Application of the principle of minimum

entropy production to shock wave

reflections. II. Pseudosteady flows
Cite as: Journal of Applied Physics 80, 2038 (1996); https://doi.org/10.1063/1.363097
Submitted: 15 January 1996 . Accepted: 04 April 1996 . Published Online: 04 June 1998

H. Li, and G. Ben-Dor

ARTICLES YOU MAY BE INTERESTED IN

Application of the principle of minimum entropy production to shock wave reflections. I.

Steady flows
Journal of Applied Physics 80, 2027 (1996); https://doi.org/10.1063/1.363096

Hysteresis effect in stationary reflection of shock waves

Physics of Fluids 7, 685 (1995); https://doi.org/10.1063/1.868591

Experimental investigation of shock oscillations on V-shaped blunt leading edges

Physics of Fluids 31, 026110 (2019); https://doi.org/10.1063/1.5084184

Journal of Applied Physics 80, 2038 (1996); https://doi.org/10.1063/1.363097 80, 2038

© 1996 American Institute of Physics.

Application of the principle of minimum entropy production to shock wave
reflections. II. Pseudosteady flows
H. Li and G. Ben-Dora)
Pearlstone Center for Aeronautical Engineering Studies, Department of Mechanical Engineering,
Ben-Gurion University of the Negev, Beer Sheva, Israel
~Received 15 January 1996; accepted for publication 4 April 1996!
The stability of both regular reflection and Mach reflection wave configurations in pseudosteady
flows has been investigated by applying the principle of minimum entropy production, in a way
similar to that done by us in Part I of this study @Li and Ben-Dor, J. Appl. Phys. 80, 2027 ~1996!#
where the stability of steady flow reflection configurations was considered. As a result, it was found
that Mach reflection wave configurations are stable inside the dual solution domain, although, based
on the state of the art when this study started, they were considered not to be stable. Consequently,
a new general transition criterion from regular to Mach reflection was proposed. With the aid of this
criterion the domains of regular, direct-Mach and stationary-Mach reflections, in the domain of the
incident shock-wave Mach number and the reflecting wedge angle, were reestablished. All the
analytical findings were supported by experimental results from different sources. © 1996
American Institute of Physics. @S0021-8979~96!02715-6#

I. INTRODUCTION reflected shock wave r, and the Mach stem m, and one slip-
stream s. They all meet at a single point known as the triple
As indicated by Ben-Dor,1 two main shock-wave- point T. The Mach stem m is not necessarily a straight shock
reflection configurations are possible in pseudosteady flows, wave. However, it always is perpendicular to the reflecting
namely, regular reflection ~RR! and Mach reflection ~MR! surface at the reflection point G. The flow states are ~0!
depending on the incident shock wave Mach number M i , ahead of i and m, ~1! behind i, ~2! behind r, and ~3! behind
and the reflecting wedge angle u w . Courant and Friedrichs2 m. Unlike the case of a regular reflection where the net de-
noted that the Mach reflection could be further divided into flection of the streamline is zero, in the case of a Mach
three types: a direct-Mach reflection ~D i MR! in which the reflection the net deflection of the streamline is not necessar-
triple point moves away from the reflecting surface, a ily zero. Since the streamlines on both sides of the slipstream
stationary-Mach reflection ~S t MR! in which the triple point must be parallel to each other, one of the two boundary con-
moves parallel to the reflecting surface, and an inverse-Mach ditions of a Mach reflection is
reflection ~I n MR! in which the triple point moves towards
the reflecting surface. Schematic illustrations of the wave u 12 u 25 u 3 . ~2a!
configurations of a regular reflection, a direct-Mach reflec-
tion, a stationary-Mach reflection, and an inverse-Mach re- The second boundary condition is that the pressures on both
flection, together with the definition of some flow param- sides of the slipstream separating regions ~2! and ~3! are
eters, are shown in Figs. 1~a!–1~d!, respectively. The I n MR equal, i.e.,
is a temporary ~unstable! wave configuration since it termi-
nates when its triple point collides with the reflecting sur- P 25 P 3 . ~2b!
face. The wave configuration which results after the termina-
Graphical solutions in the pressure-deflection plane @i.e.,
tion of an I n MR was investigated by Ben-Dor and Elperin.3
the ( P, u ) plane# have been traditionally used to illustrate
The regular reflection ~RR! wave configuration consists
and better understand the shock wave reflection phenom-
of two shock waves, namely, the incident shock wave i and
enon, in general, and possible RR↔MR transition criteria, in
the reflected shock wave r. They meet at the reflection point
particular. Examples of seven different I2R polar combina-
G, which is located on the reflecting surface. The flow states
tions for different values of f1 are shown in Figs. 2~a!–2~g!.
are ~0! ahead of i, ~1! behind it, and ~2! behind r. The angle
The locus of all the pressures achievable from the free stream
of incidence f1 of a regular reflection is sufficiently small so
of state ~0! via an oblique shock wave deflecting the flow
that the streamline deflection u1 , caused by the incident
through an angle u is given by the I polar. Thus, state ~1! in
shock wave i, can be canceled by the opposite streamline
Figs. 1~a!–1~d! maps into point ~1! in Figs. 2~a!–2~g!. The
deflection u2 , caused by the reflected shock wave r. There-
locus of all the pressures achievable from the free stream of
fore, the boundary condition of a regular reflection is
state ~1! via an oblique shock wave deflecting it by an angle
u 1 2 u 2 50. ~1! u is given by the R polar.
The boundary conditions of a regular reflection, given by
The Mach reflection ~MR! wave configuration consists Eq. ~1! implies that the solution of a regular reflection in the
of three shock waves, namely, the incident shock wave i, the ( P, u ) plane is at the point where the R polar intersects the P
axis. Two such points are obtained by the I2R polars com-
a!
Electronic mail: [email protected] bination shown in Fig. 2~a!. They are known as the weak and

