Shock-induced flow separation and the orbiter thermal protection system

Acla Astronaagca Vol. 10, No. 5--6, pp. 453--466, 1983 Printed in Great Britain.

0104-5765/83 $3.~ +.00 Pergamon Press Ltd.

SHOCK-INDUCED FLOW SEPARATION AND THE ORBITER THERMAL PROTECTION SYSTEM S.-A. WAITER Space Transportation & Systems Group, Rockwell International, 12214 Lakewood Blvd., Downey, CA 90241, U.S.A. (Received 4 January 1983) Abstract--The Space Shuttle orbiter's thermal protection system (TPS) is composed of reusable tiles separated by narrow gaps that accommodate the contraction and expansion of the aluminum structure that the tiles protect. When local pressure gradients exist, air flows through the tile gaps and releases heat energy by convection. The gaps represent a heat short to the structure, strain isolator pad (SIP), and filler bar. A typical problem is the pressure gradient created during entry by body flap deflection. After a brief description of how this problem affects the Space Shuttle orbiter, a theoretical and experimental review of the major parameters involved in gap heating are analyzed. Then, a review of well-known classical methods to resolve the gap aeroheating problem in the presence of a pressure gradient is presented, and a few solutions are illustrated to assess the sensitivity of each one. The following section starts with a basic relationship (called "eyeball" because of its simplicity) and follows the results up through the most modern engineeringapproach available in the literature. It shows that in all cases calculated significant areas of overtemperature were predicted. However, none of these methods could be correlated by experimental data. Lastly, the paper presents the solution obtained by using the most sophisticated method, based upon the Navier-Stokes equations. This approach shows excellent correlation with wind tunnel data. The application to four trajectory time points shows less severe results than the other methods. This can be explained by the introduction of a certain amount of conservatism to account for uncertainties inherent in the previous analyses. No correlation of this "exact solution" with the simple preestablished relationships has been found, indicating that more parameters than expected could be involved. However, an after-the-fact, semi-empirical engineering solution that fits the Navier-Stokes solution with good agreement was established.

1. INTRODUCTION

that the heavy burden resulting from the approximate solutions was unnecessary.

The thermal protection system (TPS) of the Space Shuttle orbiter consists of approx. 31,000 reusable surface insulation (RSI) silica tiles bounded to an aluminum alloy and graphite epoxy structure. The bonding of these delicate tiles requires extreme precautions, but damage to the tiles is not to be expected. These tiles are aerothermodynamically defined so that friction and convection from the hot and highly energetic flow field surrounding the spacecraft during flight are not likely to create severe problems for the TPS system. However, the tiles are of finite dimension (8 in. maximum for orbiter vehicle (OV)-102, for instance) and gaps between tiles are designed with a 0.045-in.-nominal width. When hot air flows into these gaps, damage to the structure or to the strain isolator pad (SIP) by convection and friction could occur. Tests have shown that the coupling of the pressure gradient (mass flows and ohm friction coefficient gradient) with the heat fluxes on the tiles could be a crucial test for Shuttle's TPS. The purpose of this paper is to present the problem for a typical location where high pressure gradients exist during entry, i.e. the bottom fuselage under the influence of body flap deflection. Once the problem is defined, the impact of quick and approximate solutions is assessed, showing a very heavy penalty. A more rigorous solution, whose excellent correlation with experimental data is promising for flight application, is then introduced. The results of the Shuttle's first flight, STS-I, have proven

2. THE INFLUENCE OF PRESSURE GRADIENT ON TILE GAP HEATING

The Space Shuttle orbiter has a complex geometry, and no exact solutions are presently available to compute the flow field surrounding the vehicle during flight, making it impossible to predict the actual aerothermodynamic environment with a single program code. Instead, a step by step approach is followed using Descartes' methodology. First, an aeroheating environment is defined assuming the orbiter is a smooth body (no gaps, steps, or cavities). Then, the influence of "roughness" is added. Figure 1 shows the design of orbiter isotherms computed for a smooth surface, and Fig. 2 shows the actual orbiter and its tiles. Strong pressure gradients will allow the flow to circulate throughout the many tile cavities and gaps; overheating of the SIP and structure may appear and problems may occur. Figure 3 shows the thermal insulation as it exists and the tolerances presently allowed for gap criteria. Actual size panels of different tile combinations have been tested in NASA Center tunnels under pressure gradients similar to those encountered in flight, and a parametric study has shown that it is a combination of pressure and pressure gradients that can produce excessive gap heating. This can be understood when one considers that a 453

