGIMMEDIATE NEEDS

• NEAR-WALL TURBULENCE

- ECONOMICAL, ROBUST LOW REYNOLDSNUMBER 2 EQ. EVM's AND RSTM

- A GENERAL AND VERSATILE NEAR-WALLTREATMENT FOR RSTM

• RSTM MODEL

- ALTERNATIVE CLOSURE OF THE WALLREFLECTION COMPONENT, WITHOUT NEED OFWALL TOPOGRAPHY PARAMETERS

• EDDY-VISCOSITY MODELS

- EXTENSION OF THE NON-LINEAR k-e TOINCORPORATE FORCE-FIELD EFFECTS

• BENCHMARKING

- A RELIABLE DATABASE OF BENCHMARK SET OFREPRESENTATIVE COMPLEX FLOWS

- BENCHMARK PERFORMANCE CLASSIFICATIONOF VARIOUS EVM's (k-s, k-_, RNG AND NON-LINEAR k-_, MULTISCALE EVM's) AND RSTMCLOSURE VARIANTS

CONCLUSIONS: TURBULENCE MODELINGPROGRAM NEEDS

• A LARGER VIEW OF THE RSTM DEVELOPMENT

TOWARDS IMPLEMENTATION IN GENERALCOORDINATE, COMPLEX GEOMETRY DOMAIN,UNSTRUCTURED CFD METHOD

. A BROADER APPLICATION OF DNS TOCOMPLEX FLOWS TO ASSIST TURBULENCEMODEL DEVELOPMENT/OPTIMIZATION

• .WELL-POSED EXPERIMENTAL DATA,OBTAINED IN THE ORIGINAL OR REDUCEDSCALE MODEL OF THE INDUSTRIALCOMPONENT FOR CFD VALIDATION

• COLLABORATIVE INDUSTRY-CFD

RESEARCH/DEVELOPMENT PROGRAMS FOREXPERIMENTATION - CFD VALIDATION

(CALIBRATION) FOR SPECIRC INDUSTRIALAPPLICATIONS

192