2038 J. Appl. Phys. 80 (4), 15 August 1996 0021-8979/96/80(4)/2038/11/$10.00 © 1996 American Institute of Physics
FIG. 1. Schematic illustration of the wave configuration of the various
shock wave reflections and definition of various flow parameters in pseu-
dosteady flows. The solid lines represent shock wave fronts. ~a! Regular
reflection ~RR!, ~b! a direct-Mach reflection D i MR, ~c! a stationary-Mach
reflection ~S t MR!, and ~d! an inverse-Mach reflection ~I n MR!.

the strong solutions. It is an experimental fact that the strong

RR, which results in the higher pressure @marked by an open
circle in Fig. 2~a!#, is unstable.
The boundary conditions of a Mach reflection, given by
Eq. ~2! imply that the solution of a Mach reflection is at the
point where the I and R polars intersect. In addition, as
shown by Ben-Dor1 D i MR, S t MR, and I n MR can be distin-
guished by the net deflection of the streamline. In a D i MR it
is positive, in an I n MR it is negative and in a S t MR it is zero.
Consequently, the intersection of the I and R polars in Fig.
2~a!, at which the net deflection is negative, represents the
solution of an I n MR; the intersection of the I and R polars in
FIG. 2. Various I-R polar combinations illustrating ~a! regular reflections
Fig. 2~b!, at which the net deflection is zero ~it takes place on ~strong and weak! and an inverse-Mach reflection for the same initial con-
the P axis!, represents a solution of a S t MR; and the inter- ditions, ~b! the von Neumann criterion at which both a regular and a
section of the I and R polars in Fig. 2~e!, at which the net stationary-Mach reflections are theoretically possible for the same initial
conditions, ~c! a regular and a direct-Mach reflection for the same initial
deflection is positive represents a solution of a D i MR. conditions, ~d! the detachment criterion at which both a regular and a direct-
Since for the combination shown in Fig. 2~b! the R polar Mach reflections are theoretically possible for the same initial conditions, ~e!
intersects the I polar on the P axis, both RR and S t MR are a direct-Mach reflection, ~f! the situation for which the von Neumann crite-
theoretically possible at this intersection point. Conse- rion and the detachment criterion meet at the same point, ~g! regular reflec-
tions ~strong and weak! and a direct-Mach reflection for the same initial
quently, this I2R polars combination represents a possible conditions for weak shock waves.
condition for the RR↔MR transition. This possible transi-
tion, which was first suggested by von Neumann4 in the early
40’s was reinitiated by Henderson and Lozzi5 who called it
u 1 2 u 2 5 u 3 50. ~3!
the ‘‘mechanical equilibrium’’ criterion. It is known nowa-
days as the von Neumann criterion and can be formulated by The reflecting wedge angle appropriate to the I2R polars
combining Eqs. ~1! and ~2! to read combination shown in Fig. 2~b! will be denoted as u Nw . Note