454

S.-A. WALTER

2,300F - 2000F ~ ' \ \ y 1,800F\ X ~ 6 / / . ~

r l I LOWERSURFACEV,EWl 2,000F

/ - - - - / ~ ~

,.=OOF" / 1 200F r-

^^_f 800F/600 - -

-

,.

~;":'~.~ i

,

/ UPPERSURFACEVIEW I

2,460F . ~

,~\%./-,

~'~1"~ ENTRY/ASCENT _ ~ J /~ ~. \ ." _700F/750F

~ ~ " ~'~'\"

GOt'tIE1,200F "/" J / . . . .

2.000F

~ . / "\" ' ~ -"' ~ - ~

~

850F750F//900F830F

* 1,000 F

1,680F .7,~?0F ~ > / ~ " ~ ' 2 ENTRY/ASCENT~ ouu - ~ [ ~ ~ . , \800F

2,665 F

oo o , - , o o , ,

/\ 750F / 830F ~ V~__
~

~.J-~l~

1 800F"

100F 1" F

~- -~'~-

2.000F

I SiDE SURFACEViEW i

-"

"

-.805F "iL~

830F

[ *DENOTESASCENTTEMPERATURES 1 IMAXlMUM YAW 8 DEal

Fig. 1. Orbiter isotherms--design trajectory.

Fig. 2. Typical tile distribution on windward fuselage.

455

Shock-induced flow separation RSI SILICA 2,300°F MAX. TEMP

OUTER BOND LINE RTV 550°F MAX. TEMP

/

/

/_

k P

•

RU

FILLER BAR / 800OF MAX. TEMP 350 FMAX. T E M P

RE

LOY 350 FMAX, TEMP

550°FMAX TEMP

SURFACE WAVINESS CRITERIA MAXIMUM WAVINESS

F A I R E :/ OML

"

'

~

-

HA = 0.0015

LOWER WING (EXCEPT ELEVON) LOWER FUSELAGE X/L ~<0.3

H/'~ = 0.0025

LOWER FUSELAGE X / L > 0.3 UPPERFUSELAGE, UPPER WING, ELEVONS, VERTICAL TAIL

MAXIMUM ALLOWABLE GAP LOWER SURFACE NEAR BODY FLAP HINGE LINE: < 0.030 IN, TPS GAP EDGE RADIUS: 0.060 IN. (MAX,)

Fig. 3. TPS temperature limits. ANAL YSIS

• DEVELOP MATH MODEL THAT WILL RELATE GAP HEATING TO PERTINENT PARAMETERS 8" ALLOW EXTRAPOLATION TO FLIGHT =PRELIMINARY MODEL BASED ON WING/GLOVE TEST ARTICLE DATA; VERIFY WITH DOUBLE WEDGE TEST ARTICLE DATA "MATH MODEL TO BE APPLICABLE TO ALL PRESSURE GRADIENT LOCATIONS THROUGHOUT TRAJECTORY

CORFtELA T/ON

~1GAP

~1CHANNEL

~1ref

~1LAMINAR

(dpldx) (p)112

11.622

(HI 3/2

G(H) 3/2

(dp/dx) (p)1/2

1 1.622

(H) 112

G(H) II2

FLAT PLATE

BETTER" CORRELATION:

~1GAP ~1ref

Fig. 4. Gap heating data analysis, correlation, and application.

driving pressure is necessary to pull the flow through the gap. It should be noted that the pressure gradient can be either positive or.negative. Figure 4 summarizes the analysis and the correlation and its application to the Space Shuttle. A pressure gradient coefficientt, G, combining the pressure gradient and the driving pressure has been defined, and its influence on the gapheating with and without pressure gradient is shown in Fig. 5 as a function of the gap depth for a 0.030-in. gap width. In the range of interest, the gap tSee Nomenclature in Appendix at end of paper.

