ISSN 0018-151X, High Temperature, 2018, Vol. 56, No. 1, pp. 109–123. © Pleiades Publishing, Ltd., 2018.

Original Russian Text © L.V. Bykov, A.M. Molchanov, D.S. Yanyshev, I.M. Platonov, 2018, published in Teplofizika Vysokikh Temperatur, 2018, Vol. 56, No. 1, pp. 104–120.

REVIEWS

Modern Approaches for Calculating Flow Parameters

during a Laminar–Turbulent Transition in a Boundary Layer
L. V. Bykov*, A. M. Molchanov**, D. S. Yanyshev***, and I. M. Platonov
Moscow Aviation Institute (National Research University), Moscow, Russia
*e-mail: [email protected]
**e-mail: [email protected]
***e-mail: [email protected]
Received June 20, 2016

Abstract—We analyze modern methods for calculating heat and hydrodynamic flow parameters in a bound-
ary layer during the laminar–turbulent transition. The main approaches for describing the phenomenon of
laminar–turbulent transition are examined. Each approach is analyzed. The manner in which different fac-
tors influence the laminar–turbulent transition is studied. An engineering model of the laminar–turbulent
transition in a high-velocity flow is presented.

DOI: 10.1134/S0018151X18010042

CONTENTS cient with the necessary lift force, while the heat pro-
Introduction tection of the structure should absorb heat loads
caused by high enthalpy of the incoming flow and by
1. The mechanisms for transition to turbulence extreme heat fluxes. In many cases the extreme heat
2. Approaches to determining boundary-layer stability and mechanical loads that occur in structural elements
2.1. Methods from stability theory of a high-velocity vehicle are caused by phenomena
2.2. Applied methods that occur in a boundary layer and in particular, by
3. Laminar–turbulent transition at high velocities transition from laminar flow conditions into turbulent
4. Factors that influence the laminar–turbulent ones.
transition Main parameters defining vehicle performance
4.1. Acoustic impacts such as frictional drag, heat fluxes at the surface of the
vehicle, and lift force depend on the boundary-layer
4.2. Pressure gradient parameters.
4.3. Mach number The heat fluxes that act on the structural elements
4.4. Heat exchange on the surface of a body with of the vehicle are most important at supersonic and
flow around it hypersonic velocities (for example, for reusable space-
4.5. Gas injection into a boundary layer craft). The structure should be protected against high
4.6. Surface state heat fluxes. Numerous experiments have shown that
5. An engineering model of laminar–turbulent in a turbulent boundary layer the heat fluxes are higher
transition in a high-velocity flow by dozens of times than in a laminar layer. That is the
reason that during the design of modern high-velocity
Conclusions vehicles it is very important to properly predict the
References place of the laminar–turbulent transition and to deter-
mine whether it is possible to affect it.
INTRODUCTION Proper prediction of the laminar–turbulent transi-
The problem of hypersonic velocity is becoming tion is also very important for investigating flows in
increasingly important due to active development of channels. The problem has been studied properly for
the aviation and space industries. Engineers and sci- simple flow geometries (pipes with a simple cross sec-
entists are making attempts to create a vehicle that tion without turbulence stimulators and artificial
operates at Mach numbers M > 6 at high altitudes with roughness); while flows in channels with artificial
the required strength parameters. Flights with hyper- roughness need further investigation [1].
sonic velocities in dense atmospheric layers cause new The results of investigations (performed in Russia
problems, which must be solved. The shape of the and abroad) on boundary-layer stability under differ-
vehicle should be characterized by a low drag coeffi- ent conditions are presented in this paper. Special

109
110 BYKOV et al.

Formation Turbulent In the receptivity area external disturbances are

T–S waves transformed into instability waves of the boundary
of the -structures spots
X layer.
The first mode of instability is waves in the form of
viscous instability, which cause low-frequency vortex
disturbances. For the case of a low-velocity flow the
first mode is presented by Tollmien-Schlichting
waves, which are most unstable in the form of 2D dis-
Recr Retr turbances. For high velocities (for supersonic flows)
Laminar Edge 3D inclined waves of the first mode are the most
Vortex decay
flow turbulization unstable.
IV V
III Disturbances of the second mode (the Mack
U∞ II
I mode) are intrinsic to compressible high-velocity
(M > 3) boundary layers. Such instabilities cause high-
frequency acoustic (not viscous) disturbances, which
rise faster than T–S waves. The second mode is the
Fig. 1. A picture of the instabilities and turbulence devel- dominant mechanism of the high-velocity flow transi-
opment in a boundary layer. tion if there are no strong transverse flows, Görtler
vortexes, and bypass transition. In this case the spe-
cific conditions for dominating second-mode distur-
attention is paid to the changes in flow conditions in bances are determined by the flow conditions (in par-
the boundary layer, to the main factors that cause such ticular, by heat conditions at the wall). As an example,
changes, and to methods for predicting such phenomena. for a heat-insulating flat plate the second-mode dis-
This review contains data on both high-velocity turbances are dominant at M > 4 according to the lin-
and low-velocity flows, since the majority of methods ear theory of stability [4].
for investigating the laminar–turbulent transition were The transverse flow becomes unstable if there is
developed initially for low-velocity flows (for an bending in the velocity profile. One example of such a
incompressible approximation). flow is the 3D boundary layer at a sweptback wing.
The nonlinear interactions are intrinsic to transverse
1. THE MECHANISMS FOR TRANSITION disturbances. According to the theory, the traveling
TO TURBULENCE waves of a transverse flow increase faster than any
other waves, but in practice transverse standing waves
The phenomenon of the laminar–turbulent transi- are studied since they are efficiently generated by
tion has not been studied up to present. It is known roughness at a flow-around surface [5].
that the main feature of turbulent flow is irregular
pulsed variations in instantaneous parameters (such as Görtler instability consists of longitudinal counter-
velocity, temperature, and pressure). rotating vortexes, which are common for Prandtl-
Meyer flow in a boundary layer. They can occur both
A flow does not change from laminar to turbulent naturally, when the streamlines are curved, and artifi-
instantly. Separate ordered vortex structures occur in a cially, after obstacles that are placed in the flow. The
flow, which disintegrate to smaller vortexes; finally, main instability that causes longitudinal counter-
the layered ordered fluid flow is destroyed. rotating vortexes is centrifugal instability, when the
Figure 1 illustrates the modern conception of the centrifugal force is higher than the radial pressure gra-
laminar–turbulent transition (hereinafter the LT tran- dient [6].
sition) in a medium with small initial disturbances at Several different instabilities can occur simultane-
relatively low velocities of the incoming flow [2] (Re is ously or separately. Which instability occurs depends
the Reynolds number). on several factors, such as the incoming flow velocity,
The first stage, stage I, of the examined process is wall curvature, roughness, and irregularity. If oscilla-
called receptivity [3]. Disturbances in the incoming tions in a boundary layer are small, the disintegration
flow such as vortexes and sonic waves act on the follows path A (Fig. 2) and the process of growth of
boundary layer in the form of stable or unstable fluc- disturbances can be described by the linear theory of
tuations. The form and intensity of these disturbances stability. Such growth is not intensive; it occurs far
determine the initial conditions of the disturbances in from the initial segment and variations in the pressure
the boundary layer (amplitude, frequency, and phase), gradient, temperature gradient and injection into the
which can cause the failure of the laminar boundary boundary layer from the surface can influence it.
layer (due to the intensification process). The recep- The second stage, stage II, of the process is the lin-
tivity examines the initiation of instability waves, not ear growth of eigenmodes of instability waves in the
their development. boundary layer, which can occur in the form of eigen-

