Shock Waves(1996) 6:275-286
9 Springer Verlag 1996
Investigation on onset of shock-induced separation
H.-D. Kim 1, K. Matsuo 2, T. Setoguchi 3
1Department of Mechanical Engineering, AndongNational University, Andong760-600, Korea
2Graduate School of Engineering Sciences, Kyushu University, Kasuga, Fukuoka 816, Japan
3Department of Mechanical Engineering, Saga University, Saga 840, Japan
Received 7 February 1996 / Accepted 6 June 1996
Abstract. A great number of experimental data indicating
shock wave/boundary layer interactions in internal or ex-
ternal supersonic flows were reviewed to make clear the
mechanism of the interaction and to decide the onset of
shock-induced separation. The interesting conclusions were
obtained for the considerably wide range of flow geometries
that the onset of separation is independent of the flow geome-
tries and the boundary layer Reynolds number. It is found
that the pressure rise necessary to separate the boundary
layer in supersonic external flows could be applied to such
internal flows as overexpanded nozzles or diffusers. This is
due to the fact that the separation phenomenon caused by
shock wave/boundary layer interactions is processed through
a supersonic deceleration. The shock-induced separation in
almost all of interacting flow fields is governed by the con-
cept of free interaction, and the onset of shock-induced sepa-
ration is only a function of the Mach number just upstream of
shock wave. However, physical scales of the produced sep-
aration are not independent of the downstream flow fields.
Key words: Supersonic flow, Shock wave, Boundary layer,
Separation, Free interaction, Compressible flow
1 Introduction
Supersonic flows usually contain shock waves generated ei-
ther by a change in the slope of a control surface or by a
back pressure. These shock waves nearly always meet wall
surfaces on which a boundary layer develops. Then a com-
plex viscous interaction between the subsonic layer, very
near to the wall, and the adjacent supersonic region occurs
in the shock foot region. Because influences of the pres-
sure waves embedded in the supersonic region propagate
upstream through the subsonic part of wall boundary layer,
the streamline directions in the subsonic layer alter to accom-
modate these pressure effects. Also the neighboring stream-
lines in the supersonic region are correspondingly altered,
but because the supersonic pressure field is very sensitive
to streamline angle, it is significantly affected by relatively
Correspondence to: H.-D. Kim
small adjustments in the subsonic layer and the resulting
compression waves can be generated from the sonic line in
the boundary layer. The compression waves interact with
the pressure waves in the supersonic region and the global
wave pattern is significantly changed from that expected by
an invicsid theory.
As the boundary layer flow, whether it is laminar or
turbulent, meets the shock wave, frequently the flow is sep-
arated from the shock foot, with a consequent serious fall
of the performance of system. If the shock wave impinging
on the boundary layer is weak, the boundary layer flow may
endure the adverse pressure gradient due to the shock wave.
When the shock wave is, however, strong over a certain
strength, a destabilizing action due to its adverse pressure
gradient is imparted into the boundary layer flow and as a
result, the shock induced boundary layer separation (SIBLS)
can be occurred. The SIBLS is usually the cause of increas-
ing the aerodynamic drag and heating, and thus adversely
affects the inlet or control surface performance. Therefore
the ability to predict and suppress the SIBLS should lead to
substantial improvements in the performance, reliability and
costs of a wide range of fluid dynamic devices.
In the supersonic external flows past a body, the effects
of SIBLS on the global flow field have considerably inves-
tigated from the earlier time and now they can be predicted
for most circumstances, or to be avoided by a suitable de-
sign. However the onset condition of SIBLS in such flow
fields could not be successfully obtained in spite of many
successive investigations.
Recently the sustained interests in predicting the onset
condition were due to the phenomenological as well as aero-
dynamical and industrial importance of the mechanism caus-
ing the SIBLS. The onset of SIBLS in many external flows
may be correlated with the data of internal flows such as su-
personic diffusers or overexpanded nozzles if the mechanism
of SIBLS is fully understood and the governing parameters
are appropriately found. The onset of SIBLS may be signif-
icantly changed by the characteristics of the boundary layer
flow. Because of the greater ability of the turbulent boundary
layer to endure an adverse pressure gradient, the boundary
layer flow may be still attached for a strength of the shock
wave necessary to cause the SIBLS in laminar flows.
276

The present paper is a review of experimental and theo-

retical works published in the past. The focus of the present
work is the assessment of the experimental data on the
SIBLS in external and internal flows in order to determine
the onset of SIBLS. We will therefore mainly treat the onset
of SIBLS in shock wave boundary layer interaction flows.
Criterion of SIBLS will be obtained for both external and
internal flows. We will also discuss the effects of boundary
layer flow on the SIBLS. All of the discussions are lim- Peak pressure
ited to steady, two-dimensional or axisymmetric interacting
flows, and wall curvature (see Kim et al. 1993), heat transfer P2
(see Raghunathan et al. 1995), or mass transfer effects (see Ps
Hingst et al. 1983) are assumed to be absent. Separation pressure
Streamwise distance

2 Free interaction of shock wave/boundary layer

Historically the investigation of shock wave/boundary layer

interactions has been experimental and theoretical but by
far experiments have contributed most to the current under-
standing of the interaction flow fields. Some of the earliest
studies (Green 1970, Hankey et al. 1975, Delery et al. 1985)
of the shock wave/boundary layer interactions were mostly
concerned with the boundary layer separation by an inci-
dent shocks or a flow obstacle. Many later investigations of
quasi-two dimensional interactions generated by compres-
sion ramps, forward facing steps, sudden enlargements and
shocks in diffusers or overexpanded nozzles were also car-
ried out with the same objective. Recently the detailed com-
pilations of experimental and computational studies of three-
dimensional interactions in hypersonic inlet models, were
presented by some researchers (Garrison et al. 1996, Knight
et al. 1995, Liao et al. 1996). Nearly all research efforts were
concentrated on the symmetric and asymmetric crossing-
shock interactions. The detailed structure of shock-induced
separation was elucidated by means of flowfield visualization
(Garrison et al. 1993, Knight et al. 1992). Many have been
learned, but there still exist a large discrepancy between the
information to be needed and the information actually avail-
able. For example, what are the fluid mechanical parameters
governing the interaction phenomenon? How can we effec-
tively control the interaction to improve the performance of
the fluid devices? At precisely what condition does the on-
set of SIBLS appear? There are countless other questions
about flow unsteadiness and appropriate turbulence model.
Further progress to answer these unsolved problems awaits
advances in the fundamental understanding about the inter-
action physics.
A large body of experimental data in supersonic flows
exists on boundary layer separation due both to obstacles
and shock waves. The cases of forward-facing steps, curved
surfaces, and compression comers or ramps have been com-
prehensively studied for laminar and turbulent flows (Chap-
man et al. 1957, Pearcey 1955, Kuehn 1959, Liu et al. 1988).
Various theoretical attempts (Gadd 1953, Inger et al. 1976)
have been made in order to describe the mechanism and
onset of the SIBLS. It can be found that in every exper-
imental model investigated, reattachment of the boundary
layer follows quite soon after the separation, and the early
part of the pressure rise associated with the separation has
; ~ ! ~ .....
Peak p r e s ~ --'~
P2
Psi
Separation pressure
b Streamwise distance
Fig. 1. Wall pressure distributionof shock wave/boundarylayer interaction
on forward-facingstep and compressionramp
features closely similar between all models, while the later
part associated with the reattachment has not.
Mager (1955, 1956) argued that the boundary layer does
not know what combination of circumstances creates the
pressure rise leading to the separation. It only knows what
pressure rise is required at given conditions of Mach number
and Reynolds number to cause the boundary layer to sepa-
rate. Chapman et al. (1957) used the termed free interaction
with respect to regions of flow which are free from direct
influences of downstream geometry, and in other words, in-
dependent of the mode of inducing the separation. Many
experimental results showed that, at least as far as the sepa-
ration point (in the sense of two-dimensional flow) the inci-
dent shock waves and all the forms of the obstacle mentioned
above are indeed free interaction. But, once separated, the
effects of geometry which the boundary layer must negotiate
during the process of reattachment put the later phase out-
side the category of the frce interaction, and the similarity
of the various models can be no longer ascertained.
Figure 1 shows schematics of wall pressure distribu-
tions of the interaction flow fields between shock waves
and boundary layers over a forward-facing step and a com-
pression ramp. In general, all the models investigated have
been exhibited the pressure rise characteristics which cor-

