You are on page 1of 6

Boundary layer receptiwity theory

E J Kerschen
Department of Aerospace and Mechanical Engineering, The University of
Arizona, Tucson, AZ 85721
The receptivity mechanisms by which free-stream disturbances generate
instability waves in laminar boundary layers are discussed. Free-stream
disturbances have wavelengths which are generally much longer than those
of instability waves. Hence, the transfer of energy from the free-stream
disturbance to the instability wave requires a wavelength conversion
mechanism. Recent analyses using asymptotic methods have shown that
the wavelength conversion takes place in regions of the boundary layer
where the mean flow adjusts on a short streamwise length scale. This
paper reviews recent progress in the theoretical understanding of these
phenomena.

1. INTRODUCTION
The important influence of free-stream disturbances on boundary layer transition has been appreciated for many years. Experiments clearly showed
that increased free-stream
disturbance levels
enhanced the amplitudes of instability waves in the
boundary layer, hastening transition. However, the
physical mechanism by which energy is transferred
from the long wavelength, free-stream disturbances
to the short wavelength, boundary layer instabilities
was not understood. This came to be known as the
receptivity problem.
In order for an external disturbance to generate an
instability wave, energy must be transferred to the
unsteady motion in the boundary layer at an appropriate combination of frequency and wavelength. To
simplify the discussion, it is useful to consider the
situation where the external disturbance is of small
enough amplitude that the unsteady motion can be
represented as a linear perturbation of the mean
flow. Attention can then be restricted to a single
time harmonic, with the results for general time
dependence obtained by superposition. This paper
focuses on linear, time harmonic unsteady flows.
The earliest theoretical analysis of instability wave
generation in a boundary layer was presented by
Gaster (1965). He considered the case of twodimensional Tollmien-Schlichting wave excitation in
a parallel boundary layer by a time harmonic disturbance at the wall. The wall disturbance was localized
in the streamwise direction, and hence the wave-

Appl Mech Rev vol 43, no 5, Part 2, May 1990

number spectrum of this disturbance was broad.*


Thus, the input disturbance contained energy at the f
appropriate frequency-wavelength combination t;.. i
directly excite an instability wave.
<
The first attempts to predict receptivity to natur-1
ally occurring free-stream disturbances (Rogler and I
Reshotko, 1975; Tarn, 1981; Mack, 1975) were based \
on a parallel flow formulation similar to that of;
Gaster. However, naturally occurring free-stream}
disturbances (sound waves, turbulence, etc.) travel a! j
much higher speeds than instability waves. Thus, the j
wavenumber spectrum of the free-stream distur-;
bance at a given temporal frequency is concentrated!
at wavenumbers which are substantially differed j
from the wavenumber a 0 of the instability wave. I
Hence, these parallel flow analyses succeeded only in [
finding "particular solutions" which are unrelated to \
the instability waves. In order to transfer energy j
from a naturally occuring free-stream disturbance to \
an instability wave, a wavelength conversion process:is required. A more complete discussion of the j
differences between instability wave generation by
localized disturbances and by naturally occuring dis- \
turbances can be found in Kerschen (1989).
j
Experimental investigations in the 1970s showed |
that receptivity can occur in the vicinity of the lead- j
ing edge and at localized downstream locations, j
Reviews of these and more recent experiments arc j
presented by Kachanov, Kozlov and Levchenko j
(1982) and Nishioka and Morkovin (1986). The;
experimental results stimulated theoretical investiga- =
tions, which showed that the wavelength conversion I
process takes place in regions of the boundary layer >

S152

Copyright 1990 American Society of Mechanical Engineers j

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 02/07/2015 Terms of Use: http://asme.org/terms

Appl Mech Rev vol 43, no 5, Part 2, May 1990

Kerschen: Boundary layer receptivity theory

S153

where the mean flow exhibits rapid changes in the


streamwise direction. This occurs (a) near the body
leading edge and (b) in any region farther downstream where some local feature forces the boundary
layer to adjust on a short streamwise length scale.
The rapid streamwise adjustment requires that nonparallel mean flow effects be included at leading
order, in contrast to the parallel flow assumption of
classical Orr-Sommerfeld stability theory. This paper
discusses recent progress in theories for receptivity
produced by small-amplitude acoustic and vortical
free-stream disturbances interacting with the leading
edge of a flat plate and with a localized mean flow
adjustment downstream of the leading edge.
Reviews of receptivity by other authors are presented in Kozlov and Rizhov (1987) and Goldstein
and Hultgren (1989). A discussion of the role of
receptivity in transition prediction can be found in
Heinrich, Choudhari and Kerschen (1988).

earized unsteady boundary layer equation (LUBLE).


