Ada 030015

AEDC-TR-7S-127 Ill
,U II
(i 11
[97t)
c':.-~\ ,4- "
ANALYTICAL MODEL
OF SUPERSONIC, TURBULENT, NEAR-WAKE FLOWS
ENGINE TEST FACILITY

ARNOLD ENGINEERING DEVELOPMENT CENTER
AIR FORCE SYSTEMS COMMAND
ARNOLD AIR FORCE STATION, TENNESSEE 37389
September 1976
Final Report for Period July 1973 - December 1975
Approved for public release; distribution unlimited,
t'~j
F'~uu:",\'
Prepared for
DIRECTORATE OF TECHNOLOGY (DYR)

ARNOLD ENGINEERING DEVELOPMENT CENTER
ARNOLD AIR FORCE STATION, TENNESSEE 37389
NOTICES
When U. S. Government drawings specifications, or other data are used for

any purpose other than a definitely related Government procurement 1
operation, the Government thereby incurs no responsibility nor any
obligation whatsoever, and the fact that the Government may have J
formulated, furnished, or in any way supplied the said drawings,
specifications, or other data, is not to be regarded by implication or
otherwise, or in any manner licensing the holder or any other person or
corporation, or conveying any rights or permission to manufacture, use, or
sell any patented invention that may in any way be related thereto.
Qualified' users may obtain copies of this report from the Defense
Documentation Center.
\
References to named commercial products in this report are not to be

considered in any sense as an endorsement of the product by the United
States Air Force or the Government.
This report has been reviewed by the Information Office (01) and is releasable
to the National Technical Information Service (NTIS). At NTIS, it will be
,
available to the general public, including foreign nations.
APPROVAL STATEMENT
This technical report has been reviewed and is approved for publication.
FOR THE COMMANDER
//JL ~. ...t'7 _ /2 -..#-

,?
~t::: l/?r~ ~~'1!!<J
ELTON R. THOMPSON ROBERT O. DIETZ

Research & Development Director of Technology
Division
Directorate of Technology
UNCLASSIFIED
READ INSTRUCTIONS
REPORT DOCUMENTATION PAGE BEFORE COMPLETING FORM
I. REPORT NUMBER GOVT ACCESSION NO. 3. RECI~IEN'''S CATALOG NUMBER
\2.
AEDC-TR-76-l27
4. TITLE (and Subtitle) 5. TyPE OF REPORT 1\ PERIOD COVERED
Final Report
ANALYTICAL MODEL OF SUPERSONIC, TURBULENT, 1973 July
December 1975 - -
NEAR-WAKE FLOWS 6. PERFORMING ORG. REPORT NUMBER
7. AUTHOR(s) 8. CONTRACT OR GRANT NUMBER(s)
C. E. Peters and W. J. Phares, ARO, Inc.

9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT, PROJECT, TASK
AREA a WORK UNIT NUMBERS
Arnold Engineering Development Center (DY)
Air Force Systems Command Program Element 65807F
Arnold Air Force Station, Tennessee 37389
II. CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATE
Arnold Engineering Development September 1976

Center (DYFS) , Air Force Systems Command 13. NUMBER OF PAGES
Arnold Air Force Station, Tennessee 37389 74

14: MONITORING AGENCY NAME /I ADDRESS(II dillerent Irom Controlllns Office) 15. SECURITY CL ASS. (of this report)
UNCLASSIFIED
ISa. DECL ASSI FICATION/ DOWNGRADING
SCHEDULE /
N A
16. DISTRIBUTION STATEMENT (01 thl. Report)
Approved for public release; distribution unlimited.
17. DISTRIBUTION STATEMENT (of the abstrsct entered In Block 20, II dillereni Irom Report)
18. SUPPLEMENTARY NOTES
Available in DDC
19. KEY WORDS (Continue on reverse side Jf necessary and Identify by block number)
;ma thema-ticai
-
model boundary layer transport properties
supersonic free stream turbulence
turbulent flow Mach number interactions
flows (near-wake) base pressure viscous flow
20. ABSTRACT (Continue on reverse side II necessary and Identify by block number)
An analytical model for planar and axisymmetric, supersonic,

turbulent, near-wake flows is presented. The viscous region
behind the blunt base is described by the integral form of the
boundary-layer equations, and the inviscid outer-flow region,
including the remnant of the initial turbulent boundary layer, is
computed with the rotational method of characteristics. The
solution of the two regions is fully coupled. A saddle-point
FORM
DO I JAN 73 1473 EDITION OF 1 NOV 65 IS OBSOLETE
UNCLASSIFIED
UNCLASSIFIED
20. ABSTRACT (Continued)
singularity, similar to the Crocco-Lees critical point, occurs

downstream of the rear stagnation point. The base pressure is
obtained by iteration of the initial conditions until the flow-
field solution will pass through the singularity. Base bleed of
a gas different from the outer-stream gas is included in the
formulation, and provision is made to treat equilibrium chemical
reactions in the viscous wake. However, an unresolved problem
has been encountered in the solution of the species conservation
equations. Therefore, results for only single gas flows are
presented. The analytical model is shown to adequately predict
the effect of free-stream Mach number and initial boundary layer
on the planar base pressure. In addition, the planar flow-field
structure is well predicted. Axisymmetric' base pressure and
flow-field structure are reasonably well predicted for free-
stream Mach numbers greater than 2.0, but the turbulent transport
model used yields only fair results for Mach numbers less than
1.7. The effect of base bleed on the axisymmetric base pressure
is well predicted.
AFSC
Arnold AFS Til""
UNCLASSIFIED
AEDC-TR-76-127
PREFACE
The work reported herein was conducted by the Arnold Engineering Development
Center (AEDC), Air Force Systems Command (AFSC). The results were. obtained by
ARO, Inc. (a subsidiary of Sverdrup & Parcel and Associates, Inc.), contract operator of
AEDC, AFSC, Arnold Air Force Station, Tennessee. The Program Element number was
65807F. The work was done under ARO Project Nos. RF423, R33P-60A, and R33A-02A.
The authors of this report were C. E. Peters and W. J. Phares, ARO, Inc. The manuscript
(ARO Control No. ARO-ETF-TR-76-69) was submitted for publication on July I, 1976.
1
AE DC-TR-76-127
CONTENTS
1.0 INTRODUCTION
1.1 Problem Definition 7
1.2 Review of Earlier Analytical Work . . . . . . . . . 11
1.3 Important Features of a Realistic Analytical Model 15
2.0 DEVELOPMENT OF ANALYTICAL MODELS
2.1 Basic Assumptions 16
2.2 Basic Equations . . . . . . . . . . . . . . 18
2.3 Shear Layer Profiles .......... . 20
2.4 Computation of Density in the Viscous Flow 21
2.5 Turbulent Transport Terms 21
2.6 Transformation of the Integral Equations 22
2.7 Solution of the Outer Flow and Coupling with the Viscous Layer 23
2.8 Expansion of the Initial Boundary Layer 24
2.9 Characterization of the Initial and Boundary Conditions 26
2.10 Two-Layer Treatment of the Outer Flow 28
2.11 Method of Solution ......... . 28
2.12 Model for the Turbulent Reynolds Number 31
3.0 DISCUSSION OF RESULTS
3.1 Concentration Field Solution Problems 38
3.2 Planar Base Pressure 38
3.3 Planar Flow-Field Structure 42
3.4 Axisymmetric Base Pressure 48
3.5 Effect of Base Bleed on Axisymmetric Base Pressure 50
3.6 Axisymmetric Flow-Field Structure 51
3.7 Dividing Streamline Properties 57
4.0 CONCLUDING REMARKS 59
REFERENCES 60
ILLUSTRATIONS
Figure
1. Features of the Near-Wake Flow Field 8

2. Examples of Supersonic Near-Wake Flows 10
3. Nomenclature for Near-Wake Analysis 16
4. Coupling of the Shear Layer and Outer-Flow Solutions 24
3
AEDC-TR-76-127
Figure
5. Corner Expansion of the Initial Boundary Layer 25

6. Boundary-Layer Profile Exponent for Adiabatic Flow
along a Flat Plate . . . . . . . . . . . . . . . . . . . . ........ 27
7. Boundary-Layer Momentum Thickness Ratio for Adiabatic Flow
along a Flat Plate . . . . . . . . . . . . . . . . . . 27
8. Two-Layer Treatment of the Supersonic Flow Field 28
9. Sub critical and Supercritical Solutions ...... . 29
10. Turbulent Reynolds Number for Planar Shear Layers with No
Secondary Flow .......... 32
11. Configurations for Low-Speed Separating and Reattaching
Flow Experiments ............ 33
12. Turbulent Reynolds Number Distribution Upstream of the
RSP in the Reattachment Regime . . . . . . . . . . . 34
13. Turbulent Reynolds Number Downstream of the RSP 35
14. Influence of the Turbulent Reynolds Number at the
RSP on the Predicted Base Pressure . . . . . . . . .. ............ 36
15. Turbulent Reynolds Number at the RSP as a Function of
the Pressure Gradient Parameter ............. . .......... 37
16. Effect of the Initial Boundary Layer on the Base Pressure in
Planar Flow (Mal = 1.5) ............................. 39
Planar Flow (Mal = 2.0) ............................. 40
Planar Flow (Mal = 3.0) ................. 40
19. Mach Number Effect on the Base Pressure for Planar Flow 41
20. Predicted Base Pressure for Planar Flow ........ . 42
21. Total Pressure at the Edge of the Shear Layer for Planar Flow 43
22. Centerline Pressure Distribution for Planar Wake
Flow (Mal = 2.0) ................................ 44
23. Centerline Pressure Distribution for Planar Wake
Flow (Mal = 3.05) .............. 45
24. Centerline Mach Number Distribution for Planar Wake
Flow (Mal = 3.05) .................. 46
25. Pitot Pressure Profiles for Planar Wake Flow (Mal = 3.05) 47
Axisymmetric Flow (Mal = 2.0) ......................... 48
4
AEDC-TR-76-127
Figure

Axisymmettic Flow (Mal = 3.9) . . . . . . . ______ . 49
28. Mach Number Effect on the Base Pressure for Axisymmetric Flow 50
29. Effect of Base Bleed on Axisymmetric Base Pressure 51
30. Total Pressure at the Edge of the Shear Layer for
Axisymmetric Flow ............. 52
31. Centerline Pressure Distribution for Axisymmetric Wake
Flow (Mal = 3.91) ................. . ............ 53
32. Pitot Pressure Profiles for Axisymmetric Wake Flow
(Mal = 3.91) .................... . ............ 54
Flow (Mal = 1.92) ................... . ......... 55
34. Centerline Mach Number Distribution for Axisymmetric Wake
Flow (Mal = 1.92) ................... . ......... 56
Flow (Mal = 2.03) ................... 57
36. Total Pressure along the Dividing Streamline for Planar Flows 58
37. Total Pressure along the Dividing Streamline for Axisymmetric
Flows ....................................... 59
APPENDIX
A. COEFFICIENTS OF THE INTEGRAL EQUATIONS . . . . . . . . . . . . . . . 65
NOMENCLATURE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5
AEDC·TR·76·127
1.0 INTRODUCTION
1.1 PROBLEM DEFINITION
Turbulent near wakes behind blunt bases in supersonic external flow occur in a
variety of flows of practical interest. The near-wake flow process (Fig. I) is characterized
by a strong interaction between the inviscid external flow and the turbulent portion of
the flow; therefore, computation of the entire flow field requires a coupled solution for
the viscous and inviscid regions.
Several features of the near-wake flow field are shown in Fig. 1. First, a
recirculation region exists just downstream of the base. With no bleed flow from the
base, the velocity on the centerline is zero at the base plane. When flow is bled into the
base, the onset of recirculation is delayed until the bleed flow has been entrained into
the turbulent shear layer. If the bleed flow is sufficiently large, the recirculation region
can be eliminated. As shown in Fig. 1, the pressure just downstream of the base is
considerably lower than the pressure on the body surface upstream of the base. The
centerline static pressure rises slightly with distance until the bleed flow is completely
entrained, then tends to decrease slightly with distance as the recirculation velocities
become larger. At some distance downstream of the base, the shear layer merges with the
axis and the flow begins to recompress. The recompression region is characterized by
large pressure gradients and by acceleration of the flow along the centerline. Downstream
of the rear stagnation point (RSP) the velocities throughout the viscous layer are positive.
If the flow is planar, and if the external flow upstream of the base is effectively
infinite and parallel to the centerline, then the centerline static pressure in the
recompression region monotonically rises to the free-stream static pressure. However, in
an axisymmetric flow, the pressure in the recompression region "overshoots" the
free-stream pressure (Fig. 1). Downstream of the recompression region the flow relaxes
toward a self preserving turbulent far-wake structure, in which static pressure gradients
are negligible.
Experience has shown that the base pressure and the near-wake flow structure are
dependent upon (1) the thickness and shape of the boundary-layer velocity profile just
upstream of the base, (2) the inviscid external flow properties (Mach number, flow angle,
etc.) at the base plane, and (3) the gas properties and flow rate of the base bleed. The
near-wake flow structure also depends on the lateral extent of the supersonic external
flow. If the external flow is bounded by a solid wall or by a free boundary, then waves
which are generated by the expansion process at the base plane will be reflected from the
boundary. These reflected waves can influence the near-wake structure if they intersect
the viscous wake upstream of the point where the flow on the axis has reaccelerated to
supersonic speed.
7
AEDC-TR-76-127
Recompression Shock Wave

