Recent Advances in the Mechanics of Boundary Layer Flow

BY HUGH L. DRYDEN]
National Bureau of Standards

CONTENTS
Page
I. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
11. Laminar Flow in Boundary Layers . . . . . . . . . . . . . . . . . . 4
Bibliography on Laminar Flow in Boundary Layers. . . . . . . . . . 5
111. Stability of Laminar Boundary Layer Flow . . . . . . . . . . . . . . 8
1. Status of Problem in 1938. . . . . . . . . . . . . . . . . . . . 8
2. Observation of Tollmien-Schlichting Oscillations. . . . . . . . . . . 10
3. Relation of Tollmien-Schlichting Waves to Transition . . . . . . . 14
4. Effect of Curvature . . . . . . . . . . . . . . . . . . . . . . 16
5. Effect of Pressure Gradient . . . . . . . . . . . . . . . . . . . 16
6. Effect of Compressibility . . . . . . . . . . . . . . . . . . . . 22
Bibliography on Stability of Laminar Boundary Layer Flow and Transition 22
IV. Boundary Layer Suction. . . . . . . . . . . . . . . . . . . . . . . 23
Bibliography on Boundary Layer Suction . . . . . . . . . . . . . . 24
V. TurbuIent Flow in Boundary Layers . . . . . . . . . . . . . . . . . 25
1. Development of Empirical Methods of the Buri-Gruschwitz Type . . 25
2. Developments Proceeding from Reynolds Theory of Turbulent Stresses 27
3. National Bureau of Standards Experiments . . . . . . . . . . . . 28
Bibliography on Turbulent Flow in Boundary Laiers. . . . . . . . . 38
Bibliography of Related Papers. . . . . . . . . . . . . . . . . . . 39

NOTATION
2 = length
m = parameter in velocity distribution U = kzm
p = static pressure
r = radius of curvature of surface
u = local mean velocity, always parallel to surface
u' = root mean square component of turbulent velocity fluctuation parallel to the
surface and to the mean flow
u' = root mean square component of turbulent velocity fluctuation normal to the
mean flow and to the surface
20' = root mean square component of turbulent velocity fluctuation normal to the
mean flow but parallel to the surface
x = distance along surface from leading edge
XI = distance from leading edge to transition point
y = distance normal to the surface
1 Now Director of Aeronautical Research, National Advisory Committe? for
Aeronautics.
1
2 HUGH L. DRYDEN

z = Mangler parameter U6*/vt

C , = ~ / i p U a coefficient
, of turbulent shearing stress based on U
CI,= ~ / i p U ~coefficient
e, of turbulent shearing stress based on UQ
D = reference dimension of body, diameter of sphere, minor axis of elliptical
cylinder
H = &*/& a parameter describing turbulent velocity distribution in boundary
layers with pressure gradient
Z = scale of turbulence, defined as the integral of the correlation coefficient
between longitudinal velocity fluctuations a t points at distance y apart in a
direction transverse to the flow for g varying from zero to infinity
R = US*/v, Reynolds number based on displacement thickness
R, = critical Reynolds number below which all oscillations in a laminar boundary
layer are damped
RCRIT= U D / v , critical Reynolds number of a sphere, i.e. that for which the drat
coefficient ia 0.3
ue
-
Re = - 2 Reynolds number baaed on momentum thickness
Y

U mean speed in free stream just outside boundary layer, usually varying
with x
U’ dU -
-
P
dx
U O reference speed a t fixed point; for National Bureau of Standards turbulent
boundary layer a t x = 17 f t
V o = suction velocity normal to wall in boundary layer suction through a porous
wall
ct == 2~ times wave length of oscillation; also angle of principal axis of turbulent
stress tensor to flow direction
fl = Hartree parameter in velocity distribution U = b m , equal to 2rn/(l + m)
@, 2r times frequency of oscillation
@i a amplification coefficient
6 = boundary layer thickness; if otherwise unspecified, that of the Pohlhausen
four-term approximation

8’ = displacement thickness of boundary layer = J*(1 -); dY

8 = momentum thickness of boundary layer =

=Pohlhausen parameter U ’ ~ * /for V four-term approximation
b-6
p -
= Pohlhausen parameter
viscosity of fluid
p = density of fluid
for six-term approximation

