Linear stability analysis on supersonic streamwise vortices

Toshihiko Hiejima

Citation: Physics of Fluids (1994-present) 25, 114103 (2013); doi: 10.1063/1.4828055

View online: http://dx.doi.org/10.1063/1.4828055
View Table of Contents: http://scitation.aip.org/content/aip/journal/pof2/25/11?ver=pdfcov
Published by the AIP Publishing

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 PHYSICS OF FLUIDS 25, 114103 (2013)

Linear stability analysis on supersonic streamwise vortices

Toshihiko Hiejimaa)
Department of Aerospace Engineering, Osaka Prefecture University,
1-1 Gakuen-cho, Nakaku, Sakai, Osaka 599-8531, Japan
(Received 11 August 2013; accepted 16 October 2013; published online 5 November 2013)

In this study, the spatial growth rates of supersonic streamwise vortices were inves-
tigated by inviscid linear stability analysis. The freestream Mach numbers were 2.5,
5.0, and 7.5. In previous measurements taken to define the streamwise vortices, the
stagnation temperature profile of supersonic flows is approximately uniform. This
study found that the growth rate of vortices at the uniform stagnation temperature is
smaller than that of isentropic vortices. The instability properties of the streamwise
vortices can be explained by the ratio of the circulation to the axial velocity deficit,
and also by the Mach number. Moreover, it is found that the compressibility effect, by
which the instability reduces as the Mach number increases, is caused by the negative
energy arising from the entropy gradient of supersonic vortices that accompanies
the axial velocity deficit-like wake. From an energy perspective, the effect may rea-
sonably be correlated with the large density perturbations in supersonic flows. This
study also proposes a general convective Mach number for supersonic streamwise
vortices. The normalized growth rates are shown to be a function of convective Mach
number within the investigated range of ratio parameters.  C 2013 AIP Publishing

LLC. [http://dx.doi.org/10.1063/1.4828055]

I. INTRODUCTION
Efficient mixing and turbulence control of supersonic flows are essential for realizing supersonic
combustion ramjet (scramjet) engines for airbreathing hypersonic aircraft. To generate sufficient
firing pressure to maximize the thrust, the fuel and air must be completely mixed within the
combustor. The short time-scale of combustion requires rapid molecular mixing, which must be
accommodated in scramjet combustor design. However, numerous studies have shown that the growth
rate of the mixing layer decreases with increasing Mach number.1 In other words, compressibility can
strongly suppress the development of vortical activity under supersonic conditions. To overcome this
difficulty, enhancing the fuel and air mixing is of vital importance. The most promising approach for
escaping the compressibility effects and thereby improving mixing is the application of streamwise
vortices or swirling flows.
Supersonic streamwise vortices are generated in various ways from injector struts or device
configurations. Vortices are also generated by adding swirls to jets or wake flows. Vortex generation is
detailed in Gutmark et al.2 Experimental studies3, 4 have suggested that streamwise vortices increase
the entrainment rate and the spreading rate in supersonic flows. Furthermore, as determined from
hot-wire measurements5 conducted at Mach number 2.5, the flow fluctuation spectra in supersonic
streamwise vortices satisfy Kolmogorov’s −5/3 power law. These spectra suggest that many turbulent
eddies develop from the breakup of streamwise vortices. However, most of the existing literature
focuses on Mach numbers below 2−3, while streamwise vortices at higher Mach numbers remain
poorly understood. Ultimately, turbulent transient processes in supersonic streamwise vortices must
be clarified with high Mach numbers. Resolving the features of streamwise vortices, in particular,
elucidating how small perturbations evolve to turbulent transitions, will largely contribute to this

a) Electronic mail: [email protected]

1070-6631/2013/25(11)/114103/17/$30.00 25, 114103-1 

C 2013 AIP Publishing LLC

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-2 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

Spiral modes

Supersonic streamwise vortex

FIG. 1. Transition process of a supersonic streamwise vortex at M = 2.5; the vortical structure is described by an isosurface
of the second invariant of the velocity gradient tensor.

goal. As an example of the transition, a numerical simulation of an unstable streamwise vortex

developing spiral modes is shown in Fig. 1. Accordingly, as an early step, vortex evolution may
be investigated by linear stability analysis. In incompressible flows, the stability of streamwise
vortices is known to depend on the swirl number. In supersonic flows, however, it is not clear from
stability theory why streamwise vortices enhance mixing and combustion and break into small-scale
structures.
In this way, the stability and collapse processes of streamwise vortices are essential for un-
derstanding supersonic mixing and combustion. Although compressibility can strongly affect the
characteristics of vortices in high speed flows, the stability of supersonic streamwise vortices in
compressible swirling flows has received little attention. Stott and Duck6 investigated a compress-
ible model of a trailing-line vortex. Their inviscid stability analysis showed that vortex instability is
suppressed as the Mach number increases. Yin et al.7 studied the absolute and convective instabil-
ities of incompressible and high-Mach number vortices with axial velocity deficits. However, they
assumed uniform density, pressure, and temperature,6, 7 which are not applicable to supersonic flow.
Applying temporal stability analysis, Khorrami8 showed that the mixing capacity in supersonic flows
is improved by adding a weak azimuthal velocity component to an approximately parallel-streaming
jet flow. Lu and Lele9 studied the stability of an annular swirling mixing layer whose circulation
decreased with increasing radius r. They reported that the maximum amplification rate enhanced by
the swirl is sustained under compressible conditions. Müller and Kleiser10 conducted a large-eddy
simulation of compressible swirling flows at uniform temperature. They investigated linear instabil-
ity disturbances based on viscous spatial stability analysis11 to clarify vortex breakdown at Mach
number 0.6. Instability of compressible vortices at low Mach number have also been investigated
theoretically. For instance, from the asymptotic analysis12 of large-wavenumber disturbances, the
spatial structure of compressible eigenmodes is known to be similar to the incompressible modes
obtained by Leibovich and Stewartson.13 However, little is known about the general case of large
Mach number, in which the profiles are non-uniform. At high Mach number, the entropy and density
will exert non-negligible effects on the profile.
Some studies have used acoustic waves to evaluate entropy in compressible vortices. Chan et al.14
studied stabilities of two-dimensional inviscid compressible Rankine vortices. They demonstrated the
destabilizing influence of the positive entropy gradient in the vortex core. Men’shov and Nakamura15
modeled entropy stratification as a Gaussian profile and investigated the relationship between entropy
and vorticity intensity. Reporting on incompressible swirling flows with a heavy core, Sipp et al.16
noted that the density gradients introduce centrifugal instability and the Rayleigh-Taylor instability
to the vortex. To obtain profiles of compressible viscous vortices, Colonius et al.17 and Grasso and
Pirozzoli18 evolved the entropy profile from initially uniform entropy and density. They showed
that viscous effects induce weak radial velocities in the vortex, and that entropy changes over time
because of viscous dissipation and heat conduction. However, at sufficiently high Reynolds number,

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-3 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

these effects exert negligible influence over sonic time-scales and play no important role in vorticity
decay.
This study investigates instabilities in supersonic streamwise vortices by linear stability analysis.
Section II describes the basic flow of the vortices. The entropy gradient distributions are discussed
in Sec. III. Section IV formulates the linear stability analysis for the supersonic streamwise vortices.
The stability results at different Mach number are presented and discussed in Sec. V. Section VI
concludes the paper.