/'
/'
/
,,// I
" / I ~
?~------~~~?>~~
~4
Co 7" u_ t<"'> 0")
v t
RSpJ
Recirculating Flow Turbulent Shear Layer
AXiSymmetric~
Static Pressure on
Body Ahead of Base
~~
---IIIo..A
Q)
~
::l
Base /- L Planar
V'l Pressure
V'l
Q)
~
Q)
0-
W
>
Q)
+-'
ro
c
+-'
(/')
Base Bleed ~
Q)
.~ 0
+-'
Velocity C
Q)
~
Q)
- --0 u
+-'
C
Q)
U RSP
Axial Dista nce from Base
Figure 1. Features of the near-wake flow field.
8
AEDC·TR·76·127
Several examples of supersonic near-wake flows which are of interest at the AEDC
are shown in Fig. 2. The classic turbulent near-wake flow is that behind bodies in
supersonic flight (Fig. 2a). Truncated plug exhaust nozzles (Fig. 2b) have been used for
both turbine engines and rockets. An accurate prediction of the plug base pressure is
required for the analytical assessment of the thrust performance of such nozzles. The
supersonic external flow is limited in lateral extent, and wave reflections from the free
boundary must be taken into account in the prediction of the near-wake structure.
Truncated plug nozzles are often operated with base bleed.
The annular ejector (Fig. 2c) is often used in place of the traditional ejector, which
has a centrally located supersonic driving nozzle. The supersonic external flow in the
annular ejector is bounded by a solid wall, and wave reflections from the boundary
strongly interact with the near-wake structure.
A highly complex near-wake flow structure is involved with the external burning
propulsion concept (Fig. 2d) which was first proposed by Strahle (Ref. I). According to
Strahle, waves are generated by the mixing and combustion process; these waves interact
with the near wake and cause the base pressure to be much higher than that with no fuel
injection. Indeed, Strahle suggests that the base pressure can be increased to a level
greater than free-stream static pressure, perhaps to a level where net thrust can be
obtained with a slender forebody. This external burning propulsion concept has been the
subject of several experimental investigations (e.g., Ref. 2). It is clear that an adequate
analysis of this complex flow must be based on a physically pe,rceptive analysis of the
simpler flow shown in Fig. 2a.
In view of the numerous practical flows in which supersonic near-wake mechanisms

occur, it was decided that a generalized analysis for such flows was needed at the AEDC.
The generalized analysis should include the following factors: (1) arbitrary turbulent
initial boundary layers; (2) arbitrary inviscid external flow profiles at the base plane; (3)
external flows which are bounded by a solid wall or by a constant pressure boundary; (4)
arbitrary base bleed, including gases which can react chemically with the external flow;
and (5) planar or axisymmetric geometry. The analysis should provide reasonably
accurate predictions not only of the gross features of the flow, such as the base pressure,
but also of the detailed flow structure in the viscous wake and in the adjacent inviscid
flow field.
9
»
~COWI m
o
~l /~/
()
Free Boundary ~
:0
/ ~
~/----.. C!l
~
I\J
--E -~
--.J
""-...'\
~ ~
'"
-----e.
'" "-\
---'"
"""-"'~
'- '\
\ ~
~
a. Bodies in supersonic flight b. TnJlncated plug nozzles
,...
o / '701 / / / / / I njection Shock Wave
(~~
/ / / / ////~
//////
M> 1
Pumped Flow
... /- . 1--
L/-- __
/ ~;.····"'~'-::,::~*«;s:~~.;;S~:~:::./~?:~JU;{;i·
L Fuel Mixi ng and
Generator Combustion Zone
c. Annular ejectors d. External burning propulsion
Figure 2. Examples of supersonic near-wake flows.
AEDC-TR-76-127
1.2 REVIEW OF EARLIER ANALYTICAL WORK
The analysis of turbulent near-wake flows has received much attention dUling the
past twenty-five years. Most of the work has been based either on the viscous interaction
model proposed by Crocco and Lees (Ref. 3) or on the component model first developed
by Korst for turbulent flows (Refs. 4 and 5). Most investigators have concentrated on
planar flows, which are inherently simpler than axisymmetric flows.
Crocco-Lees Model - This integral method is based on solving the integral form of
the boundary-layer equations for the viscous portion of the flow, including the reversed
flow region. The planar external flow field is coupled to the viscous region by the use of
the Prandtl-Meyer equations, which relate the local static pressure to the flow angle at
the edge of the viscous layer. The viscous layer growth rate is computed with an
empirical entrainment coefficient. A major problem in the original Crocco-Lees analysis is
establishing the relationship of the integral parameters of the viscous flow profiles to the
other variables of the problem.
A major feature of the Crocco-Lees solution is the appearance of a saddle-point

singularity (the critical point) in the recompression region. This singularity is interpreted
as being analogous to choking in the throat of a nozzle, i.e., the wake flow upstream of
. the singularity is subcritical and can transmit disturbances in the upstream direction. The
wake flow downstream of the singularity is supercritical, i.e., supersonic on the average.
The correct near-wake solution is taken as that which will pass through the singularity.
Although applicable to a broad class of strong viscous interaction problems, the

original Crocco-Lees analysis yields rather poor results for planar turbulent near wakes.
The first successful extension of the Crocco-Lees framework to the planar turbulent
near-wake problem was made by Alber (Refs. 6 and 7). Alber obtained the integral
parameters of the viscous flow profiles by using, in a transformed incompressible flow,
the Stewartson solutions of the Falkner-Skan equations. The empirical mixing coefficient
of the original Crocco-Lees analysis was replaced by the solution of the first moment of
momentum equation (the mechanical energy equation); the dissipation integral in this
equation was evaluated by the use of an empirical eddy viscosity model. Prandtl-Meyer
coupling with the external flow was retained.
It is only recently that an attempt has been made to extend the Crocco-Lees
framework to axisymmetric turbulent wake flows. Strahle and his coworkers (Refs. 8 and
9) have been developing an axisymmetric version of Alber's analysis, with the objective of
extending. it to the external burning po pulsion problem. The external flow is assumed to
be isentropic and is treated with an approximate method of characteristics technique.
(Prandtl-Meyer coupling is not applicable to axisymmetric flow). Only flows without base
11
AEDC-TR-76-127
bleed are considered. The solution for only one axisymmetric flow field has been
presented (Ref. 9).
Korst model - The Korst component model (Ref. 5) is much simpler than the
Crocco-Lees model and can be considered a first-order solution. The component processes
involved in the near-wake flow field (inviscid external flow, turbulent mixing,
recompression) are considered separately. The results from the component analyses are
then combined to yield a solution for the base pressure and for the other global
properties of the recirculating flow region. Three major assumptions are involved in the
basic Korst analysis: (1) the recirculation region is quasi-stagnant, (2) the flow along the
"discriminating" streamline (that which separates the flow which passes through
recompression from that which is recirculated upstream) isentropically recompresses, and
(3) the flow along the discriminating streamline stagnates at the peak of the
recompression pressure distribution. Although experimental evidence is available to show
that the first and third of these assumptions are quite incorrect, the basic Korst analysis
(especially the simplest version in which initial boundary-layer effects are neglected)
yields surprisingly good predictions of the base pressure for planar and quasi-planar flows.
The analysis has been extended to include base bleed, asymmetrical two-stream flows,
nonadiabatic flows, and flows with chemical reactions in the near wake (Ref. 10).
Numerous attempts have been made to improve the component analyses in the
Korst model. Nash (Ref. 11) introduced an empirical factor to account for the fact that
the discriminating streamline does not stagnate at the peak of the pressure distribution.
Other investigators (e.g., Lamb and Hood, Ref. 12) have developed control volume
analyses for the recompression region.
The basic assumptions in the basic Korst analysis are not very realistic. Because of
compensating effects, the simplest Korst analysis for negligible initial. boundary layers
yields fairly satisfactory base pressure predictions for experimental flows with initial
boundary layers of moderate thickness (0 less than the base height). Attempts to
properly include the influence of the initial boundary layers in the Korst analysis, using
techniques such as the apparent mixing layer origin shift concept of Hill and Page (Ref.
13), have not been very successful. However, these methods work reasonably well when
applied to a mixing process which does not involve a large expansion at the base plane.
The discrepancy for near-wake flows is related to the fact that the origin shift concept
includes the change in the boundary-layer mean flow profile when it passes through the'
corner expansion process, but does not take into consideration that the boundary-layer
turbulence decays rapidly when it passes through the expansion. This rapid decay of the
turbulent fluctuations in the "remnant" of the initial boundary layer was observed
experimentally by Lewis and Behrens (Ref. 14) and Page and Sernas (Ref. 15). These
12
AEDC-TR-76-127
observations lead to the conclusion that the boundary-layer flow after the corner
expansion should not be taken as the initial condition for the subsequent development of
the free turbulent mixing layer. Instead, the remnant of the boundary layer behaves as a
rotational but inviscid external flow, and a new turbulent layer begins to develop at the
edge of this rotational flow_
Several attempts have been made to apply the Korst method to the axisymmetric
base pressure problem. If the method of characteristics is used to compute the inviscid
flow adjacent to the constant pressure portion of the wake flow, then the solution
cannot be extended to the axis. Therefore, several investigators (e.g., Ref. 16) have used
the assumption that the flow recompresses, in a locally planar process, on an imaginary
sting which extends from the base. Mueller (Ref. 17) assumed that recompression occurs
either on a sting with a radius of one-half of the base radius, or on a sting equivalent in
radius to the experimental radius of the wake after recompression; the latter radius was
determined as an empirical function of the free-stream Mach number.
The simple recompression concepts of the original Korst theory are poorly suited for
the computation of axisymmetric near-wake flows. However, there are enough free
parameters in the axisymmetric extensions of the Korst model so that reasonably
adequate values for the base pressure can be obtained. But these models really are used
to correlate experimental base pressure measurements rather than to predict them.
Moreover, the Korst-type models do not provide any information about the detailed flow
structure in the recompression region.
Other methods - McDonald (Ref. 18) took a unique approach to the prediction of
the planar base pressure for flows which reattach to a solid wall. (The centerline of Fig. I
is replaced by a wall). McDonald also applied his analysis to axisymmetric flows in which
a relatively large sting extends downstream from the base (Ref. 19). The base pressure is
uniquely determined as that which causes the reattached flow downstream of
recompression to become a fully developed flat plate boundary layer. Although
McDonald produced reasonable predictions of the base pressure, his model can be
criticized on physical grounds.· Numerous experiments have shown that the planar
near-wake structure for reattachment to a plane of symmetry is not significantly different
from that for reattachment to a wall. Therefore, wall viscous effects cannot be a major
controlling factor in the near-wake structure.
Chow (Ref. 20) has developed a two-layer integral treatment of recompression in

planar flow; in many aspects, this physically realistic model combines features of both
the Crocco-Lees and Korst models, Chow found it necessary to include the lateral
pressure gradients in the recompressing viscous layer; however, this need to include lateral
pressure gradients is probably related to the assumed geometry of the dividing streamline.
13
AEDC-TR-76-127
This recompression analysis is applicable only to the region upstream of the RSP (Fig. I).
The predictions of the Chow model are discussed in Section 3.0.
Recently, Chow and Spring (Ref. 21) developed a method for predicting the wake
development downstream of the RSP. In addition, they improved the computation of the
turbulent mixing upstream of the onset of recompression, and quite good predictions of
planar base pressures were obtained.
Weng (Ref. 22) has extended the Chow recompression model to axisymmetric wake
flows in which a sting extends from the base; it seems that this model should, in
plinciple, also apply to flows without a sting. In common with the models of Chow and
of Chow and Spring, the finite recirculation velocities are neglected in the flow upstream
of the recompression region. The external flow is computed by the method of
charactelistics downstream to the onset of recompression; Prandtl-Meyer coupling is used
in the recompression region. The initial boundary-layer profile after the base expansion
process is taken as the initial condition for the development of the free shear layer. Weng
used Prandtl mixing length theory to compute the turbulent shear stresses in the flow
upstream of the onset of recompression; in the recompression region, he used an arbitrary
spatial variation of the eddy viscosity.
Weng obtained reasonably good predictions of the effect of sting-base radius ratio on
the base pressure for free-stream Mach numbers from 1.5 to 2.5.
Summary of Earlier Analytical Work - Of the available analytical models for planar
wake flows, those of Alber (Refs. 6 and 7) and Chow and Spring (Ref. 21) are the most
perceptive. Both models provide a prediction or the flow-field structure as well as of the
base pressure. The Chow and Spring model provides the best prediction of the effect of
initial boundary layer on the base pressure over the Mach number range investigated, but
Alber's model is attractive because finite reverse flow velocities are included in the region
upstream of the onset of recompression. Neither of these models has been applied to
flows with base bleed; indeed, only the Korst-type models have been used to predict the
effects of base bleed.
Of the available models for axisymmetric wake flows, only those of Strahle, et al.
(Refs. 8 and 9) and Weng (Ref. 22) are based on a reasonably adequate description of
the detailed flow processes. Strahle, et aI., have not yet included initial boundary-layer
effects in their model. As discussed in Section 3.0, Weng's method of computing the
initial boundary-layer effect will probably be inadequate for free-stream Mach numbers of
three or grea ter.
14
AEDC-TR-76-127
It is clear that even the recently developed models for axisymmetric flow do not
satisfy the critella established in Section 1_1 for a generalized near-wake analysis. The
void between the requirements and the capabilities of available models was even more
pronounced at the time that the work described in this report was initiated. Therefore, it
was decided to undertake the development of an analytical model which would be
applicable to a broad range of supersonic near-wake flows (Section 1.1).
1.3 IMPORTANT FEATURES OF A REALISTIC ANALYTICAL MODEL
When one considers the earlier analytical work, as well as the available experimental
information,' it becomes clear that a perceptive model of the near-wake structure should
include several features. First, adequate treatment of the external flow in the
axisymmetric case requires the use of the method of characteristics. Second, the evidence
on quenching of the turbulence in the expanding boundary layer suggests that the
remnant of the initial boundary layer can be computed with the rotational method of
characteristics. Third, as shown by Weinbaum (Ref. 23), the viscous forces are negligible
compared to inertial forces during the turning of the initial boundary layer, and the
expansion of the supersonic portion of the initial boundary layer can be computed with
the rotational method of characteristics. (A streamtube method can be used to compute
the expansion of the subsonic portion of the initial boundary layer). Fourth, a new
turbulent shear layer should be considered to begin after the corner expansion process. It
should be noted that the concept of a new shear layer which develops downstream of the
corner expansion has been suggested by Korst (Ref. 24) and has been used in the planar
analysis of Chow and Spring (Ref. 21).
The available experimental evidence suggests that the near-wake region between the
onset of recirculation and the beginning of recompression can be considered to be a
constant pressure region, as has been assumed by most investigators. However, the
experimental evidence does not suggest that the reverse flow velocities can be neglected,
as has been done in the Korst-type models; the finite recirculation velocities must be
included.
The analytical model which is described in Section 2.0 incorporates the features
which have been discussed. The framework of the analysis is conceptually similar to the
Crocco-Lees approach. However, the detailed treatment of the flow processes is closely
related to the method used by the authors for dueted two-stream mixing flows (Refs. 25
and 26).
15
AEDC-TR-76-127
2.0 DEVELOPMENT OF ANALYTICAL MODEL
The geometry and nomenclature for the near-wake analysis is illustrated in Fig. 3.
The flow downstream of the base is divided into three distinct regimes: (1) the "jet
regime" which extends from the base plane to the station where recirculation begins, (2)
the "wake regime II which extends from the onset of recirculation downstream to the
onset of recompression, and (3) the "fully developed" regime which begins at the onset
of recompression and extends downstream into the fully developed far-wake region.
6 ] PI /~
/</~~ //
/ / / /' /'
</ ./
7 Z Upper Boundary ofExternal
Flow (Prescribed Pressure or /
Prescribed Geometry)
~ /:~ /' ~ Profile of I
22277 ~ ~ us~/ I Axial Velocity /' /~//

