1. Introduction
Shock-on-shock interaction has important applications in scenarios where one supersonic object encounters another supersonic object. Small perturbation theory and experimental measurements were used in the early-stage studies (Smyrl Reference Smyrl1963; Blankenship Reference Blankenship1965; Brown & Mullaney Reference Brown and Mullaney1965; Gentry et al. Reference Gentry, Martin and Daly1966; Inger Reference Inger1966; Merritt & Aronson Reference Merritt and Aronson1966, Reference Merritt and Aronson1967). Numerical simulations began in the 1970s (Kutler et al. Reference Kutler, Sakell and Aiello1974,Reference Kutler, Sakell and Aiello1975; Kutler & Sakell Reference Kutler and Sakell1975; Law et al. Reference Law, Felthun and Skews2003). Apart from applications where two supersonic objects move at different speeds or directions, shock-on-shock interaction also occurs in the launch of supersonic projectile, where the projectile overtakes an outgoing shock wave that was formed earlier (Athira et al. Reference Athira, Rajesh, Mohanan and Parthasarathy2020).
Conventional theories of shock reflection or shock interference can be applied in studies of shock-on-shock interaction, since the shock-on-shock interaction problems can be made equivalent to shock reflection and shock interference problems through frame transformation. For instance, one can attach the reference frame to one body and consider the impingement or reflection of another shock on the attached shock flow configuration of this body (this equivalent problem is called the reduced problem). For shock-on-shock interaction when one shock overtakes the other, as may occur when one supersonic vehicle overtakes the other, the rightward-moving shock is of first family. Being of first family means that in the frame co-moving with the shock, the upstream flow is from the left-hand side. For shock-on-shock interaction when two shock waves collide, as may occur when two supersonic vehicles intercept, the rightward-moving shock is of second family. Being of second family means that in the frame co-moving with the shock, the upstream flow is from the right-hand side. Using two- and three-shock theories together with the geometric shock dynamics theory, Li & Ben-Dor (Reference Li and Ben-Dor1997) obtained the critical conditions and flow parameters for both regular and irregular reflections when the impinging shock is of second family. Wang & Wu (Reference Wang and Wu2022) considered the reflection of a rightward-moving normal shock of both first and second families over an attached steady oblique shock wave. The interaction between the moving shock and the steady oblique shock defines the primary reflection problem, and is connected to the six types of shock interference problem of Edney (Reference Edney1968) by using the reference frame co-moving with the nominal intersection point between the undisturbed moving shock and steady shock. One transmitted or reflected shock of the primary reflection then reflects over the wedge, and this reflection is connected to the classical pseudo-steady shock reflection as occurs when a moving shock impinges a wedge initially immersed in a still gas and for which reflections such as regular reflection, single Mach reflection and double Mach reflection may be possible. In the study of a moving oblique shock wave of first family by Wang et al. (Reference Wang, Xu and Wu2024), the initially steady oblique shock wave is decomposed into a pre-interaction part and a post-interaction part. The reflections of the pre-interaction part and post-interaction part with the moving incident shock were first considered independently, and then considered as a whole to see possible combined pre- and post-interaction reflection types.
Law et al. (Reference Law, Felthun and Skews2003) discovered the possibility that for the reduced problem, the post-interaction part of the original attached shock is an expansion fan (referred to as a post-formed expansion fan in this paper, although Law et al. (Reference Law, Felthun and Skews2003) termed it ‘shock–shock–fan interaction’). The study of Law et al. (Reference Law, Felthun and Skews2003) concerns shock-on-shock interaction that appears in two supersonic objects moving in opposite directions, such that in the reduced problem, the moving shock is of second family. In this paper, we consider shock-on-shock interaction involving a post-formed expansion fan that possibly arises when one shock overtakes the other (for the reduced problem, the rightward-moving shock belongs to the first family).
The remainder of the paper is organized as follows. Section 2 defines the problem, outlines previous work related to the present study, and provides numerical evidence of a post-formed expansion fan reflection. Section 3 is dedicated to studying the critical condition for post-formed expansion fan. The critical condition is displayed in the
$M_{S}{-}\lambda$
plane, where
$M_{S}$
is the Mach number based on the speed of the rightward-moving shock wave, and
$\lambda$
is the inclined shock angle. Section 4 focuses on the flow characteristics at the critical condition, at which there is no post-formed wave. We investigate whether both regular and Mach reflections occur at this specific condition, and explore the transition conditions between these two types of reflections if both are present. Section 5 shows numerical results with various reflection patterns for post-formed expansion fan reflection. In the case of Mach reflection, the study is connected to the work of Hillier (Reference Hillier2007), who identified three types of shock–expansion fan interaction. Complex local reflection structures will emerge, and this paper will delve into their details. Notably, we will decompose the reflection structure into substructures that can be linked to well-known steady flow structures by utilizing a reference frame co-moving with a typical point, such as a triple point, within a local flow structure. Concluding remarks are provided in § 6.
2. Problem definition and observation of post-formed expansion fan
2.1. Recall of past work with post-formed shock wave
The reflection of a rightward-moving oblique shock wave of first family over a steady state oblique shock wave is depicted schematically in figure 1(a). A wedge with angle of
$\theta _{w}$
is immersed in an initially steady supersonic flow with Mach number
$M_{0}$
, and produces this steady oblique shock wave (denoted
$AQ$
) with shock angle
$\beta _{w}$
. The incident shock wave (rightward-moving shock, RMS) moves at constant speed, denoted as
$u_{s}$
in the normal direction. We use subscripts
$l$
and
$r$
to indicate variables in the left- and right-hand sides of the RMS, respectively.
Reflection between an oblique RMS and a steady oblique shock wave attached to a sharp wedge: (a) initial state; (b) during reflection.