heating rates can increase by more than one order of magnitude [ 1]. 3. THE INFLUENCEOF BODY FLAP DEFLECTION ON FLOW SEPARATION

This section describes the influence of body flap deflection on gap heating. The geometry of the bottom fuselage with its body flap deflected can be compared to a curved plate with a downstream wedge (see Fig. 6). Shock waves are generated at separation and reattachment, producing a strong rise in static pressure. Two strong pressure gradients, both positive, are present,

456

S.-A. WALTER

© 'TESTRUN 25] 26 "0"030 FOR IN. [3 TESTRUN27] TESTRUN TESTRUN 28JGAP 0.1

[ REFERENCE1 I ©

qref

dp/dx = 0

0.01

~,---0.030 IN. ~1 qref

N\\\\\\\\\\\\ 0.002

0.1

0.01

Z (INCH)

0.0055 GZ

11.622 G= dp/dx-Jp 10.0

1.0

Fig. 5. Double-wedgehigh dp/dx gap heating data correlation. separated by a "plateau pressure". Each gradient is driven by a different pressure level and will provide different values for G. The analysis will focus on the determination of the aeroheating pressure coefficient, G, created by such a configuration. It must be remembered that the subgroup [dp/dx~/(pm.x)] is of interest and not only the pressure gradient.

DIVIDINGSTREAMLINE~ ~

M~>I •

~

~

~

N

~

Figure 7 shows the sensitivity of the pressure rise, /5 = (p2/p,), as a function of a parameter such as a wedge or cone angle, Reynolds number, or temperature ratio (wall-to-total) at a given Mach number for a laminar separation [2] (laminar flow is studied here because peak heating occurs during this flow regime). Because of the nature of the parameters playing major roles in the

~''

J / / L ~ ~ ' / . ~ \ \ N ~ S H

OCKS

SEPARATIONONA WEDGE INCOMINGSHOCK " ~ ~/w.-~"EXPANSIONFAN DIVIDINGSTREAMLINE~ ~ / S /

M~>I

~//

\

,////

///

~'////////////////,Sy/////////////////////7///////////////////~ SHOCKIMPINGEMENTSEPARATION

P

~ATTACHMENT POINT ~'SEPARATION POINT PRESSUREDISTRIBUTION Fig. 6. Laminar separation with convective heating.

Shock-induced flow separation

457

10°

Tw/To = 0.4

--~

M~ = 7

6,620 I

,,

12.5°

I

I

5,680 5,575 4,780 4,714 4,005

2oo/ ,

PlNVISCID

,,

3,300

I 2,858 , 2,420 I 2,095

OBLIQUESHOCKTHEORY TANGENTCONETHEORY o

t

0.50

I

0.75

I

1.00 1.25 Rex 10-6 Fig. 7. Laminar separationwith convective heating(sensitivity of pressurerise).

analysis, it was suggested that the boundary layer displacement thickness, 8", would be a plausible factor in the calculation of the dp/dx, and a crude analysis showed that dp/dx could be, for sample cases, approximated by dp/dx = (p2-p010~*. The choice of 10× 5" has been determined by empirical analyses based on a significant amount of experimental data for wedges and cones. These data, available upon request, are limited to super-

sonic Mach numbers below M~ = 6.0, where no real gas effect~ behind the shock exist.

4. CALCULATIONOF THE AEROHEATINGPRESSURE GRADfENTCOEFFICIENT

This section presents the results of an approximation of dp/dx. It is assumed that the flow reattaches at the trailing edge of the body flap (see Fig. 8). Selecting a /REATTACHMENT

FLOW FIELD~

//COMPR ESSIDNFAN

SEPARATION COMPRESSION

SEPARATED / SHEAR LAYER ~

/~'/ ///

BOUNDARYLA'YERED(

LPRESENTAPPROXIMATION~I

RECIRCULATING STREAMLINE

DIVIDING STREAMLINE

~/~/'///~///////////////~,~/.,.