40
{R~I~ -0.15-10.8•
0 ~ - 1.19- 1.22
@ Hil = 1.41
30 -" ~ l Q(1/It~ Distance
9 ID
.................... ............................ ~,........... .~........~............
~"a""~176 ..........i " " § ........ ~ r ..................................
20 ........................................{.
! i
O. ................i.............. "m". . . . . . .). " *
.........................9 9~
ol oo.
O
)
10 m
1.1 1.2
Mach number M1
Fig. 2. Effects of boundary layer Reynolds number and shape factor on
interaction length
respond closely to one another under similar conditions of
Mach and Reynolds numbers as far as the station 2 for lami-
nar boundary layers, and up to s for turbulent boundary layer
flows. However a difficulty appears in further pressure rise
from s to 2 in the turbulent boundary layer flow, where the
detailed pressure distributions from various models are no
longer identical. Therefore this region must be regarded to
be outside the category of the free interaction.
Typical experimental substantiation are the interaction
on a supercritical wing surface. Figure 2 shows an excellent
grouping of the experimental data (Sirieix et al. 1981, Delery
et al. 1986), with a moderate scatter due to the difficulty of
accurately defining the supersonic interaction length L* from
the wall pressure distributions. For the range of Mach num-
ber of 1.09 to 1.30, the L* is normalized by the boundary
layer displacement thickness 6{ at the start of the interaction,
and the Reynolds number based upon the boundary layer dis-
placement thickness is varied in the range of 0.15 • l04 to
1.08 x 105 and the value of the incompressible shape fac-
tor Hi1 for the whole set of data is close to about 1.20,
where Hu means the ratio of the boundary layer displace-
ment to momentum thickness at the start of the interaction.
It can be found that the influence of the Reynolds number
on the physical extent L* and on the thickness 6~ disappears
when these two variables are normalized one by the other.
For a given value of the shape parameter, the displacement
thickness of the incoming boundary layer can be a proper
scale for the supersonic interaction length L*, and the ratio
L*/~5~ is not very sensitive to the effect of the upstream
Mach number M1. Although the scatter observed when the
Ml comes close to 1.30 may correspond to a situation in the
verge of separation, the L* is practically independent of M1.
For different situations of the state of the incoming bound-
ary layer, the experimental data shown are regularly spaced
as a function of the shape parameter. For example, we ob-
serve that the normalized interaction length increases almost
277
110
) i
) . O H =1.20
1.28
9 Hil = 1.28 90 ............................................................................................................................. 1.31 ...................
Averaged exp. [] 1.41
Nil = 1.31
/ )
L o/o,,i oo 8 o
70
~7 ......~,,-~.~"Or
~" ~ i _ -4.O.~-..rp~.-.
-_
~'rTe~-O(Y~-O.B.O.-..<~
U)~.~_~ O ~ ~ "
0<> )u 0 C~. ) u,J-, i ,, 0
) 0 0 0
50
~ ............
30
1.1 1.2 1.3
~ Ob,....q .. o Mach number M1
Fig. 3. Interaction length of shock wave and boundary layer
1.3
by twofold when the Hi~ increases from about 1.20 to 1.41.
This can easily understand by considering that when the H.il
is high, the boundary layer is less full and its subsonic part
is thicker, and consequently the distance for the propagation
of upstream influence is longer. Also the influence of shape
parameter on the interaction length can be disappeared by
using an appropriate variable (see Fig. 3). As indicated in
these figures, the normalized interaction length at low su-
personic speeds can be provided in terms of the boundary
layer displacement thickness and shape parameter at the start
of the interaction.
Similar conclusions for the other supersonic flows, i.e.,
the forward-facing steps, the compression ramps, or the in-
cident shock waves can be also deduced by surface pressure
distributions. Figures 4a and 4b show the pressure coefficient
of the interaction fields, where Cps and Cp 2 refer to the pres-
sure coefficient at separation point and peak pressure point,
respectively, as illustrated in Fig. 1. The data shown refer
to the flow conditions with a variety of interaction geome-
tries but with a zero pressure gradient turbulent boundary
layer. For the range of Mach number employed, the data of
both the pressure coefficients seem to collapse onto a sin-
gle curve, and to decrease with increasing Mach number.
The pressure rises up to separation point or even up to peak
pressure point are independent of an external action to cause
themselves and are only a function of the upstream Mach
number. This is just the concept of free interaction that has
described earlier.
Many theoretical or semi-empirical equations (Inger et
al. 1980, 1984, Mager 1956) concerned with the free in-
teraction show good agreement with the experimental data
obtained in the external flows. According to an analytical
methods by Mager (1955, 1956), it is interpreted that the
SIBLS occurs whenever the turbulent boundary layer finds
itself at a certain pressure rise. Under such a consideration,
the pressure gradient in the shock wave boundary layer in-
teraction was assumed to be sufficiently large compared with
the friction force on the wall surface, and the pressure rise
to the separation was thus a function of Mach number be-
fore and after the oblique shock wave. This results in the
278

0.4 0.8
i A Sterret et a1.(1962)
0.7 -..~....-..-@.---.-- Averaged exp. -.................i......................................
i......................................
~"...~...,...i-.~"
i . O Bogdonoff et a1.(1955)
~ 0.3
g
~ff~/z~"~.r I Gadd et a1.(1954)
0.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8 0.2 0.5
..