HIGH TEMPERATURE Vol. 56 No. 1 2018

 MODERN APPROACHES FOR CALCULATING FLOW PARAMETERS 111

solutions of homogenous linearized equations for dis- Disturbance growth

turbances.
If the amplitude of the oscillations increase, 3D External disturbances
and nonlinear interactions in the form of secondary
instabilities occur in the boundary layer. At this stage
the waves propagate with the same phase velocity, Receptivity mechanisms
which makes it possible to transfer energy from pri-
mary (basic) waves to secondary waves; as a result, the
secondary waves grow rapidly. A Unsteady growth
In area III the mechanisms of secondary instability B
generate oblique waves by means of nonlinear interac- Growth of eigenmodes C D E
tions. In the area of tertiary instability IV layers con-
nected with instantaneous bending in the velocity pro-
file generate high-frequency fluctuations (oscilla- Parametric instabilities Bypass
tions, peaks, and striped structures) and turbulent and interaction of modes mechanisms
spots. Turbulent spots are accumulated and associ-
ated; as a result a turbulent area forms (area V).
The amplitude of the initial disturbances increases Decay of vortexes
from left to right. The primary disturbances can be too
small for measurement and it is possible to estimate
Turbulence
them only indirectly according to the parameters of
the incipient instability.
Since it is possible to calculate the boundary layer Fig. 2. A mechanism for transition to turbulence.
behavior in the case of linear instability, very often it is
thought that the process follows path A and only linear
growth is examined. Such an assumption is true for the 2.1. Stability-Theory Methods
flows with small disturbances of the free flow and with For generating the laws of instability spreading in a
extensive segments of linear growth with respect to boundary layer all physical values that characterize the
nonlinear growth. flow conditions are presented in the form of two com-
At times, the flow disturbances are so large that the ponents: averaged and pulse components:
turbulent mode occurs significantly earlier, skipping
the stage of linear growth via the so-called bypass q = q ( x, t ) + q ' ( x, t ) . (1)
mechanism [7]. This corresponds to path E. It has
been historically established that only two scenarios If we substitute values of the form (1) into Navier–
for transiting to turbulence: А and Е are investigated. Stokes equations we can obtain the precise equations
We have begun to better understand which path the for fluctuations. For an incompressible fluid these
process of laminar–turbulent transition follows only equations can be written as follows
recently as a result of a large volume of scientific work. ∇ u ' = 0, (2)
The stage of unsteady growth occurs if the instabil-
∂ u ' − 1 ∇ 2u ' + u ' ⋅ ∇ u
ity waves interact with each other. ( )
Scientific investigations in this field have shown ∂ t Re (3)
that at the respective initial conditions unsteady + ( u ⋅ ∇ ) u ' + ∇ p ' = ( u ' ⋅ ∇ ) u '.
growth can cause high amplitudes of disturbances. For small disturbances the right part of Eq. 3 can be
Such initial conditions depend on the receptivity of accepted as equal to zero. In this case Eqs. (2), (3) are
the boundary layer. Depending on the amplitude the linear.
unsteady growth, it can cause the modulation of 2D
waves (path В), disturb the initial state up to secondary These equations are solved in the wave form (in the
(nonlinear) instabilities (С) and/or transition accord- form of normal modes). For the parallel flow
( u = [u,0, w] ) the solution is as follows:
ing to the bypass mechanism (D). T

2. APPROACHES FOR DETERMINING q ' = qˆ ( y ) exp [i ( α x + β z − ω t )] , (4)

THE BOUNDARY-LAYER STABILITY
where q ' = [u ',v ', w ', p '] is the vector of the variables that
T
Here, we investigate the main approaches for
describe the flow and qˆ ( y ) = [uˆ ( y ) ,vˆ ( y ) , wˆ ( y ) , pˆ ( y )]
T
determining the conditions of the laminar–turbulent
transition. For simplification we examine the flow of is the vector of functions that describe the shape of the
an incompressible fluid. High-velocity flows are profile (amplitude) along the axis normal to the flow-
examined in part 3. around plane (axis y); α and β are the wave numbers;

HIGH TEMPERATURE Vol. 56 No. 1 2018

112 BYKOV et al.

ω the problem as follows: it is necessary to determine the

relationship between Im ( α ) (and/or Im (β)) and the
disturbance frequency and the Reynolds number. For
Im ( α ) < 0 unstable flow occurs, while for Im ( α ) > 0
the flow is stable. The boundary of the two areas is
determined by the neutral curve, whose equation is as
follows
Im [α ( ω,Re )] = 0.
Generalizing, it is possible to say that there are
three main approaches to examining flow instability: a
temporal theory that examines disturbances that
Recr Re increase with time (α and β are real and ω is complex);
a spatial theory that examines the growth of distur-
bances in space (α and β are complex and ω is real),
Fig. 3. The general form of a neutral curve.
and a spatial and temporal theory (α , β and ω are
complex numbers).
ω is the frequency (in general case α, β, and ω are It is necessary to point out that in the general case
complex numbers). the hypothesis on small disturbances is not correct.
From (4), it is seen that the disturbances grow In this case it may be impossible to solve equations (2),
infinitely, if Im ( α ) < 0 , Im (β) < 0 , or Im ( ω) > 0. (3) since it requires large computational power (per se,
the solution of equations (2), (3) is equivalent to direct
If there is a pronounced direction of flow (i.e., if numerical solution of the Navier–Stokes equations).
the mean velocity of the flow is sufficiently high), all
disturbances drift downstream and two types of flow However, there is an intermediate solution: on the
instability occur. one hand it is not very sensitive to computation power,
while on the other hand it is free of many disadvan-
If (in spite of the fact that disturbances drift down- tages that are intrinsic to the linear theory.
stream) at any fixed point in space for the smallest ini-
tial disturbances their amplitude rises infinitely with This approach is called parabolized stability equa-
time, absolute instability occurs. tions (PSE). The set of PSE equations makes it possi-
ble to calculate the behavior of disturbances both for
If the disturbances drift downstream so fast that at parallel flows and for systems with complicated geo-
a fixed point in space for t → ∞ the intensity of the metric configurations (for example, to calculate the
disturbances goes to zero then convective instability flow around blades in impeller machines).
occurs.
We do not present a statement of the method here,
If we substitute (4) in the linearized set of equa- since there are many variants of its formulation. The
tions (2), (3), we obtain the equation of the linear sta- parabolized equations of stability were examined in
bility theory (LST) detail in [8].
L qˆ = 0. (5) LST and PSE describe the behavior of instability
The differential operator L includes derivatives waves in a flow, but they do not determine the transi-
only with respect to the y coordinate. If we apply dif- tion point.
ferent transformations to operator L, it becomes pos- At present, the semi-empirical eN-method is used
sible to obtain different forms of this set of equations, for determining when laminar flow transforms into
for example, a set of equations that consist of the Orr– turbulent flow.
Sommerfeld and Squire [8] equations (equations for It is accepted that transition to turbulence occurs
the velocity disturbance and vorticity, respectively) or when the amplitude of the instability wave increases by
the Orr–Sommerfeld equation separately. eN times with respect to the initial amplitude at the
Usually LST equations are solved by determining neutral point (the first point, where Im ( α ) = 0).
the ω(Re) or α(Re) relationships (Fig. 3). The curve It is possible to show that for the boundary layer at
separating the crosshatched region is called the neutral a flat plate
curve. In this curve Im ( α ) = 0 or Im ( ω) = 0. The
R
crosshatched region is an area of instability and all of ⎛A ⎞
the other region is an area of stability (all disturbances N = ln ⎜ tr ⎟ = − 2 Im ( α ) dR,
∫
in it die). ⎝ A0 ⎠ R 0
In the case of convective instability the distur-
ρ ux
bances increase only if the x coordinate increases where R = Re x = .
downstream. In this case it is reasonable to formulate μ