Pb ~ --_ Us /' -;:/ / / /
~
_ /'/'//1
uil - b- _ /' /' /' / / /
~ ~f--rs--- s_
/' / / / , u
r r·~ ----
U I rm I ~ -- ~ -------~--
c
--'---------ll~ - ~ - /-
Beginning of Recirculation--'-/ ~ L RSP

Beginning of
~ ~~~ime
Press ure Rise
Wake Fully Developed
Regime ----to+-- Regime
Figure 3. Nomenclature for near-wake analysis.
2.1 BASIC ASSUMPTIONS
The major assumptions used in the analysis are:
1. The flow is steady and is either planar or axisymmetric.
2. All gases obey the perfect gas law.
3. The integral form of the boundary-layer equations describes the flow

between the axis and the outer edge of the turbulent mixing layer, rs.
4. The turbulent Prandtl and Schmidt numbers are unity.
16
AEDC-TR-76-127
5. The free shear layer which forms downstream of the base expansion process
is fully turbulent (negligible laminar transport) and has a negligible initial
thickness_
6. The flow outside of the free turbulent shear layer (r > rs) is supersonic,
adiabatic, and inviscid.
7. The upper boundary of the supersonic external flow is inviscid and is either
a prescribed geometry or has a prescribed constant pressure.
8. The normalized profiles of the axial component of velocity in the turbulent

shear layer (between ri and rs) are similar and are represented by a cosine
function.
9. The element species and axial component of velocity between the axis and
ri are one-dimensional.
10. The centerline static pressure in the jet regime is an isentropic function
of u e .
11. The centerline static pressure is constant in the wake regime.
12. The initial boundary layer is turbulent and is characterized by its thickness
and a power law velocity profile.
13. Pressure feedback effects in the boundary layer upstream of the base are
negligible.
14. The expansion of the initial boundary layer at the base plane is inviscid.
15. The base bleed flow is uniform and subsonic at the base plane.
16. Chemical reactions in the viscous wake flow, if they occur, are assumed
to be in eqUilibrium.
17. If the planar flow reattaches to a wall rather than to a plane of symmetry,
then viscous effects along the wall are negligible.
The use of the boundary-layer approximations (assumption 3) is consistent with

Crocco-Lees and implies that (1) there are negligible lateral pressure gradients in the
viscous wake flow, and (2) streamwise turbulent transport mechanisms are negligible. The
assumption of negligible lateral pressure gradients is reasonably valid in the jet and wake
regimes because the curvature of the streamlines is small in the high-speed portion of the
17
AEDC-TR-76-127
viscous layer; this assumption is less valid in the recompression region of the fully
developed regime. The use of a cosine profile in the shear layer, along with a uniform
reverse flow profile, was suggested by Green (Ref. 27) for incompressible wake flows. It
should be noted that, even though the common term "base bleed" is used in this report,
the momentum of the stream issuing from the base is not neglected; therefore, the
. analysis is applicable to "bleed" velocities in the high subsonic range.
2.2 BASIC EQUATIONS
The basic boundary-layer equations are:
Continuity
o (1)
where
a 0 for planar flow and
a 1 for axisymmetric flow

Axial Momentum
pur
a au
ax + pvr
a au
a-; = ara (rr a) (2)
where
T is the turbulent shear stress
Element Species Conservation
pur a - ac + pvr a _ac a ( a)

ax ar = -
ar qr (3)
where q is a turbulent species flux.
Equations (1) through (3) are integrated (Ref. 25) to obtain five integral equations:
(1) a continuity equation for the flow between the centerline and r s , (2) a momentum
equation for the flow between the centerline and r s , (3) a momentum equation for the
flow between the centerline and rm (where fro == fi + b/2), (4) a species conservation
equation for the flow between the centerline and r s , .and (5) a species conservation
equation for the flow between the centerline and rm. The integral equations are:
18
AEDC-TR-76-127
Overall Continuity
_ p v ra
s s s (4)
Overall Momentum
-p v u r a - (5)
s s s s ~ (a + I)
Half-Radius Momentum
(6)
dx (a + I)
Overall Species
o (7)
Half-Radius Species
(8)
Because· the flow is one-dimensional for 0 < r < ri, the integral equations can be
rewritten as follows:
Overall Continuity
r
a
! 1
s
ax (pu)ra dr (9)
Overall Momentum
d Pc rs
a+1 a+ 1
ri d 2
- P s vu r
a
- - - - -(a+ 1) dx(Pcuc)
(10)
s s s dx (a + 1)
Half-Radius Momentum
1
dp r a+
T ra _ c m
m m d;- (a + 1)
(11)
a+ 1 a+1
rid
_r_i_ ~( u 2 ) (p u )
(a + 1) dx Pc c + urn (a + 1) ~ c c
19
AE DC-TR-76-127
Overall Species
a+1
f i d
-- - (p U C ) (12)
(a + 1) dx e e e
Half-Radius Species
a+1 a+1
fi d fi d
qmr~ - (a + 1) dx (PeueCJ + Cm (a + 1) dx (Peu e )
(13)
Equations (9) through (13) apply to all three regimes of Fig. 3. Of course, ri goes to
zero in the fully developed regime, the pressure gradient terms are zero in the wake
regime, and Ce = Cj = I in the jet regime.
2.3 SHEAR LAVER PROFILES
The free shear layer velocity profile is given by
u - ue
= -
1
-
1
- cos
(.
1T-
f- f i) (14)
u - u 2 2 b
s e
The velocity at the half-radius surface, rro (= ri + b/2), is given by

1
+-(u
2 s -u)
e
(15)
If the shear layer were at constant pressure, with constant values of Us and u e , then
the element species concentration profile would be given by
or
c - Ce u-u
e
cs - ce us - ue
(16)
which is the well known Crocco integral solution obtained from Eqs. (2) and (3). In the
near-wake problem, however, the pressure gradient terms are negligible only in the wake
regime. In addition, Ce , ue , and Us vary with x, even in a region of constant pressure.
Therefore, Eq. (16) cannot be used. The approach which has been taken in this study is
to assume that the deviations from Eq. (16) can be expressed as
c - Ce
(17)
where K can vary with x. It should be noted that a power law relation similar to Eq.
(17) has been used to compensate for the deviations from the Crocco relation which are
20
AEDC-TR-76-127
caused by non unity transport coefficient ratios in free turbulent flows (Ref. 28). Since Cs
is defined to be zero, Eq. (17) becomes
C=c c - cc , u
{U_U)K (18)
_ uc
s c
2.4 COMPUTATION OF DENSITY IN THE VISCOUS FLOW
For unity turbulent Prandtl and Schmidt numbers, the stagnation enthalpy and the
gas properties are uniquely related to C, and the density field is calculated with the
techniques described in Ref. 25. The stagnation temperature, specific heat, and gas
constant are specified as a function of C, which varies from zero (pure outer-stream gas)
to one (pure base bleed gas). For chemically frozen flow, only the To, cp and R values at
C = 0 and at C = 1 need be specified. For flow with equilibrium chemical reactions, To,
cp and R are specified as functions of C.
At any point in the shear layer, u is given by Eq. (14), C is given by Eq. (18), and
To, c p , and R are then determined. The density can be obtained from the energy
equation and the perfect gas law. Therefore, the density can be expressed functionally as
P = P (u, c, p) (19)
2.5 TURB,ULENT TRANSPORT TERMS
The turbulent shear stress, T m , which appears in Eq. (11) is given by
r
m -
-
Pmt ar
- au Im (20)
where € is the turbulent eddy viscosity. The eddy viscosity is given by
t = _1_
RT
b lu - ucl
S
(21)
where RT is a turbulent Reynolds number based on the eddy viscosity, the shear layer
width and the velocity difference across the layer; the model used to obtain RT is
described in Section 2.12.
The equation for Tm becomes, for the cosine velocity profile,
r
m
= _TT_
2RT
P
m
lu s - uc I(u - uc ) S
(22)
21
AEDC-TR-76-127
The turbulent species flux, qm, is given by
or
(23)
2.6 TRANSFORMATION OF THE INTEGRAL EQUATIONS
Jet regime - Assuming that the flow conditions along rs can be specified, sufficient
information is available to relate the terms in Eqs. (9) through (13) to four flow-field
variables. Since there are only four unknowns in the jet regime, Eq. (13) is not used. The
four dependent variables were selected to be Pc, ri, band K. Equations (9) through (12)
can be transformed into the following system (Refs. 25 and 29)
(24)
where N == F for the continuity equation, N == G for the overall momentum equation, N
== H for the half-radius momentum equation, and N == I for the overall species
conservation equation. The equations for the coefficients are presented in Appendix A.
The numerous auxiliary equations necessary for computation of the coefficients are
developed in the same manner as described in Ref. 25.
Equations (24) are solved for the derivatives (dPc/dx, drddx, db/dx, dK/dx) by use
of a matrix factorization technique. The resulting equations for the derivatives are
numerically integrated with a modified Euler technique (variable step size).
Wake regime - The five dependent variables selected in the wake regime are u c , ri, b,
K and Ce . Equations (9) through (13) are transformed into the following system:
duc dri db dK de c
N 1 dx + N2 dx + N3 dx + N4"";h + N5 dx = N6 (25)
where N == J for the half-radius species equation. The equations for the coefficients are
presented in Appendix A.
22
AEDC·TR·76·127
Fully developed regime - The five dependent variables selected in the fully developed
regime are Pc, U c , b, K and Cc . Equations (9) through (13) are transformed into the
following system:
(26)
The equations for the coefficients are presented in Appendix A.
2.7 SOLUTION OF THE OUTER FLOW AND COUPLING

WITH THE VISCOUS LAVER
The supersonic flow in the region r > rs is computed with the rotational method of
characteristics. The "reference plane" technique, in which the supersonic flow parameters
are established along planes normal to the axis, is used. The axial location of the
reference planes coincides with the axial stations used in the numerical integration of the
viscous layer equations.
Many terms which appear in the coefficients,N, of the viscous layer equations
depend on the flow conditions (and their derivatives with respect to x) along r s, the
interface between the viscous region and the outer supersonic flow. Therefore, the
solution of the two regions must be coupled. The technique used for coupling the two
regions is illustrated in Fig. 4. At some station, x, the flow in both regions is fully
specified. In the integration technique for the viscous layer equations, the step size, &, is
given, as are the tentative values of p and rs at (x + L'lx). A left running characteristic,
along which the flow is fully defined, is constructed from the the last boundary point
(Fig. 4). The right running characteristic and the streamline, each of which passes through
the new boundary point and intersects the left running characteristic, are constructed; the
equations for the streamline and the right running characteristic are solved in the usual
iterative manner to define the entropy and flow angle at the new boundary point. Then
the density and velocity at the new boundary point are computed, and the derivatives
with respect to x of these flow properties along rs are computed with an upstream
differencing technique. (These derivatives are assumed to be zero at x = 0). Therefore,
sufficient information is available about the flow properties along rs so that the
coefficients of the viscous layer equations at (x + L'lx) can be evaluated.
Within the framework of the Euler integration technique, this boundary point
procedure is repeated, with different values of Pc and rs at (x + L'lx); until the solution of
the viscous equations converges. The solution then proceeds downstream to a new axial
station. In this downstream marching procedure, development of the solutions for the
two flow regions is fully coupled.
23
AEDC-TR-76-127
/'
./
/'
Streamline Which Passes through /'
,/
New Boundary Point

/,,/"'''-- Left Running Characteristic
/' /' along Which Flow Is Fully Defined
,/ '"
"': / Right Running Characteristic
/
'" "-
"-
"-
"-
~
Upper Edge of Turbulent Shear
Last Boundary Point Layer, rs(x)
-------f~--fj.x
X
New Boundary Point
Figure 4. Coupling of the shear layer and outer-flow solutions.
Even though the outer flow is treated with the rotational method of characteristics,
no shock waves are allowed to develop. However, two shock waves exist even in simple
near-wake flow fields, the lip shock and the recompression shock. The lip shock, which
has been experimentally investigated by Hama (Ref. 30), results from the overturning of
the low Mach number portion of the initial boundary layer in the base expansion process
(Ref. 23). In neglecting the lip shock, we have treated it as an' isentropic compression
which is justified since experiments have shown that the lip shock is rather weak, at least
for free-stream Mach numbers up to 2.0 or 3.0.
The series of compression waves which result from the recompression process
coalesce to form the recompression shock wave (Fig. I). However, the shock is formed
too far downstream to affect the near-wake closure mechanism. Therefore, neglecting the
recompression shock wave influences only the downstream portion of the recompression
process.
Inclusion of the lip and recompression shock waves in the present analysis is quite
feasible, but would require a significant increase in the computation time.
2.8 EXPANSION OF THE INITIAL BOUNDARY LAYER
The expansion of the supersonic portion of the initial boundary layer (Fig. 5) is
computed with the rotational method of characteristics. Since pressure fccdback effects
are neglected, the supersonic expansion process is initiated at x = 0 with a Prandtl-Meyer
24
AE DC-TR-76-127
turn (from PI to Pb) at the sonic point in the boundary layer. Expansion of the subsonic
portion of the initial boundary layer is computed with an isentropic stream tube model.
The subsonic layer is divided into ten streamtubes, each of which expands isentropically
from PI to Pb. The flow direction of the stream tubes after expansion is taken to be the
same as that of the initially sonic streamline after it passes through the Prandtl-Meyer
turn. The flow in the wall streamtube, after expansion, is the initial edge condition (at x
= 0) for the new turbulent shear layer which develops downstream of the base.
ual r--- Velocity Profile in Initial