J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor 2039
that for u w . u Nw only an inverse Mach reflection is theoreti-
cally possible. However, since the wave configuration of an
I n MR is unstable, only a RR is valid for u w . u Nw .
A further decrease in the reflecting wedge angle eventu-
ally results in the I2R polars combination shown in Fig.
2~d! which, in fact, corresponds to the smallest value of u w
for which the R polar and the P axis have a common point
and hence a regular reflection is theoretically possible. Note
that a further decrease of u w below the value appropriate to
that of Fig. 2~d! results in a situation similar to that shown in
Fig. 2~e!, in which the R polar does not intersect the P axis
and hence a regular reflection is theoretically impossible.
Consequently, the I2R polars combination of Fig. 2~d! rep-
resents another possible condition for the RR↔MR transi-
tion. This possible transition, which was also suggested by
von Neumann4 in the early 40’s, is known in the literature as
the detachment criterion because it corresponds to the case in
which the streamline deflection through the reflected shock FIG. 3. Domains of various types of reflections in the (M i , u w ) plane: the
wave is maximal. Its mathematical formulation is reflection in domain A is D i MR; the reflection in domain B is either D i MR
or RR for weak shocks (M i ,M Ki ); the reflection in domain C is either
u 1 2 u D2 50, ~4! D i MR or RR for strong shocks (M i .M Ki ); the reflection in domain D is
either I n MR or RR.
where u D 2 is the detachment deflection angle. The reflecting
wedge angle appropriate to the I2R polars combination
shown in Fig. 2~d! will be denoted as u D w . Note that for plane into three domains: a domain D in which only RR is
u w , u Dw , a regular reflection is theoretically impossible, theoretically possible ( u w . u Nw ), a domain A in which only
hence u D w is the smallest reflecting wedge angle for which a MR ~i.e., D i MR! is theoretically possible ( u w , u D w ), and a
regular reflection is theoretically possible for a given inci- domain in which both RR and MR ~i.e., D i MR! are theoreti-
dent shock wave Mach number M i . cally possible ( u Nw > u w > u D
w ). Note that this latter domain
Based on the foregoing discussion, for all the reflecting consists of two subdomains B and C, which are separated at
wedge angles in the range u Nw > u w > u D w both regular and point K where the u w 5 u Nw and u w 5 u D w lines are tangent to
Mach reflection wave configurations are theoretically pos- each other. The shock wave Mach number appropriate to
sible. A typical I2R polars combination appropriate to this point K is M Ki 51.46 and its corresponding reflecting wedge
dual-solution domain is shown in Fig. 2~c!. The point where angle u Kw 548.6°. Traditionally, point K has been used to
the R polar intersects the P axis indicates a possible RR distinguish between weak and strong incident shock waves.
solution ~only the weak RR is shown!, while the point where The I2R polars combination appropriate to the conditions of
it intersects the I polar indicates a possible MR solution. ~as point K is shown in Fig. 2~f!.
mentioned earlier the MR at this point is a D i MR!. The I2R polars combinations shown in Figs. 2~a!–2~e!
It should be noted here that since both RR and MR wave are all appropriate to strong incident shock waves ~i.e.,
configurations are theoretically possible in the dual-solution M i .M Ki !. When the incident shock wave is weak, ~i.e.,
domain which is bounded by u Nw > u w > u D w , the RR↔MR M i ,M Ki ! the I2R polars combination shown in Fig. 2~g! is
transition could take place at any value u w inside that range. obtained. It represents two possible RR and one MR ~i.e.,
Consequently, the von Neumann criterion, u w 5 u Nw is, in fact, D i MR! solutions.
the largest possible value of u w for transition, while the de- It should be noted here that the I2R polars combination
tachment criterion, u w 5 u Nw is, in fact, the smallest possible shown in Fig. 2~g! is not the only one possible combination.
value of u w for transition. Similar to the case of strong incident shock waves @i.e., Figs.
It should also be noted here that while transitions at in- 2~a!–2~e!# here one may have various combinations. How-
cident shock wave angles inside the range u Nw . u w > u D w in- ever, since they are not relevant to this article they are nei-
volve a sudden pressure jump behind the reflected shock ther shown nor discussed.
wave ~i.e., a positive jump for the MR→RR transitions and a A long-time dispute exists in the literature regarding the
negative jump for the RR→MR transitions!, a transition at exact RR↔MR transition criterion in pseudosteady flows.
u w 5 u Nw is continuous as far as the pressure behind the re- While experimental evidence over single wedges clearly in-
flected shock wave is concerned. For this reason, Henderson dicates that the detachment criterion, i.e., u w 5 u D
w , is the cor-
and Lozzi5 considered the transition at this point to be in rect one, Henderson and Lozzi5,6 claimed that the experimen-
mechanical equilibrium. tal results were misinterpreted and that, in fact, the von
The dual-solution domain for which both regular and Neumann criterion, i.e., u w 5 u Nw , should be regarded as the
Mach reflection wave configurations are possible in the correct one.
(M i , u w ) plane is shown in Fig. 3 for g51.4, where g is the Since a similar dispute in steady flow reflections was
specific heat capacities ratio. The detachment ( u w 5 u Dw ) and resolved recently by applying the principle of minimum en-
the von Neumann ( u w 5 u Nw ) criteria divide the (M i , u w ) tropy production ~for details see Li and Ben-Dor7!, it was