HIGH TEMPERATURE Vol. 56 No. 1 2018

 MODERN APPROACHES FOR CALCULATING FLOW PARAMETERS 113

According to the empirical data the turbulence

( )
2
occurs at N = 7–9 [8]. is the conjugation operator, and Δ = d 2 − α 2 + β 2
dy
This method has a disadvantage: it cannot be is the Laplace operator.
applied for all of the examined mechanisms of
destruction of a laminar flow. In particular, it is For β = 0 the equations of set (6) are independent.
impossible to predict laminar–turbulent transition via Their eigenmodes are Orr–Sommerfeld modes (for
the bypass mechanism with its use. 2D definition of Tollmien–Schlichting waves) and
Squire modes. Orr–Sommerfeld modes are solutions
Intensification conditions or conditions for the of Orr–Sommerfeld equations (equations for velocity
decrease of spontaneous small-amplitude instability disturbance) and of Squire equations (equations for
waves are examined using the methods of the stability vorticity disturbance), while Squire modes are solu-
theory described above. The process of amplitude tions of Squire equations for uˆ = 0 or β = 0.
growth of spontaneous instability waves is a natural
mechanism for the laminar–turbulent transition. As In all cases where β ≠ 0 operator С ≠ 0 , i.e., the
mentioned above, it occurs only if the level of exter- operator of the Navier–Stokes set of equations is not
nal-flow turbulence is low. conjugated. This means that Orr–Sommerfeld modes
and Squire modes can interact with each other under
In practice, the level of turbulence of the external flow conditions of a non-flat flow. It is possible to show
is large and the stability theory approaches described that for the case where Orr–Sommerfeld modes and
above cannot be used directly. Mathematically, this is due Squire modes decrease individually, their unstationary
to the fact that in the stability theory the growth of the dis- interaction can cause significant growth of distur-
turbances is examined in an asymptotic approximation bance amplitude and it can cause the laminar–turbu-
passing from the problem of initial and boundary condi- lent transition.
tions to the problem of searching for eigenvalues [9, 10]. Usually, the concept of an intensification factor (a
We examine the flow in terms of the stability the- growth factor) is used for analyzing unsteady growth
ory; we do not examine the dynamics of the system, we
estimate the ability of the system to continue the G = E, (7)
infinite growth of disturbances. E0
However, in the general case such a problem defi- where E is the energy norm, which is determined dif-
nition is not very correct mathematically when we ferently for different flow types; the 0 index is the ini-
speak about a system described by Navier–Stokes tial value.
equations (due to the properties of the operators of this As an example, for an incompressible flow this
system). This can occur as follows: a certain external norm is an integral with respect to the kinetic energy of
disturbance that should decrease to zero as t → ∞ the disturbance
(according to asymptotic analysis of the system) in ∞

∫ (u ' )
reality increases greatly at a specific moment of time
causing a transition to turbulence. E = 2
+ v ' 2 + w ' 2 dy.
In other words, there are cases where a flow in 0
which external disturbances should die out with time To analyze unsteady growth one determines the so-
(according to stability theory) can transit to turbulence called optimal disturbances for which the intensification
due to unsteady growth of a certain disturbance. factor (7) is maximal. This makes it possible to determine
In the general case, the concept of unsteady growth the upper boundary of the intensification of disturbances
means local development of a disturbance in contrast in a flow and to understand which type of disturbance is
to the stability theory, where the possibility of distur- the most likely to grow in a specific type of flow.
bance development separating cases of its exponential Very often unsteady growth is connected with a
growth or decrease within a certain range is examined. laminar–turbulent transition caused by surface rough-
ness. However, there is no unified theory of the lami-
For mathematical illustration let us write set (5) in the nar–turbulent transition caused by roughness or by its
form of the Orr–Sommerfeld and Squire equations separate elements.
⎡L 0 ⎤ ⎡ uˆ ⎤
⎢C S ⎥ ⎢Ω ⎥ = 0, (6) 2.2. Engineering Methods
⎣ ⎦⎣ˆ⎦
The methods described above, in which equations
where Ω̂ is the vorticity of the velocity disturbance, for instability waves are used, are in good agreement
2 with the experimental data; however, it is difficult to
L = S Δ − i α d u2 is the Orr–Sommerfeld operator, use them for solving engineering problems for different
dy reasons.
S = −i ω + i α u − Δ is the Squire operator, C = iβ du First, it is necessary to be well trained in mathe-
Re dy matics to use the methods of stability theory correctly.

HIGH TEMPERATURE Vol. 56 No. 1 2018

114 BYKOV et al.

Second, it is difficult to use these methods for The boundary conditions recommended by Men-
flows with complicated geometric configurations. ∂γ
ter are as follows: on the wall = 0 (n is the normal
Third, due to the characteristics of the described ∂n
methods, it is very problematic to use them in modern to the wall surface); at the input γ = 1.
CFD software. The input condition is chosen mainly according to
From the practical point of view, approaches for numerical stability [12]. Its physical sense can be given
determining the stability of laminar flow based on by examining the bypass mechanism for transition to
numerical simulation, such as direct numerical simu- turbulence when the disturbances in the boundary
lation, Langtry–Menter methods, and k − k L - transi- layer are introduced from the external flow.
tion models, are more suitable. Let us point out that the intermittency factor is the
Reynolds averaged Navier-Stokes equations (for “indicator” of flow conditions (it is equal to zero in
compressible flows—Favre averaged) play the central laminar flow and to unity in turbulent flow); it does
role in the modern methods for calculating turbulent not regulate the turbulence level in the flow. As an
flows. example, if we use the value of the kinetic energy of the
turbulence k → 0 as a boundary condition, the turbu-
These equations are the basis of modern applied lence level in the external flow also goes to zero in spite
software such as Ansys CFX, Ansys Fluent, Star-CD, of the set value γ = 1.
FlowVision, and LOGOS.
The form of the source component Fγ depends on
If the turbulent fluctuations go to zero, these equa-
tions transform into ordinary Navier–Stokes equa- the model version (Menter et al. suggested several
tions. modifications (see, for example, [13, 14]). This practi-
cally determines the model behavior.
For such transformation it is convenient to intro- It is evident that it is most convenient to determine
duce the intermittency factor [11] between the laminar the source type by using the generalized experimental
flow model (q → q ) and the model of turbulent flow data on the laminar–turbulent transition as a function
(q = q + q ' ) of local variables. Jointly with the chosen form of
τt transport equation for γ , it removes the necessity to
γ= , determine the concrete type of instability that causes
τ L + τt transition.
where τ L , τ t are the times within which the flow at a The main problem is as follows: the majority of the
given point is in the laminar and turbulent states, experimental data on the laminar–turbulent transition
respectively. are obtained in the form of relationships for integral
By using the intermittency factor and the Boussin- quantities, in particular, for the Reynolds number
esq hypothesis for turbulent stresses, it is possible to Re θt , under which transition to turbulent flow occurs
determine the mean viscosity coefficient μ eff at a given and which depends on the turbulence level in the
flow point external flow Tu, pressure gradient parameter, etc.:
ρ U ∞θ t
μ eff = μ + (1 − γ ) μ nt + γμT , (8) Re θt = = f (Tu...) ∞ ,
μ
where the index T indicates the turbulent flow, while nt
is related to non-turbulent disturbances (transitive). where θ t is the thickness of the impulse loss in the
boundary layer at which the flow mode changes (it is
Alternately, it may be possible to impose a respec- the integral performance of the boundary layer and
tive restriction to the turbulence source in the trans- cannot be determined at a point), Tu =
( 2 3 k u) × 100%.
port equation for the kinetic energy of the turbulence
(which effects the viscosity coefficient).
The main disadvantage of this approach is that it is Menter and Langtry suggested a method for solving
impossible to derive the respective equation for γ with this problem.
the appropriate physical strictness. Instead of the integral Reynolds number Re θ it is
It is only possible to assume that such an equation suggested to use the so-called vortex Reynolds number
should have the standard form of the transport equa- Rev , which is calculated using only the local parame-
tion with a special source component: ters of the flow:
∂ ργ + ∂ ρ u γ = ∂ ⎡⎛ μ + μT ⎞ ∂γ ⎤ + F , ρy S
2
( ) ( j ) ⎢⎜ ⎟ ⎥ γ Rev = .
∂t ∂x j ∂ x j ⎣⎝ σ γ ⎠ ∂x j ⎦ μ
where σ γ is the Prandtl–Schmidt number for the Here, y is the distance from the wall, S = S ij S ij is
intermittency factor. the invariant of the strain rate tensor