~0.4
8
0.1
0.3
............................. iEq.(7),J=0.60 ~1[...........
Eq.(8),J=0.60
" ~
0 i I I I r I i I l
0.2
0.4 2 3 4 5
Mach number M]
Fig. 5. Variation of separation pressure with Mach number
i a i i i
J to give quite good fit to experiment. A revise of this ex-
pression was proposed by Arens et al. (1963a, 1963b) for
0.2 .... the case in which the stagnation was conducted by a normal
~ " ' ",,,el rn~ i shock wave followed by an isentropic compression when the
1 5<H/61<3 0 characteristic streamline was initially supersonic. This leads
0.1 [] Sterret et a1.(1962)
O I to the following relation,
a~ ..................... Love(1955) ....................t!......................................
~.....................................
A Bogdonoffet a1.(1955) i i
I[
p, { ~2 ~ , .L(@j_~) , (4)
0 b i i , i , i , i , I
2 3 4 5 6
Mach number M 1
where good agreement with experimental results for the
Fig. 4. a Variation of separation pressure with Mach number b Variation forward-facing steps, the compression ramps, and the in-
of pressure rise coefficient with Mach number (H is the height of forward- cident shock waves was claimed to use the value J = 0.56.
facing step and ~1 the boundary layer thickness at the start of interaction)
Figure 5 shows the pressure rise at the separation point
against the Mach number, together with the averaged exper-
following relations between Mach number M] just upstream imental data for the forward-facing steps, the compression
of the interaction and Ms at separation point. ramps, and the incident shock waves. The value of specific
heat ratio was taken by 1.4 throughout. It can be found that
M2s -=KM 2 , (I) when J = 0.60, there is little difference between Eqs. (3) and
(4), and quite good agreement with the experimental data is
In order to determine the pressure ratio across the sepa-
given by Eq. (1) w i t h / s = 0.67 and Eq. (4) with J = 0.56.
ration, the oblique shock approximation can be used as
These data indicate that the flow process up to the separa-
p~ ~(1 - K ) ~- ~ tion point is dependent on only the Mach number and the
1+ ~2 , (2) pressure rise (Pl/Ps) decreases with increase in the Mach
Pl 1 + ('y - 1)~- number. This can be concerned with the separation process
being in supersonic. Therefore the free interaction can be
where K is constant to be assumed. Gadd et al. (1954)
reasonably interpreted as being independent of the external
considered a 1/7 power velocity profile on a supersonic
actions or downstream geometries causing the pressure rise
plane flow and assumed some characteristic fraction J of
in interaction field between shock wave and boundary layer.
the freestream velocity to be stagnated due to the pressure
gradient of shock wave. The friction force was also negli-
gible by comparison with one due to the pressure gradient.
In order to evaluate the pressure rise necessary to bring the 3 Onset of SIBLS in external flows
flow on a streamline to rest isentropically, Eq. (3) is drawn
from one-dimensional gas dynamic equations, The boundary layer separation in external supersonic flows
"V
was well established by the concept of free interaction. When
p~={ ~--1)II+~M
2__~ ?}~-1
(3)
the pressure rises due to external forces or downstream ge-
ometries causing the separation is, however, quite weak or
small, the separation is unlikely to occur and the concept of
where p~/pi corresponds to the pressure rise at the separa- free interaction breaks down. This situation appears when the
tion point. It is recommended 0.6 or 0.54 as the value of shock Mach number is so low that the pressure disturbance
 279

Normal shock wave

Flow ~ ]
b, Sepa.at
on Reattachmeo*
Pto /
o

p* Pressure corresponding to local sonic state

.eparat--:o attachmcn* P2 kink pressure

Pte

Streamwise distance
Fig. 6. Definition of separation onset '~ p.
,~ P2 ................ -...... ]--

,
~ Separation
is not sufficient to destabilize the boundary layer. There-
fore it is necessary to determine the limit of the strength of
disturbance applicable to the concept of free interaction. Un- Pl
fortunately the authors cannot find any theoretical study or
even any systematic experiment concerning with this prob-
lem. This section devotes to such a subject in external tows Distance X
at low supersonic speeds. Fig, 7, Wall pressure of shock/boundary layer interaction on aerofoil
By definition, the incipient separation is the situation in
0.65
which the minimum of the wall shear stress ~-~ in the inter-
action region is exactly equal to zero (see Fig. 6). A further
increase of the shock strength beyond that point leads to a
o
. ~ 0.60 ..................... Normal shock
i ...............................

change in the sign of the shear stress, the region where 7-~
is negative being called separated. At laboratory the incip- 0.55
ient separation is usually detected from visualizations such
as surface oil-tracers, shadowgraphs, schlieren photographs, e.

wall pressure or temperature distributions, and boundary ~2 0.50 ............ P/Pto~ " i" - ~ ' ........................................
layer velocity profiles, since the direct measurement of the
wall shear stress is difficult and inaccurate under the non- 0.45 I I I I I I I
quilibrium boundary layer where exists a strong pressure 0.8
gradient.
According to the above definition of separation, the on- ~ 0'7 1
E
set of SIBLS in what follows is the situation in which the
streamwise distributions of surface friction coefficient has a
0.6 ....
minimum exactly equal to zero. We define the local pres-
sures on the airfoil surface to characterize the separation
onset. Figure 7 indicates the surface pressure distribution in 0.5 ..................................
i..................... Mach number = 1.32 ..............i . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