2040 J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor
decided to try and resolve the dispute in pseudosteady reflec- and
tions by applying this principle in a similar manner. The M i 5M 0 sin f 1 . ~5b!
results of this effort are presented in the following sections.
Inserting these relations into the appropriate equations given
II. PRESENT STUDY in Li and Ben-Dor7 results in the following set of governing
equations by which the line u w 5 u w* , whose significance is
Since the principle of minimum entropy production is discussed subsequently, can be calculated.
applicable to steady cases only, its use for unsteady cases The stability condition reads
should first be justified. If the unsteady reflection is limited
4
to constant-velocity-incident shock waves, i.e., 2 g ~ g 21 !
M i 5constant, and straight reflecting wedges, it can be made G5
2
~ t2M 21 ! (
j51
A j t j >0, ~6!
steady by means of an appropriate Galilean transformation
~see Ben-Dor1!. In order to distinguish between true steady where
shock wave reflections and those which become steady only A 0 5 g ~ g 21 ! M 21 ,
by means of a Galilean transformation, the latter ones, which
are the subject of this study, are termed pseudosteady shock 1
A 1 5 ~ 5 g 21 ! 2 g ~ g 21 ! M 21 ,
wave reflections. However, as mentioned earlier, they are in 2
fact identical in all their features to steady shock wave re- 3 2
flections. Consequently, the principle of minimum entropy A 2 523 ~ g 21 ! M 22
1 2 ~ g 14 g 21 ! ,
4
production could be applied to investigate them.
1 2
III. STABILITY OF REGULAR REFLECTION A 3 5 ~ g 21 ! M 24
1 1 ~ g 14 g 29 ! M 22
1 ,
4
The following equations are, in fact, simply obtained g 21 24
from those developed by Li and Ben-Dor7 for steady shock A 4 52 M1 ,
wave reflections by applying the above-mentioned Galilean 2
transformation, i.e., 1
t5 .
p sin f 2
2

u w5 2 f 1 ~5a!
2 The oblique shock wave relations are1

sin 2u w ~ M 2i 21 ! 2 cot f 2 ~ M 21 sin2 f 2 21 !

5 , ~7!
M 2i ~ g 2cos 2u w ! 12 cos2 u w M 21 ~ g 1cos 2f 2 ! 12

~ 11 ~ g 21 ! M 2i 1 $ @~ g 11 ! /2# 2 2 g cos2 u w # M 4i sec2 u w !

M 21 5 . ~8!
$ g M 2i 2 @~ g 21 ! /2# %$ @~ g 21 ! /2# M 2i 11 %

Figure 4 is a reproduction of Fig. 3. In addition to the M Ki <M i <M Q i in which RR is also stable in the range u w
von Neumann transition line, u w 5 u Nw , and the detachment > uD w 5 u *
w However, for strong shock waves in the range
.
transition line, u w 5 u D
w , there is an additional line, the u w M i >M Qi , there is a very narrow domain inside the dual-
5 u w* line. In the range 1<M i <M Q i this line coincides with solution domain in which RR wave configurations are un-
the detachment transition line. For M i .M Q i this line sepa- stable. This domain is defined by u D
w , uw , uw *.
rates from the detachment transition line while extending The fact that RR wave configurations were observed in
slightly above it. M Qi is the incident shock wave Mach num- this narrow domain should not be used to discard the above
ber appropriate to point Q which marks the location where conclusions. This is due to the fact that the principle of mini-
these two lines separate. For g51.4, M Q i 51.70. The maxi- mum entropy production was developed under ‘‘free bound-
mum separation between these two lines, which is obtained ary’’ conditions, as well as other assumptions, e.g., inviscid
when M i →`, is about 0.7°. nonconducting perfect gas, actual experimental results do not
Based on the principle of minimum entropy production, necessarily fulfill these conditions. For this reason, RR wave
RR wave configurations are stable for configurations were observed even in the domain u w , u D w
~see Smith8!, where RR is theoretically impossible.
u w . u w* .
IV. STABILITY OF MACH REFLECTION
Consequently, it could be concluded that, for weak shock
waves, i.e., M i <M Ki , RR wave configurations are stable for As mentioned in the introduction, the MR can in
u w > u Dw 5 u w* . For strong shock waves, there is a small range general be divided into three types: inverse-~I n MR!,

J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor 2041
 FIG. 5. Schematic illustration of a wave configuration of a S t MR and the
FIG. 4. Reproduction of Fig. 3 with the addition of the line u w 5 u w* which control volume.
separates the ( P 0 / P 1 , u w ) domain into a domain in which RR is stable u w
. u w* and a domain in which RR is unstable u w , u w* .