Farther downstream where x = e2X = O(l), the unsteady flow satisfies the classical large Reynolds
number, small wavenumber approximation to the
Orr-Sommerfeld equation (OSE). Goldstein examines the LUBLE for X 1 and the OSE for x 1,
and shows that the first asymptotic eigenfunction of
the LUBLE, with amplitude C l 5 matches onto the
T-S wave which eventually exhibits growth farther
downstream. Hence the amplitude of the T-S wave
is linearly proportional to C l 5 which we call the
"receptivity coefficient." To find C 1 ; a numerical
solution of the LUBLE is required. Unfortunately,
for X 1 the first eigenfunction makes only an
exponentially small contribution to the solution. In
order to extract C^ from the numerical calculation,
the equation is analytically continued into the complex plane and solved along a ray where the first
eigenfunction dominates the solution. Further details
may be found in Heinrich (1989).

2. LEADING EDGE RECEPTIVITY

Figure 1 is a plot of the receptivity coefficient for


a convected gust as a function of disturbance orientation. The velocity and frequency of the freestream disturbance are held constant as the orientation is varied. Free-stream convected gusts can be
interpreted as a two-dimensional model of weak
free-stream turbulence. The convected gust is vortical, has no pressure fluctuations, and the velocity
fluctuations are perpendicular to the wavevector.
The maximum and minimum levels of receptivity are
found for free-stream velocity perturbations parallel
(9 = 90) and perpendicular to the plate surface,
respectively. The weak receptivity levels produced
by the perpendicular component of free-stream
velocity are rather surprising, since this component
produces both a singular flow field near the leading
edge and a significant slip velocity in the doubly infinite plate limit far downstream. Essentially, 0(1)
receptivity levels are produced by the individual
contributions (such as the singular flow near the
leading edge) to the slip velocity, but for the convected gust the amplitudes and phases of these various contributions lead to a high degree of cancellation. Hence, the receptivity for convected gusts is
produced mainly by the parallel component of the
free-stream disturbance velocity.

In this section, we examine receptivity in the leading edge region of the Blasius boundary layer.
Results are presented for both free-stream acoustic
waves and convected gusts. The analysis considers
low Mach number, two-dimensional flow. The flow
far from the plate consists of a small amplitude, harmonic disturbance superposed on a uniform mean
flow U 0 . The Reynolds number is assumed large,
and hence the outer problem for the unsteady flow
corresponds to the inviscid interaction of the small
amplitude free-stream disturbance with the semiinfinite flat plate. This outer solution provides the
distributions of pressure and slip velocity which
drive the unsteady motion in the boundary layer on
the plate. Since the free-stream disturbances are
assumed to be of small amplitude, linear superposition is valid.
Hence, the incident disturbance
velocity can be separated into components parallel
and perpendicular to the plate surface. The parallel
component contributes directly to the slip velocity on
the plate surface, while the perpendicular component
contributes to the slip velocity via its scattering by
the plate surface. Near the leading edge, this scattered component has a square root singularity corresponding to inviscid flow around the sharp edge,
while far downstream the scattered component takes
a simple form appropriate to a doubly infinite plate.
Goldstein's (1983) analysis of the unsteady flow in
the boundary layer shows that the asymptotic solution for e 1 (T6 = Ug/wv) contains two distinct
streamwise regions. The receptivity occurs in the
first region near the leading edge, where
X= x'w/U0 = O(l) and the motion satisfies the lin-

The receptivity produced by free-stream acoustic


waves is illustrated in Fig. 2, for two values of the
Mach number, M = 0.01 and 0.1. Here, 0 = 0 corresponds to incident acoustic waves propagating
downstream parallel to the plate surface, and was the
case considered by Goldstein, Sockol, and Sanz
(1983). The receptivity for this case is found to be
only about 1/4 that produced by a convected slip
velocity parallel to the plate surface (Fig. 1,0 = 90).