-- -

ELEVON ~HOCK

dx ~ 106'-

1.50

~

~

v . ~ / / , j B O D Y FLAP

I~IVIDINGSTREAMLINE ~ Fig. 8. Flow field and approximation.

~

/

t

458

S.-A. WAITER Tw = constant, M ~ 0, ~ = t~= 0

1. FLAT PLATE

3,

0.591X

(Ix

2. SIMILARITY SOLUTION

~ = COHEN & RESHOTKO

3. INTEGRAL TECHNIQUE

$'~ = V, KARMAN

4. SIMILARITY SOLUTION + INTEGRAL TECHNIQUE (2. + 3.)

8 ~= 2D/AXISYM (S-A W.)

5. FINITE DIFFERENCE TECHNIQUE

S-x-=2D/AXISYM (S-A W.)

6. MORE EXACT SOLUTIONS . . . . 7. NAVIER-STOKES?

NASA (ARC) . . . . .

Fig. 9. Determinationof 8'.

given position, X/L, of separation on the bottom fuselage and assuming a 2-D wedge flow, the pressure rise through the shock is easily computed as a function of the body flap deflection angle, and 3' is obtained by some well-known relationship (see Fig. 9). In the present analysis, the integral technique is used. A machine program for a desk computer (HP9020) has been written using the aerodynamic parameters derived from Rockwell International's aerodynamic heating program so that the resulting G's will be consistent with the aeroheating analysis. The results of the integral technique (3*) have been compared with finite difference

solutions for a few trajectory times and was shown to be adequate for this type of analysis. Figure 10 presents the variation of the G coefficient for two trajectories (STS-1 and the design nominal trajectories) versus the entry time for a given location of separation, X/L = 0.9. It can be seen that the influence of the trajectory is not a strong driver. The influence of the separation location, [0.5 ~
100 ~

VALID FOR LAMINAR FLOW ONLY AT X/L = 0.9

10

G(x)

[STS- TRA.'CTORYI O O'.CII

0.1

200

] 500

I 1,000

I 1,500

TIME (SEC) Fig. |0. Variation of G due to separation during entry (STS-! and design nominal trajectories).

459

Shock-inducedflow separation

COMPRESSIONF A N ~

///

SEPARATED

///

/'~)~ ACTUAL

L

A

~

Y

~

\\"

\ \ ' DIVIDING STREAMLINE

~RECIRCULATING STREAMLINE

BOTTOMFUSELAGE

1.0"/.,.

~

~

,/////~/,,~.00,

....

/ / / / / ~ / ~ ~OOP~"(.~,o

APPROXIMATION 1 SEPARATION AT B/F T. E

X/L =

BOTTOM FUSELAGE

y j / / / / / / / /• / / / / ~ / / / / / / / / / / / / / / / Z ~ ~ ~ ~

~~,,

APPROXIMATION 2 ANALYTICAL

, , . CALCULATION

Fig. 11. Flow field models.

separation, the weaker is the created shock and the higher the G coefficient. (In this analysis, large G's mean weak pressure gradients.) As expected, the G coefficient is shown to be a significant parameter and its determination could be improved. The influence of the body flap deflection angle illustrated in Fig. 12 is shown to be insignificant in the range of deflection to be used during entry. Increasing the body flap deflection angle increases the shock strengths and the plateau and reattachment pressures, but the results are hardly noticeable because of the limited deflection range analysis herein (14.6~ 8BF o ~< 20.6). A better approximation (Approximation 2)[3] is proposed to define more accurately the only unknown as yet, the separation onset location. An iterative process is used, and of the restrictive assumptions included (perfect gases, flat body, etc.), only the peak heating trajectory point was computed (i.e. T = 850sec) because of the rather long calculations involved. The results, presented in Fig. 13, show that separation onset would occur as early as X]L = 0.74, or, to be more specific, all tiles downstream of the main landing gear would experience excess gap heating problems from peak heating to landing as early in flight as T = 800 sec. Gap filler would be required for several thousand tiles with severe consequences such as increased weight, in-

creased labor costs, and significant launch delay. This solution would involve 336 in. of the fuselage's length (0.26 x 1293), or approx. 72 rows of tiles, most of them on the wing. This approach was deleted because of the lack of credibility of the assumptions involved. However, the problem was critical and the flow field mechanics so poorly known that a new approach had to be defined. 5. CALCULATIONOF THE AERODYNAMICHEATING PRE~URE GRADIENT