The shock speed Mach number
$M_{S}=u_{s}/a_{r}$
, and the shock inclined angle
$\lambda$
of the RMS are the primary input parameters. Note that
$\lambda$
is not the shock angle, but the angle between the RMS and the horizontal direction. For convenience and without loss of generality, the flow direction in the right-hand side of the RMS is set to be horizontal (
$\theta _{r}=0$
) in this paper.
We use
$p$
,
$\rho$
,
$T$
,
$a$
and
$\gamma$
to denote the pressure, density, temperature, sound speed, and ratio of specific heats, respectively. If the flow parameters in region (
$r$
) are given, then the flow parameters in regions (
$l$
) depend on
$M_{S}$
and
$\lambda$
, and can be obtained from the moving oblique shock relations (see Wang et al. Reference Wang, Xu and Wu2024)
\begin{equation} \left \{ \begin{array} [c]{l}\dfrac {{{p_{l}}}}{{{p_{r}}}}=\psi, \\ \dfrac {\rho _{l}}{\rho _{{{{r}}}}}=\dfrac {{1+\dfrac {{\gamma +1}}{{\gamma -1}}}\,\psi }{\psi +\dfrac {{\gamma +1}}{{\gamma -1}}},\\ \dfrac {a_{l}}{a_{r}}=\sqrt {\psi\, \dfrac {\psi +\dfrac {{\gamma +1}}{{\gamma -1}}}{{1+\dfrac {{\gamma +1}}{{\gamma -1}}}\,\psi }},\\ {M_{l,n}}=\dfrac {{M_{r,n}}-\dfrac {{1}}{\gamma } ( \psi -1 ) {\left ( {\dfrac {{\gamma +1}}{{2\gamma }}\,\psi +\dfrac {{\gamma -1}}{{2\gamma }}}\right ) ^{-{1}/{2}}}}{\sqrt {\psi\, \dfrac {\psi +\dfrac {{\gamma +1}}{{\gamma -1}} }{{1+\dfrac {{\gamma +1}}{{\gamma -1}}}\,\psi }}},\end{array} \right . \end{equation}
where
and
In (2.1) and afterwards, the subscripts
$n$
and
$\tau$
represent flow parameters normal to and tangent to the moving shock, respectively.
The upstream flow direction angle
$\theta _{l}$
displayed in figure 2 satisfies
where the shock angle
$\beta _{lr}$
satisfies
$\beta _{lr}=\arctan (u_{l,n}/u_{l,\tau })$
, with
$u_{l,n}=M_{l,n}a_{l}$
(representing normal velocity magnitude) and
$u_{l,\tau }=u_{r,\tau }=u_{r}\cos \lambda$
(representing tangential velocity magnitude). The velocity
$(u_{l},v_{l})$
in region (
$l$
) is given by
Wang et al. (Reference Wang, Xu and Wu2024) utilized the equivalent problem, defined on a frame co-moving with the nominal intersection point between the unperturbed part
${I_{S}}$
of RMS and the pre- or post-interaction part of the steady-state shock, to investigate the reflection types and transition conditions. They divided the initial steady oblique shock into three parts (see figure 1 for notation): (i) pre-interaction shock
${S_{OR}}$
, (ii) post-interaction shock
${S_{NE}}$
and (iii) the shock between these two parts. Various types of shock reflection types were identified. For the purpose of showing where we will have new types of reflection, we reproduce in figure 3 the transition conditions of Wang et al. (Reference Wang, Xu and Wu2024) for
$M_{r}=6$
and
$\theta _{w}=10^{\circ }$
.
Schematic illustration of the relations between
$\theta _{l}$
,
$\beta _{lr}$
and
$\lambda$
: (a)
$\lambda \gt 90^{\circ }$
; (b)
$\lambda \lt 90^{\circ }$
.

Transition conditions in the
$M_{S}{-}\lambda$
plane, with
$M_{r}=6$
and wedge angle
$\theta _{w}=10^{\circ }$
(Wang et al. Reference Wang, Xu and Wu2024).

Here, the reflection is classified according to the six types (types I to VI) of shock interference, which is the classification of Edney (Reference Edney1968) for shock interaction between an incident shock and a bow shock. An illustration of type I to type VI shock interferences due to the impingement of an incident shock on a bow shock is displayed in figure 4. When the incident shock and the weak part of the bow shock are from the opposite sides seen at the point of intersection, type I or type II shock interferences occur. Type I and type II are also called regular reflection and Mach reflection. When the incident shock and the weak part of the bow shock are from the same sides seen at the point of intersection, type IV or type V or type VI shock interferences occur. When the incident shock intersects the strong part of the bow shock, type III–type IV shock interferences occur. For more details and further study, see Edney (Reference Edney1968), Bramlette (Reference Bramlette1974), Frame & Lewis (Reference Frame and Lewis1997), Grasso et al. (Reference Grasso, Purpura, Chanetz and Délery2003) and Windisch et al. (Reference Windisch, Reinartz and Müller2016).
In figure 3, ‘Pre-IV’, ‘Pre-V’ and ‘Pre-VI’ denote that the pre-shock reflection results in type IV, type V and type VI shock interferences, respectively. ‘Post-I’ and ‘Post-II’ denote that the post-shock reflection results in type I and type II shock interferences, respectively. ‘Pre-DAR’ refers to a pre-shock reflection where, if the pre-interaction part of the attached shock is considered as an incident shock, no solution is produced due to deflection angle reversal – a phenomenon identified by Wang & Wu (Reference Wang and Wu2022). ‘Post-DSD’ signifies that the post-shock reflection occurs within the double solution domain, where both type I and type II reflections are possible.
Edney’s six types of shock interaction.

2.2. Numerical evidence of shock reflection with post-formed expansion fan
Now we present numerical evidence of post-formed expansion fan reflection. The numerical simulation follows the same methodology as that used in the study on post-formed shock reflection (Wang et al. Reference Wang, Xu and Wu2024). We solve the compressible Euler equations for an ideal gas using the second-order advection upstream splitting method (AUSM) (Liou Reference Liou1996), within a domain that adheres to the boundary conditions outlined in figure 5(a). According to Liou (Reference Liou1996) , the AUSM features some properties such as: (i) exact resolution of one-dimensional contact and shock discontinuities; (ii) positivity preserving of scalar quantities such as density; (iii) free of ‘carbuncle phenomenon’; (iv) free of oscillations at the slowly moving shock; (v) algorithmic simplicity. The steady solution with Mach number
$M_{r}$
is first calculated, leading to a steady flow with an attached oblique shock wave. Then the incident shock is imposed (see § 2.1 for how to set up the flow parameters in the left-hand side of the incident shock for a given shock speed and inclined shock angle), and the unsteady flow field is calculated.
Due to lack of scales in the local interaction, the flow is self-similar. The self-similar evolution of flow structure will be examined.
(a) Computational domain and boundary conditions. (b) Number of grids along two directions, with three blocks of structured mesh separated with dashed lines.