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 02/07/2015 Terms of Use: http://asme.org/terms

S154

MECHANICS USA 1990

Appl Mech Rev 1990 Supplement

4.0-

180k0

Fig. 1. Leading edge receptivity coefficient for a


convected gust.
The explanation for this may be the slower phase
speed of the convected gust compared to the acoustic
wave. However, the receptivity to acoustic waves
rises rapidly as the incidence angle increases, reaching a maximum value at 6 = 180, which corresponds
to an acoustic wave propagating from downstream
infinity toward the leading edge. The high receptivity for oblique acoustic waves is caused by the perpendicular component of the free-stream disturbance
velocity. This component produces a flow around
the leading edge whose strength increases as M"1/2 as
the Mach number decreases and the acoustic wavelength becomes large compared to the hydrodynamic
length scale of the problem.

waves for the case of a semi-infinite flat plate centered within a channel. The slip velocity was calculated using a Wiener-Hopf analysis and confirmed
for low frequencies by a matched asymptotic expansion analysis (MAE). Figure 3 contains results as a
function of channel width H for a free-stream Mach
number of O.J. The receptivity coefficient grows
rapidly with increasing H and reaches a maximum
Cx = 35 at uH/c = TT. For higher values of wH/c, Cj
exhibits a regular pattern of weakly damped oscillations, gradually approaching the isolated plate result
Cj = 20 (see Fig. 2). The explanation for this behavior can be found in the alternate cut-on of upstream and downstream traveling acoustic modes.
40-.
Wiener - Hopf
saSU

Fig. 3. Effects of finite channel width on the leading edge receptivity coefficient for an upstream
traveling acoustic wave.
3. LOCALIZED RECEPTIVITY MECHANISMS ,

180.0

Fig. 2. Leading edge receptivity coefficient for a


plane acoustic wave.
The above result applies only for cases where the
acoustic wavelength is short compared to both the
plate length and the distance to any surrounding surfaces. To address the effect of nearby surfaces, we
have analyzed the receptivity to upstream travelling

In this section, we examine receptivity mechanisms f


which occur when some local feature downstream of j
the leading edge causes the boundary layer to adjus! j
on a short streamwise length scale. Examples of local j
features are short scale wall humps or suction strips i
or shock - boundary layer interaction. The asymp- \
totic descriptions of the mean and unsteady flow \
components in the local region involve triple deck j
structures. For sufficiently weak mean flow distur-1
bances (small hump height, for example), the mean j
flow triple deck equations can be linearized and \
solved in closed form. The unsteady flow can then \
be solved via Fourier transforms, and the instability j
wave is given by a residue in the complex wave-1
number plane.
!
For localized receptivity mechanisms, the wave- ;
length conversion process is produced by the un- [
steady flow conforming to the local short-seal';
geometry. Essentially, the long wavelength, unstead; j

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 02/07/2015 Terms of Use: http://asme.org/terms