HIGH TEMPERATURE Vol. 56 No. 1 2018

 MODERN APPROACHES FOR CALCULATING FLOW PARAMETERS 115

S ij = 1 2 ( ∂ ui ∂ x j + ∂ u j ∂ x i ) (in several works The main idea is to present the energy of fluctua-
tion k TOT in the flow by two components
Ω = Ω ij Ω ij the invariant of the vorticity tensor
Ω ij = 1 2 ( ∂ ui ∂ x j − ∂ u j ∂ x i ) is used). k TOT = k L + kT ,
The number Rev has a maximum that is propor- where k L is the kinetic energy non-turbulent fluctua-
tional to Re θ approximately in the middle of the lam- tions (“laminar” energy) and kT is the kinetic energy
inar boundary layer. of the turbulence.
In [15] it was shown that for a laminar boundary It is clear that non-turbulent fluctuations are of a
layer without a pressure gradient large scale while the turbulent ones are of a small scale.
The transport factor for non-turbulent fluctuations is the
Rev max molecular viscosity, while for turbulent fluctuations it is
Re θ = . the molecular and turbulent viscosities.
2.193
It was also proven that for calculating the laminar– The model consists of three equations: for k L , kT ,
turbulent transition this relationship can be used with- and the vortex dissipation rate ω.
out alteration, as well as for the cases of more compli- D (ρ kT )
cated flows. = ρ ( PkT + RBP + RNAT − ω kT − DT )
Dt
Here, Re θ is replaced by the local Rev . Moreover,
⎡⎛ ρα ⎞ ∂ k ⎤
Re θt is a function of the integral parameters of the + ∂ ⎢⎜ μ + T ⎟ T ⎥ ,
external flow (such as Tu∞ ). θ x j ⎣⎝ σ k ⎠ θx j ⎦
Here, it is necessary to introduce the local analog D (ρ k L ) ⎡ ∂k ⎤
of Re θt . For this purpose in [12, 14, 15] a certain mod- = ρ ( PkL − RBP − RNAT − DL ) + ∂ ⎢μ L ⎥ ,
Dt θx j ⎣ ∂x j ⎦
ified local Reynolds number Re  θt for which the trans-
D (ρω) ρω ⎛C ⎞ ρω
port equation is derived artificially was introduced: = C ω1 PkT + ⎜ ωR − 1⎟ ( RBP + RNAT )
∂ ρRe Dt kT ⎝ fW ⎠ kT
(  θt ) + ∂∂x (ρu j Re θt ) 2 ρ kT ⎡⎛ ρα ⎞ ⎤
∂t − C ω2ρω + C ωJ ωαT fW + ∂ ⎢⎜ μ + T ⎟ ∂ω ⎥ .
2
j

⎡  ⎤ d 3
∂ x j ⎣⎝ σ ω ⎠ ∂x j ⎦
= ∂ ⎢σ θt (μ + μT ) ∂Re θt ⎥ + Pθt .
∂x j ⎣ ∂x j ⎦ This model is in good agreement with the experi-
mental results. The issue of whether it is possible to cal-
According to [16], the empirical relationship culate the supersonic flows was not examined in [17].
between Re θt and the integral critical Reynolds num-
ber is obtained via numerical experiments: 3. THE LAMINAR–TURBULENT TRANSITION
AT HIGH VELOCITIES
⎧4
Re θc ∑
= ⎨ n=0
 θnt , Re
⎪ An Re  θt ≤ 1870, Flows at high velocities are characterized by some
peculiarities that complicate the picture of the lami-
⎪   nar–turbulent transition.
⎩0.518Re θt + 308.23, Re θt > 1870, 1. At velocities with M > 0.3 the compressibility is
⎛ − 396.035 × 10 −2 ⎞ important. In particular, it becomes more difficult to
⎜ −2 ⎟
generate the equations of the stability theory (see, for
⎜ 101.207 × 1 0 ⎟ example, [18]).
where A = ⎜ − 868.230 × 10 −6 ⎟ . 2. During the transonic transition and during a fur-
⎜ ⎟ ther velocity increase, the wave structure of the flow
⎜ 696.506 × 10 −9 ⎟ changes greatly; the character of the laminar–turbu-
⎜⎜ ⎟
− 12 ⎟ lent transition also changes, respectively.
⎝ − 174.105 × 1 0 ⎠
3. If the flow velocity increases, its energy also
According to [16] the model is not suitable for increases and it can affect the molecular structure and
highly compressible flows. thermodynamic properties of the flow. Here, in the
The Menter–Langtry approach described above vicinity of boundary layer for certain Mach numbers of
can be characterized as a pure phenomenological the external flow it is impossible to consider the gas as
approach based on experimental data. the ideal one. Moreover, under certain conditions it is
Another approach, which is an attempt to examine impossible to characterize the gas by sole temperature.
processes that occur in the laminar–turbulent transi- As a result of all of these issues the problem of
tion in terms of the transport of disturbance energy, determining gas flow stability at high velocities
was described in [17]. becomes more complicated.

HIGH TEMPERATURE Vol. 56 No. 1 2018

116 BYKOV et al.

y mechanism of the natural laminar–turbulent transi-

u(ya)  > 1.
tion in the high-velocity flow for M
ya Sonic line u(ya) = Cr – a(ya) The behavior of Mack modes has been examined in
detail by solving stability-theory equations written for
p'(y) compressible flow (in the linear and parabolic defini-
Acoustic-wave tions) and experimentally. As an example, it has been
propagation found that the Mack modes are stabilized by the body
front edge bluntness [20], and destabilized by wall
Wall x cooling [21].
Fig. 4. Acoustic waves in a boundary layer [19].
The problem of the variation of the chemical and
thermal properties of a gas at hypersonic velocities
At the present time only the theoretical approaches (first of all due to reactions of dissociation and recom-
for solving the problem have been developed. These bination) requires detailed investigation.
are based on stability theory and on the model of It is known that in a hypersonic flow intensive
chemically nonequilibrium high-enthalpy flow. Uni- physical and chemical processes are connected with
versal applied methods have not been developed as yet the excitation of a molecule’s internal degrees of free-
in this field. dom, dissociation, chemical reactions between gas
The main peculiarity of high-velocity flows with components, and the ionization of an atom.
respect to incompressible flows in terms of their stabil- Here, the set of equations for gas flow in the case of
ity is a collection of persisting instability modes; these hypersonic flow should be combined with equations
were classified for the first time by Mack [4]. that describe these physical and chemical processes.
Let us note that in the incompressible fluid exam- Let us present the set of gas dynamic equations for a
ined above, the natural mechanism of laminar–turbu- hypersonic air flow.
lent transition is determined by Tollmien-Schlichting
waves of a viscous nature. ⎯The equation of continuity
∂ρ
+ ∂ (ρ u j ) = 0,
In the case of compressible fluid flow with a suffi-
ciently high velocity (M > 3) nonviscous disturbances ∂t ∂x j
(Mack modes) of an acoustic nature begin to develop
in the flow. where ρ is the gas mixture density and u j is the velocity
Based on the example of an equation for nonvis- component in the j th direction.
cous pressure disturbances, Mack showed that the
occurrence of acoustic disturbances that did not die ⎯The momentum equation
out in a flow is determined by the velocity of the dis- ∂ ρ u + ∂ ρ u u + δ p − τ = 0,
turbance phase with respect to the flow velocity and ( i) ( j i ji ij )
the sound velocity of the external flow. ∂t ∂x j
According to this, the respective Mach number can where p is the pressure and τ ij is the tensor of viscous
be written as follows tensions.
M  = ue − c r , ⎯The equation of the total energy
ae
∂ ρ E + ∂ ⎡ρ u ⎛ E + p ⎞ + q + q
where the index e is the external boundary of the ( ) j⎜ ⎟
∂t ∂ x j ⎢⎣ ρ⎠
tr, j v, j
boundary layer. ⎝
 > 1, an area is formed in the flow where the
If M ⎤
mean velocity of the flow with respect to the phase + ∑V s, j ρ s hs − ui τ ij ⎥ = −QR .
⎥⎦
velocity of of disturbances is higher than the sound s
velocity. The line (surface) that limits this area is called Here, E is the total energy per mass unit; hs specific
the sonic line. If this line (surface) exists, acoustic waves (per mass unit) enthalpy; s is the bottom index that
reflect from it and become trapped inside (Fig. 4). Here,
show the membership to component s; ρ s is the den-
u is the velocity profile, c r is the phase velocity of distur-
bances, and p ' is the profile of the pressure disturbances.
sity; V s, j is the diffusion velocity in the j th direction;
Such trapped acoustic waves form Mack modes. qv, j is the density of heat flux of vibrational energy in
The amplitude of Mack modes increases more rap- the j th direction; q tr, j is the density of the heat flux of
idly than the disturbance amplitude of the first mode. the tranlsational and rotational energy in the j th direc-
Due to this, the Mack modes are prevalent in the tion; and QR is the losses via radiation W/(m3).