Sou ndary Layer
T [) Supersonic
---,/
./..---
1:;//
/./
~ Prandtl-Meyer Expansion
./
./
..---
t
___ ...______.~ at x :: 0 (from PI to Pb)
Sublayer Velocity
Profile after Turn
I nitial Val ue of Velocity at

Edge of Shear Layer
Figure 5. Corner expansion of the initial boundary layer.
For values of Pb /PI greater than the sonic static-total-pressure ratio, one or more of
the streamtubes will be subsonic after the expansion process. In such cases, the Mach
number of the subsonic streamtubes is arbitrarily adjusted to be slightly greater than
unity. The effect of this adjustment process is very small for typical initial
boundary-layer thicknesses because the mass flow in the adjusted streamtubes is small and
is quickly entrained by the turbulent shear layer.
25
AE DC-TR-76-127
2.9 CHARACTERIZATION OF THE INITIAL AND BOUNDARY CONDITIONS
The turbulent near-wake flow field is fully determined by the following conditions,
which must be specified:
1. The upper boundary of the supersonic external flow,
2. The base bleed flow rate,
3. The total temperatures and gas properties of the external flow and the
bleed flow,
4. Mixture gas properties and total temperature as a function of C, if chemical

reactions occur (Section 2.4),
5. Thickness, 8, and velocity profile shape of the adiabatic initial boundary

layer, and
6. The profiles at the base plane of the static pressure, total pressure and
flow angle, between the edge of the boundary layer and the upper boundary.
For computation of near-wake flows embedded in external flows of large lateral

extent (effectively infinite), the upper boundary is specified as a wall parallel to the
centerline and located far enough from the near wake so that reflected waves from the
corner expansion process do not interact with the recompression region.
The initial boundary-layer velocity profile is assumed to be given by a power law
where Ual is the velocity at the edge of the boundary layer, 8 is the thickness of the
layer, y is the distance from the wall (Fig. 3), and n is specified.
The initial boundary-layer characteristics for near-wake expeIiments are usually

specified in terms of the momentum thickness and the momentum thickness Reynolds
number, Reo. Therefore, a method was developed to estimate n as a function of Mal and
Reo. Maise and McDonald (Ref. 31) showed that supersonic boundary-layer profiles on
an adiabatic flat plate can be well represented by the Coles
law-of-the-wall/law-of-the-wake profiles, along with the Van DIiest transformed velocities
and the Spalding-Chi skin fIiction values. The 8/8 values shown by Maise and McDonald
as a function of Ma 1 and Reo, were matched with 8/8 values for power law profiles (Ref.
32) to determine n as a function of Mal and Reo (Fig. 6). The relation of n, Mal, and
8/8 is shown in Fig. 7.
26
AEDC·TR·76·127
The power law profile, as well as the profile used by Maiseand McDonald, does not
smoothly join the external stream velocity profile. The resulting II corner ll on the
total-pressure profile was found to cause computational problems in the viscous layer
solution when the corner is entrained into the shear layer. To alleviate this problem, the
outer edge of the initial boundary-layer velocity profile is smoothed with a cubic
function which extends from (rb + 0.88) to (rb + 1.28). This smoothing has a negligible
effect on the momentum thickness.
0.20F--_ _
0.18
U/U
al = (y/o)n
)'a = 1.4
c
c· 0.16
Cl>
c
0
c.
><
UJ
.....
Cl> 0.14
>-
'"
-'
I
>-
'-
'"c
'0 0.12
::l
0
c:>
0.10
0.08
0 2 3 4 5
Free-stream Mach Number, Mal
Figure 6. Boundary-layer profile exponent for adiabatic flow along a flat plate.
0.20
0.18
c
'E. 0.16
Cl>
c
o
c.
x
UJ
~ 0.14
>-
:3
&
-E O. 12 (y/o)n
::l
o
U/U
al =
CCl
)'a = 1.4
0.10
Momentum Thickness Ratio, 8/0
Figure 7. Boundary-layer momentum thickness ratio for

adiabatic flow along a flat plate.
27
AEDC-TR-76-127
2.10 TWO-LAYER TREATMENT OF THE OUTER FLOW
It was discovered early in this investigation that the entropy profile in the initial
boundary layer tends to spread into the outer flow, which is usually irrotational. Of
course, the diffusion of the entropy field across streamlines is incorrect since the flow is
assumed to be inviscid. Apparently, this false diffusion in the method of characteristics
flow field is related to the rapid expansion of the highly rotational boundary-layer flow.
The false diffusion effect is not unique to the present method of characteristics
formulation; other widely used method of characteristics computer programs were found
to yield similar results_
To eliminate the spurious entropy diffusion, a two-layer technique is used (Fig. 8).
The streamline which originates at the outer edge of the inital boundary layer is
established, and the total pressure on this streamline is specified to be Po a' Therefore, the
entropy field in the supersonic flow is divided into two layers, and the entropy gradients
from the initial boundary layer are confined to the lower layeL
Left Running Characteristics Which

Originate at Prandtl-Meyer Expansion ,
~
Velocity Profile in
/
/ // // / /
/,/ /
Initial Boundary Laye7 ;' / / ,// / / Streamline Which Originates at

/ / / // Outer Edge of Initial Bou ndary Layer
---'------:/-"---,t._ "
/:;/
//
1///
/1//
Subsonic Layer 7
Figure 8. Two-layer treatment of the supersonic flow field.
2.11 METHOD OF SOLUTION
As in the Crocco-Lees analysis, a saddle-point singularity (the critical point), which

is intrinsic to the solution, occurs in the reattachment region. The behavior of the
solution is illustrated by the axial distribution of centeriine velocity (Fig. 9). The correct
solution is that which passes smoothly through the singularity. If the assumed base
28
AEDC-TR-76-127
pressure is too low the solution will be supercritical and "blow up", i.e., in terms of
Cramer's rule, the determinant of the coefficients for Eqs. (26) will go to zero before the
numerator determinants do. If the assumed base pressure is too high, the solution will be
sub critical and is characterized by the centerline velocity reaching a maximum, then
decreasing with distance (Fig. 9). By iterating on the initial conditions (base pressure),
one can converge on the set of initial conditions which will cause the flow-field solution
to pass smoothly through the singularity.
(+)
Critical Poi nt
U Supercritical
:::J
Solutions (Base
~
....... Pressure Too Low)
'g
Q,)
>Q,)
c:
....
Q,) Subcritical Sol utions (Base
.......
c: Pressure Too High)
Q,)
u
0
/ Axial Distance from Base, x
(-) /Figure 9. Subcritical and supercritical solutions.
The sub critical solutions are identified by dUe /dx going to zero. The supercritical
solutions are identified by an inflection in the centerline velocity distribution, i.e., by
d 2 Uc /dx 2 going to zero, which implies that the determinant is approaching zero. A
halving technique, with the bounds established by the last subcritical and the last
supercritical values of Pb, is used to converge to the correct value of Pb. With a
reasonable first guess on the bounds for Pb, say within 20 percent of the correct value,
the converged value of Pb can be identified to within 0.5 percent with approximately ten
iterations of the flow-field solution.
Although it is not difficult to define the initial conditions with practical engineering
accuracy, establishing the solution through and beyond the critical region is very difficult.
The nature of a saddle-point singularity is such that very small changes in the initial
conditions result in large changes in the solution as the singularity is approached. Two
29
AE DC-TR-76-127
techniques have been used to obtain solutions which pass through the critical region. The
first is a "brute force II approach; the correct initial condition is established with a very
small tolerance, typically to within one part in 10 5 . At a station just upstream of the
critical region, the centerline velocity is perturbed upward by a small amount (typically
one percent) and the solution will usually proceed smoothly through the critical region.
This procedure has been found to work fairly well for planar flows, and for axisymmetric
flows with high external stream Mach numbers. However, for axisymmetric flows with
moderate external stream Mach numbers (2.0 or 3.0), this procedure has not been
successful, even when the initial conditions are converged to within one part in 106 .
An alternate approach for developing the solution through the critical region is to
converge the initial conditions fairly tightly, perhaps to within one part in 10 3 or 10 4 ,
then to extrapolate the centerline velocity distribution. Results at two axial stations near
the RSP, which is well upstream of the critical point, are used to establish an exponential
extropolation equation for uc(x). One equation, the continuity equation, is removed from
Eqs. (26), and the singular behavior of the solution is removed. Experience has shown
that the exponential extropolation can be extended downstream, with good results, to
the peak of the centerline pressure distribution in axisymmettic flow.
Results obtained with both of the techniques for developing the solution through
the critical region are presented in Section 3.0.
Fortunately, the engineer is usually interested only in a prediction of the initial

conditions (the base pressure). The initial conditions, and the flow field downstream to
the vicinity of the RSP, can be determined with reasonable accuracy without developing
the solution through the critical region.
The occurrence of the saddle-point singularity in the solution is a result of the

elliptic nature of the near-wake flow field. That is, an essentially elliptic problem is
treated as an iterative initial value problem, and the critical point is the mechanism
whereby the downstream wake closure conditions are manifested.
The near-wake analysis has been programmed for numetical solution with an IBM
370-165 computer. A typical computation, for one assumed set of initial conditions,
which proceeds from the base plane to downstream of the RSP requires a computation
time of from 30 to 60 seconds. The determination of the base pressure to practical
engineering accuracy (less than one percent uncertainty) involves a computation time of
five to ten. minutes.
For near-wake flows in which there is no base bleed, or in which the base bleed gas
has the same total temperature and gas properties as the outer flow, a simplified version
30
AEDC-TR-76-127
of the computer program is used. The concentration equations are not solved, and K and
Cc are set at unity. This simplified analysis is referred to as the "single gas" version.
2.12 MODEL FOR THE TURBULENT REYNOLDS NUMBER
As with any turbulent flow problem, the near-wake analysis requires an adequate
model for the turbulent transport properties. Early in this investigation, an attempt was
made to incorporate an integral formulation of the turbulent kinetic energy (TKE)
equation into the near-wake analysis. This TKE formulation has been shown to yield very
good results for a variety of constant pressure-free turbulent flows (Ref. 33). But the
results for supersonic near-wake flows were very poor; the turbulent shear stress levels in
the reattachment region were grossly underpredicted. The failure of the TKE formulation
is attributed to the "extra strain" effects described by Bradshaw (Ref. 34). He points out
that additional strains in the flow, in addition to the basic au/ar strain, have a
disproportionately large effect on the turbulence structure. In the recompression region,
the strains which result from both streamline curvature and dilatation (bulk compression)
effects are significant.
Because of the failure of the TKE approach, we decided to develop an eddy

viscosity model for the near-wake analysis, even though the history of eddy viscosity
models (Ref. 35) indicates that such a model would probably have a limited range of
applicability.
The eddy viscosity model described in Section 2.5 is the Prandtl model with the
coefficient expressed as the inverse of the turbulent Reynolds number, R T . In the jet and
wake regimes, RT is assumed to be a function only of the local Mach number, Ms , at the
supersonic edge of the shear layer. The predictions of the integral TKE method (Ref. 33)
agree well with the available experiments on the growth of planar shear layers with zero
secondary velocity; the corresponding RT values from the TKE analysis (Fig. 10) are well
described by the function
(27)
which is used in the jet and wake regimes.
The RT model used in the fully developed regime is based on the experimental
results of Mueller (Ref. 36) and Tani (Ref. 37) for incompressible reattaching planar
flows. Their experimental configurations are shown in Fig. 11. The experimental velocity
profiles were approximately fitted with a cosine function to estimate the width of the
shear layer and the effective wall slip velocity. The results, along with the measured shear
stress at the center of the velocity profile, were used to cbmpute R T . Two characteristic
RT values are of interest, the value at the beginning of reattachment, R Tp and the value
31
AE DC-T R-76-127
at the RSP, RT r. (The beginning of reattachment is defined as the axial station where the
wall static pressure is equal to the base pressure.) The values for RTp and RTr are shown
below.
Experiment RTp RTr
Mueller, h = 0.75 in. 276 119

Mueller, h = 0.50 in. 216 106
Mueller, h = 0.25 in. 219 101
Tam, h = 2 cm 295 108
Because of the consistent values for R Tr , from 100 to 120, we speCUlated that RTr
would be a good parameter to characterize the level of the eddy viscosity in the
reattachment region.
400
II-
~ 300
E
::l
Z
VI
"0
"0
c:
»
Cl)
0:::
o TKE Predictions for Fully
Developed Shear Layers (Ref. 33)
~ 200
::l RT = 390 - 333 exp (-0. 495 Ms)
.0
I-
::l
I--
lOO~------~------~------~------~------~------~
o 1 2 3 4 5 6
Mach Number at Edge of Shear Layer, Ms
Figure 10. Turbulent Reynolds number for planar shear layers with no secondary flow.
32
AEDC-TR-76-127
Velocity Profile
a. Mueller's configuration
Dividing
7 7 777
--
/
,
Streamline
b. Tani's configuration
Figure 11. Configurations for low-speed separating and reattaching
flow experiments.
From the experimental results of Mueller and of Tani, it was found that the
distribution of RT in the reattachment region could be taken to be a unique function of
uc/u s ; Us is the local velocity at the edge of the viscous layer and Uc is the effective wall
slip velocity. Results are shown in Fig. 12 for the region between the beginning of
reattachment and the RSP. The data are well represented by
(28)
where u= (uc/us)/(uc/us)p' Downstream of the RSP, the data (Fig_ 13) are well represented
by
RT/RT r
= exp
e (-2.5 s
u lu ) (29)
In the near-wake analysis, Eqs. (28) and (29) are used to express the axial variation
of R T , and RT p is given by Eq. (27). Therefore, the level of the eddy viscosity
throughout the fully developed regime is scaled by R Tr .
33
AEDC-TR-76-127
1.0
Experi ment:
Mueller (Ref. 36)
o h=0.75in.
0.8 o h = 0.50 in.
6. h = O. 25 in.
Ta ni (Ref. 37)
0.6
o h = 2 em
0.4
ro
E
:b.
o
O. 2
Z
O----------------~------~------~------
o O. 2 0.4 0.6 0.8 1.0
Figure 12. Turbulent Reynolds number distribution upstream of the

RSP in the reattachment regime.
34
AEDC·TR·76·127
1.0 Experiment:
Mueller (Ref. 36)
o h = 0.75 in.
o h" 0.50 in.
6. h = O. 25 in.
Ta ni (Ref. 37)
\.-- o h = 2 em
~
E 0.6
:J
z
VI
"C
o
C
~ 0.4
e::::
.....
c
Q)
:J
.c
\.-
~ 0.2
o
O. 1 O. 2 0.3 0.4 0.5 0.6 O. 7

Centerli ne Velocity Ratio, ue/u s
Figure 13. Turbulent Reynolds number downstream of the RSP .