Appl

Mech Rev vol 43, no 5, Part 2, May 1990

Kerschen: Boundary layer receptivity theory

free-stream disturbance temporally modulates the


local short scale mean flow, generating frequencywavelength combinations that match the instability
wave.
The detailed physics of the modulation
processes are somewhat different for acoustic and
convected free-stream
disturbances, and are
described separately.
First consider the case of acoustic free-stream disturbances (Goldstein, 1985; Ruban, 1985). The freestream pressure fluctuations penetrate the boundary
layer and produce velocity fluctuations, which are
brought to rest at the wall by a thin Stokes wave of
the same thickness as the lower deck. The receptivity to acoustic waves arises in the adjustment of the
Stokes wave to the local geometry, and hence the
source of the instability wave is near the wall, deep
inside the mean boundary layer. This basic mechanism is equally applicable to other disturbances which
produce free-stream pressure fluctuations, such as
the concentrated vortex considered in Kozlov and
Rizhov (1987). Receptivity can also be generated by
an acoustic wave scattering off a short scale variation
in wall impedance (Heinrich, Choudhari, and
Kerschen, 1988), even in the absence of a mean flow
adjustment. The latter mechanism is closely related
to the action of a vibrating piston on the wall.
Now consider the localized receptivity process for
convected gusts (Kerschen, 1989). In contrast to an
acoustic wave, a convected gust contains no pressure
fluctuations. Hence, away from the local region, the
free-stream velocity fluctuations are transmitted to
the boundary layer only by viscous diffusion. At
high Reynolds numbers, this process is very weak
and the unsteady motion is mainly concentrated at
the outer edge of the boundary layer. In fact, the
amplitude of the unsteady disturbance in the lower
deck near the wall is exponentially small. Hence, the
near wall interaction is not a significant source of
receptivity in this case. However, the hump produces short scale mean flow gradients in the upper
deck just outside the mean boundary layer. The inviscid interaction of the convected gust with these
mean flow gradients produces short scale pressure
fluctuations, which penetrate the boundary layer and
generate instability waves.
For free-stream disturbances interacting with small
amplitude, localized, two-dimensional mean flow
disturbances, we have
U

TS = e " "sup F ( a 0 )

K ) ^<y) e '

[OfgX " Wt)

(l)

where u TS is the Tollmien-Schlichting wave streamwise velocity fluctuation with wavenumber aQ and

S155

mode shape <Ky), uaiip is the amplitude of the slip


velocity produced by the free-stream disturbance,
F(Q 0 ) is the spatial spectrum of the local inhomogeneity (wall hump, etc.) evaluated at the wavenumber
of the instability wave, and A(a0) is the characteristic
function for the specific type of receptivity mechanism under consideration. The parameter e = Ref;1/8,
where Re L is the Reynolds number based on distance
from the leading edge; for an acoustic wave n = 0,
while for a convected gust n = 1. The receptivity to
acoustic waves is larger because the boundary layer is
more sensitive to forcing at the wall than to forcing
at its outer edge. For an acoustic wave, the slip
velocity is lu^cosf?, where 8 = 0 corresponds to
propagation parallel to the wall and 6 = 90 corresponds to an acoustic wave normally incident on the
wall. For a convected gust, it can be shown that the
slip velocity equals the velocity of the upstream gust
for all orientations of the gust wavevector.
The A functions for free-stream acoustic waves
and convected gusts interacting with a localized wall
hump are plotted as a function of the normalized
frequency in Fig. 4. The functions are quite similar
for low frequencies. At moderate frequencies the A
associated with acoustic waves is somewhat larger
than that associated with convected gusts. The
acoustic A approaches a constant at high frequencies,
since the amplitude of the Stokes wave is independent of frequency, while the convected gust A decreases at high frequencies, since only weak pressure
fluctuations are produced by the inviscid distortion
of a very short wavelength gust.

2-D n e u t r a l

S= Re^'Z-t vL/U

Fig. 4. The efficiency functions A^ and Acg for an


acoustic wave and a convected gust interacting with
a wall hump (S 0 = Re^""1/4 wL/U).
Equation (1) also shows that the geometry of the
wall hump or other local disturbance significantly influences the receptivity coefficient. For example, a
Mylar strip produces the highest receptivity coeffi-

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 02/07/2015 Terms of Use: http://asme.org/terms