The only suitable theoretical approach is to solve the exact Navier-Stokes equations. Many models exist, but one of the most satisfying has been developed by NASA/ARC. It can be modified to include the boat tail geometry. Only one month was allocated for the analysis so the results could at that time be incorporated in STS-1 without delaying the STS-I launch. The program was modified very quickly, and its first application was to existing oil flow wind tunnel data (OH 25B)[4] where separation and reimpingement could be accurately defined. These tests were run in the ARC/3.5ft tunnel at M= = 7.3 and several Refit and angles of attack. A similar pressure test (OH 38) [5] was available and provided the input conditions necessary to run the program. Six wind tunnel cases were analyzed--three angles of attack at 2 (Refit) each. Figure 14 presents the

S.-A. WALTER

460 100 --

INFLUENCE OF BODY FLAP DEFLECTION AT X / L = 0.9 ( V A L I D FOR L A M I N A R FLOW ONLY)

BODY FLAP DEFLECTION (6BF) 14.6 o 16.6 °

18.6o~.~

20 • 6 °

=

10--

G(x)

1--

0.1

I 500

I 1,000

I !,500

TIME (SEC)

Fig. 12. Variation of G due to separation during entry (body flap deflection angle).

at XIL, 8*, and Me, were defined theoretically. Convergence to the exact solution should be quick, considering that the starting condition (X/L =0.8) was selected well upstream of the disturbance created by the onset of separation. The results are shown in Fig. 15. Even though dp/dx does not appear to vary significantly

results for a = 40 deg (2 Refit), and compares theory with experiment. The comparison is excellent for all cases analyzed and gives great confidence for the flight analysis. Four trajectory time cuts were computed: T = 500, 700, 900 and 1200 sec. The starting conditions, such as p

REFERENCE 3 ~

Lo

Lp DI

a

•

/

7///////////////////////////////~

"eL

=

1.32

P3 L__L Lo

0,362

L o~

0.74

~

"

~BF

SEp

~

L CONCLUSION: SEPARATION AT X/L = 0.74

Fig. 13. Results of analytical calculations at T = 850 sec.

/

461

Shock-induced flow separation

= 1.11 X 106

1.5--

= 0.538 X 106 1.4--

WIND T U N N E L A N A L Y S I S c( = 40 ° ~BF = 16.5°

1.3-P P ( X / L ) = 0.5

[ REFERENCE

41

1.2--

1.1--

1.0--

0.9--

B/F TRAILING\

HINGE, LINE ~

I

0.8 0.95

I

0.96

1

0.97

E

0.99

0.98

I

1.00

I

1.01

I

1.02

1.03

1.04

1.05

X/L

Fig. 14, Analysis of OH 25B test.

3.0

STS-1 T R A J E C T O R Y A N A L Y S I S T = 500 SEC

T = 700 SEC

2.0 T = 900 SEC T = 1,200 SEC P ( X / L ) = 0.5

900 SEC 700 S E C . ~ 1.0

-

I

"1,200 SEC 500 SEC

0 0.95

[

I

I

L

I

I

I

I

I

I

0.96

0.97

0.98

0.99

1.00

1.01 X/L

1.02

1.03

1.04

1.05

Fig. 15. Actual trajectory case 80701.14 (STS-I).