.
A series of computational experiments was carried out to establish the values of
RT r required to give a reasonably accurate prediction of experimental base pressures in
supersonic flow (say to within five percent). The following matrix of test cases was
selected:
Planar, Mal 2.0, e/h 0.016
Axisymmetric, Mal = 2.0, e/h = 0.024
Axisymmetric, Mal 2.0, e/h 0.053
Axisymmetric, Mal 4.0, e/h O.Oll
Typical effects of RTr on the base pressure, for planar flow at Mal;::; 2, are shown in
Fig. 14. As would be expected, an increase in RTr (a decrease in the turbulent shear
stress level) causes an increase in the base pressure.
35
AEDC-TR-76-127
0.55
Planar Flow
Mal = 2.0
·0.50
- .-I
Cl..
..c
Cl..
o·
= 0.062
:;::;
0.45
ro
0:::
«l)
I-
::J
VI
VI
«l)
l-
0.40
e.. = 0.016
«l)
VI
ro
co
0.35
O. 30 L----I--~..I-..---I.---'---~'!.-------!18U-
60 80 100 1ZU 14U IbU
Turbulent Reynolds Number at RSP, RTr
Figure 14. Influence of the turbulent Reynolds number at the RSP on the
predicted base pressure.
From the matrix of base pressure test cases, it was found that RTr, must decrease as
the initial boundary-layer thickness is increased and as Ma 1 is increased. In addition, it
was found that accurate prediction of axisymmetric flows requires a larger value of R Tr
than is required for planar flows_ Since Bradshaw's curvature and dilatation strains are
both related tv a pressure gradient parameter, we decided to correlate RT r with a
parameter proportional to the pressure gradient at the beginning of recompression. After
some trial and error, the pressure gradient parameter was selected to be
p = 2 1-a_y1M
sp 0 pdpc/dx)
p /Pcp (30)
wherc op is the distance between the axis and the streamline which originates at the edge.
of the initial boundary layer. The results of the correlation of R Tr with P are shown in
Fig. 15. Both curves which have been fitted to the required RT r values can be expressed
by the following single equation:
RTr = 60 + 15a + (40 + 135a) exp (-0.024 p2.3) (31 )
36
AE DC-TR-76-127
250 Approximate Required

RTr Values:
b
0 Axisymmetric
l-
0::: 200
t:;Planar
Cl.. RTr = 75 + 175 exp (-0.024 p2. 3)
V'l
0::: (Axis ym metric)
.....
1'1:1
b. Low-Speed Planar
~ 150
E Experiments
:::J
:z (Refs. 36 and 37)
VI
"0
"0
c
>-
100
-
Q)
0:::
c 0
Q)
:::J
.0
b. 50
:::J
I-
RTr = 60 + 40 exp (-0.024 p2. 3)
(Planar)
0
0 2 4 6 8 10
Pressure Gradient Parameter, P
Figure 15. Turbulent Reynolds number at the RSP as a function of the pressure
gradient parameter.
It is interesting that the required RT r values for planar flows (Fig. 15) at low values
of P are consistent with the low-speed experimental results of Mueller and of Tani.
It will be shown in Section 3.0 that the RT model developed in this section yields
good results for planar near-wake flows; in addition to the base pressure, the flow
structure is predicted with adequate accuracy. However, the RT model has been found to
yield only fair results for axisymmetric flows with Ma 1 less than about 1.7; the
predictions are fairly adequate at higher Mach numbers.
3.0 DISCUSSION OF RESULTS
In this section, the predictions of the near-wake model are compared with
experimental results for planar and axisymmetric flows of the type shown in Fig. 2a; i.e.,
the supersonic external flow upstream of the base and outside of the wall boundary layer
is uniform and parallel to the axis and is effectively infinite in lateral extent.
37
AEDC-TR-76-127
3.1 CONCENTRATION FIELD SOLUTION PROBLEMS
When attempts were made to compute near-wake flows for which the concentration
equations must be solved, a singularity in the wake regime solution was encountered.
Downstream of the onset of recirculation, the predicted centerline concentration, Cc ,
decays with axial distance. At some axial station upstream of the onset of recompression,
the system of equations becomes singular; i.e_, the determinant of the system of
equations [Eq. (25)] goes to zero and the solution "blows up". This singularity is not
related to the Crocco-Lees saddle-point singularity which occurs in the recompression
region. Indeed, the new singularity does not exhibit saddle-point behavior; the solutions
for all of the derivatives (dCc/dx, etc.) remain smooth and monotonic as the singularity is
approached.
The wake regime singularity does not appear to have physical significance, but is
probably related to the particular formulation of the half-radius species equation or of
the concentration profile shape function. The singularity also occurs when the
concentration equations are solved for a single gas base bleed problem, even though no
wake regime singularity occurs when the single gas version of the analysis is used for the
same problem.
Strahle and Mehta (Ref. 8) encountered an extraneous singularity in their analysis of

near-wake flows without base bleed. They were able to remove the singular behavior by
changing their velocity profile functions in the region upstream of recompression (Ref.
9). Various modifications of the concentration profile function in the present analysis
have been tried, but to no avail; the wake regime singularity occurs for both single gas
flows and for flows in which the base bleed gas is different from the outer-stream gas.
An obvious next step in attempting to solve this concentration field solution problem is
to alter the formulation of the half-radius species equation.
Because of the unresolved problems in solving the concentration equations, only

results from the single gas version of the near-wake model are presented in this report.
3.2 PLANAR BASE PRESSURE
The effect of initial boundaly-Iayer momentum thickness on base pressure in planar

flow is shown in Fig. 16 for Mal = 1.5 and in Fig. 17 for Mal = 2.0. The scatter shown
in the experimental results is typical for planar flows and is attributed to sidewall effects
which vary from experiment to experiment. At both free-stream Mach numbers, the
present analysis predicts the experimental base pressures very well for eIh less than 0.03
or 0.04. However, the analysis overpredicts the base pressure for higher values of eIh; the
predicted values of Ph Ipl are about ten percent too high at eIh : : 0.1. In agreement with
38
AEDC-TR-76-127
experimental observations (Ref. 38) at free-stream Mach numbers of 1.5 and 2.0, the
near-wake model predicts a negligible influence of Reo on the base pressure for a given
value of 8/h. (Computations were made for Reo from 10 3 to lOS).
The trend of Pb /Pl versus 8/h predicted by the present analysis is similar to that
predicted by Alber (Figs. 16 and 17); however, the present analysis provides significantly
better predictions of the level of base pressure. The Chow theory (Ref. 20) predicts an
incorrect trend of Pb /Pl versus 8/h; however, the Chow and Spring extension of this
theory provides very good predictions of the base pressure at Mach numbers of 1.5 and
2.0 (Figs. 16 and 17).
O. 7
0.6 / S:_T::t.61 __----

-
..e-
.0
c.. 0.5
_~
.......... .........-0 I
I / ---0 =g-~~=:~--------------------~-----
-R..=-...:--===-~-::.=-.---------
0
o·
:;::: "...- -...---- 0 \.- 0
~
c:::
...
CI.l 0.4 Chow and Spring
~~:s:n~Jht~O~~
::::::J
VI
/\ Theory (Ref. 21)
...as
c..
CI.l 0.3 Experiment:
VI
~
co Chow Theory, 0 Compiled by McDonald (Ref. 38)
Ma 1 = 1. 56 (Ref. 20) Rea = 103 to 4 x 104
O. 2
- - - Chapman (Ref. 39)" Rea = 6.5 x 103
O. 1
0 O. 02 0.04 O. 06 0.08 O. 10
I nitial Boundary-Layer Momentum Thickness, ath
Figure 16. Effect of the initial boundary layer on the base pressure in planar
flow (M a1 = 1.5).
Experimental base pressure results for Ma 1 = 3.0 (Fig. 18) show much more scatter
than was observed at the lower Mach numbers. In agreement with the trend of the
experimental results, the present analysis predicts a much larger effect of 8/h on the base
pressure at Mal = 3.0 than at Mal = 2.0. In addition, a significant effect of Reo on the
base pressure is predicted.
The effect of free-stream Mach number on the base pressure is shown in Fig. 19a for
8/h = 0.02 and in Fig. 19b for 8/h = 0.04. In view of the scatter in planar base pressure
data, the predictions shown in Fig. 19 are very satisfactory.
39
AEDC-TR-76-127
0.6
0.5
AlberTheory IRef. 61 ~ _ _ S,ChOW~~d~pri ng :eor y IRef. 211
__ - - 0 \
~
~-;--o
!>-II
- o
~=-~-
'\ :=0.-_---
~ 0.3 I Present Theor~f:
~ I ReS = 103 to llY
~
Experi me nt:
o Compiled by McDonald (Ref. 38)
Q 2 / ' \ Chow Theory IRef. 201 ReS = 1. 5 x 103 to 4 x 104
0.1 o Chow (Ref. 20)
~ - Chapman (Ref. 39), Ree = 6 x 103
O~--~----~----~---d----~----~----~--~----~----
o 0.02 0.04 0.06 0.08 O. 10
I nitial Boundary-Layer Momentum Thickness. 9th
Figure 17. Effect of the initial boundary layer on the base pressure in planar
flow (Mal = 2.0),
0.4
o
o
Res = loS 0
,..... O. 3
CI-
£:,
ReS = 103
--
0..
6
:;:::;
ro
cr:
E 0.2 o --- -- --
::J
Vl
Vl
(])
>-
0...
Q.)
Vl
Exper i me nt:
ro
co 0.1 o Compiled by McDonald (Ref. 38)
ReS = 1. 7 to 40 x 103
- - -Chapman (Ref. 39), Ree = 6 x 103
o Reda and Page (Ref. 40), Re e = 9 x 104
-Theory
O~I----~----~~----~----~----~
o 0.02 0.04 0.06 0.08 0.10
I nitial Boundary-Layer Momentum Thickness, eth
Figure 18. Effect of the initial boundary layer on the base pressure in
planar flow (M a 1 = 3.0).
40
AEDC-TR-76-127
0.6
0.5 Theory --LI

o ReS = ll!
o 8/h = 0.02
...... o
c. 0.4
-.0
c.
o
i'" o
0:::
~ 0.3
:J
~
a>
'-
D-
e; 0.2
'"
co o Experiments Compiled by
McDonald (Ref. 38)
8/h = 0.016 to O. 020 o
0.1 Re8:::; 5 x 103
0
1.0 1.5 2.0 2.5 3.0
a. 8/h = 0.02
0.6
o
0.5 o 0 Theory --LI
Re8 = ll!
8/h = 0.04
c:
-.0
0.4
c.
o
:;:;
'"
0:::
~ 0.3
:J
V>
V> o
~
D-
e; o Experiments Compiled by
'"
co
0.2
McDonald (Ref. 38)
8/h = O. 03 to O. 05
Re8 = 5 to 30 x 103
0.1
O~I----~----~----~--~
1. 0 1.5 2. 0 2. 5 3. 0
b. e/h = 0.04
Figure 19. Mach number effect on the base pressure for planar flow.
41
AEDC-TR-76-127
Most near-wake flows of engineering interest involve initial boundary layers of

moderate thickness, with 8/h values less than 0.05. The base pressure predictions of the
present analysis for this range of 8/h are shown in Fig. 20 for Mal from 1.2 to 4.0. Also
shown in Fig. 20 are the predictions of the Korst theory for no initial boundary layer.
Because of a fortunate combination of physically unrealistic assumptions, this simplest
version of the Korst theory provides fairly satisfactory predictions of planar base pressure
in the range of initial boundary layers most often encountered in engineering
applications.
0.6
- - Present Theory
Re8 = 104
0.5 ..... - - Korst Theory (Ref. 5)
8th = a
0.4
~- 8th = 0.05
0.3
Figure 20. Predicted base pressure for planar flow.
3.3 PLANAR FLOW-FIELD STRUCTURE
Predicted axial distributions of the total pressure at the supersonic edge of the
turbulent shear layer are shown in Fig. 21 for three typical planar wake flows. A value of
Po s /Po a less than unity indicates that the shear layer is growing into the rotational
remnant of the initial boundary layer, and the point at which Pos/Poa becomes unity is
42
AEDC-TR-76-127
the point where the initial boundary-layer flow has been completely entrained. The
results shown in Fig. 21 indicate that, even for relatively thin initial boundary layers (8/h
= 0.013 - 0.014), the entire viscous near-wake structure downstream to the critical point
is immersed in a rotational external flow.
1.0
o Beginning of Recompression
o Rear Stagnation Poi nt
8 f::. Critical Point
..e-
;p 0.8
.....
~ Mal = 2, 8th = 0.014
.5
.... Re8 = 104
~ 0.6
.s:::
--
VI
o
Q,)
c::n
-....
"'0
I.J..I
0.4
rn
CI)
::J
Vl
Vl
Mal'" 2, 8th'" 0.079
ReS = 104
CI)
5:.. O. 2
JI'''''- Mal:; 3.05, Sth
ReS = 9 x 104
= 0.013
o ~------~------~------~~------~------~
o 1 2 3 4 5
Axial Dista nce from Base, xth
Figure 21. Total pressure at the edge of the shear layer for planar flow.
Predicted and experimental centerline (wall) static pressure distributions are shown
in Fig. 22 for Mal = 2.0. The distance to the beginning of recompression is somewhat
overpredicted by the present theory; however, the complete predicted pressure
distribution is considered to be in satisfactory agreement with the experiment, as is the
predicted location of the RSP. The critical point in this flow occurs at a centerline Mach
number of 0.64.
The predicted pressure distribution of Chow and Spring, which is shown in Fig. 22,
is fairly satisfactory. Their analysis is calTied out in two parts: (1) for the recompression
process up to the RSP and (2) for the redeveloping flow downstream of the RSP. As
shown, the two segments of the predicted pressure distribution do not join smoothly at
the RSP.
43
AEDC-TR-76-127
The experimental centerline (wall) static pressure distribution obtained by Reda and
Page (Ref. 40) for Mal = 3.05 is shown in Fig. 23, along with the predicted pressure
distribution. The overprediction of the distance to the onset of recompression is
somewhat greater than at Mal = 2.0; however, the axial distance to the experimental RSP
is predicted to within ten percent. The centerline Mach number distribution for the Reda
and Page flow is shown in Fig. 24. As a result of neglecting the streamwise transport
terms in the viscous wake flow, the predicted centerline Mach number distribution always
exhibits the sharp minimum at the beginning of recirculation. However, the solution
seems to recover from this localized unrealistic behavior, and the downstream portion of
the predicted Me distribution is reasonably good. Detailed lateral profiles of the pitot
pressure in the Reda and Page flow are shown in Fig. 25 for four axial stations; the first
two stations are upstream of recompression, and the last two are downstream of the
point where the wake flow is completely supersonic. The overall shapes of the predicted
profiles agree fairly well with the experiment, but it is evident that the experimental
shear layer growth rate in the wake regime is underpredicted. Part of the deviation
between prediction and experiment at the two downstream stations may be attributed to
the presence of the strong recompression shock wave, which is treated as an isentropic
compression in the analysis.
The "brute force" technique described in Section 2.11 was used to obtain the planar
flow-field solutions which extend through the critical region.
l.0 o Experiment (Ref. 20) ---0--'-0 0