The flow with two sets of conditions for cases (a) and (b) presented in table 1 is computed. The numerical results of Mach contours at a typical instant are displayed in figure 6(a) for case (a), and in figure 6(b) for case (b). We observe reflection with post-formed shock reflection for case (a) (the post-formed shock is denoted
${S_{NE}}$
in figure 6
a) and post-formed expansion fan reflection for case (b) (the post-formed expansion fan is denoted
${R_{NE}}$
in figure 6
b). Figure 9 is a sketch of the flow before and during reflection with post-formed expansion fan.
In figure 7, the variation of pressure coefficient
$C_{p}=(p-p_{l})/(0.5\rho _{l}(u_{l}^{2}+v_{l}^{2}))$
along four streamlines named s
$_{1}$
, s
$_{2}$
, s
$_{3}$
and s
$_{4}$
are displayed below the pressure contours for case (a) (with post-formed shock wave) and case (b) (with post-formed expansion fan). For case (a), the streamline s
$_{1}$
near the wall goes through the post-formed shock at the left, and goes through shock
$\mathrm {P^{\prime }}$
G and shock GQ with increase of pressure coefficient at the intersection point. After going through the uniform region after GQ, the flow goes through a region of compression and expansion waves, with the pressure coefficient increasing slightly then decreasing gradually. The flow along streamline s
$_{2}$
does not go through post-formed shock
$\mathrm {S_{NE}}$
, but goes through shock
$\mathrm {P^{\prime }}$
G and shock GQ with same change of pressure coefficient as along streamline s
$_{1}$
. The flow along streamlines s
$_{3}$
and s
$_{4}$
goes through only incident shock
$\mathrm {I_{S}}$
and pre-formed shock
$\mathrm {S_{OR}}$
. The variation of pressure coefficient along streamlines s
$_{3}$
and s
$_{4}$
increases with two steps, and decreases gradually after going through pre-formed shock
$\mathrm {S_{OR}}$
. For case (b), the streamlines s
$_{1}$
and s
$_{2}$
near the wall go through the post-formed expansion fan and the shock P
$_{1}$
G. The expansion wave
$R_{NE}$
is very weak, thus the decrease of pressure is negligible here. The pressure coefficient increases across shock P
$_{1}$
G, then wave-like rises along streamline s
$_{1}$
. The pressure coefficient increases across shock P
$_{1}$
G and shock QR along streamline s
$_{2}$
. The streamline s
$_{3}$
goes through the incident shock
$\mathrm {I_{S}}$
, shock P
$_{1}$
Q and shock QR. The streamline s
$_{4}$
goes through the incident shock
$\mathrm {I_{S}}$
and the pre-formed shock
$\mathrm {S_{OR}}$
. The pressure coefficient along streamlines s
$_{3}$
and s
$_{4}$
increases across each shock.
To demonstrate the self-similar nature of the current problem, we present results on the evolution of shock structures, similar to those in Martínez-Ruiz et al. (Reference Martínez-Ruiz, Huete, Martínez-Ferrer and Mira2019). Here, the evolution of the length of structures for cases (a) and (b) is displayed in figure 8, in terms of the non-dimensional time t* defined as
where
$L$
is the horizontal length of the wedge. The self-similar or pseudo-steady nature of case (a) with post-formed shock can be observed in figure 8(a). The lengths of
$\mathrm {P^{\prime }}$
Q,
$\mathrm {P^{\prime }}$
G and GQ for case (a) are proportional to the non-dimensional time t*. The self-similar or pseudo-steady nature of case (b) with post-formed expansion fan can be observed in figure 8(b). The evolution of
$\mathrm {P_{1}G}$
and
$\mathrm {P_{1}Q}$
is linear with respect to time t*.
Four sets of conditions for numerical simulation.

Mach contours for numerical results: (a) case (a) with post-formed shock; (b) case (b) with post-formed expansion wave.

Pressure coefficient along typical streamlines for (a) case (a) and (b) case (b).

Time evolution of the structure lengths for cases (a) and (b).

Case (a) lies in the region Post-I in figure 3(b), while case (b) does not lie inside any regions marked in figures 3(a) and 3(b). This means that post-formed expansion fan reflection may occur outside the domains with post-formed shock reflection. In the following, we will search for the region in the
$M_{S}{-}\lambda$
plane to have a post-formed expansion wave.
Reflection between a rightward-going incident shock wave and a steady oblique shock wave attached to a sharp wedge: (a) initial state; (b) during reflection.

3. Domain with post-formed expansion wave in the
$M_{S}{-}\lambda$
plane
3.1. Method to determine the critical condition
The critical condition is the line in the
$M_{S}{-}\lambda$
plane that connects the region with post-formed expansion fan and the region with post-formed shock wave. This critical condition can be determined through the magnitude of the flow deflection angle
$\theta _{l}$
that is computed by (2.4).
Obviously, post-formed shock reflection occurs if
$\theta _{l}\lt \theta _{w}$
, and post-formed expansion fan reflection occurs if
$\theta _{l}\gt \theta _{w}$
. The critical condition of the shock inclined angle,
$\lambda =\lambda _{cri}$
, for transition between post-formed shock reflection and post-formed expansion fan reflection, is thus given by
At this critical condition
$\theta _{l}=\theta _{w}$
, the upstream flow direction is parallel with the wedge surface. As shown in figure 2, to have
$\theta _{l}\gt 0$
, it is required that
$\lambda \gt 90^{\circ }$
.
From the notations displayed in figure 2, it is easy to see
Thus the transition condition (3.1) can be written as
where
$\beta _{lr}$
is the shock angle (figure 2
a) for
$\lambda =\lambda _{cri}$
. The critical value of
$\beta _{lr}$
, denoted as
$\beta _{lr}|_{cri}$
, is a function of
$M_{S},\ M_{r}$
and
$\lambda _{cri}$
. Substituting (2.1)–(2.2) into the shock angle relation for
$\beta _{lr}$
, we obtain
\begin{equation} \beta _{lr}|_{cri}=\arctan \left ( \frac {M_{r}\sin \lambda _{cri}-\dfrac {\psi _{cri}-1}{\gamma }\left ( \dfrac {\gamma +1}{2\gamma }\,\psi _{cri}+\dfrac {\gamma -1}{2\gamma }\right ) ^{-{1}/{2}}}{M_{r}\left \vert \cos \lambda _{cri}\right \vert }\right ) , \end{equation}
with
The critical value of
$\lambda _{cri}$
can be obtained by substituting the expressions of
$\beta _{lr}|_{cri}$
in (3.4) and (3.5) into (3.3), for a given set of parameters
$M_{r}$
,
$M_{S}$
and
$\theta _{w}$
.
3.2.
Domain with post-formed expansion fan in the
$M_{S}{-}\lambda$
plane for specific values of
$M_{S},\ M_{r}$
and
$\theta _{w}$
First, we need to note that when solving for
$\lambda _{cri}$
, we find that the boundary
$\lambda =\lambda _{cri}$
lies between
$\lambda _{\textit{min}}$
and
$\lambda _{\textit{max}}$
, where
$\lambda _{\textit{min}}$
and
$\lambda _{\textit{max}}$
are the lower and upper boundaries for the incident shock wave to be of first family. According to Wang et al. (Reference Wang, Xu and Wu2024),
and
We also find that the boundary
$\lambda =\lambda _{cri}$
intersects the boundary
$\lambda =\lambda _{\textit{max}}$
at some value of
$M_{S}$
, and this value will be denoted by
$M_{S}^{\ast }$
. Setting
$\lambda _{cri}=\lambda _{\textit{max}}$
and
$M_{S}=M_{S}^{\ast }$
in (3.7) gives the following expression for
$M_{S}^{\ast }$
:
\begin{equation} \pi -\arcsin \left ( \frac {M_{S}^{\ast }+\sqrt {({\gamma -1})/{2\gamma }}}{M_{r}}\right ) =\lambda _{cri}. \end{equation}
Thus in the
$M_{S}{-}\lambda$
plane, the domain to have post-formed expansion fan lies in the region
$\lambda _{cri}\lt \lambda \lt \lambda _{\textit{max}}$
, for
$M_{S}\lt M_{S}^{\ast }$
. This will be made clear below using specific conditions. Now we superimpose the new region with post-formed expansion fan onto the plane, with the critical conditions for post-shock reflection found by Wang, Xu & Wu (Reference Wang, Xu and Wu2024) for completeness. Figure 10(a) is for
$M_{r}=6$
and
$\theta _{w}=5^{\circ }.$
The region for post-formed expansion fan, marked in red, is a narrow stripe that is above the region for Post-I shock reflection, which lies in the region where
$\lambda _{\textit{min}}\lt \lambda \lt \lambda _{cri}$
. An enlarged view for this region is displayed in figure 10(b), for clarity. It can be seen that as
$M_{S}$
increases, the critical condition to have post-formed expansion fan reflection gets narrow and occurs at smaller
$\lambda$
. When
$\theta _{w}$
increases, the regions for post-formed expansion fan reflection become smaller, as can be seen from figures 10(c,d) for
$M_{r}=6$
with
$\theta _{w}=10^{\circ }$
, and figures 10(e,f) for
$M_{r}=6$
with
$\theta _{w}=15^{\circ }$
. The critical value
$M_{S}^{\ast }$
decreases as
$\theta _{w}$
increases.
Domains of various reflection types in the
$M_{S}{-}\lambda$
plane, where the red narrow stripe means reflection with post-formed expansion wave: (a) for
$M_{r}=6$
and
$\theta _{w}=5^{\circ }$
; (b) enlargement of (a) with
$\lambda _{\textit{min}}$
,
$\lambda _{\textit{max}}$
and
$\lambda _{cri}$
; (c) for
$M_{r}=6$
and
$\theta _{w}=10^{\circ }$
; (d) enlargement of (c) with
$\lambda _{\textit{min}}$
,
$\lambda _{\textit{max}}$
and
$\lambda _{cri}$
; (e) for
$M_{r}=6$
and
$\theta _{w}=15^{\circ }$
; (f) enlargement of (e) with
$\lambda _{\textit{min}}$
,
$\lambda _{\textit{max}}$
and
$\lambda _{cri}.$