the vicinity and downstream of the separation, together with

a schematic description of interaction flow field. The on- 0.4 ' , i , i , i ,
set of SIBLS can be diagnosed by considering the variation 1.1 1.2 1.3 1.4 1.5
Mach number M 1
of the normalized pressure Pz/Pto and Pte/Pto as a func-
tion of Mach number Mj, i.e., the shock strength (Pearcey Fig. 8. Criterion of shock-induced separation on aerofoil surface
1955, 1961). Figure 8 shows that as the Mach number Ml
increases, the P2/Pto increases but it then decreases with
further increase in the Mach number. Also, when the Mach value. From these considerations, the value of Mach number
number is increased, Pte at the trailing edge decreases and of 1.32 is interpreted as the bare minimum shock strength
abruptly drops off at Mach number of 1.32. Variation of necessary to cause the separation.
P2/Pto at this point is related to the divergence of pressure The same conclusion can be also obtained in the separa-
at the trailing edge Pte/Pto, which is typical of the develop- tion measurements on an airfoil carried out by Stanewsky
ment of a large separation bubble. That is, the data shown et al. (1960) and Stanewsky (1981). Figure 9 shows the
in Fig. 8 show that the dramatic change in the flow struc- variation of the boundary layer displacement thickness at
ture takes place when the kink pressure P2 is equal to sonic certain appropriate stations versus the Mach number M1. In
280
~o 3
"C'rdl" ''>:200mm'Si .,
.,
............
0 o
1.2
Mach number M 1
Fig. 9. Criterion of shock-induced separation by Stanewsky
1.6
i 16~
[ p2/Pto= 0.528
I,
=~
=
t
...................................
........................
Z
/i
1.01. 0 1.2
Fig. 10. Criterion of shock-induced separation by Alber et al.
experiment, three stations on the airfoil upper surface (i.e.,
upstream of shock wave: 1, immediately downstream of it:
2, and at the trailing edge: te) were allocated as measure-
ment position of the boundary layer displacement thickness.
The data included refer to various transition techniques to
trip the boundary layer and Reynolds number based upon
the chord length of 1.95 to 3.5 • 106. In fact, the boundary
layer tripping mainly affects the thickness of the boundary
layer at the position of shock wave. The large difference
between the corresponding variations in its thickness can
constitute the scaling effect in viscous phenomena on the
shock wave boundary layer interaction. However, the point
of interest here is the kink in the curves which is observed
for the Mach number M1 of 1.32. The kink of variation in
its thickness is related to the fast thickening due to the onset
of SIBLS. Mach number at the kink remains constant, ir-
respective with variation of the state of incoming boundary
layer. This is consistent with the conclusion drawn in Fig. 8.
Another way to determine the onset of SIBLS was made
by Alber et al. (1973). They employed surface oil-tracers to
detect separation on a wing surface. The conclusions were
derived from the wall pressure distributions. In the vicinity
of the shock and before the separation, these distributions
exhibited a kink change in the pressure gradient. Figure 10
1.6
low / i
1.5 [- " Hi/ / .....i.....................................{.....................................
.............................
e~
1.4 I-"................................!"Averagedexperiment .........}........ With separation ...............
..,w-',,z , o t.95
"~ 1.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ ..................
i; N
1.2
Inger i
.................................~......................................~......................... Without separation ...........................
g i i
0~oo i ii [ [ i i i Ii ii ii ii ii .... ,,,,,I,,,,,,,,,I,,,,,,,,
1.3 1.4 111.1 1.2 1.3 1.4 1.5 1.6
Shape factor Hil
Fig. 11. Criterion of shock-induced separation by Inger
shows the situations in which the pressure rises are plotted
as a function of Mach number at the start of interaction. The
Meyer compression
A i / i experimental data points fall exactly onto the curve corre-
sponding to a sonic state. Furthermore, it can be shown that
~ -.....
k/ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
the turning angle of the inviscid outer flow during the in-
teraction process cannot exceed 6.6 deg. The pressure rise
calculated from this turning angle by assuming a simple
compression wave agrees well with experimental data. That
is, the separation onset occurs when the Mach number M1
just upstream of the shock wave is 1.32 and the local outer
Mach number at the separation point is sonic. Although this
criterion of the SIBLS did not include a possible influence
of Reynolds number, it is quite consistent with the two con-
clusions earlier.
1.4 1.6 1.8
Mach number M1 As mentioned carlier, the separation onset and its pres-
sure rise can be influenced by the incoming boundary layer
shape parameter which represents the fullness of the bound-
ary layer, and the Reynolds number. The effect of Reynolds
number can be commonly included in the variation of the
shape parameter. But these two parameters are not necessar-
ily linked, since it is possible to change the shape parameter
at a fixed Reynolds number by an external agent, i.e., pres-
sure gradient, suction and blowing. Figure 11 shows another
data (Inger 1980, 1984) representing the separation onset.
We find that the data all nearly collapse on a single curve
defining the boundary between the interactions without and
with separation. The averaged data somewhat decrease in the
limit Mach number of separation when the shape parameter
is increased. Such a trend can be interpreted since as the
Hil is lower, the boundary layer velocity profile becomes
fuller so that the resistance of the boundary layer to sepa-
ration is greater. However the effect is not as significant as
could be expected. When the boundary layer velocity profile
becomes less filled, i.e., when the H , ii increases, the inter-
action length L* will be increased rapidly. Such an increase
in the interaction length reduces the intensity of streamwise
pressure gradient, the supersonic part of the compression
being spread over a longer distance. This reduction of the
extent of the adverse pressure gradient allows the bound-
ary layer separation to be delayed or avoided, in spite of
the less filled velocity profile at the beginning of the inter-
action. This tendency of the shape parameter to the limit
Mach number is well in agreement with other experimental
 281

j
10 i
M1:4.0 ii iiii i i iiiii
8 ....................
i........................
i....................Compressio .................
i...................... ' ' " " .....
7 ..................................

6
~ i Turbulentbound~~ i

"~
~ 4 i i ~"Incident shock .......................
~L i ! ~ i impinging on flat "~ 4 ~ ~ i ~.~...??N~.;..~.~Q.~..
face 3 ............. i- . . . . .

2 ...................
i. . . . . . . . . . . . . . . . . .

~l 0
1.0
i
1.5
i
i
I
2.0
I I
2.5
i

3.0
I
3.5
Mach number M 1
Laminar boundary layer
I J I
4.0
[ . . . . . ~ i
4.5 5.0 1104
I I I
105
I

Reynolds number (Re~)

106

Fig. 12. Separation pressures under turbulent and laminar boundary layer 9
i
: i i !ilil
: : : : : : :
: ; : ; : : : =

.......................................i....................i.................iii .....

(Sterrett et al. 1960, 1962, Love 1955, Guman 1959, Cooke - 7 .......................
!-- M1'=3:85"---'i'"---i-"4"--i--g ..... M1 P "i---i'-
1963) and theoretical data (Lees et al. 1964, Inger 1984). i ! r i [ [ iii Flow ~ / M s Ps ii
! ! : i i if!.-" :~ ~ ii
On the other hand, the effect of Reynolds number on
the SIBLS is frequently ambiguous rather than the shape
parameter or the boundary layer thickness. In the interac- = 1293! ~ i i !! ! i i i i [ii
tion flows of shock wave and boundary layer, the boundary
layer thickness is usually employed as a characteristic length '8~ 4~ 2..........................
.2: 3i-'-';"-i . '. '. ' . . . . O . ~ . . i...-i-..i..j..:
scale. In that case, the interaction can be strongly dependent
on the way by how is the variation of Reynolds number
given. For example, the pressure rise up to the separation
point decreases as the boundary layer thickness increases 2 ........................................
i...........
iii i i i i i i il
..........................
....i..i.i.
: ; ; ; : ; :
at a fixed unit Reynolds number, while it decreases as the I10 4 105 10 6
unit Reynolds number is increased at a fixed boundary layer. Reynolds number (Re~)
Furthermore, the variation in the Reynolds number can be Figs. 13 and 14. Effect of Mach number and boundary layer Reynolds
induced to change the Mach number. Figure 12 indicates number on separation pressure rise
the pressure rise up to the separation point under both cir-
cumstances of turbulent and laminar boundary layers (Kuehn
1959, 1961). For the turbulent boundary layer, the separa- made too large giving rise to sizable adverse pressure gradi-
tion pressure rise is nearly the same for both interactions ents in the subsonic flow downstream of the shock wave. As
of a compression ramp and an incident shock wave. How- mentioned in the previous section, many experimental works
ever the separation pressure rise under the laminar bound- on the SIBLS in external flows have shown that the pres-
ary layer is significantly small compared with the turbulent sure rise to the separation point is apparently independent of
boundary layer, being dependent on the Mach number. Fur- the geometry of the interaction, and dependent on only the
ther discussions can be given in Figs. 13 and 14. For the upstream Mach number. Such a deduction may be also ap-
wide range of the boundary layer Reynolds number of 104 plicable to overexpanded nozzle flows, where it is generally
tO 106 , the separation pressure rise is insensitive to variation expected that a strength of oblique shock wave is required
of Reynolds number as the Mach number becomes relatively to separate the boundary layer. The present section devotes
small, and essentially independent of Reynolds number for to a discussion of the onset of SIBLS in such overexpanded
the minimum Mach number necessary to separate the bound- supersonic nozzles.
ary layer.
The various ways in which the flow through a nozzle
reacts in order to attain the back pressure conditions are
presented in Fig. 15. Curve A on the plot shows the rela-
4 Onset of SIBLS in internal flows tionship between the overall pressure ratio Pb/Po and the
nozzle area ratio Ae/At for the correct isentropic expansion
Many earlier investigations (Campbell et al. 1960, Bloomer conditions. If for a given nozzle the back or ambient pres-
et al. 1961, McKenny 1954, Frazer et al. 1959, Fradenburch sure is reduced below the correct expansion value (curve
et al. 1953) on the SIBLS in overexpanded nozzles or dif- A), the change is met by the formation of Prandtl-Meyer
fusers have showed that it could be reasonably assumed that expansion waves at the nozzle exit. Conversely if the back
the boundary layer separates whenever the normal shock re- pressure is increased, compression waves at the nozzle exit
compression exceeds the exit pressure. Generally the similar raise the flow pressure to meet the new boundary condition.
effects would occur if the divergence angle of nozzle were However there is a limit to the compression shock strength
282

1.0 0.7

0.6 ..... Surnmerfield et al. ~ . . ~

0.5 g

~0.5 ta.i ............. : ............ i.............. i..............