HIGH TEMPERATURE Vol. 56 No. 1 2018

 MODERN APPROACHES FOR CALCULATING FLOW PARAMETERS 117

⎯The equation of the vibrational energy of the mth where hs0 is the heat of formation for s and CVtr,s is the
vibration mode
specific (per mass unit) translational-rotational heat
∂ E
( v,m ) + ∂ ( Ev,mu j + qv,m, j + Ev,mV m, j ) = Sv,m, capacity at a constant volume.
∂t ∂x j The vibrational energy is intrinsic only for two-
m = 1,2,..., N M , atom and multi-atom molecules; it is equal to zero for
atoms.
where E v,m is the specific (per volume unit) vibrational
Here, in contrast to the flow with lower velocities,
energy of the mth vibration mode; qv,m, j is the density the hypersonic flow is described by a larger number of
of the heat flux of the vibrational energy of the mth vibra- parameters. For an incompressible flow we have four
tion mode in the j th direction; Sv,m is the source of vibra- main parameters (the components of the velocity vec-
tional energy caused by the V–T, V–V energy transition, tor and pressure); for a compressible flow at M < 6 we
the generation rate of the vibrational energy due to chem- have five parameters (the components of the velocity
ical reactions, as well as by the vibrational energy loss due vector, temperature, and pressure/density) and for a
to spontaneous deactivation of radiation; and N M is the hypersonic flow we have more than eight depending
number of modes of the vibrational energy. on the chemical composition of the media (the com-
ponents of the velocity vector, total energy, vibrational
Here, V m, j is the diffusion velocity of the compo- energy, and the concentrations of the components).
nent to which the mth vibration mode is related.
Therefore, the set of equations of the stability the-
A detailed discussion of the problem of how to calcu- ory that are derived with the use of decomposition (1)
late the radiation parameters was presented in [22, 23]. becomes more complicated: equations for disturbances,
⎯The equation of mass conservation for a chemi- the vibration temperature, and the densities of the com-
cal component s ponents are added. As a result, the number of parameters
∂ ρC + ∂ ρC u + ρC V that influence the flow stability and the position of the
( s) ( s j s s, j ) = w
 s, laminar–turbulent transition becomes larger.
∂t ∂x j
s = 1,2,..., N C − 1, In a linear approximation the set of stability equa-
tions for a flat layer can be written as follows [24]:
where C s = ρ s ρ is the mass fraction, w s is the gener-
ation rate for s due to chemical reactions, and NC is the ⎛ d2 d ⎞
⎜ A 2 + B + C ⎟ qˆ = 0.
number of components in the gas mixture. ⎝ dy dy ⎠
The following assumptions are used in this set:
Here qˆ ( y ) = [ρˆ, uˆ ( y ) ,vˆ ( y ) , wˆ ( y ) , Tˆ,Tˆv1...Tˆv m,
(1) the rotational energy modes are in equilibrium
with the translational ones and they are determined by Cˆ1 ( y ) ...CˆN C ( y )]T is the vector of the unknown values
one translational-rotational temperature T = Ttr ; (form functions); m is the number of molecular vibra-
(2) the energy for exciting the electron states of a tional modes; and A, B, C are complex matrixes with a
molecule is negligibly small with respect to other dimensionality minimum of NC + m + 5 (if the mixture
energy modes; can be characterized only by the vibration temperature).
(3) it is accepted that the heat losses for radiation in The set of equations is solved jointly with the equa-
the energy equations are caused mainly by deactiva- tions for the averaged values. Methods for solving
tion of the vibrational modes; equations for averaged characteristics at high velocities
(4) electron-to-ion energy transitions are not taken were discussed in detail in [25, 26] (for structured
into account. meshes) and in [27] (for unstructured ones).
The gas mixture pressure p is described by Dalton’s The described approach makes it possible to inves-
law and is equal to the sum of the partial pressures for tigate hypersonic laminar flow stability with respect to
the components ps the receptivity to unstable waves and to different fac-
tors that influence the laminar–turbulent transition.
Nc Nc
In particular, by using the stability-theory equa-
∑ ∑ρ
RU
p= ps = s Ttr , tions using the definition described above it is possible
s =1 s =1
Ms
to explain the effect of flow stabilization in case of car-
where RU is the universal gas constant and Ms is the bon-dioxide gas injection into the boundary layer.
molecular mass.
The total energy E consists of the translational, rota- 4. FACTORS THAT INFLUENCE
tional, vibrational, kinetic energy, and energy of chemical THE LAMINAR–TURBULENT TRANSITION
components
Since the experimentsof Reynolds on the stability
Nc NM of a laminar flow in a circular pipe, a large amount of
ρE = ∑ρ C s V tr, sT tr + ∑E v ,m + 1 ρ ui u i +
2
∑ρ h , 0
s s experimental and theoretical experience has been
accumulated in analyzing different factors that influ-
s =1 m =1 s

HIGH TEMPERATURE Vol. 56 No. 1 2018

118 BYKOV et al.

KS Here, we examine the most important factors that

8 influence the laminar–turbulent transition in a
boundary layer.
6
3 4.1. Acoustic Impacts
When we investigate low-turbulent flows at ground
4 conditions (in a wind tunnel) typical for the flight
1
conditions of a high-velocity vehicle, the main exter-
nal impact that causes disturbances in a laminar
2 boundary layer are acoustic waves that occur in the
2
incoming flow. When an acoustic wave arrives at a
plate, disturbances are generated in the boundary
0 15 30 45 60 75 90 layer. The largest impact causes a wave that moves to
θ∞, deg the front edge, which can be due to the small thickness
of the boundary layer in this area [28, 29]. When the
incoming acoustic wave interacts with a separated
Fig. 5. The coefficient of the intensification of different wave, three types of disturbances occur in the bound-
modes in a high-velocity boundary layer during the impact
of external acoustic vibrations as a function of their inci- ary layer: vortex, acoustic, and entropy waves. It is
dent angle [32]: 1, I stable mode, 2, II Mack mode, 3, II necessary to note that in the incoming flow two waves
stable mode. occur simultaneously: slow and fast waves with phase
velocities of 1 − 1 M ∞ and 1 + 1 M ∞ , respectively. One
Reθ wave spreads along the flow, the second wave spreads
Tu = 0
in the opposite direction, but since the flow is super-
sonic, both waves move in the flow direction. In this case
0.25% the coefficient of receptivity to the slow acoustic wave is
1000 higher than to the fast one [28]. Waves reflected from the
0.5% boundary layer edge and wall interact each other and may
(if the phase velocities are synchronized) resonate with
1% stable wave modes that exist in the flow; as a result, the
500 vibration amplitude of the latter increase and instability
1.5% waves of the first or second mode occur [30, 31].
2%
3% It has been found that boundary-layer receptivity to
an acoustic wave depends on the incident angle to the
Tu ≥ 5%
wall. It has been shown that there is a relationship
–0.1 –0.06 –0.02 0 0.04 0.08 between the receptivity factor and the incident angle of
λθ the wave (Fig. 5). The intensification coefficient K S is
the ratio between the maximal amplitude of the pressure
disturbance and the pressure in the incoming flow.
Fig. 6. The relationship between the transition Reynolds
number and the pressure gradient at different turbulization Boundary-layer behavior near a cone surface
levels of the external flow Tu [33]: the points are the expe- caused by an acoustic wave is similar to the boundary
rimental data; the lines are the empirical relationships. layer behavior at a flat plate. Acoustic waves influence
stability and transport disturbances most efficiently if
they are incident to the cone vertex.
ence the laminar–turbulent transition. Incompress-
ible and compressible flows; subsonic, supersonic,
and hypersonic boundary layers; and flows around 4.2. Pressure Gradient
bodies with different shapes have been studied. The effect of a pressure gradient has been studied
Factors that influence the prevailing (having the properly for the case of the instability of the first
largest amplification rate) mechanism of stability loss mode. Experiments are in good agreement with theo-
and laminar boundary layer destruction can be sepa- retical predictions, where a positive gradient causes a
rated into external and internal ones. The external fac- destabilizing effect and a negative gradient causes a
tors are as follows: pressure gradients, acoustic wave stabilizing one. A detailed review of experimental
impacts, and parameters of the incoming flow such as investigations related to this phenomenon was pre-
its velocity and turbulence intensity (i.e., external fac- sented in [33].
tors). The internal factors are as follows: surface Figure 6 depicts the relationship between the tran-
roughness and porosity, heat-mass exchange with the sition Reynolds number and pressure gradient at dif-
surface, and several other factors. ferent levels of turbulization of the external flow for