Y = p / p , kinematic viscosity of fluid

T = turbulent shearing stress, at 5, 21

rV 7 turbulent shearing stress at the wall

I. INTRODUCTION
In 1904 L. Prandtl proposed the concept of a boundary layer near the
surface of a body moving through a fluid, a region of comparatively small
 MECHANICS O F B O U N D A R Y L A Y E R F L O W 3
thickness within which the viscous forces are comparable to the inertial
forces and within which the relative speed between the body and the
fluid decreases rapidly t o zero. He derived the equations of flow in the
boundary layer from the Navier-Stokes equations, and his pupil, Blasius,
in 1908 obtained their solution for the special case of flow along a thin
flat plate parallel to the direction of flow in an airstream of uniform speed.
The theory gave both the skin friction and the velocity distribution
within the layer. While experimental data on skin friction were avail-
able at that time, experimental observations on the velocity distribution
were first made by Burgers and van der Hegge Zijnen in 1925.
Comparisons of skin friction data soon showed certain discrepancies
and it became clear that the boundary layer flow did not always conform
to the Prandtl equations. It had been known since the work of Osborne
Reynolds that flow in a pipe did not always correspond to the usual
solution of the Navier-Stokes equations, a type of flow usually designated
laminar $ow. The pipe flow was often turbulent. The flow in a bound-
ary layer may likewise be laminar or turbulent; and although this was
at first a mere hypothesis, it has since been abundantly confirmed.
This concept of Prandtl has been extraordinarily fruitful in the devel-
opment of fluid mechanics. Many puzzling phenomena have been
clarified by a study of the nature of the flow within the boundary layer
and its effect on the general flow around the body. Scale effect in
aerodynamic measurements, the effect of wind tunnel turbulence, flow
separation phenomena, minimum drag and maximum lift, can be under-
stood and treated, a t least qualitatively if not always quantitatively, by
consideration of the boundary layer flow. Even a t high speeds near the
speed of sound it appears that the general flow pattern and location of
shock waves are dependent on the type of flow in the boundary
layer.
The last extensive review of the mechanics of boundary layer flow
published in the English language was contained in the book Modern
Developments in Fluid Dynamics by S . Goldstein and others (Oxford,
1938). German periedicals now available refer to a series of lectures by
H. Schlichting on boundary layer theory published in 1942 but this
document of 279 pages with 116 figures is not yet generally available in
this country. The present report is an attempt to summarize or give
references to some of the more important papers published since 1936 that
may contribute to a better understanding of the mechanics of boundary
layer flow. Most of the work cited relates to incompressible flow
although references are given to such papers as are available on com-
pressible flow. Because of the current difficulties in covering the Iitera-
ture of the world, this survey is of necessity incomplete.
4 HUGH L. DRYDEN

11. LAMINARFLOWI N BOUNDARY LAYERS

The problems of laminar flow in boundary layers are largely mathe-
matical problems, since it is certain that the fundamental mechanical
principles are fully understood and that the equations adequately describe
the phenomena when the flow may be regarded as incompressible. A
bibliography of papers on this subject during the decade 1936-1946 that
have come to the author’s attention is given a t the end of this section.
Only a few highlights will be discussed, since the developments follow the
same general methods of attack as those described in Modern Develop-
ments in Fluid Dynamics.
H. Schmidt and K. Schroder (33,34) have attempted to clarify the
boundary layer theory and make it more precise from the mathematical
point of view. The discussion is limited t o two-dimensional flow and
does not attempt to deal with the question of proving the existence or
uniqueness of the solution. The papers thus become descriptions of the
boundary layer that are intellectually satisfying to the mathematician
because they are stated in the language of limits and inequalities.
The principal problem of the theory of the laminar boundary layer
is the calculation of the flow for a given pressure distribution and, as in the
past, most of the work accomplished has been limited‘to two-dimensional
flow. The problem becomes especially difficult in a region of adverse
pressure gradient as the position of separation is approached. The
method of expansion in series is not practicable, step-by-step methods are
extremely laborious, and the results obtained from application of the
momentum equation to assumed families of velocity distributions are
unreliable. The method of K&rmBn and Millikan has been frequently
used (6) but is known to be subject to some error.
The application of the momentum equation to assumed families of
velocity distributions as first carried out by Pohlhausen remains popular,
and many variations in detail have been proposed, in spite of the fact that
von Mises demonstrated long ago that accurate values of laminar skin
friction can not be determined by this method. The method is useful in
giving at least a general picture of the variation of the thickness of the
laminar boundary layer provided the point of separation is not approached
too closely. As an example of one type of variation of procedure,
Mangler (24) uses the momentum equation as an integral equation for
the quantity z = U&2/vlrather than as a differential equation for the
parameter X = U’A2/v. Many types of nomograms have been constructed
to facilitate the solution by this general method.
Further study has been made of the so-called “similar” solutions
(4,13,40,43) of the boundary layer equations for which the partial dif-
 MECHANICS OF' BOUNDARY LAYER FLOW 5

ferential equations reduce to a siLgle ordinary differential equation.

These solutions can occur only if the distribution of velocity outside the
boundary layer is proportional to a power of the distance x along the
wall or to ez. Mangler (40) discusses these solutions at some length,
giving the body shapes for which such distributions occur and investigat-
ing the asymptotic behavior of the solutions a t the outer boundary.
The physically meaningful solutions can be distinguished among the
solutions satisfying the boundary conditions by the fact that the dis-
placement thickness remains finite and the velocity in the boundary
layer remains smaller than that outside.
Discussion of the work done on laminar boundary layers along porous
or slotted walls with suction or pressure is postponed to a later section.
Very little additional work has been done during the last decade on
three-dimensional problems. There are several papers on the axially
symmetric problems for bodies of revolution (16,26,30,36), the paper of
Burgers (20) on flows with a rotational component, and papers on flow
toward a plane surface (37) and flow in a corner (45). It is understood
that the Gottingen group have succeeded in showing that the differential
equation for the axially symmetric case can be reduced to the differential
equation of the two-dimensional boundary layer by a simple transforma-
tion but details are not available. The same group had in progress the
behavior of a boundary layer under the influence of a pressure gradient
transverse t o the flow direction.
There are several papers listed on the effects of compressibility on
laminar boundary layer flow (7,1$,21,22,27,28). At supersonic speeds
the aerodynamic and thermodynamic problems become intertwined and
the thermal environment becomes of great importance. The papers of
Ackeret, Feldmann, and Rott (42) and of Liepmann (44) are of great
importance as demonst,rating the influence of the state of the boundary
layer flow on the general flow pattern and as indicating the breakdown
of the usual assumption of the boundary layer theory that the pressure
differences across the layer may be neglected.