S.-A. WAITER

~2

3.G

"r = 1.25- Cp= 1/7 Cp FLIGHT @ Cv - 9,000. ACTUAL ,5* INFLUENCE OF FLOW PARAMETERS DESIGN TRAJECTORY AT T = 700 SEe

FLIGHT CASE (~)

2.0a

/ / 7

(~)~

P

I

P(X/L) = 0.5

/

f

~ 7 = 1.117- ACTUAL FLIGHT CASE ACTUAL ~,* ~ ~.~

= 1.25-Cp( 7 = 1.25)= CP(7 = 1.117) (~) Cv = 6,300. ACTUAL ,3*

1.0

0 0.95

I 0.96

I

I

I

I

I

I

I

~

I

0.97

0.98

0.99

1.00

1.01

1.02

1.03

1.04

1.05

X/L

Fig. 16. Analysis of the flow parameters on an actual flight case.

with time, the product dp/dx "X/(P) is strongly time dependent. The very interesting feature of this solution is that shock-induced separation occurs almost at the hinge line, contrary to the previous approaches. It is assumed that the introduction of the boat tail shape into the solution has provided the significant relief expected. With the present results, only the two rows of tiles upstream of the hinge line are affected by the low pressure gradient coefficient G due to separation. In order to assess the sensitivity of the parameters of the product [dp/dx x/(Pmax)], four different cases were computed for which all input parameters, such as a*, Cp, and Cv, were dramatically changed. The results of this error analysis show that dp/dx is only affected by -+7% when these parameters are modified (see Fig. 16). It can also be seen that the large P's are associated with the lower dp/dx's, and that the final pressure coefficient G is almost not affected. This can be explained by the extremely fast convergence of the present computer code. Even though the "starting conditions" at X[L = 0.8 are voluntarily off range, the program converges extremely fast to the correct solution, and when separation occurs, the previous history of the flow has been damped and accounted for. Accordingly, it was decided that only two rows of tiles upstream of the hinge line would be affected by separation and, as a consequence, have their gaps filled. This is a significant improvement when compared with the 70 or more rows of tiles previously subjected to gap filler. During the Navier-Stokes analysis, a memo by Jimmy Carter of NASA/ARC[6] was discovered and analyzed.

In this report, experimental data about pressure distribution on wedges in the separation region at M= up to 6.06 have been collected and plotted. The range of wedge angle is 5-
dX[dx dp/dx = dp/dP dPldX dX[dx = 1 . 8 - -

dP/dp'

with dXldx, dPldp and P and X = qS~[p, x, X, 8 ' . . . ] according to flight conditions and the free interaction model. The results are presented in Fig. 18. The correlation is impressive even though the two logics are so widely different. Flight test results are shown in Figs. 19 and 20. It can be seen that no tile damage is visible upstream of the hinge line. Figure 20 shows the main landing gear door locations. Note the number of rows of gaps to be filled. 6. CONCLUSION The conclusions are summarized below. • No correlation has been found between dp/dx and any boundary layer thicknesses. A statistical analysis has been conducted where & 8 ' and 0 were compared with

Shock-induced flow separation

SCATTER IN CORRELATION .~. 1"6 -OF L E W I S ~

P-Po

I

P= 1752-~° ~

463

~-

3.-I

[dh*~

/

d~*~ 21

~+ -~- M~ ~x )o+0V%x J 04

1.2-

~ ,

0.4 -

~

o

~

-

X-Xo/'~--o,Tw,-l[

///t.,~/£~C/~M= •, DEG RO,Xo ,~g',(~ ~ 3.0 5 1.35 X 104 j(,(~/ v D 3 . 0 7 . 5 1.22 X 104 04.01003.0 10 4.511"05xX104104~_ ~'* RJ-R~,x= ~

, ~ -0.4

dd~ =l"8dP

X = 0 . 3 1 1 ~ / ~ . o M Z~ "

P 0.8

t~-)

5-3> /~",

~1 - T

(1.937 ~-+Tw 0.578-~'Y - 1 M2 - 0.207)

z~6.06 10.25 1.01 X 105 L

0

L

0.4

0.8

I

I

1.2

1.6

I REFERENCE6 I B

X dp

dp dP dX dX/dx = d~ dX d~ = 1.8 dP/dp

/ dXX = d /0.311 (x - Xo) dx dx xo

d_P dp

= d

dp

,[

I rwl

1

~

I PROPOSEDANALYSIS (RI) I

WITH M2 ~

14

M ~ - ) o +0 ( M ~

L

• ,.752--

Fig. 17. Free interaction model.