S156

MECHANICS USA 1990

cient when its length is a half integer multiple of the


instability wavelength. Equation (1) was derived
from a linear theory, and hence the receptivity is
linearly proportional to the amplitude of the mean
flow disturbance. Bodonyi et al. (1989) have examined the influence of large hump heights, and conclude that nonlinear effects increase the receptivity
level for that case. The influence of subsonic, compressible flow can be found in Choudhari (1990),
and the impact of receptivity on laminar flow control
via suction is discussed in Heinrich et al. (1988).
Nishioka and Morkovin (1986) have emphasized
the role of streamwise gradients of unsteady freestream pressure as a source of receptivity. This concept is perfectly consistent with the above discussion.
In fact, the localized receptivity to convected gusts is
a direct example of instability wave generation by
the action of streamwise gradients of unsteady pressure. In this case the unsteady pressure fluctuations
are caused by the interaction of a naturally occurring
free-stream disturbance with a local mean flow perturbation. However, the excitation of instability
waves by an external source, such as a loudspeaker
located just outside the boundary layer, can also be
analyzed in a similar manner (Kerschen, 1989;
Heinrich, Gatski, and Kerschen, 1989).
These analyses have also been extended to the case
of localized three-dimensional adjustments to a nominally two-dimensional boundary layer (Choudhari
and Kerschen, 1989; Choudhari, 1990; Choudhari
and Kerschen, 1990). For the case of an isolated,
three-dimensional mean flow disturbance, the interaction of a harmonic free-stream acoustic wave with
the local short scale mean flow gradients generates
instability waves in a symmetric, wedge shaped
region downstream of the element. The direction
0 max , corresponding to the maximum instability amplitude, and the angle <j>max, corresponding to the
orientation of the associated phase lines, are plotted
as a function of frequency in Fig. 5. At low and
high frequencies, the maximum instability amplitude
lies directly downstream of the element, while for
moderate frequencies the maximum amplitude
appears at a modest angle to the downstream direction. This angle reaches a maximum of almost 13 at
approximately twice the two-dimensional lowerbranch neutral frequency. For S0 > 1.38, #max is
identically zero.
An increase in the normalized frequency S0 can
also be interpreted as an increase in the boundary
layer thickness with the freqeuncy held constant. In
this regard, it is interesting to compare the prediction
of the present theory with the experiment of Gilev et
al. (1981), who measured the lines of constant phase

Appl Mech Rev 1990 Supplement !

downstream of a time-harmonic point source in a j


growing boundary layer. Gilev et al. found that the
lines of constant phase were very oblique just downstream of the source (corresponding to small S0), and
that the constant phase lines gradually took on a
two-dimensional form as the boundary layer thickened with downstream distance, in qualitative agreement with the present prediction.

s 0 = n&i1/* wiyu

Fig. 5. Direction <9max = tan"1 z/x of maximum instability wave amplitude and orientation <^max of
constant phase lines along this direction.
The influence of acoustic wave orientation on the
receptivity level for isolated three-dimensional mean
flow disturbances is illustrated in Fig. 6. At low frequencies, the strongest receptivity is produced by
acoustic waves propagating in the spanwise direction,
while at high frequencies the strongest receptivity is
produced by acoustic waves propagating parallel to
the flow direction. Essentially, the energy transfer
from the acoustic motion to the instability wave is
most efficient when the acoustic velocity is perpendicular to the lines of constant phase in the resulting
instability wave pattern. The influence of roughness
element size and shape has also been examined. For
roughness elements of circular planform, the maximum receptivity occurs for elements having a radius
of approximately one-third the T-S wavelength.
Roughness elements whose planform is elongated in
the flow direction produce the strongest instability
waves at low frequencies, while at high frequencies
the strongest receptivity is produced by elements
which are elongated in the spanwise direction. This
result is also related to the frequency dependence of
the phase line orientation. For additional information, see Choudhari (1990) and Choudhari and
Kerschen (1990).

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 02/07/2015 Terms of Use: http://asme.org/terms

Appl Mech Rev vol 43, no 5, Part 2, May 1990

Kerschen: Boundary layer receptivity theory

legend
So = 0,44
H

(jss4

$J

04

-90.0

-45.0

0.0

4S.0

90.0

vac (degrees)
Fig. 6. Normalized receptivity coefficient as a function of acoustic wave propagation angle $
| ACKNOWLEDGEMENTS
;