II. BASIC FLOW

In the present paper, we study the instability characteristics of supersonic streamwise vortices
with the axial velocity deficits (alias compressible swirling wake flows). The basic flow is assumed
to be steady and axisymmetric. In the cylindrical polar coordinates (r, θ , x), ur , uθ , ux , ρ, p, T, and S
represent the three components of velocity, density, pressure, temperature, and entropy, respectively.
It is known from experiments and numerical simulations that both the streamwise (namely, axial)
vorticity and the axial velocity can have Gaussian profiles, like incompressible flows. Thus, the axial
vorticity ωx∗ and the axial velocity u ∗x can be represented by Eqs. (1) and (2) (hereafter, dimensional
quantities appear with the superscript):
∗2
ωx∗ (r ∗ ) = ωmax
∗
e−Cr , (1)

  ∗2
u ∗x r ∗ = U∞
∗
−  Ux∗ e−D r , (2)
∗ ∗
where ωmax denotes the maximum axial vorticity, U∞ the freestream velocity,  Ux∗ the axial velocity
deficit. C and D are constant for the profiles. A measure of vortex radius, namely, the swirl thickness
δs∗ is given by

∗ 1
δs∗ = ∗
=√ , (3)
π ωmax C
where * represents the total circulation to the whole distributed axial vorticity. As shown in Eq. (2),
there exists a field of the azimuthal vorticity. The azimuthal vorticity thickness δω∗ θ can be defined as

u ∗ (∞) − u ∗x (0) e
δω∗ θ = x  = . (4)
∂u ∗x  2D
∂r max
Assuming that C = D we obtain. The azimuthal vorticity thickness is thus almost the same order as
the swirl thickness.
∗ ∗
Using the supersonic freestream velocity U∞ , the freestream density ρ∞ , and the swirl thickness
∗
δs , we can obtain the following nondimensional quantities:
ρ∗ u∗ u∗ u∗ p∗
ρ= ∗
, u r = r∗ , u θ = θ∗ , u x = x∗ , p = ∗ ∗2 ,
ρ∞ U∞ U∞ U∞ ρ∞ U∞
(5)
R∗ S∗ r∗ x∗ U∗ ∗
T = ∗ 2 T ∗, S = ∗ , r = ∗ , x = ∗ , t = ∞ t ,
U∞ Cv δs δs δs∗
where R* represents the gas constant, Cv∗ the specific heat at constant volume, and γ the ratio of
specific heats.
Then a freestream Mach number M and Reynolds number Re are given by
U∗ ∗
ρ∞ U∞ ∗ ∗
δs
M= ∞ , Re = ∗
, (6)
γ R ∗ T∞
∗ η∞
∗ ∗
where T∞ is the freestream temperature, and η∞ the viscosity.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-4 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

Experimental Compressible Velocity Profiles Experimental Compressible Velocity Profiles Cattafesta et al.(1992)
Cattafesta et al.(1992) Naughton et al.(1992)
1.5 2.0 1.5 Kalkhoran et al.(1998)
Naughton et al.(1992)

Azimuthal Velocity Uθ
Kalkhoran et al.(1998) Cutler et al.(1993)
Lee & Bershader (1994)

Γ/Γ∗
Metwally et al.(1989)
Axial Velocity Ux
Batchelor Vortex 1.0
(q-vortex) 1.0

Circulation
1.0 0 Cattafesta et al.(1992)
Naughton et al.(1992)
Kalkhoran et al.(1998) 0.5 Rankine Vortex
-1.0 Cutler et al.(1993) Batchelor Vortex
Batchelor Vortex (q-vortex)
(q-vortex) Self-similar
Compressible Vortex
0.5 -2.0 0
-4 -2 0 2 4 -4 -2 0 2 4 0.1 1 10
r r r/r*
(a) (b) (c)

FIG. 2. Comparison of theoretical model profiles with experimental measurements21–26 in supersonic streamwise vortices
for (a) axial velocity, (b) azimuthal velocity, and (c) normalized circulation profiles.

The azimuthal velocity uθ calculated from the axial vorticity ωx and the axial velocity ux are
derived from Eqs. (1) and (2)
q 
1 − e−r , q =
2
u θ (r ) = , (7)
r 2π

u x (r ) = 1 − μ e−r ,
2
(8)
where
∗  Ux∗ q
q= ∗ δ∗
, μ= ∗
, = . (9)
2π U∞ s U∞ μ
Here, q and μ, respectively, represent the circulation and the velocity deficit. indicates the ratio
of them. The three parameters are important on representing the present streamwise vortices. The
vortex is the same as the incompressible Batchelor vortex19 (i.e., q-vortex) that is commonly used
as a vortex model in stability studies. As shown in Fig. 2, the velocity profiles of Eqs. (7) and (8) are
consistent with measurements20–26 conducted for various supersonic conditions.
In supersonic flows, the Reynolds number is large enough. For the case of inviscid steady flow,
suppose ur = 0, so that the density ρ (r) and pressure p (r) can be determined from the following
equations, under the condition that ρ (∞) → 1 and p (∞) → 1/(γ M2 ):
dp uθ 2 dS 1 dp 1 dρ
=ρ , = −γ . (10)
dr r dr p dr ρ dr
Note that the value of the circulation should not violate the condition that the static pressure in the
vortex must be positive, p(0) > 0, with regard to the maximum of the circulation qmax
1
q < qmax = √ ,
M (γ − 1)I∞
(11)
1
∞
(1 − e−ξ ) e−(S(ξ )−S(∞))
2
γ

I∞ = dξ.
0 ξ3 γ

The swirl number Sw is used to represent the relative amount of swirl present on supersonic
freestream conditions
Jθ
Sw = ,
rs Jx
∞
Jθ = 2π ρ u x u θ r 2 dr, (12)
0
∞
Jx = 2π (ρu 2x + p) r dr,
0

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-5 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

where Jx is the axial momentum flux, Jθ is the axial component of angular momentum flux, and rs
is the radius of a streamwise vortex. The swirl ratio, SR = uθ, max /ux, max , is also used. As another
assessment of the swirl degree, the helix angle φ R is well known and it can compare the local
azimuthal velocity and the local axial velocity at the reference position

−1 u θ
φ R = tan . (13)
ux
The Sw , S R , and φ R , as well as q and μ, are useful variables in terms of the correspondence to
practical flows.