4. Flow at the critical condition
$\lambda =\lambda _{cri}$
4.1. Numerical results: regular and Mach reflections
Now we provide some numerical simulations at the transition condition
$\lambda =\lambda _{cri}$
. For
$M_{r}=6$
, the values of
$\lambda _{cri}$
determined by (3.3) are displayed in figure 11. The usefulness of this numerical simulation is twofold. First, if the numerical results indeed show no post-formed waves, then the critical condition
$\lambda =\lambda _{cri}$
determined in § 3.1 is verified. Second, numerical simulation will show if we have more than one type of reflection, and if so, we can further analyse the transition conditions.
Variation of
$\lambda _{cri}$
in the
$M_{S}{-}\theta _{w}$
plane for
$M_{r}=6$
.

For the above purposes, we have tested several conditions with
$\lambda =\lambda _{cri}$
, and two typical ones, labelled cases (c) and (d), are given in table 1. The Mach contours and pressure contours by numerical simulation are displayed in figures 12(a,b) for case (c), and in figures 13(a,b) for case (d) (see figure 11 for the locations of cases (c) and (d)). There is indeed no post-formed wave, showing that
$\lambda =\lambda _{cri}$
determined in § 3.1 is indeed the transition condition between post-formed shock reflection and post-formed expansion fan reflection. We remark that case (c) is a Mach reflection, and case (d) is a regular reflection. For Mach reflection, the incident shock (RMS) reflects over the wedge to produce a Mach stem and a triple point. The reflected shock of this Mach reflection intersects with the pre-formed shock (
$\mathrm {S_{OR}}$
) to produce another triple point. In contrast to case (c), case (d) gives a regular reflection, where the incident shock is the RMS, and the reflected shock connects to the pre-formed shock in a smooth way.
Numerical results for case (c): (a) Mach contours; (b) pressure contours.

Numerical results for case (d): (a) Mach contours; (b) pressure contours.

4.2. The critical condition between regular reflection and Mach reflection for
$\lambda =\lambda _{cri}$
Section 4.1 shows that both regular reflection and Mach reflection can occur at
$\lambda =\lambda _{cri}$
. A further consideration is for the transition condition between these regular and Mach reflections.
We know that for steady shock reflection, the transition conditions for regular and Mach reflections are determined by the von Neumann condition
$\beta =\beta ^{(N)}(M_{0})$
and the detachment condition
$\beta =\beta ^{(D)}(M_{0})$
.
For the present problem, the reflection point D moves along the wedge surface at velocity given by
$\boldsymbol{V}_{D}=(u_{D},v_{D})$
, with
Thus the Mach number and the shock angle that determine the von Neumann condition and detachment conditions are replaced by the Mach number
$\overline {M}_{0}$
and the shock angle
$\bar {\beta }_{01}$
measured in the frame co-moving with D.
Since
and
regular reflection and Mach reflection are thus determined by the conventional von Neumann condition
and detachment condition
Since here we have taken
$\lambda =\lambda _{cri}$
, the transition conditions depend on only
$M_{S}$
and
$\theta _{w}$
. For
$M_{r}=6$
, transition conditions for regular reflection and Mach reflection in the
$M_{S}{-}\theta _{w}$
plane are displayed in figure 14. The double solution domain marked as DSD is between the detachment condition and the von Neumann condition where both regular reflection and Mach reflection are possible. Regular reflection occurs for smaller
$M_{S}$
, and Mach reflection occurs for larger
$M_{S}$
, as displayed in figure 14.
Domains of regular reflection (RR), Mach reflection (MR), and double solution domain (DSD) for
$\lambda =\lambda _{cri}$
and
$M_{r}=6$
in the
$M_{S}{-}\theta _{w}$
plane.