8th = O. 014 //"" 0
--Present Theory, 8th = 0.014 /
ReS = ut' I
.-<
..e-u
0.8 --- Chow and Spring Theory /0 "
0..
(Ref. 21) I "'- Critical Poi nt (Mc = 0.64)
....
(l)
~
VI
0.6
(l)
'-
c..
u
~
Vl 0.4
(l)
c
o
l:
(l)
c: Predicted RSP
V
(l)
u 0.2
r Experimental RSP
0~1__~~~___
~~! __ ~! __ ~
o 2 3 4 5 6
Axial Distance from Base, xth
Figure 22. Centerline pressure distribution for planar wake flow (M a1 = 2.0).
44
1.0
- - - Experiment (Ref. 40)
Sth = 0.013
ReS = 9 x 104
0.8 ---Theory, Same Conditions
-
~
Cl...
L)
Cl...
Q)
!.-
::l
Vl
V)
0.6
Q)
!.-
0....
u /J Critical Poi nt (M c = 0.59)
....... /.
.......
C'O
/ /
.j::>.
Vl (/)
Q)
0.4 -
C
/
!.- /'
Q)
....... /'
C
Q)
/'
U
0.2 1-_ 07 , / / / / ~ Experimental RSP
'- V Predicted RSP
oo ,"
1 2I !
3 ~4 »
m
o
Axial Dista nce from Base, x/h ()
~
JJ
Figure 23. Centerline pressure distributnon for planar wake flow (Mal = 3.05). ~
cr>
'"
-..J
»
2.5 m
o
I 0 Experiment (Ref. 40)
()
.:.,
:0
2.0~
S/h = 0.013
ReS = 9 x 104
Theory, Same Conditions
,- ~
m
~
tv
-..J
o
u
~
1.5
I-
15
E
~ Note: Experimental Values for Me Taken at
z
.s::
u
C'O
1. 0 I r/h = 0.03
~ ~
0\ Q)
c / _____ Critical Poi nt
-
I-
Q)
..- 0.5
c
Q)
<..,)
O l r - - - - -__
-0. 5 L I v IV I I !
o 1 2 3 4
Axial Dista nee from Base, x/h
Figure 24. Centerline Mach number distribution for planar wak.e flow (Ma 1 = 3.05).
O~'----~----~--~----~
o O. 04 O. 08 O. 12 O. 16
r r T
o O. 04 O. 08 O. 12 O. 16
r-- 1 I I
o 0.1 0.2 0.3 »
m
I I i i o
o 0.1 0.2 0.3 ()
~
::D
~
Figure 25. Pitot pressure profiles for planar wake flow (M a1 = 3.05). <p
~
I\l
-.J
AEDC-TR-76-127
3.4 AXISYMMETRIC BASE PRESSURE
In order to avoid possible support interference effects on the base pressure data,
results from experiments without base stings (free flight or annular, tunnel experiments)
have been used for comparison with the present analytical model. The effect of initial
boundary-layer momentum thickness on the base pressure is shown in Fig. 26 for Mal =
2.0. As was found for planar flows, the predicted axisymmetric base pressure at Mal =
2.0 is insensitive to changes in Re6 in the range from 10 3 to 105. Although the
experimental base pressures are predicted with less than ten-percent error, the analytical
results exhibit a peculiar oscillation about the expected results, which are indicated by
the dashed line. The oscillation is attributed to the RT model used in the reattachment
region, which is overly sensitive to small changes in the flow field. The RT model yields
realistic results for higher Mach numbers, say 3.0 or 4.0, but the results at Mach numbers
less than 2.0 exhibit more oscillation than shown in Fig. 26.
0.8f
...... t::. &a-:-~------=

..e- 0.6 .~~--c~
..0
~ Theory -"
0-
o'
:;:; - Experi me nts
co
c::: Ree " 1lf'
e 3
::I
0.4 Sym Ref. Ree x 10-
V'l
VI
Q.)
0 41 2.8 to 4. 7
'-
0.. t::. 42 5.7 to 8.5
Q.)
~ O. 2 0 43 7
00 0 44 6
V" 45 2. 1
\l 46 2.5
OL-~~~~--~----~----~----~----~----~----~
o 0.04 0.08 O. 12 0.16
Boundary-Layer Momentum Thickness, e/rb
axisymmetric flow (Mal = 2.0).
48
AEDC-TR-76-127
The effect of initial boundary-layer momentum thickness on the base pressure is

shown in Fig. 27 for Mal = 3.9; the prediction is reasonably satisfactory.
0.5
Theory, ReS = 2.6 x 104
o
0.4
o
o
:;::: 0.3
m
~
w
L-
:::l
Vl
Vl
Experi me nts
E 0.2 o Ref. 42, Mal:: 3. 85, ReS = 6.9 x 103
0...
!J. Ref. 47, Ma 1 :: 3.91, ReS = 2.6 x 104

<> Ref. 43, Mal =4.0, ReS= 2.5to5x 104
O. 1 I 0 Ref. 48, I\~al '" 3. 9, ReS'" 3. 2 to 3. 8 x 104
OL........--~~
o 0.01 0.02 O. 03 0.04 O. 05

Boundary-Layer Momentum Thickness, S/rb
axisymmetric flow (M a 1 = 3.9).
The effect of free-stream Mach number on the axisymmetric base pressure is shown
in Fig. 28. In view of the considerable scatter in the experimental data, the predictions
for 81rb = 0.01 and 81rb = 0.05 are considered to be satisfactory for Mal greater than
2.0. However, at lower Mach numbers the peculiarities caused by the RT model are
evident. The prediction for 8/rb = 0,01 is realistic down to a Mach number of 1.5, below
which the the base pressure ratio is predicted to be constant. A similar flattening of the
Pb Ipl curve for 81rb = 0.05 occurs at Mach numbers less than 2.0. Because of the
unrealistic behavior of the recompression region turbulent transport model, the near-wake
analysis in its present form is not recommended for use at free-stream Mach numbers less
than about l.7.
49
AE DC-T R-76-127
EXQeriments
Sym Ref. 9/rb Rea x 10- 4

0 42 0.008 to 0.017 0.44 to 0.9
1.0 cf 47 0.011 2.6
<> 43 0.014 4
'\I 43 0.024 O. 7
V" 44 0.024 0.6
0,8 "V 41 0.009 0.47
...... o Do 45 0.053 0.21
..e-.c 9/rb = 0.05 0 48 0.017 3.8
c. 0' 48 0.037 3.2
o·
:;:;
ro
0.6 - - Theory, Re9 = 104
0:::
...
Q)
::::l
Vl
Vl
d
E 0.4
0..
Q)
Vl
ro
co
O. 2
i ________~------~I--------~------~I--------~------~
O...
1 2 3 4
Free-Stream Mach Number, Mal
Figure 28. Mach number effect on the base pressure for axisymmetric flow.
3.5 EFFECT OF BASE BLEED ON AXISYMMETRIC BASE PRESSURE
Results of the axisymmetric base bleed experiments of Sirieix, et al. (Ref. 43), are
shown in Fig. 29, along with the predictions of the near-wake analysis. The experimental
trend of increasing base pressure with increasing base bleed flow is well predicted; in
addition, the maximum difference between the predicted and experimental base pressures
is less than eight percent.
50
AEDC-TR-76-127
0.8
o
o
...-4
o
.e-.c 0.7 o
c..
0'
......
ro
a:::
E 0.6 o Experi ment (Ref. 43)
:::J
til
Ma 1 = 1. 92, S/rb = O. 024,
ReS = 7 x 103, Airbleed, Toj = Toa
til
(\.)
"-
c...
(\.) --- Theory, Sa me Conditions
~ 0.5
o:::l
0.4------~-----b----~--~~~~~~ __~
o 0.01 0.02 0.03
Figure 29. Effect of base bleed on axisymmetric base pressure.
3.6 AXISYMMETRIC FLOW·FIELD STRUCTURE
Predicted axial distributions of the total pressure at the supersonic edge of the
turbulent shear layer are shown in Fig. 30 for three typical axisymmetric wake flows. As
was found for planar flows, most or all of the viscous near-wake structure is imbedded in
a rotational external flow. The abrupt increase in the slope of the Pos versus x curves at
the onset of recompression, which is much more pronounced for axisymmetric flow than
for planar flow, indicates a sudden increase in the entrainment rate of the shear layer.
The mass flow in the shear layer increases rapidly with distance in the recompression
region, even though the thickness of the shear layer, b, initially decreases in this region;
this phenomenon is related to the rapid change in the shear layer velocity profile during
recompression.
51
AEDC-TR-76-127
1.0 Mal" 2 0
S/rb = 0.016
_ A
~ ReS = 1(J'
..e- VI
c...
0 O. 8
1--
~ Mal =3.91
ro
....J
I-
S/rb = 0.011
m
Q,) 0.6 ReS = 2. 6 x 104
L:
-
VI
0
Q,)
t:n 0 Beginning of Recompression
'0
LJ..I
....... 0.4 0 RSP
m
6. Critical Poi nt
~Ma1 =2
Q,)
I-
::l
VI
VI
Q,)
I- S/rb = 0.062
C- 0.2
m
"0
ReS = 104
I-
O~----~------~----~------~------~----~
o 1 2 3 4 5 6
Axial Dista nee from Base, x/rb
Figure 30. Total pressure at the edge of the shear layer for axisymmetric flow.
The experimental centerline static pressure distribution measured by Seiling, et al.

(Ref. 47), for Mal = 3.91 is shown in Fig. 31, along with the predictions of the
near-wake analysis. The solution through the critical region was obtained with the "brute
force" technique described in Section 2.11. The distance to the onset of recompression is
overpredicted as is the distance to the RSP. The characteristic overshoot of the static
pressure is predicted.
The length scale of the near wake, as indicated by the distance to the onset of
recompression, is strongly influenced in high Mach number flows by the rotational
boundary-layer flow at the base plane. To illustrate this point, a computation of the flow
field of Seiling, et al., was made with the assumption that there was no intial boundary
layer, i.e, that the turning of the flow at the base plane corresponds to a Prandtl-Meyer
expansion of the inviscid free stream through the pressure ratio Pb/PI' The results, which
are presented in' Fig. 31, show that the distance to the onset of recompression is nearly
twice as large as that predicted when the initial boundary layer is included.
52
AEDC-TR-76-127
1. 25
-
.e-u
C-
1.00
Q,;
'-
::::J
II> O. 75
VI
(I) I
'-
Q... I o Experiment (Ref. 47)
I
U
:;::; I S/rb = 0.011
.12
Vl 0.50 I ReS = 2. 6 x 104
(I) I
C
o I - - Theory, Same Conditions
o I - - - Theory, Experimental
O. 25 o 0 0 0 ----- --'
Experimental RSP
Pb' No Boundary Layer
o __ ____ ____
~ ~ ~ ~~ __ ~~ __ ____ __ ____
~ ~ ~ - b_ _ _ _ ~ __ ~
o 1 2 3 4 5
Axial Dista nee from Base, x/rb
FigUie 31. Centerline pressure distribution for axisymmetric wake flow (Mal = 3.91).
Detailed radial profiles of pitot pressure, which were measured by Seiling, et al., are
shown in Fig. 32. The predicted profiles agree well with the experiment, although the
experimental shear layer growth rate in the wake regime is underpredicted; a similar
result was found for planar flow (Fig. 25). Based on the results shown in Fig. 32, as well
as those shown in Fig. 25 for planar flow, it does not appear that the neglect of the
finite-strength lip shock is a serious deficiency in the present analysis, even for
free-stream Mach numbers up to 4.0.
The experimental centerline pressure distribution measured by Sirieix, et al. (Ref.