III. ENTROPY GRADIENT DISTRIBUTIONS

A perfect gas with constant specific heats is called a polytropic gas. Since real gases are closely
polytropic over wide temperature ranges, the gas model is a natural choice for air working in the range
of M < 5  7. Thus, for air, the normalized state equation and the enthalpy equation, respectively,
are
p = ρ T, (14)

γ −1 2
T = T0 − (u x + u 2θ ), (15)
2γ

γ −1 2
where T0 = 1 + M /(γ M 2 ) is the stagnation temperature, namely, the total temperature.
2
The Crocco’s equation is important in that it connects the velocity vorticity vector product with
state variables such as the entropy and the enthalpy as follows:
1 dS dh 0
T = − u θ ωx + u x ωθ , (16)
γ − 1 dr dr
γ
where h 0 = T0 is the enthalpy, ωx the axial vorticity, and ωθ the azimuthal vorticity.
γ −1
Delery,20 Metwally,21 and Cattafesta22 show that the stagnation temperature is nearly uniform
with the deficit being approximately 4%–8% by experiment, as shown in Fig. 3(a). For comparison,
the supposed isentropic and uniform stagnation temperature profiles are also described there, using
the velocity profiles Eqs. (7) and (8) fitted with the measurements. For the isentropic vortex, it is
found that the stagnation temperature deficit stands at 35% in the vortex center. Hence, we may
assume that the stagnation temperature is constant. On the right hand side of Eq. (16), provided that
T0 = constant, the first term is zero, the second and the third terms are negative so that the entropy
gradient should be negative. Then, it follows that the entropy profile illustrates a convex normal
[K]

Settles et al.(1993)
Supersonic streamwise vortex
Gaussian- fitting
0,
Stagnation Temperature T / T

Entropy distribution: S

1.0
0

1 1
dS/dr
Isentropic region

Measurements, M = 3.0 0 0
0.5 ; Cattafesta et al.(1992)
T = const. Entropy gradient,
0 using Gaussian-fitting
ΔS=0 -1 -1
Vortex core
0
0 1 2 3 4 0 1 2 3 4 5 6
y [mm] r
(a) (b)

FIG. 3. (a) Stagnation temperature and (b) entropy profiles of supersonic streamwise vortices for the measurement profiles3, 22
at M = 3.0 (note that (a) S = 0, the supposed stagnation temperature based on q = 0.2, μ = 0.29 fitted with the measurements).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-6 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

q = 0.32, μ = 0.5 q = 0.32, μ = 0.5

4
M = 0.01

Entropy gradient : dS/dr

M = 2.5 0
3 M = 5.0 M = 0.01

Entropy : S(r)
M = 2.5
2 -1 M = 5.0

1
-2
Total temperature: constant Total temperature: constant
0
-3
0 1 2 3 4 5 6 0 1 2 3 4 5 6
r r

FIG. 4. Entropy and entropy gradient distributions for streamwise vortices q = 0.32, μ = 0.5 at M = 0.01, 2.5, and 5.0,
using Eq. (17), given that stagnation temperature is uniform.

distribution. Figure 3(b) shows that this entropy is not inconsistent with the results of measurement.27
Therefore, for the case of uniform stagnation temperature, the entropy gradient can be expressed
from Eq. (16) as follows:

u 2θ du x du θ
−(γ − 1) + ux + uθ
dS r dr dr
=  . (17)
dr γ −1 2 (γ − 1)
1+ M /(γ M ) − 2 (u x + u θ )
2 2
2 2γ
Figure 4 shows the entropy and its gradient distributions of the streamwise vortices obtained by
Eq. (17). The entropy in the vortex core increases as the Mach number increases. Also the circulation
q has the similar effect on the entropy profile.

IV. LINEAR STABILITY ANALYSIS

We consider the inviscid linear stability of streamwise vortices in the cylindrical polar coordi-
nates. This assumption is reasonable since supersonic flows inevitably have a high Reynolds number.
Each flow variable is assumed to consist of its mean part ( ) and small amplitude perturbation (  ).
Utilizing normal-mode analysis, the perturbations are Fourier-decomposed. These flow variables are
written as follows:
⎧ ⎫ ⎧ ⎫
⎪
⎪ ρ (r, θ, x) ⎪ ⎪ ⎪ ρ (r ) ⎪ ⎧ R  (r, θ, x) ⎫
⎪
⎪ u (r, θ, x) ⎪ ⎪
⎪ ⎪ ⎪
⎪ ⎪ ⎪
⎪ ⎪ ⎪
⎪
⎪
⎪ ⎪ ⎪ ⎪
U (r ) ⎪ ⎪
⎪ U (r, θ, x) ⎪⎪
⎪
⎪
r ⎪
⎪ ⎨ ⎪
⎪ ⎪
⎪ ⎨⎪
⎪ ⎪
⎪
⎨ u θ (r, θ, x) ⎬ ⎬ V (r, θ, x) ⎬

V (r )
= +  ,
⎪
⎪ u x (r, θ, x) ⎪
⎪ ⎪
⎪ W (r ) ⎪
⎪ ⎪
⎪ W (r, θ, x) ⎪⎪
⎪
⎪ ⎪
⎪ ⎪ ⎪ ⎪ ⎪ ⎪
⎪
⎪
⎪ p (r, θ, x) ⎪ ⎪
⎪
⎪
⎪
⎪ p (r ) ⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
 (r, θ, x) ⎪
P ⎪
⎪
⎭
⎪
⎩ ⎪
⎭ ⎩ ⎭ 
S (r, θ, x)
S (r, θ, x) S (r )
    (18)
U
R, , V
, W
 , P, S = R, U, V, W, P, S † exp[i(αx + mθ − ωt)],

where m is the azimuthal wave number, ω the angular frequency, and α = αr + i αi , α r being
the axial wave number and α i the spatial growth rate. The entropy, axial, and azimuthal vorticity
perturbation eigenfunctions are derived from the following equations, respectively:
P R
S† = −γ , (19)
p ρ

dV V U dW
x = + −im , θ = i α U − . (20)
dr r r dr

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-7 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

A. The linearized perturbation equations

In the present study, we focus on the compressible Batchelor vortex which mostly amplifies
disturbances of inviscid unstable modes so that we may assume the flow to be inviscid as long as the
disturbance growth is concerned. Substituting Eq. (18) into the compressible Euler equations and
neglecting second-order terms in the perturbations, the linearized governing equations are obtained
in terms of perturbation eigenfunctions P and U
 2  2 
dP V 2mV 2V ω x V dS
= − P + i ρ −φ + − U, (21)
dr r c2 r 2φ rφ γ φr dr
 2
  2

dU i m2 φ mω x V 1 α dW
= α + 2 − 2 P+
2
− 2− + U, (22)
dr ρφ r c rφ rc r φ dr
     
1 p dS 1 m 1 α dW
R= 2 P−i U , V = − P + i ωx U , W = − P+i U , (23)
c φ dr φ ρr φ ρ dr

V dV V p
φ = −ω + m + αW , ω x = + , c2 = γ . (24)
r dr r ρ
As for the boundary conditions of the linearized equations, we use asymptotic solutions of
Eqs. (21) and (22) in the vicinity of the origin (r ≈ 0) and at infinity (r → ∞). Note that the
boundary condition at r = 0 depends on the azimuthal wave number.28 To solve the perturbation
equations, we first transform it to a Riccati formulation, focusing on  = P/U . Thus, the boundary
conditions can be expressed as follows:
at r  0, near the origin(the subscript 0 indicates r = 0),
⎧ ⎛ ⎞
⎪
⎪
⎪
⎪ ⎜ ⎟
⎪
⎪ ⎜ 4 φ0 ⎟1
⎪
⎪
⎪
⎪ −iγ p⎜
⎜φ −  ⎟
⎟r for m = 0
⎪
⎪
⎪ ⎝ 0 d V  ⎠
⎪
⎨ 
dr 
(r ) = 0 , (25)
⎪
⎪
⎪
⎪ −iρ φ 0 r for |m| = 1
⎪
⎪
⎪
⎪   
⎪
⎪
⎪
⎪ ρ |m| d V 
⎪
⎩ −i m m φ 0 − 2 dr  r
⎪ for |m| ≥ 2
0

at r → ∞, far away from the origin,

i ρ (ω − α)
(∞) = −  . (26)
α2 − M 2 (ω − α)2

B. The solution method of linearized equations

To determine the complex eigenvalue α, a shooting method is used to solve Eqs. (21) and (22)
numerically with boundary conditions (25) and (26). The transformed function  (r ) is integrated
for a specified value of α using the Runge-Kutta scheme from 0 to rm (1 (rm )), and from rmax to
rm (2 (rm )), at rm = 1. The Newton-Rapson method is used to search the eigenvalue α which satisfies
the condition F(α r , α i ) = |1 (rm ) − 2 (rm )| 10−8 . Note that if rmax > 20, the eigenvalues are
free of influence from rmax . Confirming eigenvalues, the maximum growth rates for linear stability
of streamwise vortices are shown in Table I. The temporal stability for the incompressible flow is

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-8 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

TABLE I. Validation of eigenvalues for stability of the Batchelor vortex.