RR in this case is the weak RR @here the term weak has

stationary-~S t MR!, and direct-~D i MR! Mach reflections. > K
nothing to do with the condition M i, M i . It arises from the
Consequently, in the following each one of these wave con- fact that the reflected shock wave of the RR is described by
figurations is treated separately. the weak solution. For more details see Li and Ben-Dor7#.
The wave configuration inside the control volume which
is added to Fig. 5 ~dashed-dotted line! is a clear regular re-
A. Inverse-Mach reflection—I n MR flection. Based on the ( P, u ) polar shown in Fig. 2~b!, the
An I n MR is a MR in which the triple point moves to- flow Mach number in state ~2! is supersonic, i.e., M 2 .1;
wards the reflecting surface, as shown schematically in Fig. consequently, the reflected shock wave is isolated from
1~d!. Thus, as soon as the triple point reaches the reflecting downstream disturbances.
surface, the I n MR terminates. Consequently, the I n MR is, by Based on the arguments presented in the previous sec-
definition, an unstable wave configuration. tion, regarding the stability of regular reflection wave con-
Note that I n MR wave configurations correspond to the figurations, it could be concluded that strong S t MR are stable
domain u w . u Nw in Fig. 3. Therefore, it is concluded that MR since the line u w 5 u Nw in the range M i .M Ki lies in the do-
is unstable in this domain. main u w > u w* for which RR wave configurations are stable
The discussion considering S t MR and D i MR wave con- ~see Fig. 4!.
figurations will be divided, for reasons of clarity, into two
parts, namely, strong incident shock waves, i.e., M i .M Ki ,
and weak incident shock waves, i.e., M i ,M Ki . C. Strong direct-Mach reflection—strong D i MR
For simplicity we therefore define S t MR and D i MR in
the range M i .M Ki as strong S t MR and strong D i MR, re- Figure 6 represents the ( P, u )-polar solutions of a typical
spectively. Similarly S t MR and D i MR in the range M i ,M Ki strong S t MR ~such as the one discussed earlier! and a strong
are defined as weak S t MR and weak D i MR, respectively. D i MR for the same incident shock wave Mach number M i .
As shown in Eqs. ~5a! and ~5b!,
M 0 5M i /cos u w . ~9!
B. Strong stationary-Mach reflection—strong S t MR
Consequently, since strong D i MR are obtained with smaller
The domain of strong-S t MR wave configurations, in reflecting wedge angles ~for the same M i ! it is clear that
fact, degenerates to the line u w 5 u Nw in the range M i .M Ki .
M 0 ~ strong D i MR! ,M 0 ~ strong S t MR! .
A schematic illustration of the wave configuration of a
strong S t MR is shown in Fig. 5. Since for a S t MR Therefore, the I polar corresponding to the solution of a
u12u25u350, it is obvious that the incident (i) and reflected strong D i MR (I 2 ) is smaller than that corresponding to the
(r) shock waves can be regarded as a regular reflection ~RR! solution of a strong S t MR (I 1 ), as shown in Fig. 6 in which
for which the triple point trajectory serves as an imaginary the points corresponding to the strong S t MR and strong
reflecting surface. This is also evident in Fig. 2~b! where the D i MR solutions are indicated.
R polar intersects, the I polar at the same point where it As indicated by these two possible solutions, the pres-
intersects the P axis. The intersection of the I and R polars sure behind the reflected shock wave of a strong D i MR is
implies a S t MR solution while the intersection of the R polar smaller than that behind the reflected shock wave of a strong
with the P axis implies a RR at the same point. Note that the S t MR. Therefore, the angle of incidence of the reflected

2042 J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor
 FIG. 8. The ( P, u ) solution of a weak D i MR.

FIG. 6. The ( P, u ) solutions of a strong S t MR and a strong D i MR for the

same strong incident shock wave Mach number, M i (M i .M Ki ) in the dual- D. Weak stationary-Mach reflection—weak S t MR
solution domain ( u Dw < u w < u Nw ).
The ( P, u )-polar solution of a weak S t MR is shown
schematically in Fig. 7. In order for a weak S t MR to exist, its
upper part which, as mentioned earlier, is a regular reflection
shock wave, f2 @see Fig. 1~b!# is smaller for a strong D i MR ~i.e., the net flow deflection through the incident and the
than that for a strong S t MR ~for the same incident shock reflected shock waves is zero! is unstable. This is due to the
wave Mach number, M i !, i.e., fact that this RR wave configuration is a strong RR which, as
strong D i MR strong S t MR proven by Li and Ben-Dor,7 is unstable. Note that had a
f2 ,f2 . ~10! weak S t MR existed, the flow Mach number behind its re-
flected shock wave would have been subsonic. As a result,
Thus, based on the principle of minimum entropy produc-
the reflected shock wave would not be isolated from down-
tion, if a strong S t MR is stable, i.e.,
stream disturbances which, upon reaching the reflected shock
strong S t MR
f2 ,f*
2 , ~11! wave, would continuously change it, thereby discarding the
possibility of its stable existence. It could be concluded from
a strong D i MR is also stable since by combining conditions these arguments that a weak S t MR is unstable.
~10! and ~11!, one obtains
strong D i MR
f2 ,f*
2 . ~12!
E. Weak direct-Mach reflection—weak D i MR
Note that f *
2 is obtained from solving the equality in Eq. ~6!, The ( P, u )-polar solution of a weak D i MR is shown
i.e., schematically in Fig. 8. As shown by Li and Ben-Dor7
G ~ g ,M i , f *
2 ! 50. ~13!
f strong
2
RR
. f 2* . ~14!

For this reason the strong RR is unstable. Figure 8 implies

that

P 2 ~ weak D i MR! . P 2 ~ strong RR! , ~15!

where P 2 is the pressure behind the reflected shock wave of

the weak D i MR or the strong RR. The inequality given by
Eq. ~15! can be satisfied only if
weak D i MR
f2 .fstrong
2
RR
. ~16!

Combining the inequalities given by Eqs. ~14! and ~16! re-

sults in
weak D i MR
f2 . f 2* . ~17!