HIGH TEMPERATURE Vol. 56 No. 1 2018

 MODERN APPROACHES FOR CALCULATING FLOW PARAMETERS 119

10–4 F (а) 10–4 F (b) 10–4 F (c)

50 5
10 2 2
40 4
8
30 2 3
6
4 20 2
1
2 1 10 1
1
0 200 400 600 0 200 400 600 0 200 400 600 800 1000
Re Re Re
Fig. 7. Curves of the neutral stability of the first (1) and second (2) modes at M = 5.35 and at different pressure gradients:
(а) λ s = 0; (b) –0.01, (c) 0.01.

low flow velocities. In this case the pressure gradient ship between the critical Reynolds number and pres-
can be written as follows: sure gradient for the first mode disturbance, i.e., the
negative pressure gradient destabilizes the first mode,
λθ = θ
2
dU ∞ while a positive pressure gradient stabilizes it [34]. At
.
ν dx λ s > 0.015 the reversal of the relationship does not
The calculations presented in [34] give an interest- occur.
ing picture of how the pressure gradient influences the The effect of a pressure gradient on the second
stabilization process of instability modes. In [34] the mode is classical, i.e., a negative gradient stabilizes the
following parameter characterizing the pressure gradi- second mode, while a positive pressure gradient
ent in a flow was used (in the authors’ terminology it strongly destabilizes it. The total effect of a pressure
is called the gradient parameter): gradient to a high-velocity flow with a tendency to sec-
ond mode domination for a small gradient parameter
dp
λs = − s . is such that for a positive pressure gradient the bound-
ρ eue ds ary layer stability is determined only by the second
For moderate Mach numbers (M ~ 2), i.e., in the mode, while for a negative pressure gradient it is deter-
area where the first instability mode is dominant, the mined by the first one (Figs. 7, 8).
effect of a pressure gradient in the boundary layer is For high Mach numbers of the incoming flow and
similar to the low-velocity case (a positive pressure gradient parameters λ s > 0.015 the effect of the first
gradient causes a destabilizing effect, while a negative mode is dominant, the second mode is unstable, and
pressure gradient causes a stabilizing effect). the resulting effect of the pressure gradient to the flow
For high Mach numbers (M ~ 5) and small gradi- is classical since the reversal effect disappears.
ent parameters ( λ s < 0.015) we see a reverse relation- Unfortunately we were not able to find data similar to
the results presented in Fig. 6 for a high-velocity flow.

Retr 4.3. Mach Number

108 The Mach number of the incoming flow can influ-
ence the position of the laminar–turbulent transition. If
the Mach number increases the turbulence moves
downflow [35].
2 Figure 8 depicts the respective experimental and
107 1 3
calculated results for flow around a cone.
4
4.4. Heat Exchange on the Surface
5
of a Body with Flow Around It
106 The effects that occur during heat exchange on a
0 4 8 12 Me surface with flow around it are important. In practice
Fig. 8. The relationship between the transition Reynolds
the most interesting situation is the case of flow cool-
number and the Mach number at the edge of the boundary ing (Tw/Tr < 1).
layer: 1, 2, calculations according to the eN (N = 10) Using stability theory it is shown that first-mode
method for adiabatic and cooling walls, respectively;
3, experimental data for ordinary wind tunnels; 4, the disturbances are stabilized by flow cooling, while sec-
results of flight experiments; 5, experimental data for low- ond-mode disturbances are destabilized. Here, in
noise wind tunnels. flows where the first mode disturbances are determi-

HIGH TEMPERATURE Vol. 56 No. 1 2018

120 BYKOV et al.

0.40 injection into boundary layer causes a destabilizing effect

and accelerates the laminar–turbulent transition.
0.35 1
On the other hand, in some cases gas injection can
0.30 stabilize the boundary layer. This concerns physical
0.25
and chemical processes connected with chemical
2 reactions and molecular vibrations. Usually these pro-
0.20 cesses take place in high-enthalpy flows.
α

3
0.15 In particular, it has been found that carbon-dioxide
4
gas injection into a high-velocity air flow stabilizes the
0.10 boundary layer by decreasing the disturbance factor N
0.05 and by increasing the transition Reynolds number
[40]. It has also been shown that if other gases are
injected (nitrogen, argon, and air), we do not observe
0 500 1000 1500 2000 the same effect [41]. Upon analyzing this phenome-
Re non it has been found that the effect caused by carbon
dioxide gas is connected with the vibration-relaxation
Fig. 9. The wall cooling effect on the neutral stability curve properties of the CO2 molecule and with the processes
at М = 5.8; 2D disturbances, T∞ = 125 K according to of energy redistribution between the translational and
[36]: 1, Tw/Tr = 0.05, 2, adiabatic wall; 3, 0.25, 4, 0.65. vibrational molecular modes in the boundary layer.
Qualitatively, the effect of the injection of carbon
native in the mechanism of boundary layer destruc- dioxide gas depends on the flow rate at which this gas
tion, the flow cooling stabilizes the flow and closes the is injected into the boundary layer; it is characterized
laminar–turbulent transition, while in flows with a by injection parameter
higher velocity, where the second instability mode is
2 Re x ρ wue
prevalent, flow cooling destabilizing the flow and fw = .
accelerates the transition to turbulence. ρ ∞u ∞
Figure 9 depicts the neutral curves for the flow in a Since (as mentioned above) gas injection into the
2D boundary layer for a flat plate when the wall is boundary layer destabilizes it, there is a definite bound-
cooled. ary value of the f w0 parameter. If f w < f w0 , the molec-
ular-relaxation effects of gas injection prevail and injec-
4.5. Gas Injection into the Boundary Layer tion of carbon dioxide gas stabilizes the boundary layer;
if f w > f w0 , the hydrodynamic effects prevail and gas
Different investigations (see, for example, [21, 37, injection destabilizes the boundary layer.
38]) have shown that gas injection into a boundary
layer can greatly influence the laminar–turbulence Figure 10 shows the calculated relationship
transition in the boundary layer. between the point of the laminar–turbulent transition
on the cone and the injection parameter.
The qualitative character of this effect depends on The effects of argon and nitrogen injection are
the chemical composition of the injected gas and the practically the same. It is seen that for the carbon
injection parameters. dioxide gas there is an optimal injection parameter at
The physical mechanisms of this effect were inves- which the stabilizing effect is maximal.
tigated numerically using the Navier–Stokes equa-
tions and stability theory. 4.6. Surface State
It is possible to separate the following factors due to The surface state greatly influences the boundary-
which the gas injection can influence laminar–turbu- layer stability.
lent transition: growth of the boundary layer thickness
and effects connected with the chemical composition The experiments and the simulation results show
of the injected gas. that both single obstacles and distributed roughness
can cause waves of instability. The main parameters that
According to the data presented in [39] (without influence the wave type and the intensity of its growth are
considering the chemical reactions and molecular the ratio between the obstacle height and boundary-layer
vibrations) gas injection increases the boundary layer thickness and the shape of the obstacle.
thickness; as a result it increases the wavelength
(decreases amplitude) of the disturbance mode and As an example, it was found that the largest desta-
deforms the velocity vector of the averaged flow. As a bilizing effect is seen if the obstacle height is greater
result it causes area displacement from the wall with large than half of the boundary-layer thickness.
frictional stresses and the stabilizing effect of viscosity In general, this problem has been investigated
decreases. Here, in terms of hydrodynamics the gas insufficiently because different roughnesses and