VALID FOR LAMINAR FLOW ONLY 100 FROM NASA TR-R-385 PAGE 81 BY JIMMY CARTER (NASA/ARC) AND SERGE-ALBERT WAITER (RI) NASA/ARC - JANUARY 1980 SOLUTION OF THE NAVIER-STOKES EQUATIONS G(45o) = 11 10 G (45o)

DESIGN CURVE

PRELIMINARY ANALY~ JANUARY 1980 1

[

500

\

j

d~*~

M=t~-x } o + 0 ~ M ~ - x )o_j

I

1,000 TIME (SEC)

Fig. 18. Variation of G due to separation during entry (G(~)--Navier-Stokes solution).

1,500

464

S.-A. WaftER

Fig. 19. Bottom fuselage after STS-I (overview).

the dx in dp/dx as predicted by the Navier-Stokes solutions. As of now, no empirical or mathematical relationships have been found. This could be because the various boundary layer thicknesses are not related to the pressure gradient, or that other parameters (not yet included) should be inserted. • As for the G coefficient or [dp/dxx/(p)], a good correlation exists between the free interaction model and the solution of the Navier-Stokes equations. A simple computer program has been written that can compute G for any trajectory within a few minutes. • For a given flow condition, (M~, Rex), the extent of the separated flow region is less severe for flight cases than for wind tunnel cases at iso-Reynolds number (per length or per foot). This could be attributed to the fact

that the energy level (enthalpy) is much greater in flight than in the wind tunnel. • Sophisticated solutions predict less severe separation onset and pressure gradient effects than could be predicted by less rigorous analysis. It is assumed that these solutions are based upon experimental wind tunnel data that are, as previously explained, more conservative than flight data. • It has not yet been possible to correlate quantitatively the flight data to the theoretical predictions because not enough pressure data are available from flight to define a good pressure variation and, thus, a good value of the pressure gradient. • However, a good qualitative correlation exists for the prediction of the separation location onset.

Shock-induced flow separation

Fig. 20. Bottom fuselage after STS-1 (closeup).

465

466

S.-A. WAITER REFERENCES

1. C. B. Blumer, Private communication. 2. A. Seka Achy, Contribution a I'etude du recollement de la couche limite laminaire avec transfert de chaleur en regime hypersonique. These di docteur 36me Cycle Specialite: Mechanique des Fluids. Faculte des Sciences di I'Universit~ des Pasteurs, 30 Oct. 1%9. 3. L. D. Miller, Separated Flow Studies on the Orbiter Body Flap and Elevon. Rockwell International, SAS-AA & T 74-280 (9 Oct. 1974). 4. Heat Trans/er Phase Change Paint Tests o/ O.Ol75-Scale Model (Nos. 21-0 and 46-0) of the Rockwell International Space Shuttle Orbiter in the AEDC Tunnel B Hypersonic Wind Tunnel (Test OH 25B). Rockwell International, NASA-CR141546 (July 1975). 5. Pressure Test for the O.OlO-ScaleRockwell International Space Shuttle Orbiter Model (No. 61-0) in the Ames Research Center 3.5-Foot Hypersonic Wind Tunnel (Test OH 38). Rockwell International, SB 74-SH-0205 (5 June 1974). 6. James E. Carter, Numerical Solution of the Navier-Stokes Equations for the Supersonic Laminar Flow Over a 2-D Compression Corner. NASA TR-R 385 (July 1972). APPENDIX

Nomenclature C Chapman-Rubesin constant Cp, Cv specific heat at constant pressure, volume

G or

G(xo pressure

gradient coefficienl {11.622/[(dpldx,)X/(pmax)]} enthalpy characteristic length Mach number pressure transformed variables (Fig. 17) P2/Pl (upstream of separation) heat flux rate Reynolds number or unit Reynolds number time or temperature Cartesian coordinate system gap depth (Fig. 5) angle of attack wedge angle, cone angle Cp/Cv boundary layer thicknesses body flap deflection angle hypersonic similitude interaction coefficient

H L M P P, X r) t~ Rex or Refit T Xi.j,k Z a aw, ctc 7 ~, ,5", 0 ~°F )~,x Subscripts max maximum e edge condition W wall condition O at separation PL plateau 1 before shock wave 2 after shock wave or ref reference conditions