Finanacial support for this research was provided


by the National Science Foundation under grant
ME A- 8351929 and by the McDonnell Douglas
Research Laboratory. The author would like to
thank Dr. M.E. Goldstein for a number of stimulating discussions.
REFERENCES
Bodonyi, R.J., Welch, W.J.C., Duck, P.W., and
Tadjfar, M. (1989), A numerical study of the interaction between unsteady free-stream disturbances and localized variations in surface geometry, /. Fluid Mech., Vol. 209, 285-308.
Choudhari, M. (1990), Boundary layer receptivity in
laminar flow control applications, Ph.D. Dissertation, University of Arizona.
Choudhari, M. and Kerschen, E.J. (1989), Receptivity to three-dimensional convected gusts,"
Proceedings of ICASE / NASA Langley Instability and Transition Workshop, Springer-Verlag.
Choudhari, M. and Kerschen, E.J. (1990), Instability
wave patterns generated by interaction of sound
waves with three-dimensional wall suction or
roughness, AIAA Paper 90-0119.
Gaster, M. (1965), On the generation of spatially
growing waves in a boundary layer, J. Fluid
Mech.. Vol. 22, 433-441.
Gilev, V.M., Kachanov, Y.S., and Kozlov, V.V.
(1981), Development of spatial wave packets in a
boundary layer, Preprint 34-81, Inst, of Theoreti-

S157

cal and Applied Mathematics, USSR Acad. Sci.,


Novosibirsk.
Goldstein, M.E. (1983), The evolution of TollmienSchlichting waves near a leading edge, J. Fluid
Mech., Vol. 127, 59-81.
Goldstein, M.E. (1985), Scattering of acoustic waves
into Tollmien-Schlichting waves by small streamwise variations in surface geometry, J. Fluid
Mech., Vol. 154, 509-529.
Goldstein, M.E. and Hultgren, L.S. (1989), Boundary-layer receptivity to long-wave free-stream disturbances, Ann. Rev. Fluid Mech., Vol. 21,
137-166.
Goldstein, M. E., Sockol, P. M. and Sanz, J. (1983),
The Evolution of Tollmien-Schlichting Waves
Near a Leading Edge. Part2. Numerical Determination of Amplitudes, J. Fluid Mech., Vol. 129,
443-453.
Heinrich, R.A.E. (1989), Flat plate leading edge
receptivity to various freestream disturbance
structures, Ph.D. Dissertation, University of Arizona.
Heinrich, R.A.E., Choudhari, M. and Kerschen, E.J.
(1988), A comparison of boundary layer receptivity mechanisms, AIAA Paper 88-3758.
Heinrich, R.A.E., Gatski, T.B. and Kerschen, E.J.
(1989), Boundary layer receptivity to unsteady
free-stream pressure gradients, Proceedings of
ICASE / NASA Langley Instability and Transition
Workshop, Springer-Verlag.
Kachanov, Yu.S., Kozlov, V.V., and Levchenko,
V.Ya. (1982), Initiation of Turbulence in Boundary Layers, Novosibirsk: Nauka Publ. Siberian
Div.
Kerschen, E.J. (1989), Boundary layer receptivity,
AIAA Paper 89-1109.
Kozlov, V.V., and Rizhov, O.S. (1987), Boundary
layer receptivity, Invited Lecture at Euromech
334, Exeter.
Mack, L. M. (1975), Linear stability theory and the
problem of supersonic boundary layer transition,
AIAA Journal Vol. 13, 278-289.
Nishioka, M. and Morkovin, M.V. (1986), Boundary
layer receptivity to unsteady pressure gradients:
Experiments and overview, /. Fluid Mech., Vol.
171, 219-261.
Rogler, H. L. and Reshotko, E. (1975), Disturbances
in a Boundary Layer Introduced by a Low Intensity Array of Vortices, SIAM J. Appl. Math., Vol.
28, No. 2, 431-462.
Ruban, A.I. (1985), On the generation of TollmienSchlichting waves by sound, Transl. in Fluid
Dyn., Vol. 19, 709-716.
Tam, C. K. W. (1981), The Excitation of TollmienSchlichting Waves in Low Subsonic Boundary
Layers by Freestream Sound Waves, J. Fluid
Mech. , Vol. 109, 483-501.

Downloaded From: http://appliedmechanicsreviews.asmedigitalcollection.asme.org/ on 02/07/2015 Terms of Use: http://asme.org/terms

You might also like