Consequently, a weak D i MR is an unstable wave configura-

FIG. 7. The ( P, u ) solution of a weak S t MR (M i ,M Ki ). tion.

J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor 2043
TABLE I. Stability of RR wave configurations. TABLE II. Stability of MR wave configurations.

Mi Mi

uw 1,M i ,M Qi M i .M Qi weak strong

uw 1,M i ,M Ki M i .M Ki
u w . u w* Stable Stable
u w* . u w . u Dw Irrelevant Unstable u w . u Nw Unstable Unstable
u w , u Dw Theoretically impossible Theoretically impossible u Nw > u w . u Dw Unstable Stable
u w , u Dw Stable Stable

F. Summary
and MR wave configurations more than a hundred years ago
Prior to proceeding to the discussion section and the ex-
verified once and again that RR wave configurations are ob-
perimental verification of the foregoing analytically deduced
tainable and hence stable in the domain u w . u D w , and that
conclusions regarding the stability of regular and Mach re-
MR wave configurations are obtainable and stable in the do-
flections, the conclusions are summarized in the following:
main u w , u D w . Consequently, in the following only the dual-
1. Stability of regular reflection—RR solution domain, i.e., u D w , u w < u w ~see Fig. 3! is discussed
N

~1! RR wave configurations are theoretically impossible in and only the conclusions regarding the stability of MR wave
the domain u w , u D w.
configurations inside it need experimental verification.
~2! The domain u w . u D It is claimed in conclusion ~e! of the previous section
w in which RR wave configurations
are theoretically possible can be divided into two parts: that strong D i MR wave configurations are stable inside the
~a! for M i ,M Q domain u w , u Nw . This conclusion seems to be contradicted by
i ~see Fig. 4!, RR wave configurations are
stable; ~b! for M i .M Q the fact that stable-D i MR wave configurations have never
i , there is a very narrow domain,
u Dw , u w , u w* , inside which RR wave configurations are been reported to be observed inside the dual-solution domain
unstable. Note that the maximal difference between u w* u Dw , u w , u Nw when planar shock waves were reflected over
and u D single wedges. On the contrary, all the evidence from these
w is about 0.7°, therefore practically u w > u w
D
* and
hence, it could be said that RR wave configurations are experiments report RR wave configurations there. Conse-
stable in the domain u w . u D quently, this ‘‘contradiction’’ should first be dealt with and
w.
resolved.
2. Stability of Mach reflection—MR
Dewey and van Netten9 showed experimentally that
when planar shock waves were reflected over small angle
The entire u w domain can be divided into three parts wedges, there was no delay in the initiation of the Mach
according to the type of the MR: reflection. However, based on their experiments, at large
—I n MR wave configurations are theoretically possible wedge angles close to those at which the RR→MR transition
in the domain p /2. u w . u Nw . takes place, i.e., close to u Dw , D i MR wave configurations ~ei-
—S t MR wave configurations are theoretically possible ther single, transitional, or double! were delayed, and prior to
on the line u w 5 u Nw . their formation the planar shock wave reflected over the re-
—D i MR wave configurations are theoretically possible
in the domain u w , u Nw .
Since I n MR are, by their definition, temporary reflec-
tions, they are unstable, and hence
~a! I n MR wave configurations are unstable in the domain
u w . u Nw .
The S t MR and the D i MR can be divided into two sub-
types depending on whether their incident shock wave is
> K
weak or strong, i.e., M i, Mi :
~b! Weak-S t MR wave configurations are unstable.
~c! Strong-S t MR wave configurations are stable.
~d! Weak-D i MR wave configurations are unstable in the
domain u D w , u w , u w and stable in the domain u w , u w .
N D

~e! Strong-D i MR wave configurations are stable in the en-

tire domain u w , u Nw .
For the reader’s convenience, the above summary is shown
in Tables I and II for RR and MR, respectively.

V. CONCLUSIONS OF THE STABILITY ANALYSIS—

DISCUSSION AND EXPERIMENTAL VERIFICATION
FIG. 9. Four digitized images from a multiple schlieren sequence of a
Shock tube experiments over single wedges which have double Mach reflection for M i 52.67, u w 548° and P 0 514 torr, as reported
been conducted all over the world since the discovery of RR by Dewey and van Netten ~see Ref. 9!.

2044 J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor
FIG. 10. Holographic interferometry photograph of a S t MR for M i 51.9 and first and second wedge angles u 1w 525° and u 2w 553°, respectively. ~Courtesy of
Professor K. Takayama, Institute of Fluid Science, Tohoku University, Sendai, Japan!.