~0.4 AJ Ai
"4 , i i i io
0.3 ......~ ' i ' " L " " ~ ; ~ ' O T ~ "..............................!""""2..........g'"
o 0.1 ' ~ AA~ O ~ T o
$02 ............,............
A
. . ,1
..................................+ ........................., .............. .............i ...........................
~x
0.05 i ~X i [ i •Scheller & Bierlein i ]
0.1 .............i.............."*.............i..............i.......... [] Foster & Cowles ~..........."l
i i Others Green i ]
0 . . . . . . . . ~ " ~ " ' - ' " i , i , !
9
0 20 40 60 80 100 120 140 160 180 200 220
Z
Nozzle pressure ratio p0/Pb

0.01 Fig. 16. Separation pressure data in supersonic overexpanded nozzle flows

0.005
l 5 10 50
Area ratio Ae/At
Fig. 15. Flow pattern with nozzle pressure ratio and possible separation
range
which can be maintained at the nozzle exit, and when the
back pressure increases sufficiently to cause this limit to be
exceeded, the shock wave moves back into the nozzle and
the flow downstream separates from the nozzle wall to form
a free jet. For example, considering a nozzle of area ratio
10, a correct expansion at the nozzle exit is obtained with
a pressure ratio Pb/Po of 0.0072. If this ratio is increased,
for example, to 0.13, it is found in practice that isentropic
expansion will only take place down the nozzle to an area
ratio of about 2.6, where oblique shock waves raise the pres-
sure ratio to the nozzle exit value on the curve B. Further
flow down the nozzle will be separated from the nozzle wall.
Thus the possible range of the SIBLS in this type of noz-
zle can be considered to be in the shaded area. It will be
noted that the chosen overall pressure ratio of 0.13 can be
attained by a nominal shock wave at the nozzle exit, but by
experimental experiences this condition is seldom normally
encountered.
Figure 16 shows comparison of various experimental val-
ues of separation pressure ratio P~/Po versus overall pres-
sure ratio Pb/Po for conical nozzles. The data indicated in-
volves the separation pressure ratio P~/Po for a variety of
divergence angles of nozzle both with unheated and heated
flows, and for different working fluids (Campbell et al. 1960,
Bloomer et al. 1961, McKenny 1954, Frazer et al. 1959,
Fradenburch et al. 1953). Strictly speaking, the divergence
angle varied from 15 to 30 degrees and the area ratio from 8
to 75. Summerfield et al. (1954) argued that in an experiment
of rocket nozzle the separation occurred when the flow was
expanded to a static pressure approximately 0.4 times the
back pressure. A similar result was also obtained by many
other investigators (Green 1953, Page 1962). In accordance
with those experimental results, the onset for flow detach-
ment in nozzles based upon a critical value of the ratio of
the static pressure at the point of separation (p~) to the back
pressure (Pb) are often cited in the literature. However the
critical pressure ratio Ps/Pb is not a true constant, but rather
may be a function of the overall pressure ratio prevailing dur-
ing the experiment. The experimental results (Green 1953,
Scheller 1953), obtained in axially symmetric nozzles with
the divergence half angles of 15 degrees, are again assem-
bled in Fig. 17, from which it can be seen that the conditions
sufficient to cause the boundary layer separation are not well
represented by the criterion Ps/Pb = constant. Furthermore,
the apparent lack of agreement between the different groups
of data is such that only with some trepidation may a single
empirical curve be fitted to all the points.
Figure 18 shows that these data on SIBLS permit a sin-
gle empirical curve to be fitted with increased accuracy, if
they are represented in terms of the ratio (Pb -- P,)/Po. We
can deduce that for a given value of overall pressure ratio
(Po/Pb), the corresponding value of separation pressure ratio
(Pb/Ps) reduces to a function of Po/Ps. In order to compare
the criterion of SIBLS with one due to the concept of free
interaction in external flows aforementioned, the separation
to back pressure ratio is again plotted against the overall
pressure ratio of the nozzle (see Fig. 19). The data indicated
represent two-dimensional and conical nozzles with half an-
gles between 7 and 30 degrees, and gas specific heat ratio
between 1.2 and 1.4. The curve is presenting the separation
pressure ratios as measured on forward facing steps, com-
pression ramps, and incident shock waves, as shown earlier.
The trend of these data is identical to that of the nozzle sep-
aration data for a wide range of Reynolds number applied.
It does not seem to be a significant effect of Reynolds num-
ber on the separation pressure ratio. It should be noted that
within the accuracy of the data the pressure rise to separa-
tion is independent of the geometry causing the shock wave
boundary layer interactions, depending only on the upstream
Mach number. If these observations are interpreted as indi-
cating that a supersonic turbulent boundary layer separates
when subjected to a critical pressure jump, whose value de-
pends only on the Mach number and not on the geometry of
the interaction, it is then reasonable to expect that separa-
tion in such internal flows as supersonic nozzles or diffusers
should occur for a pressure jump with the same magnitude
as obtained in the external flows.
 283

0.1 0.7
+a Campbell & Farley
l.-
._...4
9 ~xAblberg et al. .~..............
9 Sunley & Ferriman ! i
i [] McKenny i i
0.6 g[i .............-~.............t.............t....... [] Musial & Ward i .............!..............
"~ o(~i i i 9 Goleseorthy & Herbert t i
~-.-.&-.-.i.............i.............i......Others Herbert& Herd , ...............i..............
o
~0.01
.... L--.! .............i ..... 0~ .i...........................
0.5 A i --'~B ~ ! i JMs/MI =0.67 ' i
~ 9 i .~i~i r~i ~/pb/Ps = 1.05+0.05M 1 i !
~ -J--:*R,.i.. ~.. & .i.............i..............

0.4 ............[i .............[i .............