HIGH TEMPERATURE Vol. 56 No. 1 2018

 MODERN APPROACHES FOR CALCULATING FLOW PARAMETERS 121

The relative coordinate of the transition

mutual positions can influence boundary-layer stabil-
ity in different ways. 0.90
In addition to roughness, the problem of the sur- 0.85
face porosity effect on the laminar–turbulent transi- 1
tion has been studied in detail. 0.80
In the experiments in [42–44] it was shown that 0.75
with respect to a solid impermeable wall a porous 0.70 2
coating greatly decreases the growth of disturbance of 0.65
the second mode, while it somewhat destabilizes the 3
first mode. It has been shown that at hypersonic flight 0.60
velocities a porous sound-absorbing coating can 4
0.55
increase the region of laminar flow.
0.50
0 0.1 0.2 0.3 0.4 0.5
5. AN ENGINEERING MODEL fw
OF THE LAMINAR–TURBULENT
TRANSITION IN A HIGH-VELOCITY FLOW Fig. 10. The relationship between the point of the lami-
nar–turbulent transition and the injection parameters for
Approaches based on stability theory and the eN different gases: 1, heated CO2, 2, CO2, 3, Ar, 4, N2.
method have been developed properly for high-veloc-
ity flows but they are of lesser interest from the practi-
cal point of view.
C 2ς1.5
Among engineering models, it is possible to distin- τ nt1 = eff
0.5
,
guish only [45–50]. However, this field is promising ⎡( 2E u ) 0.5 ν⎤
⎣ ⎦
due to its simplicity of implementation and the wide
application of automated software for computation of C 3 2ς eff
hydrodynamics, in which it is practically impossible to . τ nt 2 =
UP
use stability theory methods in contrast to engineering
methods whose essence is to solve additional semi- Here, ς eff is the effective length scale of the turbu-
empirical transport equations. lence [48, 49], ν is the coefficient of kinetic viscosity,
E u = 0.5 U e is the kinetic energy of the external flow,
2
In these works, the intermittency concept is used,
but there is a difference between them: models [45– and U P is the phase velocity for the second mode. This
49] are differential and model [50] is algebraic and is can be estimated as follows [45, 46]
based on the Cebici–Smith equations.
U P = 0.94U e,
If we compare these models, we see that models of
the [45–49] type are universal, while [50] is more simple. where U e is the external flow velocity.
Here, we present a model of the laminar–turbulent The following transport equation is used for deter-
transition, which we developed by generalizing the mining the intermittency factor [48, 49]
approach presented in [45–49]. In general, these ∂ ργ + ∂ ρu γ
approaches are similar to the Menter–Langtry approach; ( ) ( k )
∂t ∂x k
we add components that consider the contribution
⎡⎛ μ ⎞ ∂γ ⎤
= ∂ ⎢⎜ μ + eff ⎟
caused by non-turbulent fluctuations.
⎥ + Pγ ( Fonset ) − ε γ ,
Formula (8) is used for the effective viscosity. The ∂xk ⎣⎝ σ γ ⎠ ∂x k ⎦
turbulent viscosity is determined by the corresponding
Pγ ( Fonset ) = Pγ
turbulence model; for the non-turbulent viscosity, μ nt ,
0.5 ⎛ K ⎞d 
= C 4ρ Fonset [− ln (1 − γ )]
the following relationship is used:
⎜1 + C 5 2E ⎟ ν ∇ E u ,
μ nt = C μρ K τ nt , ⎝ u⎠
ε γ = γ Pγ ,
where C μ = 0.09 is a constant and K is the kinetic
energy of the turbulence. ⎛ ς K ∇K ⎞
0.5
Fonset = 1 − exp ⎜ −C6 eff ⎟,
The characteristic time τ nt consists of two parts: ⎝ ν ∇ Eu ⎠

τ nt = τ nt1 + τ nt 2, where d is the distance from the wall.

This equation is solved according to the recom-
where τ nt1, τ nt 2 are the characteristic times of the first mendations presented in [45–49] with zero initial and
and second mode of the disturbances, respectively. boundary conditions at the inlet of the computational
These are determined as follows: domain.

HIGH TEMPERATURE Vol. 56 No. 1 2018

122 BYKOV et al.

Table 1. Experimental parameters [51] systematizing and generalizing the accumulated mate-
Test 6911 6909, 6912, 6926 rial is of interest for further work.
6910 6913 In the present work we perform a brief review of the
existing procedures for predicting the laminar–turbulent
M 6.3 6.2 6.1 5.5 transition and the factors that influence this process.
Flow velocity, m/s 2980 3210 3370 4330 There are two different approaches for simulating
Static pressure, kPa 2.8 5.4 12.1 9.4 and analyzing the boundary-layer behavior during a
laminar–turbulent transition.
Static temperature, K 570 690 800 1560
In the classical approach the boundary layer is
Unit Reynolds number Re, 1.7 2.6 4.9 1.6 examined as a certain system that transforms external
106 m–1 disturbing signals. The aim of this approach is to
determine the physical mechanisms that are responsi-
ble for the intemittency and transition.
(а) (b) The engineering approach is based on CFD meth-
10–2 Test 6912 ods where additional semi-empirical transport equa-
Test 6913
tions are introduced artificially for characteristics that
St

10–3 determine the state of the boundary layer (for exam-

10–4
ple, intermittency). The aim of this approach is to
determine the area of the laminar–turbulent transi-
105 106 107 105 106 107 tion using computer aided engineering (computational
Rex Rex
fluid dynamics) software without a detailed analysis of
transition mechanisms.
Fig. 11. The Stanton-number distribution along a plate for Unfortunately, these approaches are developing
modes 6912 and 6913 (а) 6926 (b): experimental points independently. Both approaches have advantages and
[51], solid line, the results of calculations; dashed line,
laminar flow calculations (without considering the lami- disadvantages, which should be overcome.
nar–turbulent transition). In this work we presented an engineering model of
the laminar–turbulent transition developed by using
the intermittency concept and time scale values typical
The following constants are used [48, 49]: for different disturbance modes.
Cμ C1 C2 C3 C4 C5 C6 σγ Finally, when designing modern high-velocity fly-
ing vehicles it is necessary to make a complex exam-
0.09 0.7 0.35 0.005 8 × 10–5 0.07 1.2 1.0 ination of the phenomenon of the laminar–turbulent
transition by considering different factors that can
This is a phenomenological model; it does not con- influence the flow-around conditions.
sider the concrete mechanisms of the transition. The This problem can be solved only by using theoreti-
generation structure and constants for the model were cal and engineering methods with a large volume of
obtained in [45–49]. In this paper we generalize and experimental data.
use the approaches that were developed in [45–49].
To verify the model, the calculation results are REFERENCES
compared with the experimental data presented in
1. Tarasevich, S.E., Zlobin, A.V., and Yakovlev, A.B.,
[51], in which the hypersonic laminar–turbulent tran- High Temp., 2015, vol. 53, no. 6, p. 908.
sition is examined for the plate under conditions pre- 2. Kachanov, Y.S., Annu. Rev. Fluid Mech., 1994, vol. 26,
sented in Table 1. p. 411.
Some computation results for the Stanton number, 3. Morkovin, M.V., Critical evaluation of transition from
St, are presented in Fig. 11. laminar to turbulent shear layer with emphasis of hyper-
It is seen that the model is in good agreement with sonically travelling bodies, FFDLTech. Rep. 68-1-49,
1968.
the experimental data.
4. Mack, L.M., AIAA J., 1975, vol. 13, no. 3, p. 278.
5. White, E.B., Saric, W.S., Gladden, R.D., and Gabet, P.M.,
CONCLUSIONS Stages of swept transition, AIAA Pap. 2001-0271, 2001.
The laminar–turbulent transition in a boundary 6. Jie Ren, Song Fu, and Jianxin Liu, in Proc. 44th AIAA
Fluid Dynamics Conf., 2014.
layer is a complicated hydrodynamic phenomenon
that depends on a large number of factors. 7. Morkovin, M.V., On the many faces of transition, in
Viscous Drag Reduction, Wells, C.S., Ed., London: Ple-
A large volume of theoretical and experimental num, 1969, p. 1.
data on the regularities of this phenomenon has been 8. Herbert, T., Annu. Rev. Fluid Mech., 1997, no. 29, p. 245.
accumulated. Factors that impact the laminar–turbu- 9. Schmid, P.J., et al., Theor. Comput. Fluid Dyn., 1993,
lent transition have also been studied. The problem of vol. 4, no. 5, p. 227.