M q μ m αr ωi, max a ω (m/α r ) (ω/α r )

0.0 0.8 1.0 −2 1.19 0.298b ... −1.68 ...

0.01 0.8 1.0 −2 1.19 0.304c 0.37c −1.68 0.31

aω
i : temporal growth rate.
b Temporal stability of an incompressible swirl flow, from Ref. 29.
c Temporal stability of a swirl flow, using the present code.

conducted to verify with a known value, and then the eigenvalue obtained agrees with that of Lessen
et al.29 (Note that spatial growth rates of the Batchelor vortex have never been found.)

C. A sufficient condition for compressible vortex stability

A few remarks should be made concerning characters for the compressible vortex stability.
Howard30 first derived a sufficient condition for stability to axisymmetric disturbances of com-
pressible inviscid swirling flows. Subsequently, the stability condition has been extended by several
researchers.12, 31, 32 According to Lalas,31 the sufficient condition for stability to all small perturba-
tions is
⎡ 2  2 ⎤
1 dW d V V
N2 − ⎣ + − ⎦ ≥ 0, (27)
4 dr dr r

where
2
V dS
N2 = − . (28)
γ r dr

N is the effective Brunt-Väisälä frequency.30 This condition suggests that the isentropic streamwise
vortex might not be stable. In general, the stability of compressible flow depends on the Mach
number. However, M is not specified explicitly in Eq. (27).

D. The total energy of perturbations

The energy integral of the perturbation equation provides more useful information.33 The total
energy of the perturbation is given by

2
 = 1 ρ (U
E 2 + V  2) + 1 P .
2 + W (29)
2 2 γP

 derived from the linearized perturbation equations is as follows:

The rate of change of E
∞ ∞  
D  r dr =
E  ux +  uθ +  S +  Ru r dr. (30)
Dt 0 0

Here, the azimuthal average is defined by

2π
1
ψ ≡ ψ dθ,
2π 0

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-9 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

where

W
 ux = −ρ U  dW ,
dr
 
V d V V
 uθ = −ρ U − ,
dr r
(31)
1  d P
S =− U S ,
γ dr
1 ∂ P ).
 Ru = − (r U
r ∂r
The first term  ux expresses the perturbation product due to the axial velocity, the second term  uθ
the product due to the azimuthal velocity, the third term  S the product due to the entropy profile,
and the last term is involved with the redistribution.9 These terms represent a transfer of energy
between the basic flow and the perturbations. The terms  ux and  uθ have a part which is similar
to the Reynolds stress tensor. Thus, we can understand the source of the perturbation energy from
this equation.

E. Unstable modes for incompressible swirling flows

The instability of the Batchelor vortex has been investigated by many researchers for incom-
pressible flows.29, 34–37 There are two kinds of unstable modes. The first is strong “inviscid helical
modes,” which have negative azimuthal wavenumbers and low values of the parameter < 2.3 with
highly unstable (note that they become weak when > 1.6). The second is weak “viscous modes,”
which are present for all values of the parameter at sufficiently high Reynolds number. The growth
rates of the mode are much smaller than that of the helical modes and are not considered here.

V. RESULTS AND DISCUSSION

A. Effects of velocity profile parameters on streamwise vortices
To date, almost all studies of incompressible and compressible swirling flows have focused
on jets. Swirling flows are accompanied by streamwise vortices, whose stability characteristics
solely depend on the circulation (equivalent to the swirl intensity). However, since many streamwise
vortices generated in supersonic flows take the form of swirling wakes, the effect of velocity deficits
must be considered. As shown in Table II, the conventional parameter Sw or SR cannot distinguish
differences in the velocity deficit. Therefore, we newly investigate the relationship between the axial

TABLE II. Parameters of supersonic streamwise vortices.

M q μ Sw SR φ R (deg)

2.5 0.16 0.2 0.80 0.143 0.102 6.1

2.5 0.16 0.3 0.53 0.143 0.102 6.4
2.5 0.16 0.5 0.32 0.143 0.102 7.0
2.5 0.32 0.5 0.64 0.286 0.204 13.8
2.5 0.48 0.5 0.96 0.430 0.306 20.3
5.0 0.16 0.2 0.80 0.155 0.102 6.1
5.0 0.16 0.3 0.53 0.155 0.102 6.4
5.0 0.16 0.5 0.32 0.155 0.102 7.0
5.0 0.32 0.5 0.64 0.311 0.204 13.8
7.5 0.16 0.2 0.80 0.158 0.102 6.2
7.5 0.16 0.3 0.53 0.158 0.102 6.5
7.5 0.16 0.5 0.32 0.158 0.102 7.0

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-10 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

M = 0.01, q = 0.16, μ = 0.5: T = const M = 0.01, q = 0.16, μ = 0.3: T = const M = 0.01, q = 0.32, μ = 0.5: T = const
0.4 0 0.24 0 0.4 0
m = -10
m = -6 m = -8
m = -5
m = -10 m = -4
0.3 0.18 m = -6 m = -8 0.3 m = -3
m = -1 m = -16 m = -5
m = -4 m = -2
-α m = -2 m = -12 -α m = -3 -α m = -1
i m = -10 i m = -2 i
m = -3
0.2 m = -4 m = -8 0.12 m = -1 0.2
m = -5 m = -6

0.1 0.06 0.1

0 0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7
ω ω ω
M = 2.5, q = 0.16, μ = 0.5: T = const M = 2.5, q = 0.16, μ = 0.3: T = const M = 2.5, q = 0.32, μ = 0.5: T = const
0.4 0 0.24 0 0.4 0

0.3 0.18 0.3 m = -10

m = -10 m = -6 m = -8
m = -1 m = -6 m = -8 m = -5
-α m = -2 -α m = -5 -α m = -4
m = -3
i i m = -4 i
m = -3 m = -12 m = -3 m = -2
0.2 m = -4 m = -10 0.12 m = -2 0.2 m = -1
m = -5 m = -6 m = -8
m = -1

0.1 0.06 0.1

0 0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7
ω ω ω
M = 5.0, q = 0.16, μ = 0.5: T = const M = 5.0, q = 0.16, μ = 0.3: T = const M = 5.0, q = 0.32, μ = 0.5: T = const
0.3 0 0.18 0 0.3 0

-α 0.2 -α 0.12 -α 0.2

i m = -1 i i
m = -2 m = -2 m = -2
m = -3 m = -1 m = -3
m = -4 m = -1 m = m
-3
m = -4 m = -5 = -4
m = -5 m = -6 m = -8 m = -5
0.1 m = -6 0.06 m = -10
0.1 m = -6
m = -8 m = -8
m = -10 m = -10