Now we provide further discussion for cases (c) and (d) used for producing the numerical results shown in figures 12(a,b). Case (c) lies in the Mach reflection region; see figure 14. The corresponding relative Mach number
$\overline {M}_{0}$
determined by (4.2) is
$\overline {M}_{0}=4.69$
, and for this Mach number, the von Neumann condition is
$\beta _{01}^{(N)}(\overline {M}_{0})=31.52^{\circ }$
and the detachment condition is
$\beta _{01}^{(D)}(\overline {M}_{0})=39.29^{\circ }$
. The shock angle
$\bar {\beta }_{01}$
computed by (4.3) is
$\bar {\beta }_{01}=40.26$
. Thus
$\bar {\beta }_{01}$
is larger than
$\beta _{01}^{(D)}(\overline {M}_{0})$
and lies in the Mach reflection region. From the numerical simulation results, the incident shock reflects over the wedge surface as Mach reflection (see figure 12
b), which is consistent with the prediction from transition condition theory. Case (d) lies in the region RR of figure 14. The corresponding relative Mach number
$\overline {M}_{0}$
determined by (4.2) is
$\overline {M}_{0}=5.27$
. For this Mach number, the von Neumann condition is
$\beta _{01}^{(N)}(\overline {M}_{0})=30.23^{\circ }$
and the detachment condition is
$\beta _{01}^{(D)}(\overline {M}_{0})=39.37^{\circ }$
. The shock angle
$\bar {\beta }_{01}$
computed by (4.3) is
$\bar {\beta }_{01}=27.64^{\circ }$
, which is smaller than
$\beta _{01}^{(N)}(\overline {M}_{0})$
and lies in the regular reflection region. From the numerical simulation results, the incident shock reflects over the wedge surface as regular reflection, which is consistent with the prediction from transition condition theory.
5. Numerical simulation of flow with post-formed expansion fan reflection
5.1. Numerical results for increasing inclined shock angle
Consider seven test cases, labelled case 1 to case 7, where the parameters are fixed at
$M_{r}=6,$
$M_{S}=2.55$
and
$\theta _{w}=10^{\circ }$
, while the variable
$\lambda$
is set to
$150^{\circ },$
$150.1^{\circ },$
$150.2^{\circ },$
$150.3^{\circ },$
$150.4^{\circ },$
$150.5^{\circ }$
and
$150.60^{\circ }$
for each case, respectively. These cases fall within the domain with post-formed expansion fan reflection, as detailed in figure 15 (enlargement of figure 10
d), which illustrates the positions of the cases. Given that post-formed expansion fan occurs within the narrow region between
$\lambda _{cri}$
and
$\lambda _{\textit{max}}$
, the selection of these cases is aimed at investigating the impact of increasing
$\lambda$
values. The rationale for this approach is that the intensity of the post-formed expansion fan intensifies with higher
$\lambda$
values. Note that the critical angle
$\lambda _{cri}=149.74^{\circ }$
(case (c)) is slightly lower than
$\lambda =150^{\circ }$
for case 1.
Numerical simulation cases chosen in the
$M_{S}{-}\lambda$
plane.

The Mach contours and pressure contours for these seven cases are displayed in figures 16(a–n) and figures 17(a–n), listed in order of increasing
$\lambda$
. For case 1 (
$\lambda =150^{\circ }$
), the flow is similar to the case with no post-formed wave. For cases 2–4 (
$\lambda =150.1^{\circ },150.2^{\circ },150.3^{\circ }$
), there is a new curved shock near the surface. The new curved shock interacts with the incident shock with Mach reflection, with two triple points
$\mathrm {P_{1}}$
and
$\mathrm {P_{2}}$
observed. For cases 5–7 (
$\lambda =150.4^{\circ },150.5^{\circ },150.6^{\circ }$
), the new curved shock interacts with the incident shock with regular reflection, with two triple points
$\mathrm {P_{1}}$
and
$\mathrm {P_{2}}$
merged into a single point
$\mathrm {P_{3}}$
.
Numerical results for
$M_{r}=6,$
$M_{S}=2.55,$
$\theta _{w}=10^{\circ }$
, with Mach contours and pressure contours, respectively, for (a,b) case 1 (
$\lambda =150^{\circ }$
), (c,d) case 2 (
$150.1^{\circ }$
), (e,f) case 3 (
$150.2^{\circ }$
), (g,h) case 4 (
$150.3^{\circ }$
), (i,j) case 5 (
$150.4^{\circ }$
), (k,l) case 6 (
$150.5^{\circ }$
) and (m,n) case 7 (
$150.6^{\circ }$
).

Numerical results for
$M_{r}=6,$
$M_{S}=2.55,$
$\theta _{w}=10^{\circ }$
(enlargement of figure 16), with Mach contours and pressure contours, respectively, for (a,b) case 1 (
$\lambda =150^{\circ }$
), (c,d) case 2 (
$150.1^{\circ }$
), (e,f) case 3 (
$150.2^{\circ }$
), (g,h) case 4 (
$150.3^{\circ }$
), (i,j) case 5 (
$150.4^{\circ }$
), (k,l) case 6 (
$150.5^{\circ }$
) and (m,n) case 7 (
$150.6^{\circ }$
).

To demonstrate the self-similar nature of the current problem, we present results on the evolution of shock structures, similar to those in Martínez-Ruiz et al. (Reference Martínez-Ruiz, Huete, Martínez-Ferrer and Mira2019). In figure 18, the evolution of the lengths of shock structures with non-dimensional time t* is shown for cases 2–7. The position of
$\mathrm {Q_{2}}$
is not very clear at the early stages for case 5, thus
$\mathrm {Q_{1}Q_{2}}$
is not marked. The lengths increase linearly with non-dimensional time, indicating a self-similar or pseudo-steady nature for these six cases. However, not all lengths are proportional to non-dimensional time due to the influence of downstream sliplines and subsonic regions.
Time evolution of the structure lengths for cases 2–7.

The extension of the expansion fan certainly affects the shock reflection. This extension can be quantified by measuring the angles
$\theta _{1}$
and
$\theta _{2}$
of the leading and trailing characteristic lines of the expansion fan, respectively, as measured from the wedge surface, as shown in figure 19. With the deflection angle
$\theta _{l}$
in the upstream of the expansion fan being computed by (2.4), the deflection angle of the flow across the post-formed expansion wave can be computed by
$\theta _{exp }=\theta _{l}-\theta _{w}$
. Then for
$\theta _{1}$
and
$\theta _{2}$
, we have
where
$\mu _{1}$
and
$\mu _{2}$
are Mach angles computed by
$\mu _{1}=\arcsin ( {1/}M_{l} )$
and
$\mu _{2}=\arcsin ( {1/M}_{lw} )$
, where
${M}_{lw}$
is the Mach number in the region downstream of the expansion fan, determined by
${\theta }_{exp }{=\nu (M_{lw})-\nu (M_{l})\,}$
, where
\begin{equation} {\nu (M)}=\sqrt {\frac {\gamma +1}{\gamma -1}}\arctan \left(\sqrt {\frac {\gamma -1}{\gamma +1}\, ( M^{2}-1 ) }\right)-\arctan (\sqrt {M^{2}-1}) \end{equation}
is the Prandtl–Meyer function.
Typical Prandtl–Meyer expansion wave.