HIGH TEMPERATURE Vol. 56 No. 1 2018

 MODERN APPROACHES FOR CALCULATING FLOW PARAMETERS 123

10. Belov, I.A. and Isaev, S.A., Modelirovanie turbulentnykh 32. Yanbao Ma and Xiaolin Zhong, J. Fluid Mech., 2003,
techenii (Simulation of Turbulent Flows), St. Petersburg: vol. 488, p. 79.
Balt. Gos. Tekh. Univ., 2001. 33. Abu-Ghannam, B.J. and Shaw, R., J. Mech. Eng. Sci.,
11. Dhawan, S. and Narasimha, R., J. Fluid Mech., 1958, 1980, vol. 22, no. 5, p. 213.
no. 3, p. 418. 34. Gaponov, S.A. and Terekhova, N.M., Vest. Novosib.
12. Menter, F.R., Langtry, R.B., Likki, S.R., Suzen, Y.B., Gos. Univ., 2013, vol. 13, no. 4, p. 64.
Huang, P.G., and Völker, S., in Proc. ASME TURBO 35. Stetson, K.F., Hypersonic boundary layer transition
EXPO 2004, Vienna, 2004, ASME-GT2004-53452. experiments, Tech. rep. AFWAL-TR-80-3062, 1980.
13. Menter, F.R., Esch, T., and Kubacki, S., in Proc. 5th 36. Reshotko, E., Annu. Rev. Fluid Mech., 1976, no. 8, p. 311.
Int. Symposium on Turbulence Modeling and Measure- 37. Schneider, S.P., Hypersonic boundary-layer transition
ments, Spain, 2002. with ablation and blowing, AIAA Pap. 2008-3730, 2008.
14. Langtry, R.B. and Menter, F., Transition modeling for 38. Johnson, H.B., Gronvall, J.E., and Candler, G.V.,
general CFD applications in aeronautics, AIAA Pap. Reacting hypersonic boundary layer stability with
2005-522, 2005. blowing and suction, AIAA Pap. 2009-938, 2009.
15. Menter, F.R., Langtry, R., and Völker, S., Flow, Turbul. 39. Ghaffari, S., Marxen, O., Iaccarino, G., and
Combust., 2006, vol. 77, nos. 1–4, p. 277. Shaqfeh, E.S.G., in Proc. 48th AIAA Aerospace Science
16. Langtry, R.B. and Menter, F.R., AIAA J., 2009, vol. 47, Meeting Including the New Horizons Forum and Aero-
no. 12, p. 2894. space Exposition, 2010.
17. Walters, D.K. and Cokljat, D., J. Fluids Eng., 2008, 40. Leyva, I.A., Laurence, S., Beierholm, A.K.-W., Hor-
vol. 130, no. 12, p. 121401. nung, H.G., Wagnild, R., and Candler, G., Transition
18. Özgen, S. and Kırcalı, S.A., Theor. Comput. Fluid Dyn., delay in hypervelocity boundary layers by means of
2008, vol. 22, p. 1. CO2/acoustic instability interactions, AIAA Pap. 2009-
19. Fedorov, A.V., Annu. Rev. Fluid Mech., 2011, vol. 43, p. 79. 1287, 2009.
20. Demetriades, A., Hypersonic viscous flow over a slen- 41. Wagnild, R., High enthalpy effects on two boundary
der cone. Part III: Laminar instability and transition, layer disturbances in supersonic and hypersonic flow,
AIAA Pap. 74-535, 1974. PhD Thesis, Minneapolis: University of Minnesota, 2012.
21. Malik, M.R., AIAA J., 1989, vol. 27, no. 11, p. 1487. 42. Rasheed, A., Hornung, H.G., Fedorov, A.V., and Mal-
22. Surzhikov, S.T., High Temp., 2016, vol. 54, no. 2, p. 235. muth, N.D., AIAA J., 2002, vol. 40, no. 3, p. 481.
23. Bykov, L.V., Molchanov, A.M., Shcherbakov, M.A., 43. Fedorov, A., Shiplyuk, A., Maslov, A., Burov, E., and
and Yanyshev, D.S., Vychislitel’naya mekhanika splosh- Malmuth, N., J. Fluid Mech., 2003, vol. 479, p. 99.
nykh sred v zadachakh aviatsionnoi i kosmicheskoi tekh- 44. Gaponov, S.A. and Terekhova, N.M., J. Appl. Mech.
niki (Computational Mechanics of Continuous Media Tech. Phys., 2009, vol. 50, no. 5, p. 733.
in the Problems of Aviation and Space Technology), 45. Papp, J.L., Kenzakowski, D.C., and Dash, S.M.,
Moscow: Lenand, 2015. Extensions of a rapid engineering approach to modeling
24. Hudson, M.L. and Chokani, N., AIAA J., 1997, vol. 35, hypersonic laminar to turbulent transitional flows,
no. 6, p. 958. AIAA Pap. 2005-892, 2005.
25. Molchanov, A.M., Matematicheskoe modelirovanie 46. Papp, J.L. and Dash, S.M., A rapid engineering
zadach gazodinamiki i teplomassoobmena (Mathemati- approach to modeling hypersonic laminar to turbulent
cal Simulation of Gas Dynamics and Heat and Mass transitional flows for 2D and 3D geometries, AIAA Pap.
Transfer Problems), Moscow: Mosk. Aviats. Inst., 2013. 2008-2600, 2008.
26. Egorov, I.V., Pal’chekovskaya, N.V., and Shved- 47. Papp, J.L. and Dash, S.M., Modeling hypersonic lam-
chenko, V.V., High Temp., 2015, vol. 53, no. 5, p. 677. inar to turbulent transitional flows for 3D geometries
27. Zheleznyakova, A.L. and Surzhikov, S.T., High Temp., using a two-equation onset and intermittency transport
2014, vol. 52, no. 2, p. 271. models, AIAA Pap. 2012-0449, 2012.
28. Yanbao Ma and Xiaolin Zhong, J. Fluid Mech., 2003, 48. Wang, L. and Song Fu, Sci. China, Ser. G: Phys., Mech.
vol. 488, p. 31. Astron., 2009, vol. 52, no. 5, p. 768.
29. Maslov, A.A., Shiplyuk, A.N., Sidorenko, A.A., and 49. Wang, L. and Song Fu, Flow, Turbul. Combust., 2011,
Arnal, D., J. Fluid Mech., 2001, vol. 426, p. 73. vol. 87, p. 165.
30. Balakumar, P., Transition in a supersonic boundary 50. Gorskii, V.V. and Pugach, M.A., High Temp., 2015,
layer due to acoustic disturbances, AIAA Pap. 2005- vol. 53, no. 2, p. 223.
0096, 2005. 51. Mee, D.J., Boundary layer transition measurements in
31. Ghaffari, S., Marxen, O., Iaccarino, G., and hypervelocity flows in a shock tunnel, AIAA Pap. 2001-
Shaqfeh, E.S.G., Numerical simulations of hyper- 0208, 2001.
sonic boundary-layer instability with wall blowing,
AIAA Pap. 2010-706, 2010. Translated by Yu.V. Zikeeva

HIGH TEMPERATURE Vol. 56 No. 1 2018