0 0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7
ω ω ω
(a) Φ = 0.32 (b) Φ = 0.53 (c) Φ = 0.64

FIG. 5. Spatial growth rates −α i as a function of frequency ω for M = 0.01, 2.5, and 5.0.

velocity deficit and the vortex growth rates. We consider the effect of both the circulation q and the
axial velocity deficit μ, whose ratio is defined by , on the instability.
Figure 5 shows the spatial growth rate −α i , as a function of angular frequency ω for = 0.32,
0.53, and 0.64 at M = 0.01, 2.5, and 5.0. The ratio of the streamwise vortices affects the most
unstable mode of the azimuthal wavenumber m. First, at small Mach numbers, the linear instability
characteristics (except for the magnitude of the growth rate) depend on : (i) when is small, the
unstable mode m = −1 predominates, as in jet or wake flows; (ii) the growth rates increase and
the frequency bands spread as increases; and (iii) when is large, the most unstable mode is a
high mode of |m|. It has been demonstrated that maximum growth rates monotonically increase with
|m| under inviscid conditions in incompressible flows.13 Thus, at small Mach number (M < 2.5),
the instability properties of compressible and incompressible fluids appear very similar.29 Second,
the growth rates at large Mach number exhibit unique features. When is small, the growth rate
retains m = −1 regardless of Mach number. However, as increases, the low wavenumber m =
−2 dominates in unstable modes and the growth rate reduces drastically. For example, when =
0.64, the growth rate at M = 5.0 is considerably smaller than at M = 0.01. The important point is
that, at high Mach number, the most unstable mode is not a high wavenumber mode. (When M =
5.0, the most unstable mode is m = −1 for = 0.32, m = −2 for = 0.53, and m = −3 for =
0.80.) This result is unique because high wavenumber instabilities are considered to be responsible
for the transition from large-scale coherent structures to small-scale eddies. The compressible effect
might contribute to the decreased growth rate in the presence of high-wavenumber perturbations.
Regarding the shear-type instability in incompressible flows, if the Reynolds number is small, the

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-11 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

M = 2.5, q = 0.16, μ = 0.5, m = -2 M = 5.0, q = 0.16, μ = 0.5, m = -2 M = 7.5, q = 0.16, μ = 0.5, m = -2

0.4 0.4 0.4
S = const S = const S = const
ρ (r), p(r) = const ρ (r), p(r) = const ρ (r), p(r) = const
0.3 T0= const 0.3 T0= const 0.3 T0= const
-α Stagnation -α Stagnation -α Stagnation
i i i
temperature temperature temperature
0.2 deficit: 8 % 0.2 deficit: 8 % 0.2 deficit: 8 %

0.1 0.1 0.1

0 0 0
0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0
ω ω ω
(a) M = 2.5 (b) M = 5.0 (c) M = 7.5

FIG. 6. Growth properties of the streamwise vortices obtained by making a comparison between the isentropic flow, the
density and pressure constant, the stagnation temperature uniform, and the stagnation temperature deficit of 8%; for =
0.32 (q = 0.16, μ = 0.5), m = −2.

viscous effect prevents perturbations from developing, so that the large wavenumbers are not usually
present in the mode. Thus, it is remarkable that the Mach number effect operates as that of the
viscous effect.
As seen at high , if a single mode can grow without being outcompeted by multiple modes,
the mode could symmetrically evolve in the flow. In this case, the flow could potentially become
metastable in the nonlinear phase.35 In contrast, if is small, multiple modes compete to dominate
the growth rates. In unstable systems, the streamwise vortices may collapse to the extent that the
unstable mode is rivaled by many other modes. This suggests a connection between increased mixing
capacity and the streamwise vorticity effect, namely, the effect of adding swirls to a jet or wake flow.
Accordingly, the merits of supersonic streamwise vortices over jets or wakes become clarified. In
this way, varying is important for identifying mode selection during a disturbance.

B. Effects of entropy gradient on streamwise vortices

We also make sure from stability results that a uniform stagnation temperature profile is rea-
sonable. Figure 6 shows the stability properties of four classes of vortices: the isentropic, the
density-pressure uniform,6, 7 the case of constant stagnation temperature, and a stagnation temper-
ature deficit3 of 8% at the vortex center. Results are shown for = 0.32 (q = 0.16, μ = 0.5) and
m = −2 at M = 2.5, 5.0, and 7.5. For the first three vortex classes, the growth rate and frequency
band are significantly altered with increasing Mach number. Note that the growth rate of isentropic
vortices, which are often assumed in theoretical models, is overestimated. This agrees with the
sufficient condition for stability (Eqs. (27) and (28)) in that the negative entropy gradient contributes
to stabilization. From the result, it appears that the compressibility effects, by which the instability
reduces with increasing Mach number, are related to the entropy gradient within the vortex core. On
the contrary, the actual vortex with the stagnation temperature deficit is nearly identical to the case
of constant stagnation temperature in stability property. Accordingly, it seems reasonable to assume
a uniform stagnation temperature profile.

C. Perturbation properties of supersonic streamwise vortices

Figure 7 shows the perturbation amplitudes of the axial vorticity |x |, azimuthal vorticity
|θ |, and entropy |S† |, for m = −2 and three values of at M = 0.01, 2.5, and 5.0. From the
perturbation structures, we observe that when |θ | > |x |, the axial velocity deficit induces a shear-
type instability, and when |x | > |θ |, the instability is based on streamwise vortices. Moreover, the
amplitude of the entropy perturbation |S† | grows as both M and increase. The existence domain
of the perturbations narrows with increasing M. The unstable feature for and M is schematically
summarized in Fig. 8. At high Mach number, the amplitude of the axial vorticity perturbation is
weakened, unlike the incompressible case in which the axial vorticity fluctuation can dominate as
increases. Although the entropy perturbation becomes relatively stronger as the Mach number

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-12 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

M = 0.01, q = 0.16, μ = 0.5, m = -2 M = 0.01, q = 0.32, μ = 0.5, m = -2 M = 0.01, q = 0.16, μ = 0.2, m = -2

1.5 |Ω | 1.5 |Ω | 1.5 |Ω |

Amplitude distributions

Amplitude distributions
Amplitude distributions
x x x

|Ω | |Ω | |Ω |
θ θ θ

1.0 |S | 1.0 |S | 1.0 |S |

0.5 0.5 0.5

0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
r r r

M = 2.5, q = 0.16, μ = 0.5, m = -2 M = 2.5, q = 0.32, μ = 0.5, m = -2 M = 2.5, q = 0.16, μ = 0.2, m = -2

1.5 |Ω | 1.5 |Ω | 1.5 |Ω |
Amplitude distributions

Amplitude distributions

Amplitude distributions
x x x
|Ω | |Ω | |Ω |
θ θ θ

1.0 |S | 1.0 |S | |S |
1.0

0.5 0.5 0.5

0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
r r r

M = 5.0, q = 0.16, μ = 0.5, m = -2 M = 5.0, q = 0.32, μ = 0.5, m = -2 M = 5.0, q = 0.16, μ = 0.2, m = -2

1.5 |Ω | 1.5 |Ω | 1.5 |Ω |
Amplitude distributions

Amplitude distributions

Amplitude distributions
x x x

|Ω | |Ω | |Ω |
θ θ θ

1.0 |S | 1.0 |S | 1.0 |S |

0.5 0.5 0.5

0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
r r r

(a) Φ = 0.32 (b) Φ = 0.64 (c) Φ = 0.80

FIG. 7. Amplitude profiles of disturbances; axial vorticity |x |, azimuthal vorticity |θ |, and entropy |S† | for m = −2 at M
= 0.01, 2.5, and 5.0.