The variations of
$\theta _{1}$
and
$\theta _{2}$
with
$\lambda$
computed by (5.1) are displayed in figure 20. The shaded region between
$\theta _{1}$
and
$\theta _{2}$
indicates the area within expansion fan. The angle of first Mach line
$\theta _{1}$
first increases with the increase of
$\lambda$
, reaches a maximum value at
$\lambda =150.26^{\circ }$
, and then decreases. The angle of last Mach line
$\theta _{2}$
decreases monotonically with the increase of
$\lambda$
. The difference between the directions of the first and last Mach lines,
$\theta _{1}-\theta _{2}$
, increases with the increase of
$\lambda$
.
Variation of the angle of the first Mach line
$\theta _{1}$
and the angle of the last Mach line
$\theta _{2}$
with
$\lambda$
for
$M_{r}=6,$
$M_{S}=2.55$
and
$\theta _{w}=10^{\circ }$
.

The locations of the triple points
$\mathrm {P_{1}}$
and
$\mathrm {P_{2},}$
as well as the intersection point
$\mathrm {P_{3}}$
(see figure 17), are also marked in figure 20. The open circle represents the angle of P
$_{1}$
, the black solid circle represents the angle of P
$_{2}$
, and the red solid circle represents the angle of P
$_{3}$
. For
$\lambda =150.10^{\circ }$
, the two triple points P
$_{1}$
and P
$_{2}$
are both below the expansion wave. For
$\lambda$
values
$150.20^{\circ }$
and
$150.30^{\circ }$
, P
$_{1}$
is situated within the expansion wave, while P
$_{2}$
remains below the expansion wave. For
$\lambda$
values
$150.40^{\circ }$
and
$150.50^{\circ }$
, the curved shock that originates from the wedge surface intersects the incident shock at point P
$_{3}$
.
5.2. Local reflection patterns
Hillier (Reference Hillier2007) studied the interaction between an upstream expansion fan and the shock reflection structure, in a steady supersonic flow, and classified the configurations into three types. He categorized these configurations based on the relative positioning of the triple point and expansion fan for type-1 and type-2. Type-1 and type-2 are defined by whether the triple point is behind or within the expansion wave. Additionally, for type-3, the flow becomes unsteady and eventually collapses due to roll-up of various slip surfaces. Type-3 appears to be when the triple point is behind the expansion fan; however, this is not stated explicitly.
However, due to the results of current numerical simulation, cases 5–7 show a new structure with regular reflection between asymmetric incident shocks at point
$P_3$
, which is not observed in Hillier (Reference Hillier2007). Now we consider the cases observed here in light of the classification by Hillier (Reference Hillier2007). When considering the definitions of type-1 and type-2 as when the triple point is behind or within the expansion wave and the flow is stable (steady or pseudo-steady), cases 1 and 2 are classified as type-1, and the cases 3–6 as type-2.
The pseudo-steady motion inherent to the present problem is not the same as the large-scale unsteadiness observed by Hillier (Reference Hillier2007, figure 21g) for his type-3 case, which is characterized by the motion of the Mach stem up to the expansion corner. For the classical wedge-generated shock wave (Hillier Reference Hillier2007, figure 2), a similar phenomenon, known as the nonstart phenomenon, occurs when the wedge height is below a certain limit (Vuillon, Zeitoun & Ben-Dor Reference Vuillon, Zeitoun and Ben-Dor1995). For the configuration studied by Hillier (Reference Hillier2007, figure 1), before considering the large-scale unsteadiness, we should first understand why a stable steady-state solution exists (such as type-1 and type-2 reflections in Hillier Reference Hillier2007). Fortunately, Hillier (Reference Hillier2007) already provided the mechanism by which a stable Mach reflection configuration is possible.
At the beginning of § 4.1, Hillier (Reference Hillier2007) stated: “In Mach reflection, stabilization of the Mach stem requires the formation of a sonic throat in the post-Mach-stem streamtube; otherwise, downstream disturbances can propagate upstream to the Mach stem and alter its scale and location. In classical Mach-reflection studies, the required low pressure is generated by the expansion wave from the trailing edge of the wedge used to generate the incident shock wave. In the present study, the reflection of the initial expansion wave from the upper surface of the duct back onto the streamtube is the equivalent mechanism. Thus, the position and scale of the Mach stem depend critically on the axial location of the expansion corner relative to the upper leading edge.”
In contrast to the nonstart phenomenon in classical shock reflection, where the trailing edge is too low so that the reflected shock impinges on the wedge’s lower surface before the trailing edge, the large-scale unsteady motion observed in figure 21(g) of Hillier (Reference Hillier2007) would be due to the inappropriate position of the expansion wave-producing corner, preventing the reflected expansion wave from providing the necessary pressure drop in the post-Mach-stem streamtube to stabilize the type-3 shock reflection.
For the present problem, the shocks exhibit pseudo-steady motion and do not need to be stabilized, so the large-scale motion discovered by Hillier (Reference Hillier2007) should not exist. Note that the type-3 case shown in figure 21(g) of Hillier (Reference Hillier2007) also displays roll-up of the various slip surfaces, which may further contribute to small-scale unsteadiness superposed on the large-scale unsteadiness. As is well known, unsteady vortices due to the Kelvin–Helmholtz instability of the shear layer or slipline streams are often captured using a high-resolution numerical scheme with high-order accuracy or with a very fine grid (Ivanov et al. Reference Ivanov, Ben-Dor, Elperin, Kudryavtsev and Khotyanovsky2002; Gao & Wu Reference Gao and Wu2010). The roll-up of the various slip surfaces observed by Hillier (Reference Hillier2007) for their figure 21(g) case (type-3) would be further initiated by the interaction of these streamlines with the expansion waves during their relative motion.
If we consider a new classification that is more suitable for the cases observed, it would be as follows. Type-a: Mach reflection of incident shock without a new triple point, as shown in figures 17(a,b). Type-b: case with a new triple point, as shown in figures 17(c–h). Type-c: asymmetric regular reflection, as shown in figures 17(i–n). This classification, ranging from type-a to type-c, is solely based on the flow structure. By deduction, we can infer the existence of subcategories within type-b, such as type-b-1 and type-b-2, which are characterized by their flow structure. These subcategories are distinguished by the position of the triple point, either behind the expansion wave or within the expansion fan. Cases 2 and 3–4 can be classified as type-b-1 and type-b-2, respectively.
There are multiple local reflection patterns that can be identified by using the reference frame co-moving with each local flow structure. In this new reference frame, the flow velocity is given by
$U=u-u_{T}$
and
$V=v-v_{T}$
, where
$(u_{T},v_{T})$
represents the velocity of a typical point (such as triple point) associated with the local flow structure. Though the flow is complex, the reflection structure can be divided into two parts, one upstream (close to the unperturbed
$\mathrm {I_{S}}$
) and the other downstream (near the unperturbed
$\mathrm {S_{OR}}$
).
For case 2, the Mach contours and streamlines based on the velocity
$(U,V)$
for the upstream structure are shown in figure 21 when
$(u_{T},v_{T})$
represents the velocity of the triple point P
$_{1}$
on
$\mathrm {I_{S}}$
. The flow is similar if
$(u_{T},v_{T})$
represents the velocity of the point P
$_{2}$
. We observe an asymmetric Mach reflection, with P
$_{1}$
and P
$_{2}$
being the two triple points. However, the flow downstream of the shock P
$_{2}$
G is not supersonic, indicating that P
$_{2}$
G should be an extension of the Mach stem from triple point P
$_{1}$
. In this case, P
$_{2}$
represents a kink, characterized by a slope discontinuity on the Mach stem.
Numerical results for case 2 with the reference frame attached to point P
$_{1}$
: (a) Mach contours, with
$M=1$
marked in red; (b) pressure contours and streamlines.