flecting surface as an RR. This is shown in Fig. 9 which is a to bypass the initially established RR, i.e., to start with an
reproduction from their results. The actual wedge angle in MR, and thereby eliminate altogether the stable RR. This can
this experiment u w is slightly smaller than the RR→MR be done and, in fact, has been done by reflecting planar
transition wedge angle, u Dw . Consequently, an RR wave con- shock waves over double wedges ~see, for example, Sysh-
figuration is theoretically impossible. However, in spite of chikova and Krassovskaya,10 and Ben-Dor et al.11!.
this fact, a clear RR wave configuration is seen to be formed Although Ben-Dor1 categorized this reflection process as
at the early stages of the interaction. Only after a distance of unsteady, it was proven experimentally by Syschchikova and
about 10 000 molecular mean free path ~which corresponds Krassovskaya10 and Ben-Dor et al.12 that the MR wave con-
to about 5 cm in the example! the D i MR appears. figurations, which were established over the second reflect-
We hypothesize that the evolution of the reflection pro-
ing surface of a double wedge, reached pseudosteady condi-
cess inside the dual-solution domain ~where u w is larger than
tions when they moved away a distance equal to about the
that in Fig. 9! is similar. Again, an RR is formed initially.
length of the first reflecting surface, from the leading edge of
However, unlike in the above described example where an
RR wave configuration was initially formed in a domain the second reflecting surface. Consequently, beyond that dis-
where it cannot theoretically exist, here the initially estab- tance, the wave configurations are practically pseudosteady.
lished RR does not have to terminate and give rise to other In Syshchikova and Krassovskaya10 and Takayama and
wave configurations, since in the dual-solution domain RR Ben-Dor13 there are experimental photographs showing both
wave configurations are stable. For this reason, only RR strong-S t MR and strong-D i MR wave configurations which
wave configurations have been observed in the dual-solution were obtained by reflecting planar shock waves over double
domain when planar shock waves were reflected over single wedges. Holographic interferograms of strong S t MR is
wedges. shown in Fig. 10 for M i 51.9, u 1w 525°, and u 2w 553°. Figure
Consequently, in order to establish D i MR wave configu- 10~a! shows the reflection over the second surface at it’s
rations inside the dual-solution domain, one must find a way early stages and Fig. 10~b! shows the same reflection at a

J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor 2045
 FIG. 10. ~Continued.!

later time. The evaluation of these two photographs indicated bility of strong-D i MR and strong-S t MR wave configurations
that the length of the Mach stem remained practically un- inside the dual-solution domain could be considered as veri-
changed. Hence, it was concluded that the recorded reflection fied.
was a S t MR. The various relevant angles as calculated using It is the authors’ belief that the fact that experimental
the three-shock theory for M i 51.9 and g51.4 are results showing weak-S t MR wave configurations or
u Dw 550.47° and u Nw 553.53°. Consequently, the MR reflec- weak-D i MR wave configurations inside the dual-solution
tion shown in Fig. 10 occurred at u w . u D w ~note u w 553°!.
2
domain have never been published, could be used as a veri-
Furthermore, since u w > u w , it is clear that the wave configu-
2 N
fication of the earlier-mentioned conclusions regarding the
ration practically fulfills the conditions for the existence of a instability of these wave configurations.
strong S t MR ~note M i 51.9.M Ki !.
VI. THE TRANSITION CRITERIA
A shadowgraph of a strong D i MR in CO2 , whose con-
ditions are inside the dual-solution domain, is shown in Fig. Owing to the fact that MR wave configurations were
5 of Syshchikova and Krassovskaya.10 The initial conditions found to be stable in the u Nw > u w . u D
w domain, it is clear that
of the reflection were M i 52.9, u 2w 550°. The corresponding the RR↔MR transition criterion accepted so far for pseu-
relevant angles are u D dosteady flows, namely, u trw 5 u Dw , is incorrect. Consequently,
w 548.7° and u w 562.2°. Consequently,
N

since u w , u w , u w , the double-Mach reflection shown in the

D 2 N the following RR↔MR transition criteria are proposed:
photograph is inside the dual-solution domain. In addition, Weak Incident Shock Waves (M i ,M Ki )
since M i 52.9.M Ki ~CO2!51.42, the reflection shown in the In the weak incident shock wave domain, i.e., M i ,M Ki
photograph verifies the existence of strong-D i MR wave con- ~see Fig. 3! the RR↔MR transition criterion is
figurations inside the dual-solution domain. u trw 5 u Dw .
Based on this and additional experimental evidence and Recall that in the domain u w . u D
w , MR wave configurations
the earlier discussion regarding the fact that they could be are unstable.
considered pseudosteady, the conclusions regarding the sta- Strong Incident Shock Waves (M i .M Ki )
2046 J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor
TABLE III. Transition criteria.