increases, the total fluctuation intensities weaken, which should also reduce the velocity fluctuations.
Thus, it is probable that stabilization is a natural consequence of decreased velocity fluctuations.
According to Fig. 6, the growth property of an isentropic vortex is close to that of a vortex
at uniform stagnation temperature when M < 2.5. However, the two vortex classes might widely
differ in their perturbation structure. Figure 9 plots the eigenfunctions of the perturbations of density

shear instability region

due to an axial velocity deficit
M
|Ω θ | > |Ω x |
unstable region
based on streamwise vortex
|Ω θ | < |Ω x |

FIG. 8. Conceptual diagram of instability features for vs. M.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-13 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

Phase R Phase R
Phase U Phase U
1.5 1.5 2π Phase V 2π Phase V
|R | |R |
Amplitude distributions

Amplitude distributions
Phase W Phase W
|U | |U | Phase P Phase P
|V | |V |

Phase distributions

Phase distributions
1.0 |W| 1.0 |W|
π π
|P | |P |

0.5 0.5 0 0

0 0 -π -π
0 1.0 2.0 0 1.0 2.0 0 1.0 2.0 0 1.0 2.0
r r r r

(a) T0 = const (b) S = const (c) T0 = const (d) S = const

FIG. 9. Eigenfunction profiles of the disturbance; density R, velocities U, V, W , and pressure P, for M = 2.5, m = −2,
= 0.64 (q = 0.32, μ = 0.5), (a) and (b) amplitude, (c) and (d) phase.

R; velocities U, V, W ; and pressure P for the cases of constant T0 and S with m = −2, =
0.64, and M = 2.5. In both cases, the dominant perturbation is the axial velocity perturbation
|W | and the phase profiles change near the inflectional point of the axial velocity. In addition, |U|
and |V | are large inside the core of the isentropic vortex. These large velocity perturbations are
likely associated with an instability imposed by the large stagnation temperature deficit in the core
as shown in Fig. 3(a). It is noteworthy that the density perturbation |R| dominates in the case of
uniform stagnation temperature. This major |R| is consistent with the density fluctuations reported
in supersonic experiments. It follows that the density perturbation amplifies while the velocity
perturbations weaken. Moreover, we can surmise that one influence of the compressibility on the
flow is the energy transfer from kinetic energy to internal energy. Clearly, velocity perturbations are
related to kinetic energy, and density perturbations are linked to the internal energy. Therefore, this
result may be relevant to the compressibility effect, in terms of the total perturbation energy. This
concept will be considered next.
Figure 10 shows the profiles of the source term of the perturbation energy obtained from Eq. (31).
Since the term  ux is overall large, the perturbation product based on the axial velocity deficit exerts
a clear dominant influence on the streamwise vortices. The instability causes a spiral mode with a
negative azimuthal wave to develop in the vortex core. According to Mao and Sherwin,38 these modes
correspond to core modes (i.e., helical modes) in the energy distribution. When = 0.64 and 0.80,
the negative perturbation energy  S imposed by the entropy profile enlarges as the Mach number
increases, so that the total energy invested in the perturbation decreases and the region of perturbation
energy narrows at high Mach number. Consequently, the energy of the generated perturbations is
suppressed when and M are large. (Since isentropic vortices lack negative perturbation energy;
 S = 0, they can retain high growth rates at high Mach numbers.) This result indicates that the
entropy profile in the vortex core influences stabilization. On the other hand, when = 0.32, the
entropy-induced negative perturbation energy is slightly affected by increasing the Mach number.
In this case, the Mach number has little impact on perturbation energy production. We observe that
when is small, streamwise vortices relieve the compressibility effect to some extent.
As mentioned above, both M and are important factors of compressibility effects in supersonic
streamwise vortices. Figure 11 plots the normalized maximum growth rates as functions of the ratio
at various Mach numbers. This figure reveals that an optimal exists in supersonic flows. In
terms of the growth rate, the unstable modes (situated between the two dotted lines) can effectively
enhance supersonic mixing. For instance, the streamwise vortex with = 0.32 at M = 7.5 maintains
approximately 50% growth rate, compared to the corresponding incompressible vortex. In summary,
the features of growth rates are divisible into three categories: (i) the bending mode due to m = −1,
and when is small, viz., a streamwise vortex is weak, this property characterizes a jet or a wake
flow without the streamwise vortex component; (ii) the shear-type unstable mode that depends on
the intensity of the streamwise vortices, whereby moderate markedly affects the growth rates; and
(iii) the weak unstable mode, which behaves like a viscous mode, and the growth rate dramatically
declines at high . (Note that category (iii) modes also impact the upper limit of the parameter q.)
Additionally, since the magnitude of the growth rate decreases to a low value irrespective of in
hypersonic conditions, additional instability elements are needed, such as isentropic vortices.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-14 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

Source terms of Disturbance Energy

Source terms of Disturbance Energy

Source terms of Disturbance Energy

M = 0.01, q = 0.16, μ = 0.5, m = -1 M = 2.5, q = 0.16, μ = 0.5, m = -1 M = 5.0, q = 0.16, μ = 0.5, m = -1
1.5 1.5 1.5
Ψ Ψ Ψ
ux ux ux
Ψ Ψ Ψ
1.0 uθ 1.0 uθ 1.0 uθ
Ψ Ψ Ψ
s s s

0.5 0.5 0.5

0 0 0

-0.5 -0.5 -0.5

0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
r r r
(a) Φ = 0.32
Source terms of Disturbance Energy

Source terms of Disturbance Energy

Source terms of Disturbance Energy

M = 0.01, q = 0.32, μ = 0.5, m = -2 M = 2.5, q = 0.32, μ = 0.5, m = -2 M = 5.0, q = 0.32, μ = 0.5, m = -2
1.5 1.5 1.5
Ψ Ψ Ψ
ux ux ux
Ψ Ψ Ψ
1.0 uθ 1.0 uθ 1.0 uθ
Ψ Ψ Ψ
s s s

0.5 0.5 0.5

0 0 0

-0.5 -0.5 -0.5

0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
r r r
(b) Φ = 0.64
Source terms of Disturbance Energy

Source terms of Disturbance Energy

Source terms of Disturbance Energy

M = 0.01, q = 0.16, μ = 0.2, m = -2 M = 2.5, q = 0.16, μ = 0.2, m = -2 M = 5.0, q = 0.16, μ = 0.2, m = -2
1.5 1.5 1.5
Ψ Ψ Ψ
ux ux ux
Ψ Ψ Ψ
1.0 uθ 1.0 uθ 1.0 uθ
Ψ Ψ Ψ
s s s

0.5 0.5 0.5

0 0 0

-0.5 -0.5 -0.5

0 1 2 3 4 0 1 2 3 4 0 1 2 3 4
r r r
(c) Φ = 0.80

FIG. 10. Source terms to the perturbation total energy, ux , uθ , and s .