43), for Mal = 1.92 is shown in Fig. 33. For Mach numbers in the neighborhood of 2.0,
the "brute force" technique for obtaining the solution through the critical region is
unsuccessful, so the extrapolation technique described in Section 2.11 was used to obtain
the solution shown in Fig. 33. The distance to the onset of recompression is somewhat
overpredicted, and the distance to the RSP is overpredicted by 25 percent. An unusual
inflection in the centerline pressure distribution is predicted in the region of the RSP, but
the characteristic overshoot in the pressure distribution is preserved in the solution.
53
»
m
o
2.0 - - ..... Experiment (Ref. 41)
()
~
:0
--Theory ~
...
I
C')
"1) " ..... '"

-..J
1.6 \
- .c
:>-
:>-
,I
/
I
c:
.2
:!::: 1.2
-- "
In
0
~
.~
"'0
ro
VI CI:::::
.j::>. 0.8 .~ rj + b·
~r·+b
I
0.4 L- ..
o O. 04 O. 08 O. 12 O. 16
I I I ! I
o O. 04 O. 08 O. 12 O. 16
I I I I I
o 0.04 0.08 0.12 0.16
Pitot Pressure Ratio, p~/poa
Figure 32. Pitot pressure profiles for axisymmetric wake flow (Mal = 3.91).
1.2
- - - Experiment (Ref. 43)
S/rb = 0.024 /'
./
1.1 ReS 7 x 103 /
~critical Point
= /
/
--Theory, Sa me Conditions /
.-I
/
...e- 1.0
L) /
0..
/
<l)
!....
/
:J
VI
I
VI
<l)
O. 9 /
!....
~
0...
VI
VI
----
U
co
V) 0.8
Uc Extrapolated
<l)
C
!....
--
<l)
u
c
<l)
O. 7
I /
/
/ Experimental RSP
1
----- -- -- ---_/ /
0.6
0.5 L--
o 1
2
r
.1
I~
I'j
3
Predicted RSP
4
I
5
~
m
o
o
Axial Dista nce from Base, x/rb .!.oj
JJ
~
q>
Figure 33. Centerline pressure distribution for axisymmetric wake flow (Ma 1 = 1.92). ~
N
-...I
AEDC-TR-76-127
The centerline Mach number distribution for the flow of Sirieix, et aI., is shown in
Fig. 34. As was shown for planar flow, the predicted Mach number distribution has a
sharp minimum; this unrealistic behavior is localized, and the slope of the predicted
centerline Mach number distribution agrees well with the experiment at the RSP,
although the entire distribution is shifted downstream from the experimental distribution
by the same distance as the RSP shift.
The experimental centerline pressure distribution measured by Reid and Hastings for
Mal = 2.03 is shown in Fig. 35. The inflection shown in Fig. 33 does not appear in this
solution, which was also obtained with the centerline velocity extrapolation technique.
Based on the results shown in Figs. 33 through 35, it can be concluded that the
extrapolation technique for obtaining solutions through the critical region is reasonably
successful.
1.0
---Experiment (Ref. 43)
S/rb = 0.024
ReS = 7 x 103
....- 0.5 --Theory, Same Conditions
~
E
::l
Z
.c
u
~
CJ.)
c
I-
-E -O. 5
CJ.)
u
c Extrapolated
U
Axial Distance from Base, x/rb
Figure 34. Centerline Mach number distribution for

axisymmetric wake flow (M a 1 = 1.92).
56
AEDC-TR-76-127
Critical Poi nt (Mc = 0.37)
1.2
o o o
u o Experiment (Ref. 41)

:;:::;
ro
.......
Vl 0.8 S/rb = O. 0061
ReS =3. 2 x 103
Q)
c: o
~
Q)
-I-'
--Theory, Same Conditions
c:
~ 0.6
P~ "c Extrapolated
o 0 0
Predicted RS
0.4
o 1 2 3 4 5 6 7
Axial Dista nce from Base, x/rb
Figure 35. Centerline pressure distribution for axisymmetric wake flow (Ma 1 = 2.03).
3.7 DIVIDING STREAMLINE PROPERTIES
Although dividing streamline concepts play no role in the present near-wake analysis,
the location of this streamline and its velocity are computed. Predicted distributions of
total pressure on the dividing streamline are shown in Fig. 36 for three typical planar
flows_ Somewhat surprisingly, the predicted dividing streamline total pressure remains
nearly constant from the base plane to the RSP.
The simple Korst theory for no initial boundary layer yields base pressure results
which are similar to those of the present analysis for 8/h from 0.01 to 0.05 (Fig. 20).
Korst's theory is based on three fundamental assumptions: (1) an isentropic
recompression along the dividing streamline, (2) stagnation of the dividing streamline at
the peak of the recompression pressure rise, and (3) negligible reverse flow velocities_
According to the results shown in Fig. 36, the first assumption is a good one for planar
flows. It has long been known that the dividing streamline does not stagnate at the peak
of the pressure rise, i.e., the RSP occurs at a station where the pressure is much less than
57
AEDC-TR-76-127
the peak value (Figs_ 22 and 23). But Korst's assumption of negligible reverse flow
velocities leads to overprediction of the dividing streamline total pressure (Fig. 36). This
high total pressure in Korst's analysis, combined with the unrealistically high pressure at
which the dividing streamline is assumed to stagnate, leads to base pressure predictions
which are about correct.
1.2
o Beginning of Recompression
o Rear stagnation Poi nt
--------~-
Korst Theory
CL>
c: Mal =2, Sth =0
E 0.8
-...
n:I
CL>
V ')
Mal = 2, S/h = 0.079
:.c 0.6 ReS = 104
:~
Cl
c:
o
...
(I)
::l 0.4
Mal = 2, S/h = 0.014
VI
ReS = 104
I
VI
CL>
>-
a..
~
~.I. . . . .
2
-5 Mal = 3. 05, S/h = 0.013
ReS = 9 x 104
f- 0'0 L _______ _ _ _ _ _ _ _ _..I.__ _ _ _ _ _ __ J
!
o 1 2 3
Axial Dista nce from Base, x/h
Figure 36. Total pressure along the dividing streamline for planar flows.
Predicted distributions of the total pressure on the dividing streamline are shown in
Fig. 37 for three typical axisymmetric wake flows. In contrast to the results for planar
flows, the dividing streamline total pressure is predicted to increase significantly during
the recompression process. However, the quality of the detailed predictions of the present
model for axisymmetric reattaching flows is not considered to be good enough so that
strong conclusions can be drawn about the validity of isentropic recompression of the
dividing streamline.
One point is clear from the results discussed in this section: any wake analysis which
is based on dividing streamline concepts (e.g., Ref. 21) should include the effect of finite
reverse flow velocities on the dividing streamline properties.
58
AEDC-TR-76-127
1.0
........ / Mal = 2, S/rb = 0.062
.e-"0
a.
0 ReS = 104
(1)-
e:
Ero o Beginning of Recompression
Q.)
O. 8
I-
.....-
(/") o Rear stagnation Point
en
e: Mal = 2, S/rb = 0.016
"'Cl
:~
Cl
ReS = 104
e:
0
Q.)
I- 0.6
~
V>
Vl
Q.)
'-
0... Ma 1 = 3. 91, S/rb = O. 011
ro
"0 ReS = 2.6 x 104
I-
0.4
0 2 3 4
Axial Distance from Base, x/rb
Figure 37. Total pressure along the dividing streamline for axisymmetric flows.
4.0 CONCLUDING REMARKS
The correlations of the present near-wake analysis with a variety of planar and
axisymmetric experiments indicate that the overall framework of the analysis is physically
realistic. Because both the detailed flow-field structure and the base pressure can be
predicted reasonably well, it can be concluded that the arbitrariness in the selected
turbulent transport model is not used to mask the effect of physically unrealistic aspects
of the analysis. The method of computing the inviscid remnant of the initial boundary
layer is considered to be a very important feature of the analysis, as is the inclusion of
finite reverse flow velocities.
In any analysis of turbulent flow, the selection of an adequate turbulent transport

model is a major problem. The transport model used in the present analysis yields
acceptable base pressure results for planar flows, but yields inconsistent results for
axisymmetric flows with free-stream Mach numbers less than about 2.0. The concept of
scaling the recompression region turbulent transport properties with the value of the
turbulent Reynolds number at the rear stagnation point seems to be satisfactory. But the
model used to relate this turbulent Reynolds number to the flow-field properties must be
improved if adequate predictions of axisymmetric flows are to be obtained over the
59
AEDC-TR-76-127
entire supersonic Mach number range of practical interest. The correlations of the
analytical results with the flow-field expeliments of Reda and Page and of Seiling, et aI.,
show that the turbulent mixing rate is somewhat underpredicted in the region upstream
of recompression and indicate that the RT model for this region should be improved.
However, based on limited numerical experimentation, it is not expected that an
improved RT model for the wake regime would greatly change the predicted base
pressures; the predicted base pressure is much more strongly dependent on the turbulent
transport properties in the recompression region than on those properties in the wake
regime.
Because of the unresolved problems in the solution of the species conservation

equations, the near-wake model as presently formulated cannot be used for flows in
which the base bleed gas is different from the external stream gas. Many practical wake
flows involve two gases; therefore, a major emphasis in future work on the near-wake
model should be placed on developing a satisfactory technique for solving the species
conservation equations.
Although the overall framework of the near-wake analysis seems to be physically

realistic, the behavior of the solutions in the vicinity of the onset of recompression is not
satisfactory. The length of the constant pressure region is overpredicted and an unrealistic
sharp minimum in the centerline velocity is predicted. Both of these characteristics of the
solution can probably be avoided by abandoning the assumption that the pressure is
constant in the wake regime. In order to solve for the pressure distribution, however, an
additional equation is required. A reasonable approach would be to add an axial
momentum equation for the flow along the axis; this equation would include an
approximate term to represent the streamwise turbulent transport of momentum.
Even though the present formulation of the near-wake model is deficient in several
respects, the model provides a significant improvement in the capability to analyze
supersonic turbulent near-wake flows. Indeed, the present model for axisymmetric flows
without base stings is the first to realistically include the effect of the initial
boundary-layer thickness and shape and the effects of base bleed.
REFERENCES
1. Strahle, W. C. "Theoretical Consideration of Combustion Effects on Base Pressure in

Supersonic Flight." Twelfth Symposium (International) on Combustion, The
Combustion Institute, Pittsburgh, 1969, pp. 1163-1173.
2. Wasson, R. A., Jr. and Darlington, C. R. "Performance of an External Burning

Rocket Configuration in a Supersonic Flight Environment." AEDC-TR-75-102
(ADB005762L), July 1975.
60
AE DC-TR-76-127
3. Crocco, L. and Lees, L "A Mixing Theory for the Interaction between Dissipative
Flows and Nearly Isentropic Streams." Journal of the Aeronautical Sciences,
Vol. 19, October 1952, pp. 649-676.
4. Korst, H. H. "A Theory for Base Pressures in Transonic and Supersonic Flow."
Journal of Applied Mechanics, Vol. 23, December 1956, pp. 593-600.
5. Korst, H. H., Chow, W. L., and Zumwalt, G. W. "Research on Transonic and

Supersonic Flow of a Real Fluid at Abrupt Increases in Cross Section - Final
Report." ME Technical Report 392-5, University of Illinois, December 1959.
6. Alber, 1. E. "Integral Theory for Turbulent Base Flows at Subsonic and Supersonic
Speeds." Ph.D. Thesis, California Institute of Technology, June 1967.
7. Alber, 1. E. and Lees, L. "Integral Theory for Supersonic Turbulent Base Flows."
AIAA Journal, Vol. 6, No.7, July 1968, pp. 1343-1351.
8. Strahle, W. C. and Mehta, G. "Turbulent Axisymmetric Base Flow Studies for

External Burning Propulsion." in 11th JANNAF Combustion Meeting, Vol. II,
CPIA Publication 261, December 1974, pp. 441-448.
9. Strahle, W. C., et al. "Turbulent Axisymmetric Base Flow Studies for Externai
Burning Propulsion." Paper Presented at 12th JANNAF Combustor Meeting,
Newport, Rhode Island, August 1975.
10. Davis, L. R. "Experimental and Theoretical Determination of Flow Properties in a

Reacting Near Wake." AIAA Journal, Vol. 6, No.5, May 1968, pp. 843-847.
11. Nash, J. F. "An Analysis of Two-Dimensional Base Flow including the Effect of the
Approaching Boundary Layer." National Physical Laboratory Aero. Report
1036, July 1972.
12. Lamb, J. P and Hood, C. G. "An Integral Analysis of Turbulent Reattachment

Applied to Plane Supersonic Base Flows." Transactions of the ASME, Series C,
Vol. 90, 1968, pp. 553-560.
13. Hill, W. G. and Page, R. H. "Initial Development of Turbulent Compressible Free

Shear Layers." Journal of Basic Engineering, Vol. 91, 1969, pp. 67-73.
14. Lewis, J. E: and Behrens, W. "Fluctuation Measurements in the Near Wake of a

Wedge with and without Base Injection." AIAA Journal, Vol. 7, No.4, April
1969, pp. 664-670.
61
AEDC-TR-76-127
15. Page, R. H. and Sernas, V. "Apparent Reverse Transition in an Expansion Fan."

AIAA Journal, Vol. 8, No.1, January 1970, pp. 189-190.
16. Zumwalt, G. W. "Analytical and Experimental Study of the Axially-Symmetric

Supersonic Base Pressure Problem." Ph.D. Dissertation, University of Illinois,
1959.
17. Mueller, T. J. "Determination of the Turbulent Base Pressure in Supersonic

Axisymmetric Flow." Journal of Spacecraft and Rockets, Vol. 5, No.1,
January 1968, pp. 101-107.
18. McDonald, H. "Turbulent Shear Layer Reattachment with Special Emphasis on the
Base Pressure Problem." Aeronautical Quarterly, Vol. XV, August 1964, p. 247.
19. McDonald, H. "An Analysis of the Turbulent Base Pressure Problem in Supersonic
Axisymmetric Flow." Aeronautical Quarterly, Vol. XVI, May 1965, pp. 97-121.
20. Chow, W. L. "Recompression of a Two-Dimensional Supersonic Turbulent Free

Shear Layer." Develo£ments in Mechanics, Vol. 6, Proceedings of the Twelfth
Midwestern Mechanics Conference, 1971, pp. 319-331.
21. Chow, W. L. and Spring, D. J. "Viscous Interaction of Flow Redevelopment after

Flow Reattachment with Supersonic External Streams." AIAA Journal, Vol.
13, No. 12, December 1975, pp. 1576-1584.
22. Weng, C. H. "Base Pressure Problem Associated with Supersonic Axisymmetric

External Flow Configurations." Ph.D. Dissertation, University of Illinois, 1975.
23. Weinbaum, S. "Rapid Expansion of a Supersonic Boundary Layer and Its

Application to the Near Wake." AIAA Journal, Vol. 4, No.2, February 1966,
pp.217-227.
24. Korst, H. H. "Dynamics an~ Theormodynamics of Separated Flows." in Heat and

Mass Transfer in Boundary Layers, Volume 2, Edited by N. Afgan, et al.,
Pergamon Press, 1972, pp. 781-825.
25. Peters, C. E. "Turbulent Mixing and Burning of Coaxial Streams Inside a Duct of
Arbitrary Shape." AEDC-TR-68-270 (AD680397), January 1969.
26. Peters, C. E., Phares, W. J., and Cunningham, T. H. M. "Theoretical and

Experimental Studies of Dueted Mixing and Burning of Coaxial Streams."
Journal of Spacecraft and Rockets, Vol. 6, No. 12, December 1969, pp.
1435-1441.
62
AE DC-TR-76-127
27. Green, J. E. "Two-Dimensional Turbulent Reattachment as a Boundary Layer