Weak Strong

Type of transition 1,M i ,M Ki M Ki ,M i ,M Qi M i .M Qi

RR→MR u w 5 u Dw u w 5 u Dw u w 5 u w*
MR→RR u w 5 u Dw u w < u w < u Nw
D
u w* < u w < u Nw

The strong incident shock wave domain, i.e., M i .M Ki

~see Fig. 3! should be divided into two parts ~see Fig. 4!. In
the first part for which
M Ki ,M i ,M Q
i

the RR→MR transition criterion is

u trw 5 u Dw
and the MR→RR transition criterion is
u Dw < u trw < u Nw .
In the second part for which
M i .M Q
i
FIG. 11. Domains of RR, SMR, TMR and DMR in the (M i , u w )-plane and
the RR→MR transition criterion is their transition criterion lines.
u trw 5 u w*
and the MR→RR transition criterion is
~1! In the very narrow domain u D w , uw , uw* RR wave con-
u w* < u trw < u Nw . figurations are unstable and hence this narrow domain
Recall that practically corresponds to double-Mach reflection ~DMR!.
~2! In the domain u w* < u w < u Nw both RR and MR wave con-
u w* > u Dw .
figurations are stable. Note that this domain is subdi-
Note that these proposed criteria agree with Ben-Dor vided into two parts by the line M L1 51, where M L1 is the
et al.’s11 findings when they experimentally investigated the incident shock induced flow Mach number in a labora-
RR→MR and the MR→RR transition processes over convex tory frame of reference. ~a! In the domain M L1 ,1 the
and concave double wedges, respectively. It should also be reflection is a single-Mach reflection ~SMR!, and ~b! in
noted that the above-proposed criteria agree with the ideas the domain M L1 .1 the reflection is a double-Mach re-
forwarded by Ben-Dor et al.14 who suggested to call them flection ~DMR!. Note that a transitional-Mach reflection
the ‘‘inertia criteria.’’ For the reader’s convenience the fore- ~TMR! wave configuration is impossible inside the dual-
going newly established transition criteria are summarized in solution domain.
Table III. ~3! In the domain u w . u Nw stable MR wave configurations
are theoretically impossible and, hence, this domain cor-
responds to RR only.
VII. DOMAINS OF DIFFERENT TYPES OF
REFLECTIONS

Based on the foregoing presentation which implied that VIII. CONCLUSIONS

MR wave configurations are stable inside the dual-solution
domain ( u D w < u w < u w ), it is natural to reexamine the do-
N
The stability of regular ~RR! and Mach ~MR! reflection
mains of different types of reflection wave configurations in wave configurations in pseudosteady flows of perfect gases
the (M i , u w ) plane. was studied using the principle of minimum entropy produc-
Figure 11 is a reproduction of a similar figure which was tion. It was found that in contrast to the accepted state of the
published recently by Li and Ben-Dor.15 Note that the D i MR art that only regular reflection ~RR! wave configurations are
domain is divided in this figure to the three subdomains of obtainable in the dual-solution domain, Mach reflection
single ~SMR!-, transitional ~TMR!-, and double ~DMR!- ~MR! wave configurations are also stable there. This analyti-
Mach reflections, which as mentioned by Ben-Dor1 are the cal findings were supported by experimental results. As a
three subtypes of a direct Mach reflection. Unlike that figure, consequence, new transition criteria from regular to Mach
in which the entire domain which corresponds to u w . u D w reflections ~RR→MR! and from Mach to regular reflections
was defined as an RR domain, in Fig. 11 the u w . u D w domain ~MR→RR! were proposed. To the best of the authors’
is divided into three domains: knowledge, the proposed criteria agree with all the available

J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor 2047
 8
experimental results. Finally, as a result of the new transition L. G. Smith, OSRD Rep. No. 6271, Off. Sci. Res. Dev., Washington, DC,
criteria, the domains of different types of reflection wave 1945.
9
configurations in the (M i , u w ) plane were redefined. J. M. Dewey and A. A. van Netten, in Shock Waves, edited by K.
Takayama ~Springer, Heidelberg, 1991!, pp. 227–232.
10
M. P. Syshchikova and I. V. Krassovskaya, Arch. Mech. 31, 135 ~1979!.
1
G. Ben-Dor, Shock Wave Reflection Phenomena ~Springer, New York, 11
G. Ben-Dor, J. M. Dewey, and K. Takayama, J. Fluid Mech. 176, 483
1991!.
~1987!.
2
R. Courant and K. D. Friedrichs, Supersonic Flow and Shock Waves ~In- 12
G. Ben-Dor, J. M. Dewey, D. J. McMillin, and K. Takayama, Exp. Fluids
terscience, New York, 1948!.
3
G. Ben-Dor and T. Elperin, Shock Waves 1, 237 ~1991!. 6, 429 ~1988!.
4
J. von Neumann, Collected Works ~Pergamon, Oxford, 1963!, Vol. 6.
13
K. Takayama and G. Ben-Dor, Sadhana 18, 695 ~1993!.
14
5
L. F. Henderson and A. Lozzi, J. Fluid Mech. 68, 139 ~1975!. G. Ben-Dor, K. Takayama, and T. Kawauchi, J. Fluid Mech. 100, 147
6
L. F. Henderson and A. Lozzi, J. Fluid Mech. 94, 541 ~1979!. ~1981!.
7
H. Li and G. Ben-Dor, J. Appl. Phys. 80, 2027 ~1996!. 15
H. Li and G. Ben-Dor, Shock Waves 5, 59 ~1995!.

2048 J. Appl. Phys., Vol. 80, No. 4, 15 August 1996 H. Li and G. Ben-Dor