D. Convective Mach number of supersonic streamwise vortices

This study organizes the relationships between the growth rate and the compressibility effect
on streamwise vortices. The compressibility effect in turbulent mixing is chiefly governed by the
convective Mach number.1, 39, 40 The convective Mach number in swirling flows is discussed by
Naughton et al.,4 who estimated the growth rates of compressible jets by adding swirls to the flow.
Normalized Spatial growth rate

1.5
M = 0.01

M = 2.5
1.0
M = 5.0

M = 7.5
0.5

0
0 0.5 1.0 1.5 2.0
Φ=q/μ

FIG. 11. Normalized maximum growth rates as a function of the ratio for various Mach numbers.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-15 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

Supersonic Batchelor vortices Langley experimental curve

Normalized Maximum Growth Rate:

Normalized Maximum Growth Rate:
Day et al. (1998)
Φ = 0.26 Papamoschou & Roshko (1988)
1.0 Goebel & Dutton (1991)
1.0 Φ = 0.32 Clemens & Mungal (1995)
Φ = 0.40 Naughton et al. (1997)

i M=0
i M=0

Φ = 0.48 Naughton et al. (1997);

weak swirling jet (0.02 < S < 0.10)

(-α )/ (-α )
Φ = 0.53
(-α )/ (-α )

Present study; streamwise vortices

Φ = 0.64
0.5 Φ = 0.80
0.5

i
i

0 0
0 1.0 2.0 0 1.0 2.0
Convective Mach number Mc Convective Mach number Mc
(a) (b)

FIG. 12. Normalized maximum growth rates as a function of convective Mach number Mc ; (a) the eigenvalues of supersonic
streamwise vortices with the uniform stagnation temperature, (b) the comparison of a single curve organized the vortices
(a) with measurements of supersonic mixing layers1, 41, 42 (- - - broken line, linear stability analysis of a mixing layer;44
- - - - dashed line, Langley experimental curve40 ) and compressible swirling flows.4

In a different way, we consider the convective Mach numbers of streamwise vortices from the same
perspective as mixing layers. As spiral modes develop, large-scale structures form in the streamwise
vortex. Thus, we may naturally introduce a convective frame of reference that travels with these
structures. To this end, we define Mr as the Mach number in the direction of the wavenumber vector
relative to the phase velocity as follows:
|W (r ) cos  + V (r ) sin  − C ph |
Mr (r ) = , (32)
a(r )

ωr αr
C ph = cos , cos  =  , (33)
αr αr2 + (m/r )2
where Cph is the phase velocity of the perturbations, tan  is the ratio of m/r to α r for the two
wavenumbers, and a(r) is the speed of sound. In the traveling reference frame, the relative Mach
numbers are defined, respectively, with respect to the reference position of the basic flow Mc1 =
Mr (r† ), and the freestream far from the origin Mc2 = Mr (∞). The position of the maximum amplitude
in perturbation is suitable for r† . Accordingly, we define the general convective Mach number Mc as
the geometric average of the two Mach numbers

Mc = Mc1 · Mc2 . (34)
Assuming that the circulation q is zero, this definition is consistent with that of Bogdanoff39 with
zero azimuthal velocity.
Figure 12 plots the maximum growth rate, normalized by the corresponding incompressible
growth rate, versus Mc for the case of uniform stagnation temperature. The effects of the streamwise
vortices on compressibility are demonstrated in Fig. 12(a). Using Eq. (34) within the range of
(Fig. 11), the growth rates can be summarized as a single curve. On the basis of this result, it is
important to compare the normalized maximum growth rates of streamwise vortices with those
of well-known mixing layers.1, 41, 42 In mixing layers, the most unstable mode identified by linear
stability analysis relates to the layer growth rate;43 specifically, −α i, max ∝δ  (x), δ  (x) = d(δ)/dx,
where δ is the experimental local layer thickness. If both sides are normalized by the growth rate
corresponding to an incompressible case, we obtain
[−αi,max ] M δ  (x) M
=  . (35)
[−αi,max ] M=0 δ (x) M=0
Using Eq. (35), we can compare the results of linear stability analysis with experimental data.
Figure 12(b) shows that, in the range = 0.26–0.80, the growth property of streamwise vortices is

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-16 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

stronger than in experimental mixing layers. Furthermore, a streamwise vortex with a weak swirling
jet,4 = 0.02–0.10, is superior to straight jets and mixing layers. This finding is partly attributable
to growth of 3D instability waves incited by streamwise vortices and swirling flows. Hence, although
streamwise vortices can reduce the compressibility effects on perturbation growth, the ratio must
be selected appropriately.

VI. CONCLUSIONS
This study has investigated instability in streamwise vortices, with a view to enhance supersonic
mixing and improving combustion efficiency. Experiments2, 3, 5 have demonstrated that streamwise
vortices can effectively enhance mixing, even in supersonic flows. Therefore, to effectively apply
vortex control, we must elucidate the effects of circulation, axial velocity deficit, and entropy
gradient of the vortices on the instability characteristics. Previous measurements3, 22 have shown that
the stagnation temperature profile of supersonic streamwise vortices is approximately constant. Thus,
Batchelor-type vortices with uniform stagnation temperature profiles were assumed as basic flows
in this paper. The inviscid linear stability of supersonic streamwise vortices was analyzed at various
Mach numbers. The instability of streamwise vortices depends on the profiles of three variables: V ,
W , and S. The first and second of these variables, the azimuthal velocity (axial vorticity) and the
axial velocity, respectively, have been actively researched in incompressible vortices,34–37 although
the effect of the velocity deficit has been less investigated. This paper examined the first and second
profiles in supersonic vortices, in addition to the entropy profile, which is important in high-speed
systems.
First, the relationship between the growth rate and , the ratio of the circulation to the axial
velocity deficit, was investigated in detail. At high Mach number, the instability growth rate remains
moderate if is small, but reduces considerably (relative to that of subsonic vortices) as increases.
This finding is quite contrary to reports of incompressible flows. Instability was also found to be
suppressed by the magnitude of the entropy gradient, derived from the stagnation temperature, and
the selected growth modes varied with and M. In addition, the growth rate of the isentropic vortex,
which is often adopted as a model vortex, was higher than that of real vortices with a uniform
stagnation temperature profile.
Second, the vorticity perturbations indicate that when |θ | > |x |, the dominant instability
is a shear-type instability introduced by the axial velocity deficit, whereas when |x | > |θ |, the
streamwise vortices induce the dominant instability due to the helical velocity field. As may be
reasonably expected, the perturbation region narrows as the Mach number increases. Moreover,
the amplitude of density perturbation increases and the entropy perturbation grows with increasing
Mach number, consistent with previous measured trends. Furthermore, by investigating the source
term of the perturbation energy, we clarified that the compressibility effect is caused by the negative
energy arising from the entropy profile.
Finally, the study demonstrated that the growth rates of supersonic streamwise vortices can be
organized in terms of the general convective Mach number, newly defined within = 0.26–0.80.
The convective Mach number clarifies that the growth rates of streamwise vortices exceed those of
well-known mixing layers and jets. Therefore, it follows that streamwise vortices can reduce the
effects of compressibility on perturbation growth.