For case 2, the Mach contours and streamlines based on the velocity
$(U,V)$
for the downstream structure are shown in figure 22 when
$(u_{T},v_{T})$
represents the velocity of the triple point P
$_{4}$
on
$\mathrm {S_{OR}}$
. We observe two
$\lambda$
shocks, a lower one and an upper one. The shock wave P
$_{6}$
P
$_{5}$
acts as a recompression shock due to deceleration of the flow impinging on the wall, and an additional recompression shock P
$_{7}$
P
$_{5}$
is generated to align the flow parallel to the wedge. For the lower
$\lambda$
shock structure, the two recompression shock waves P
$_{6}$
P
$_{5}$
and P
$_{7}$
P
$_{5}$
near the wedge join at point P
$_{5}$
to form a stronger shock P
$_{5}$
P
$_{4}$
. For the upper
$\lambda$
shock, the reflected shock P
$_{1}$
P
$_{4}$
and the shock P
$_{5}$
P
$_{4}$
merge at point P
$_{4}$
to create the shock connecting smoothly to the unperturbed shock
$\mathrm {S_{OR}}$
.
Numerical results for case 2 with the reference frame attached to point P
$_{4}$
: (a) Mach contours, with
$M=1$
marked in red; (b) pressure contours and streamlines.

The variations of pressure coefficient along five streamlines s
$_{1}$
, s
$_{2}$
, s
$_{3}$
, s
$_{4}$
and s
$_{5}$
are shown in figure 23. Along streamline s
$_{1}$
, the pressure coefficient first increases across shock P
$_{2}$
G, then exhibits a wave-like rise across expansion and compression waves. Along streamlines s
$_{2}$
and s
$_{3}$
, the pressure coefficient first increases across shock P
$_{1}$
P
$_{2}$
, then increases across shock P
$_{4}$
P
$_{7}$
. Along streamlines s
$_{4}$
and s
$_{5}$
, the pressure coefficient first increases across shock
$\mathrm {I_{S}}$
, then increases across shock
$\mathrm {S_{OR}}$
or shock P
$_{4}$
P
$_{7}$
.
Pressure coefficient variation along typical streamlines s
$_{1}$
, s
$_{2}$
, s
$_{3}$
, s
$_{4}$
and s
$_{5}$
for case 2.

Numerical results for case 4 with the reference frame attached to point P
$_{1}$
: (a) Mach contours, with
$M=1$
marked in red; (b) pressure contours and streamlines.

Numerical results for case 4 with the reference frame attached to point P
$_{4}$
: (a) Mach contours, with
$M=1$
marked in red; (b) pressure contours and streamlines.

For case 4, the Mach contours and streamlines based on the velocity
$(U,V)$
for the upstream structure are shown in figure 24 when
$(u_{T},v_{T})$
represents the velocity of the triple point P
$_{1}$
on
$\mathrm {I_{S}}$
. The flow pattern is similar when
$(u_{T},v_{T})$
represents the velocity of the point P
$_{2}$
. We still observe an asymmetric Mach reflection, with P
$_{1}$
and P
$_{2}$
being the two triple points. The flow downstream of the incident shock P
$_{2}$
G is supersonic in the vicinity of P
$_{2}$
, but it becomes subsonic in the vicinity of G on the surface.
For case 4, the Mach contours and streamlines based on the velocity
$(U,V)$
for the downstream structure are shown in figure 25 when
$(u_{T},v_{T})$
represents the velocity of the triple point P
$_{4}$
on
$\mathrm {S_{OR}}$
. We observe two
$\lambda$
shocks, a lower one and an upper one. For the lower
$\lambda$
shock structure, the two recompression shock waves P
$_{6}$
P
$_{5}$
and P
$_{7}$
P
$_{5}$
near the wedge, produced for reasons similar to those case 2, join at point P
$_{5}$
to form a stronger shock P
$_{5}$
P
$_{4}$
. Compared with case 2, the recompression shock P
$_{6}$
P
$_{5}$
is stronger. For the upper
$\lambda$
shock, the reflected shock P
$_{1}$
P
$_{4}$
and the shock P
$_{5}$
P
$_{4}$
join at point P
$_{4}$
to form the shock connecting smoothly to the unperturbed shock
$\mathrm {S_{OR}}$
.
The variation of pressure coefficient along five streamlines s
$_{1}$
, s
$_{2}$
, s
$_{3}$
, s
$_{4}$
and s
$_{5}$
is shown in figure 26. Along streamline s
$_{1}$
, the pressure coefficient first increases across shock P
$_{2}$
G, then wave-like rises across expansion and compression waves. Along streamlines s
$_{2}$
and s
$_{3}$
, the pressure coefficient first increases across shock P
$_{1}$
P
$_{2}$
or shock P
$_{2}$
G, then increases across shock P
$_{5}$
P
$_{7}$
. Along streamlines s
$_{4}$
and s
$_{5}$
, the pressure coefficient first increases across shock
$\mathrm {I_{S}}$
, then increases across shock
$\mathrm {S_{OR}}$
or shock P
$_{4}$
P
$_{7}$
.
Pressure coefficient variation along typical streamlines s
$_{1}$
, s
$_{2}$
, s
$_{3}$
, s
$_{4}$
, s
$_{5}$
for case 4.