Problem." in Separated Flows, Part 1, AGARD Conference Proceedings No.4,
May 1966, pp. 399-427.
28. Schubauer, G. B. and Tchen, C. M. "Turbulent Flow." in High Speed Aerodynamics

and Jet Propulsion, Volume V, Turbulent Flows and Heat Transfer, Edited by
C. C. Lin, Princeton University Press, 1959, p. 171.
29. Phares, W. J. and Loper, F. C. "A Technique for Solving Integro-Differential

Equations with Application to Turbulent Mixing." AEDC-TR-64-209
(AD450730), November 1964.
30. Hama, F. R. "Experimental Investigations of Wedge Base Pressure and Lip Shock."
NASA CR8l03l and JPL/CIT TR-32-l003, December 1966.
31. Maise, G. and McDonald, H. "Mixing Length and Kinematic Eddy Viscosity in a
Compressible Boundary Layer Flow." AIAA Journal, Vol. 6, No.1, January
1968, pp. 73-80.
32. Tucker, M. "Approximate Calculation of Turbulent Boundary Layer Development in

Compressible Flow." NACA TN 2337, April 1951.
33. Peters,C.E. and Phares, W. J. "An Integral Turbulent Kinetic Energy Analysis of
Free Shear Flows." in Free Turbulent Shear Flows, Volume I, Conference
Proceedings NASA SP 321,1973, pp. 577-624.
34. Bradshaw, P. "Effects of Streamline Curvature on Turbulent Flow." AGARDograph

AG169, August 1973.
35. Harsha, P. T. "Free Turbulent Mixing: A Critical Evaluation of Theory and

Experiment." AEDC-TR-7l-36 (AD7l8956), February 1971.
36. Mueller, T. J. "On Separation, Reattachment and Redevelopment of Turbulent

Boundary Layers." Ph.D. Dissertation, University of Illinois, 1961.
37. Tani, I. "Experimental Investigation on Flow Separation Over a Step." Paper

Presented at Symposium on Boundary Layer Research, Freiburg, Germany,
August 26-29, 1957.
38. McDonald, H. "The Turbulent Supersonic Base Pressure Problem: A Comparison

Between a Theory and Some Experimental Evidence." Report No. Ae 194,
British Aircraft Corporation, April 1965.
63
A E DC-T R-76-127
39. Chapman, D. R., Wimbrow, W. R., and Kesler, R. H. "Experimental Investigation of

Base Pressure on Blunt Trailing Edge Wings at Supersonic Velocities. ". NACA
TN 2611, 1952.
40. Reda, D. C. and Page, R. H. "Supersonic Turbulent Flow Reattachment Downstream

of a Two-Dimensional Back Step." AFOSR Scientific Report 69-1592 TR
(Rutgers University RU-TR l25-MAE-F), May 1969.
41. Reid, J. and Hastings, R. C. "Experiments on the Axi-symmetric Flow over

Afterbodies and Bases at M = 2.0." RAE Report Aero. 2628, October 1959.
42. Charters, A. C. and Turetsky, R. A. "Determination of Base Pressure from

Free-Flight Data." BRL Report No. 653, March 1948.
43. Sirieix, M., Delery J., and Monnerie, B. "Etude Experimenta1e du Proche Sillage de
Corps de Revolution en Ecoulement Supersonique." ONERA TP-608, 1968.
44. Donaldson, 1. S. "The Effect of Sting Supports on the Base Pressure of a

Blunt-Based Body in a Supersonic Stream." Aeronautical Quarterly, August
1955, pp. 221-229.
45. Badrinarayanan, M. A. II An Experimental Investigation of Base Flows at Supersonic

Speeds." Journal of the Royal Aeronautical Society, Vol. 65, July 1961, pp.
475-482.
46. Rebuffet, R. "Effects de supports sur l'ecoulement a l'arrierre d'un corps.". AGARD
RAPPORT 302, March 1959.
47. Seiling, W. R., Przirembel, C. E. G., and Page R. H. "Axisymmetric Turbulent

Near-Wake Studies at Mach Four: Blunt and Hemispherical Bases." AFOSR
Report No. 68-2465 (Rutgers University RU-TR l22-MAE-F), November, 1968.
48. Sinha, R. P, "Experimental and Theoretical Investigation of the Near Wake in an

Axisymmetnc Supersonic Flow with and without Base· Injection." Ph.D.
Dissertation, New York University, 1968.
64
AEDC-TR-76-127
APPENDIX A
COEFFICIENTS OF THE INTEGRAL EQUATIONS
The following parameters are defined:
f s fi + b
a o Of 1
Slg = ;g (pu)
Sn = aga 2
(pu )
S3g =
a
ag (puC)
The jet regime coefficients (F, G, H, and I), where g = p, ri, b, K, and x, are (the
subscript c on Pc has been omitted for convenience):
1 a+1
2 1 r·1 d
F1 = ab f Slpydy + r~b f S lp dy +
(a + 1)
-:- (p.u.)
dp J j'
0 0
2 1 1
F2 ab f Slr.ydy + rib
1
f Slr.dy
1
0 0
2 1 1
F3 ab f SIb ydy + rr b f Slbdy
0 0
2 1 1
F4 ab f SlKydy + rib f SlK d y
0 0
1 1
2
FS =_ab f Slx ydy + rib f Slx dy - PsV sr~
0 0
1 r sa+1 r·1
a+1
2 1 d (p.u ~)
G1 ab f S2pydy + rib f S2p dy + -- +--
(a + 1) (a + 1) dp 1 1
0 0
1 1
2
G2 ab f S2rydy + rr b f S2r. dy
1 1
0 0
2 1 1
G3 ab f S2bydy + rib f S2b d y
0 0
65
AEDC-TR-76-127
o o
H3
o o
o o
66
AEDC-TR-76-127
The wake regime coefficients (F, G, H, I, and J), where g = lie, rb b, K, Cc and x are:
a+l
2 1 1 ri a
Fl a b f Slu ydy + r%f
1
Slu e dy + (a + 1) a;-;- (p elie)
e 0
e
0
2 1 1
F2 ab f Slr.ydy + r~b f Slr. dy
1
0 1 0
2 1 1
F3 ab f Slbydy + r~b f Slbdy
0 0
2 1 1
F4 ab f SIKydy + r~b f SlKdy
0 0
a+l
2 1 1 ri a
F5 ab f SIC ydy + r%f
0 e 1
0
SIC e dy +-(--1)
a+
ac (p
e
eli e)
2 1 1 a
F6 =-a b f Slxydy - r~b f Slxdy - P s v s r s
0 0
2 1 1 d
ab
o
f S2r.ydy + r%
1 I
f
0
So
~ri
y
o o
o o
1 a+l
2 1 ria 2
G5 = ab f S2C ydy + r~b f S2C dy + (a + 1) aCe (Pelie)
0 e 0 e
2 I 1
G6 -ab f S2x ydy r~b f S2x dy Pslisvs r as
0 0
2 Yo a Yo
ab f S2u ydy + rib f S2u dy
o e 0 e
67
AEDC-TR-76-127
o o o o
o o o o
o o
Yz
+ uill r0b
I
J S1 X d), + rill r aill
o
o o
o o
o o
2 Yz Ii
ab J S3u ydy + r~b J 2 S3u dy
o c 0 c
a+1 a+1
ria
+ -- - (p u C ) - C _,'__,_) -
r· a
(a + 1) au
C
c c c ill\a + 1 au C (0. C u C )
2 Yz a Yz
ab J S3r.ydy + rib J S3r.dy
o 1 0 1
68
AEDC-TR-76-127
2
y, y, 2 Y, Y,
ab f S3bydy + rib f S3b dy Cma b f S lbydy Cmri b f SIbdy
0 0 0 0
2
y, y, y, a y,
2
ab f S3Kydy + rib f S3K dy C rna b f SIKydy Cmribf SIKdy
0 0 0 0
2
y, a - y, 2 Y, Y,
ab f S3C ydy + rib f S3C dy Cma b f SIC ydy - Cmri b f SIC dy
0 c o c 0 c 0 c
o o o o
In the fully developed regime define
y rib
rs b
The fully developed regime coefficients (F, G, H, I, and J), where g == p, u c , b, K, Cc and x are:
1
FI bba f SIp yady
0
1
F2 bba f Slu yady
0 c
1
F3 bba f Slbyady
0
1
F4 bba f SIKyady
0
a 1
FS bb f SIC yady
0 c
1
F6 bba f Slxyady - p s v s r sa
0
69
AEDC-TR-76-127
1 r a+ 1
bba f S2 yady + _s_ _
o P (a + 1)
1
bba f S2u yady
o c
1
- bbaf S2xyady - p S v S u S raS
o
o {)
o o
70
AEDC-TR-76-127
1 a
a y dy
IS bb f S3C
c
0
1
16 - bba f S3x yady
0
y. y.
1r bba f S3pyady - Cmbba f SIpyady
0 0
y. y.
h bba f S3u yady _ Cmbb
a f Slu yady
0 c 0 c
y. y.
bba f a Cmbba f Slbyady
13 S3by dy -
0 0
y. y.
a
14 bba f S3Ky dy - Cmbba f SlKyady
0 0
y. y.
15 bba f S3C yady _ Cmbb
a f SIC yady
0 c 0 c
y. y.
16 - bba f S3x yady + Cmbb
a
f a
SIx yady + qmr m
0 0
71
AE DC-T R-76-127
NOMENCLATURE
b Shear layer thickness
cp Specific heat at constant pressure
C Element fraction of bleed gas
F,G,H,I,J Coefficients in system of ordinary differential equations
g Generalized variable (Appendix A)
h Base height in planar flow
K Parameter in species profile equation
M Mach number
n Boundary-layer profile exponent
N Generalized coefficient in system of ordinary differential equations
p Static pressure
Pb Base pressure
Po Total pressure
Po I Pitot pressure
P Pressure gradient parameter
q Turbulent species flux
r Lateral coordinate
rb Base radius
rj Lateral distance from centerline to inner edge of shear layer
rs Lateral distance from centerline to outer edge of shear layer
R Gas constant
RSP Rear stagnation point
72
AEDC-TR-76-127
Reo Momentum thickness Reynolds number
RT Turbulent Reynolds number
S Generalized integrand (Appendix A)
To Total temperature
u Axial velocity component
v Lateral velocity component
x Axial coordinate
6x Integration step size
y Distance from wall in initial boundary layer; dimensionless mixing zone

coordinate (Appendix A)
a Geometric parameter (one for axisymmetric flow, zero for planar flow)
'Y Ratio of specific heats
o Initial boundary-layer thickness
Op Distance at beginning of recompression between centerline and streamline

which orginates at outer edge of initial boundary layer
e Momentum thickness of initial boundary layer
p Density
7 Turbulent shear stress
SUBSCRIPTS
Conditions before base expansion
a Outer free stream
c Centerline
d Conditions on dividing streamline
Conditions in base bleed flow
73
AE DC-TR-76-127
m Conditions at middle of shear layer
p Conditions at beginning of recompression
r Conditions at rear stagnation point
74

Ada 030015

Uploaded by

Copyright:

Available Formats

Ada 030015

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ada 030015

Uploaded by

Copyright:

Available Formats

AEDC-TR-7S-127 Ill

ENGINE TEST FACILITY

Final Report for Period July 1973 - December 1975

Approved for public release; distribution unlimited,

DIRECTORATE OF TECHNOLOGY (DYR)

When U. S. Government drawings specifications, or other data are used for

References to named commercial products in this report are not to be

FOR THE COMMANDER

//JL ~. ...t'7 _ /2 -..#-

ELTON R. THOMPSON ROBERT O. DIETZ

7. AUTHOR(s) 8. CONTRACT OR GRANT NUMBER(s)

C. E. Peters and W. J. Phares, ARO, Inc.

Arnold Engineering Development September 1976

Arnold Air Force Station, Tennessee 37389 74

Approved for public release; distribution unlimited.

18. SUPPLEMENTARY NOTES

An analytical model for planar and axisymmetric, supersonic,

singularity, similar to the Crocco-Lees critical point, occurs

1. Features of the Near-Wake Flow Field 8

5. Corner Expansion of the Initial Boundary Layer 25

27. Effect of the Initial Boundary Layer on the Base Pressure in

A. COEFFICIENTS OF THE INTEGRAL EQUATIONS . . . . . . . . . . . . . . . 65

1.1 PROBLEM DEFINITION

Recompression Shock Wave

Axial Dista nce from Base

Figure 1. Features of the near-wake flow field.

In view of the numerous practical flows in which supersonic near-wake mechanisms

a. Bodies in supersonic flight b. TnJlncated plug nozzles

1.2 REVIEW OF EARLIER ANALYTICAL WORK

A major feature of the Crocco-Lees solution is the appearance of a saddle-point

Although applicable to a broad class of strong viscous interaction problems, the

Chow (Ref. 20) has developed a two-layer integral treatment of recompression in

1.3 IMPORTANT FEATURES OF A REALISTIC ANALYTICAL MODEL

2.0 DEVELOPMENT OF ANALYTICAL MODEL

22277 ~ ~ us~/ I Axial Velocity /' /~//

Beginning of Recirculation--'-/ ~ L RSP

Figure 3. Nomenclature for near-wake analysis.

2.1 BASIC ASSUMPTIONS

The major assumptions used in the analysis are:

1. The flow is steady and is either planar or axisymmetric.

2. All gases obey the perfect gas law.

3. The integral form of the boundary-layer equations describes the flow

4. The turbulent Prandtl and Schmidt numbers are unity.

8. The normalized profiles of the axial component of velocity in the turbulent

11. The centerline static pressure is constant in the wake regime.

The use of the boundary-layer approximations (assumption 3) is consistent with

2.2 BASIC EQUATIONS

The basic boundary-layer equations are:

a 1 for axisymmetric flow

T is the turbulent shear stress

Element Species Conservation

pur a - ac + pvr a _ac a ( a)

where q is a turbulent species flux.

2.3 SHEAR LAVER PROFILES

The free shear layer velocity profile is given by

The velocity at the half-radius surface, rro (= ri + b/2), is given by

2.4 COMPUTATION OF DENSITY IN THE VISCOUS FLOW

2.5 TURB,ULENT TRANSPORT TERMS

The turbulent shear stress, T m , which appears in Eq. (11) is given by

where € is the turbulent eddy viscosity. The eddy viscosity is given by

...... t::. &a-:-~------=