ACKNOWLEDGMENTS
The author gratefully acknowledges helpful discussions with Emeritus Professor M. Nishioka
on several points in the paper. This work is partly supported by a Grant-in-Aid for Scientific Research
(Grant Nos. 20560739 and 21226020) from the Ministry of Education, Culture, Sports, Science and
Technology, Japan.
1 D. Papamoschou and A. Roshko, “The compressible turbulent shear layer experimental study,” J. Fluid Mech. 197, 453–477

(1988).
2 E.J. Gutmark, K. C. Schadow, and K. H. Yu, “Mixing enhancement in supersonic free shear flows,” Annu. Rev. Fluid
Mech. 27, 375–417 (1995).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27
 114103-17 Toshihiko Hiejima Phys. Fluids 25, 114103 (2013)

3 G. S. Settles, “Supersonic mixing enhancement by vorticity for high-speed propulsion,” Report No. NASA CR-188920,
1991.
4 J. W. Naughton, L. N. Cattafesta, and G. S. Settles, “An experimental study of compressible turbulent mixing enhancement

in swirling jets,” J. Fluid Mech. 330, 271–305 (1997).

5 M. Nishioka, S. Sakaue, K. Komada, H. Sakoshi, and I. Furukawa, “On the mixing transition in supersonic streamwise

vortices,” in Proceedings of the IUTAM Symposium on Elementary Vortices and Coherent Structures: Significance in
Turbulence Dynamics (Springer, 2006), pp. 249–258.
6 J. A. K. Stott and P. W. Duck, “The stability of a trailing-line vortex in compressible flow,” J. Fluid Mech. 269, 323–351

(1994).
7 X.-Y. Yin, D.-J. Sun, and M.-J. Wei, “Absolute and convective instability character of slender viscous vortices,” Phys.

Fluids 12, 1062–1072 (2000).

8 M. R. Khorrami, “Stability of a compressible axisymmetric swirling jet,” AIAA J. 33, 650–658 (1995).
9 G. Lu and S. K. Lele, “Inviscid instability of compressible swirling mixing layers,” Phys. Fluids 11, 450–461 (1999).
10 S. B. Müller and L. Kleiser, “Large-eddy simulation of vortex breakdown in compressible swirling jet flow,” Comput.

Fluids 37, 844–856 (2008).

11 S. B. Müller and L. Kleiser, “Viscous and inviscid spatial stability analysis of compressible swirling mixing layers,” Phys.

Fluids 20, 114103 (2008).

12 S. Leblanc and A. Le Duc, “The unstable spectrum of swirling gas flows,” J. Fluid Mech. 537, 433–442 (2005).
13 S. Leibovich and K. Stewartson, “A sufficient condition for the instability of columnar vortices,” J. Fluid Mech. 126,

335–356 (1983).
14 W. M. Chan, K. Shariff, and T. H. Pulliam, “Instabilities of two-dimensional inviscid compressible vortices,” J. Fluid

Mech. 253, 173–209 (1993).

15 I. Men’shov and Y. Nakamura, “Instability of isolated compressible entropy-stratified vortices,” Phys. Fluids 17, 034102

(2005).
16 D. Sipp, D. Fabre, S. Michelin, and L. Jacquin, “Stability of a vortex with a heavy core,” J. Fluid Mech. 526, 67–76 (2005).
17 T. Colonius, S. K. Lele, and P. Moin, “The free compressible viscous vortex,” J. Fluid Mech. 230, 45–73 (1991).
18 F. Grasso and S. Pirozzoli, “Asymptotic scaling of a decaying turbulent compressible vortex,” Phys. Fluids 11, 1636–1649

(1999).
19 G. K. Batchelor, “Axial flow in trailing line vortices,” J. Fluid Mech. 20, 645–658 (1964).
20 J. M. Delery, “Aspects of vortex breakdown,” Prog. Aerosp. Sci. 30, 1–59 (1994).
21 O. Metwally, G. S. Settles, and C. C. Horstman, “An experimental study of shock wave/vortex interaction,” AIAA Paper

No. 89-0082, 1989.

22 L. N. Cattafesta and G. S. Settles, “Experiments on shock/vortex interaction,” AIAA Paper No. 92-0315, 1992.
23 I. M. Kalkhoran, M. K. Smart, and F. Y. Wang, “Supersonic vortex breakdown during vortex/cylinder interaction,” J. Fluid

Mech. 369, 351–380 (1998).

24 J. W. Naughton and G. S. Settles, “Experiments on the enhancement of compressible mixing via streamwise vorticity,

Part 1 – Optical measurements,” AIAA Paper No. 92-3549, 1992.

25 A. D. Cutler, B. S. Levey, and D. K. Kraus, “An experimental investigation of supersonic swirling jets,” AIAA Paper No.

93-2922, 1993.
26 S. Lee and D. Bershader, “The structure of compressible starting vortices,” Exp. Fluids 16, 248–254 (1994).
27 G. S. Settles and L. Cattafesta, “Supersonic shock wave/vortex interaction,” Report No. NASA CR-192917, 1993.
28 G. K. Batchelor and A. E. Gill, “Analysis of the stability of axisymmetric jets,” J. Fluid Mech. 14, 529–551 (1962).
29 M. Lessen, P. J. Singh, and F. Paillet, “The stability of a trailing line vortex. Part 1. Inviscid theory,” J. Fluid Mech. 63,

753–763 (1974).
30 L. N. Howard, “On the stability of compressible swirling flow,” Stud. Appl. Math. 52, 39–43 (1973).
31 D. P. Lalas, “The Richardson criterion for compressible swirling flows,” J. Fluid Mech. 69, 65–72 (1975).
32 K. S. Eckhoff and L. Storesletten, “A note on the stability of steady inviscid helical gas flow,” J. Fluid Mech. 89, 401–411

(1978).
33 E. G. Broadbent and D. W. Moore “Acoustic destabilization of vortices,” Philos. Trans. R. Soc. London 290, 353–371

(1979).
34 S. Ragab and M. Sreedhar, “Numerical simulation of vortices with axial velocity deficits,” Phys. Fluids 7, 549–558 (1995).
35 I. Delbende, J.-M. Chomaz, and P. Huerre, “Absolute/convective instabilities in the Batchelor vortex: A numerical study

of the linear impulse response,” J. Fluid Mech. 355, 229–254 (1998).

36 L. Jacquin and C. Pantano, “On the persistence of trailing vortices,” J. Fluid Mech. 471, 159–168 (2002).
37 M. Abid, “Nonlinear mode selection in a model of trailing line vortices,” J. Fluid Mech. 605, 19–45 (2008).
38 X. Mao and S. J. Sherwin, “Spectra of swirling flow,” in Proceedings of the 7th IUTAM Symposium on Laminar-Turbulent

Transition (Springer, 2010), pp. 247–252.

39 D. Bogdanoff, “Compressibility effects in turbulent shear layers,” AIAA J. 21, 926–927 (1983).
40 C. Pantano and S. Sarkar, “A study of compressibility effects in the high-speed turbulent shear layer using direct simulation,”

J. Fluid Mech. 451, 329–371 (2002).

41 N. T. Clemens and M. G. Mungal, “Large-scale structure and entrainment in the supersonic mixing layer,” J. Fluid Mech.

284, 171–216 (1995).

42 S. G. Goebel and J. C. Dutton, “Experimental study of compressible turbulent mixing layers,” AIAA J. 29, 538–546 (1991).
43 S. Ragab and J. Wu, “Linear instabilities in two-dimensional compressible mixing layers,” Phys. Fluids A 1, 957–966

(1989).
44 M. J. Day, W. C. Reynolds, and N. N. Mansour, “The structure of the compressible reacting mixing layer: Insights from

linear stability analysis,” Phys. Fluids 10, 993–1007 (1998).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
93.180.53.211 On: Fri, 31 Jan 2014 09:04:27