...........[i.." ' " " 'i"q:'"'"
: ' ) ...........
: )~ ...........
" - ".............
~ t'...........
~~,"............)..............
"~"""-" -

0.001
0.004 0.01 0.05 0.1 0.2
Nozzle pressure ratio pb/P0
0,07 . . . . . . . . . . . 0.3 I f ) I t i I I ) I I I
3 5 7 9 11 13 15 17 19 21 23 25 27
o 0.06 ....... i.............".............i..............i-- , , //~.~,~&,-" -...... Nozzle pressure ratio p0/Pb

Fig. 19. Comparison of separation pressures in supersonic nozzle and theory

,~0.05 ...........~.-..-...~.............i..............i.. p0 ) / ........ due to free interaction

i i "

[ -'~ [ [ ~ O Scheller & Bierlein i Neglecting the additional pressure rise in the m i x i n g re-
= 0 03 .............
9 " " "~.................
". l. '. .. r~t.ri~i o.-i
' ( ..............[i............."i. . .a Foster
. . . &. Cowles
. . . l!.............
gion d o w n s t r e a m of the separation point, the ratio of the
=9 I [ ['.*~ei iOthersGreen , , i 9 separation to b a c k pressure is calculated as a function of
0,0 t ............ ........................................ r ............. i.............. i............ t
overall pressure ratio of the nozzle. For M s less than unity,
0.01 .-Av ' aged ..-".............."..." ~ ~ - . ~ ............~.............i?.......... the separation pressure ratio is obtained explicitly as
"v
0 ' --1 2 ,3,-1
0 10 20 30 40 50 60 70 80 90 100 110
Nozzle pressure ratio p0/Pb Ps _ P o • -1- ~ - - - - ~-
-- (9)
Figs. 17 and 18. Separation pressure data in supersonic overexpanded noz- Pb Pb ~P~) ~ -~ __ ( u2~~) '
zle flows
and the separation pressure ratio for M s larger than unity is
given by
G o o d agreement with the experimental data illustrated
above can also be obtained by using the assumption (Arens
1963a, 1963b) that the pressure rise must be sufficient to
stagnate a characteristic velocity u* in the boundary layer. (~) ,(10)
A s s u m i n g a constant stagnation temperature in the bound-
ary layer, we obtain the separation pressure ratio from the
f o l l o w i n g equations; where D is given by Eq. (11).

h~ + = h8 + u~ (5) 9 - 1 (11)
2
T~ hs u~ { 1 - (us~2~= (a-s 2 (6) The separation pressure ratios calculated from the above
\V1/ j \a,/ ' equations are presented against the M a c h n u m b e r (see Fig. 20)
and overall pressure ratio (see Fig. 21)9 For M a c h n u m b e r
~s %s %1 al
(7) lower than 1.13, the pressure j u m p required to separate the
as 7LI al as boundary l a y e r is larger than that p r o v i d e d b y normal shock
where h is enthalpy. F o r M8 less than unity, it is assumed compression. Therefore, this situation is not real. The overall
that stagnation occurs isentropically. F r o m isentropic rela- pressure ratio (Po/Pb) of the nozzle necessary for the bound-
tion between a state '1' and state 's', the pressure ratio was ary layer separation, which is caused by oblique shock w a v e
given by Eq. (3), where us~u1 means J that m e n t i o n e d ear- at a M a c h n u m b e r ( M s ) of 1.13, is calculated by about 1.7
lier. F o r M8 larger than unity, it is assumed that an isentropic for % / u l = 0.6, neglecting the pressure rise due to m i x i n g
stagnation is preceded b y a normal shock compression. If d o w n s t r e a m of the separation point 9 It is unlikely that the
the flow in the nozzle up to the separation point is a s s u m e d oblique shock separation will occur for overall pressure ra-
isentropic, then the separation Mach n u m b e r is given by tios lower than 1.7. For nozzles with very small exit to throat
"v area ratios, normal shock flow can be obtained for nozzle
P0=PO.Pb= ( 1 + ~@~M2) "/ 1 (8) pressure ratios up to 1.89, the limiting value corresponding
P8 Pb P8 to ideally e x p a n d e d convergent nozzle flow. F o r rather large
284

1.0 1.0
i~ i i ir ! i i ~!i!i
Us/U 1 = 0.56 i ~ = 1.4 . ~ d + i i ~,~*.[ii +•
.................. thick line M <1 ..................................j .......................................... la= i o i P l ! i i l [] o Reid & Hastings
.= 0.8 ~ ...... i thinline M > . I i 0.8 ........................~ ..............i...........i o i - i ! " ! H o o Goleseorthy&-Herbert
i i ~ ! iAi i i i uA[]Herbert&Herd
e-~ Us/U 1 = 0.60
0.6 / ~ ~ ...........7/.......................i.......................... thickline Ms<l ......... 0.6 ........................O...............!...........i.........i_...~x..~...~...~..~.............................~...............~...........~.....~....~...~..~..~..

i i i i i i ~i~.i ! ! i i i!i
o
0.4 ..... ........................................... ~........... L....L.L.L L..L.Lx ................ o~:-~........... b--i-...+-++-.!.

..j ...................................~ . of normal s ? c k

0.2 ..........................i Laminar B.L. ~ ~ Turbulent B.L. i-}-@!@i

i 1 !il,ii i i i ii!i
0 I i I ~ r I r I I i~l i i iii~ii i i i ii ~
2 3 4 5 05 106 107
M a c h n u m b e r M~ Reynolds number p u X / g
Fig. 20. Variation of separation with M a c h n u m b e r in overexpanded nozzles 1.0
(U.~/U~denotes the extent of velocity stagnation)
1.0 0.8 .......................... i""~I-'i'1Z1"i"""'i""i'"!'"'i"'i + ............................... I ................ b'"""t""'!"""i''!"i'"}"

0.8 ................................................................~................... Us/U 1 = 0.6 ~.................................. 0.6 i l i i i iiiil

.......................9 ..............................................................................................................................
N i i thick line M s < l i
i i thinline Ms>l i
9
0.6
o~ 0.4
"= i
..............................................................
i...................i"i , iiiii "
, "....... ,""';"T"~"Y"
'~ L a m i n a r B . L . ~ i] = Turbulent B.L. ! ! ~ !