For case 6, the Mach contours and streamlines based on the velocity
$(U,V)$
for the upstream structure are shown in figure 27 when
$(u_{T},v_{T})$
represents the velocity of the triple point P
$_{3}$
on
$\mathrm {I_{S}}$
. We observe apparently an asymmetric regular reflection, with P
$_{3}$
as the intersection point of two incident shock waves. The flow downstream of the incident shock P
$_{3}$
G is supersonic in the vicinity of P
$_{3}$
, but it becomes subsonic in the vicinity of G on the surface.
Numerical results for case 6 with the reference frame attached to point P
$_{3}$
: (a) Mach contours, with
$M=1$
marked in red; (b) pressure contours and streamlines.

Numerical results for case 6 with the reference frame attached to point P
$_{4}$
: (a) Mach contours, with
$M=1$
marked in red; (b) pressure contours and streamlines.

For case 6, the Mach contours and streamlines based on the velocity
$(U,V)$
for the downstream structure are shown in figure 28 when
$(u_{T},v_{T})$
represents the velocity of the triple point P
$_{4}$
on
$\mathrm {S_{OR}}$
, and in figure 29 when
$(u_{T},v_{T})$
represents the velocity of the triple point P
$_{5}$
. We observe two
$\lambda$
shocks, a lower one and an upper one. For the lower
$\lambda$
shock structure, the two recompression shock waves P
$_{6}$
P
$_{5}$
and P
$_{7}$
P
$_{5}$
, formed near the wedge for reasons similar to those in case 2, converge at point P
$_{5}$
to create a stronger shock P
$_{5}$
P
$_{4}$
. Compared with case 2, the recompression shock P
$_{6}$
P
$_{5}$
is stronger. For the upper
$\lambda$
shock, the reflected shock P
$_{3}$
P
$_{4}$
and the shock P
$_{5}$
P
$_{4}$
merge at point P
$_{4}$
to create the shock connecting smoothly to the unperturbed shock
$\mathrm {S_{OR}}$
.
Numerical results for case 6 with the reference frame attached to point P
$_{5}$
: (a) Mach contours, with
$M=1$
marked in red; (b) pressure contours and streamlines.

The variations of pressure coefficient along five streamlines s
$_{1}$
, s
$_{2}$
, s
$_{3}$
, s
$_{4}$
and s
$_{5}$
are shown in figure 30. Along streamline s
$_{1}$
, the pressure coefficient first increases across shock P
$_{3}$
G, then exhibits a wave-like rise across expansion and compression waves. Along streamline s
$_{2}$
, the pressure coefficient first increases across shock P
$_{3}$
G, then increases across shock P
$_{4}$
P
$_{5}$
. Along streamlines s
$_{3}$
and s
$_{4}$
, the pressure coefficient first increases across shock P
$_{3}$
G, then increases across the reflected shock of the interaction between
$\mathrm {I_{S}}$
and shock P
$_{3}$
G, finally increasing across shock P
$_{4}$
P
$_{5}$
. Along streamline s
$_{5}$
, the pressure coefficient first increases across shock
$\mathrm {I_{S}}$
, then increases across shock P
$_{3}$
P
$_{4}$
, finally increasing across P
$_{4}$
P
$_{5}$
.
Pressure coefficient variation along typical streamlines s
$_{1}$
, s
$_{2}$
, s
$_{3}$
, s
$_{4}$
and s
$_{5}$
for case 6.

In summary, the complex shock reflection pattern is essentially composed of an upstream structure (near the unperturbed
$\mathrm {I_{S}}$
) and a downstream structure (near the unperturbed
$\mathrm {S_{OR}}$
). Under the conditions investigated, the upstream structure is characterized by a Mach reflection of the incident shock over the wall, potentially featuring a triple point formed within the Mach stem. Below this triple point, there is a curved segment of the Mach stem that is close to the wall. As the inclined angle
$\lambda$
increases, the curved shock may expand and interact with the incident shock, leading to an asymmetric regular reflection, which is a phenomenon that has not been observed previously. The downstream structure is a double
$\lambda$
shock structure, with the lower shock resulting from the generation of recompression shock waves. Each
$\lambda$
shock structure is similar to type VI shock interference, though here at least one recompression shock is a strong shock.
6. Conclusion
In this paper, we consider the reflection of a rightward-moving shock over an attached oblique shock with post-formed expansion fan. The region having a post-formed expansion wave is given in the
$M_{S}$
–
$\lambda$
plane for
$M_{r}=6$
and
$\theta _{w}=10^{\circ }$
. We find that the reflection with the post-formed expansion fan occurs within a narrow band that lies above the region associated with the Post-I shock reflection.
Numerical simulations for cases with the critical condition with no post-formed wave (
$\lambda =\lambda _{cri}$
) show that both Mach reflection and regular reflection may occur. The transition condition for between these two types of reflection is obtained. When increasing the inclined angle
$\lambda$
of the incident shock from the critical value, a post-formed expansion wave is generated. Numerical simulation with a post-formed expansion wave for increasing
$\lambda$
shows that the unsteady shock reflection is complex. The complex shock reflection pattern is essentially composed of an upstream structure linked to the moving shock wave, and a downstream structure linked to the unperturbed oblique shock wave. The upstream structure is originally a Mach reflection structure, while the downstream structure comprises a double
$\lambda$
shock structure, with the lower one being due to the generation of recompression shock waves.
The upstream structure can be categorized. Hillier (Reference Hillier2007) studied the interaction of an upstream expansion fan with the shock reflection structure in a steady supersonic flow, and classified three types of configuration. He defined type-1 and type-2 by whether the triple point is behind or within the expansion wave. For type-3, the flow is unsteady and finally collapses due to roll-up of various slip surfaces. Similarly, the post-formed expansion fan may intersect with different parts of the incident shock or Mach stem, and can be classified a similar manner. Additionally, the shock structures observed in this paper can also be classified based on shock structure into three types: Mach reflection of an incident shock without a new triple point, Mach reflection with a new triple point and a curved shock below the triple point, and asymmetric regular reflection between the extended curved shock and incident shock.
The present study investigates the overtaking shock-on-shock interaction with post-formed expansion fan instead of post-formed shock wave. This situation may occur in a real application when one supersonic vehicle overtakes another upstream supersonic vehicle, under proper conditions of shock speed Mach number and shock inclined angle (measured from the reference frame attached to one body). Thus the present study expands our knowledge for unsteady shock reflection.
Acknowledgements.
The authors wish to thank the referees, whose comments and suggestions were very useful and valuable for improving the paper.
Funding.
This work was supported by the National Natural Science Foundation of China (grant no. 12402269). The authors also wish to acknowledge the support from Qinghang Scholar Funding Program and National Postdoctoral Researcher Funding Program (no. GZC20240846).
Declaration of interests.
The authors report no conflict of interest.





























































