[] • i ! i i ! ! !
0.4 0.2 ................................ A r~ R e i d & H a s t i n g s ..............L......@......L.@...L..~...J
o
+ 0 0 Goleseorthy & Herbert
...~
n A O Herbert & Herd . . . . . . .
i = 9 i
i i i i i i I l i i i i i ii
o.2 05 106 107
Reynolds numberpuH/~t
...............................
r r
!.................................
f I r
i.................................
~ t
i.....................................................................
[ r t I [ I I I i I I
Figs. 22 and 23. Effects of boundary layer on separation pressure in axi-
0 20 40 60 80 100 symmetric supersonicnozzles (-7 = 1.4, half angle of divergence =10 deg)
Overallpressure ratio p0/Pb
Fig. 21. Variationof separation with nozzle pressure ratio in overexpanded
nozzles (U~/U1denotes the extent of velocity stagnation) until now. Assuming equal shock intensity and, thus, equal
pressure disturbances imposed into the stream by the shock
wave, in laminar interaction the disturbances typically prop-
area ratios, however, oblique shock flow will be established agate upstream through the boundary layer over rather long
for overall pressure ratios exceeding 1.7. It should be noted distance, estimated by some 10 to 100 times the bound-
that, for flow characterized by a normal shock at the lim- dry layer thickness at the start of interaction, thus causing
iting Mach number, the nozzle pressure ratio is lower than a wedge-like increase in the effective geometry thickness.
the nozzle pressure ratio required for separated flow at the The abrupt change of the effective geometry thickness at
limiting Mach number. the separation point leads either to an oblique shock wave
The value of Ms = 1.13 is significantly lower than M~ or to a gradual compression wave, which together with the
obtained in external flows and the same value can be also rear shock wave form the typical lambda pattern. Thus the
obtained for a smaller u~/u~. This may be related to the intensity of the first wave is always smaller than that of the
boundary layer flow along the nozzle wall, as implied in the rear shock wave.
abrupt variation of P~/Pb of Figs. 22 and 23. The experi- On the other hand, the length of the turbulent interaction
mental data of Figs. 22 and 23, in which the abscissa means region is much smaller of the order of some 10 times the
the Reynolds number based on the distance from the noz- boundary layer thickness at the start of interaction. The rea-
zle throat to the interaction point and the height of nozzle son is in the greater ability of the turbulent boundary layer to
throat, respectively, show that the value of P~/Pb is higher withstand the adverse pressure gradients in the shock waves.
under laminar boundary layer than under turbulent bound- This is due both to a smaller thickness of the subsonic part
ary layer. This can be due to the difference in character of of the boundary layer through which the disturbances prop-
laminar and turbulent interaction which described in what agate upstream, and to a distinct turbulent mixing in the
follows. To the authors' knowledge, the laminar and turbu- interaction region. For the most typical turbulent interaction
lent interaction can differ as to the extent of the interaction patterns, a rapid increase in the boundary layer thickness
region, although there has not been any description on it appears due to the strong adverse pressure gradient in the
 285

0.8 , , , , , , i , i , , , , , , ,
tion on whether the interacting boundary layer is laminar or
turbulent can be answered from an optical observation of
0.7 6' ~ i .................. F....................... i........................ flow field.
On the other hand, in reality the value of Ps/Pb varies
0.6 ~ ......................
with the nozzle geometries, which can affect the pressure
recovery downstream of the shock wave. As described pre-
: ........Wh o& esui i : viously, we assumed the negligible pressure recovery down-
0.4 .............................................. stream. In order to investigate the effect of the pressure re-
c~ ol
zxst covery on the Ps/Pb, we have conducted the experiments
0.3 ............................ [] • Previous study ........... using a two-dimensional overexpanded nozzle. The exper-
,-I-
iments were made using wall pressure measurements and
o Present study
0.2 ................................... Schlieren observations. Figures 24a and 24b show the com-
parison of the experimental data with the result by the
0.l ................................................................................................ above theoretical model. The solid and broken lines indicate
the theoretical results mentioned previously. The separation
0 i i i i i , i , i i i i , , , ,
pressure rise in the present overexpanded nozzle well agrees
1.0 1.5 2.0 2.5 3.0
M1 with that obtained from the external flows. If we assume to
0.3 be Pb = P~, the agreement with the theoretical result im-
proves (see Fig. 24b), and the present experimental data on
---J:0.56 the SIBLS well agree with the concept of free interaction.
i "~ .... J = 0.60 These results describe that the free interaction of shock wave
;: l\ 9 pb=Pd
boundary layer in the external flows can be applied to the
0.2 ...................:""-'""~'-'"~.................. o Pb = Ps internal flows, and that the onset of SIBLS can be predicted
I \ by only the Mach number just upstream of the shock wave.

=z i, ),o
i i ~
0.1 ....................i:...........,.........'..i.........'...,..
: 9 ~!.......................i....................
Distance
0 t I I I T I r I i
0 2.0 4.0
P0/Pb
Fig. 24. Comparison of separation pressure by previous and present exper-
iment
shock wave. Therefore the gradual compression due to this
abrupt change of the effective boundary is rather strong, so
that the oblique compression waves very soon coalesce into
a strong shock wave, which at a certain distance from the
geometry becomes almost normal to the wall surface. The
extent of the whole pattern depends on the character and
history of the boundary layer.
Usually weak and almost normal shock waves appear be-
yond the main shock, thus resembling the rear branch of the
lambda shock wave in the laminar case, because of rather
weak rear shock in the turbulent case. Consequently in lam-
inar interaction the maximum height of the separated region
almost coincides with the position of the strong rear shock
wave, and shock-induced separation can be appear at the
weaker pressure disturbances than in turbulent interaction,
while in the turbulent case the maximum is shifted far to
the rear, sometimes even beyond the distance many decades
times the boundary layer thickness at the start of interaction.
The front shock wave in laminar interaction is rather weak
and the main rear shock wave is strong. In the turbulent in-
teraction, however, the front shock wave is strong, while the
rear normal shock waves are very weak. Therefore the ques-
i i 5 Conclusion
i i
A great number of experimental data indicating shock wave/
boundary layer interactions in internal or external supersonic
flows were reviewed to make clear the mechanism of the
interaction and to decide the onset of shock-induced sep-
b aration. The interesting conclusions were obtained for the
. . . . . . .
i
6.0 K0 10.0 considerably wide range of flow geometries that the onset
of separation is independent of the flow geometries and the
boundary layer Reynolds number. It is found that the pres-
sure rise necessary to separate the boundary layer is greater
in turbulent flow than in laminar flow, regardless of exter-
nal or internal flows, and that the separation pressure rise in
supersonic external flows could be applied to internal flows
such as overexpanded nozzles or diffusers. This is due to
the fact that the separation phenomenon caused by shock
wave/boundary layer interactions is processed through a su-
personic deceleration. The shock-induced separation in al-
most all of interacting flow fields is governed by the concept
of free interaction, and the onset of shock-induced separa-
tion is only a function of the Mach number just upstream of
shock wave.
References
Ahlberg JH, Hamilton S, Migdal D, Nilson EN (1961) Truncated perfect
nozzles in optimum rocket design. ARC J 31:5
Alber IE, Bacon JW, Masson BS, Collins DJ (1973) An experimental in-
vestigation of turbulent transonic viscous-inviscid interactions. AIAA
J 11:5
Arens M (1963a) The shock position in over-expanded nozzles. J Roy Aero
Soc 67
Arens M, Spiegler E (1963b) Shock-induced boundary layer separation in
overexpanded conical exhaust nozzles. AIAA J 1:3
Ashwood PF (1957) A review of the performance of exhaust systems for
gas turbine aero-engines. Proc Inst Mech Eng London 171
286
Bloomer HE, Antl RJ, Renas PE (1961) Experimental study of effects of
geometric variables on performance of conical rocket engine exhaust
nozzles. NASA TN-D-846
Bogdonoff SM, Kepler CE (1955) Separation of a supersonic turbulent
boundary layers. Journal of Aerospace Sci 22 pp 414-424
Campbell CE, Farley JM (1960) Performance of several conical convergent-
divergent rocket type exhaust nozzles. NASA TN D-467
Chapman DR, Kuehn DM, Larson HK (1957) Investigation of separated
flows in supersonic and subsonic stream with emphasis on the effect
of transition. NACA TN 3869
Cooke JC (1963) Separated supersonic flow. ARC J 24935
Delery J, Marvin JG (1986) Shock-wave boundary layer interactions.
AGARDograph 280
Foster HG, Cowles FB (1949) Experimental study of gas-flow separation in
over-expanded exhaust nozzles for rocket motors. Jet Propulsion Lab,
California Inst Tech, Progress Report 4-103
Fradenburch EA, Gorton GC, Beke A (1953) Trust characteristics of a series
of convergent-divergent exhaust nozzles at subsonic and supersonic
flight speeds. NASA RM E 53L23
Frazer RP, Eisenklam P, Wilkie D (1959) Investigation of supersonic flow
separation in nozzles, Journal of Mechanical Engineering Science 1
Gadd GE (1953) Interactions between wholly laminar or wholly turbulent
boundary-layers and shock waves strong enough to cause separation.
J Aero Sci 20:11
Gadd GE, Holder DW, Regan JD (1954) An experimental investigation of
the interaction between shock waves and boundary layers. Proc Roy
Soc A 226
Garrison T, Settles G, Narayanswami N, Knight D (1993) Structure of
crossing shock wave/turbulent boundary-layer. AIAA J 31:12 pp 2204-
2211
Garrison T, Settles G, Narayanswami N, Knight D, Horstman C (1996)
Flowfield surveys and computations of a crossing-shock wave/bounda-
ry-layer interaction. AIAA J 34:1 pp 50-56
Golesworthy GT, Herbert MV (1963) The preformance of a conical
convergent-divergent nozzle with area ratio 2.9 in external flow. NGTE
Memo M 371
Green JE (1970) Interactions between shock waves and turbulent boundary
layers. Prog in Aerospace Sciences 11. Pergamon Press
Green L (1953) Flow separation in rocket nozzles. ARS J pp 34
Guman WJ (1959) On the plateau and peak pressure of regions of pure
laminar and fully turbulent separation in two-dimensional supersonic
flow. I Aero Sci 26:1
Hankey WL, Holden MS (1975) Two-dimensional shock wave-boundary
layer interactions in high speed flows. AGARDograph 203
Herbert MV, Herd RJ (1966) Boundary-layer separation in supersonic pro-
pelling nozzles. ARC R&M 3421
Herbert MV, Martlew DL (1963) The design-point performance of model
internal-expansion propelling nozzles with area ratios up to 4. ARC
26495
Hingst WR, Tanji FT (1983) Experimental investigation of a two-dimensio-
nal shock-turbulent boundary layer interaction with bleed. NASA TM
83057
Inger GR, Mason WH (1976) Analytical theory of transonic normal shock-
turbulent boundary layer interaction. AIAA J 14:9
Inger GR (1980) Some features of a shock-turbulent boundary layer inter-
action theory in transonic flow field. AGARD Symposium on Compu-
tation of Viscous-Inviscid Interactions
Inger GR (1984) The modular application of a shock/boundary layer in-
teraction solution to supercritical viscous-inviscid flow field analysis.
Computational Methods in Viscous Flow, Vol. 3
Kim HD, Hong JW, Matsuo K (1993) Criterion of turbulent boundary-layer
separation induced by shock-wave. Engineering Sci Reports, Kyushu
Univ 15:1
Knight D, Badekas D, Horstman C, Settles G (1992) Quasi-conical flowfield
structure of the three-dimensional single fin interaction. AIAA J 30:12
Knight D, Garrison T, Settles G, Zheltovodov A, Maksimov A, Shevchenko
A, Vorontsov S (1995) Asymmetric crossing-shock-wave/turbulent-
boundary-layer interaction. AIAA J 33:12 pp2241-2249
Kuehn DM (1959) Experimental investigation of the pressure rise re-
quired for the incipient separation of turbulent boundary layers in two-
dimensional supersonic flow. NACA Memo 1-21-59A
Kuehn DM (1961) Turbulent boundary layer separation induced by flares
on cylinders at zero angle of attack. NACA TR R-117
Lees L, Reeves BL (1964) Supersonic separated and reattached lami-
nar flows, 1. general theory and application to adiabatic boundary-
layer/shock-wave interactions. AIAA J 2:11
Liu X, Squire LC (1988) An investigation of shock/boundary layer inter-
actions on curved surface at transonic speeds. Journal Fluid Mech 187
pp 467-486
Liao JH, Deng XY (1996) A physical interpretation of incipient separation
criterion. AIAA Paper 96-0321
Love ES (1955) On the effect of Reynolds number upon the peak pressure
rise coefficient associated with the separation of a turbulent boundary
layer in supersonic flow. J Aero Sci pp 345
Mager A (1955) Prediction of shock-induced turbulent boundary-layer sep-
aration. J Aero Sci 22:3 pp201
Mager A (1956) On the model of the free, shock-separated, turbulent bound-
ary layer. J Aero Sci 23:2 pp 181
McKenney TD (1954) Flow separation in overexpanded supersonic exhaust
nozzles. ARS J pp 320
Musial NT, Ward JJ (1959) Overexpanded performance of conical nozzles
with area ratios 6 and 9 with and without supersonic external flow.
NASA TM X-83
Page RH (1962) Flow separation in nozzles. J Aero Sci, pp 110
Pearcey HH (1955) Some effects of shock-induced separation of turbulent
boundary layers in transonic flow past aerofoils. ARC J 3108
Pearcey HH (1961) Shock-induced separation and its"prevention by design
and boundary-layer control, In Boundary-Layer and Flow Control. In:
Lachanan GV (ed), Vol. 2. Pergamon Press
Raghunathan S, Mitchell (1995) Computed effects of heat transfer on the
transonic flow over an aerofoil. A1AA J 33:11
Reid J, Hastings RC (1959) The effect of a central jet on the base pressure
of a cylindrical after-body in a supersonic stream. ARC R&M 3224
Scheller K, Bierlein JA (1953) Some experiments on flow separation in
rocket nozzles. ARS J 23 pp 28
Schuh H (1955) On determining turbulent boundary-layer separation in
incompressible and compressible flow. J Aero Sci 22:5 pp 343
Sirieix M, Delery J, Stanewsky E (1981) High-Reynolds number boundary-
layer shock-wave inteructiun in transonic flow. Lectures Notes in
Physics 148, Springer-Verlag, pp 149-214
Stanewsky E (1981) Wechselwirkung zwischen Aussenstr6nmng und Gren-
zschicht an trans-sonischen Profilen. Doctor-Engineer Dissertation
D.83 Berlin
Stanewsky E, Little BH (1960) Studies of separation and reattachment in
transonic flow. Journal of Aircraft 8:12 pp952
Steffett JR, Emery JC (1960) Extension of boundary-layer separation crite-
ria to a Mach number of 6.5 by utilizing flat plates with forward-facing
steps. NASA TN-D-618
Sterrett JR, Emery JC (1962) Experimental separation studies for two-
dimensional wedges and curved surfaces at Mach numbers of 4.8 to
6.2. NASA TN-D-1014
Summerfield M, Forster CR, Swan WC (1954) Flow separation in over-
expanded supersonic exhaust nozzles. Jet Propulsion 24 pp 319
Sunley HLG, Ferriman VN (1962) Jet separation in conical nozzles. J Royal
Aeronautical Society 68 pp 808-817
Szaniawski A (1965) Transonic approximations to the flow through a noz-
zle. Archives of Mechanics 17:1