Bibliography
on
Laminar Flow in Boundary Layers
1. BLAISDELL, A. H., Boundary layer flow over flat and concave surfaces, A.S.M.E.
Trans., 68, 343-348 (1936).
2. HOMANN, F., Der Einfluss grosser Ziihigkeit bei der Stromung um den Zylinder
und urn die Kugel, 2.angew. Mnlh. Mech., 16, 153-164 (1936).
3. HOWARTII, Id., Velocity and temperature distribution for a flow along a flat plate,
PTOC. Roy. SOC. (London), A , 164, 364-377 (1036).
6 HUGH L. DRYDEN

4. HARTREE,D.R.,On an equation occurring in Falkner and Skan’s approximate

treatment of the equations of the boundary layer, Proc. Cambridge Phil. SOC.,
31,223-239 (1937).
5. SUTTON, W. G. L., Approximate solution of the boundary layer equation for a
flat plate, Phi2. Mag., 28, 1146-1152 (1937).
6. VONDOENHOFB, A. E., A method of rapidly estimating the position of the laminar
eeparation point, N.A.C.A. Tech. Note 671 (1938).
7. VON KLRMLN,T., and TSIEN,H. S., Boundary layer in compressible fluids,
J . Aeronaut. Sci., 6, 227-232 (1938).
8. HOWABTH, L., Solution of the laminar boundary layer equation, Proc. Roy. Soc.
(London),A , 104,547-579 (1938).
9. MILLS, R. H.,A note on some accelerated boundary layer velocity profiles,
J . Aeronaut. Sci., 6, 325-327 (1938).
10. P R A N DL.~,,Zur Berechnung der Grensschichten, Z . angew. Math. Mech., 18,
77-82, 1938;also N.A.C.A. Tech. Memo. 969 (1940); also J. Roy. Aeronaut.
SOC.,44,795-801 (1940).
11. G~RTLER, H.,Weiterentwicklung eines Grensschichtprofils bei gegebenem Druck-
verlauf, Z . angem Math. MA., 19, 129-140 (1939);also J. Roy. Aeronaut.
Sot., 46,35-40 (1941).
12. FALKNER, V. M.,A further investigation of solutions of the boundary layer
equations, Brit. A.R,C., Repta. and Memo. No. 1884 (1939).
13. GOLDSTEIN, S., A note on the boundary layer equations, Proc. Cambridge Phil.
Soc., 36, 338-340 (1939).
14. SAKUIMI, T., New method of evaluating the flow in boundary layer which varies
with time, Proc. Phye.-Mdh. Soc. Japan, 21,632637 (1939).
16. SAKURAI, T., Two-dimensional boundary layer equations for motion of viscous
fluids near moving obstacles, Proc. Phys.-Math. SOC.Japan, 21,707-712 (1939).
10. FBUSSLINQ, N., Verdunatung, Wibmelibergang und Geschwindigkeitsverteilung
bei sweidimensionaler und rotationssymmetrimher laminarer Grenzschicht-
strtimung, Lunde Univ. Areekr. N.F. Avd 2,30 (1940).
17. HANTESCHE, W., and WENDT,H., Zum Kompressibilittits-Einflussbei der laminaren
Grensechicht der ebenen Platte. Jahrb. deul. Luflfahrlforech., 1940,p. 517.
18. WADA,K.,Theory of laminar boundary layer, Rept. Aeronaut. Reeearch Inet.
Tokyo Imp. Univ., 190 (Sept. 1940).
19. FALKNER, V. M., Simplified calculation of the laminar boundary layer, Brit.
A.R.C., Repte. and Memo. No.1696 (1941).
20. BUROERS, J. M., Some considerations on the development of boundary layers in
the case of flows having a rotational component, Proc. Koninkl. Nederland.
Akad. Wetensclr., 44, 13-25 (1941).
21. EMMONS, H.W.,and BRAINERD, J. G., Temperature effects in a laminar com-
pressible fluid boundary layer along a flat plate, J. Applied Mechanics, 8,
A105-Al10 (1941).
22. HANTZSCHE, W., and WENDT,H., Die laminare Grensschicht bei einem mit
Vberschallgeschwindigkeit angestromten nichtangestellten Kreiskegel, Jahrb.
deut. Luftfahrlforach., 1941,Vol. I, 76.
23. LOITSIANSKII, L. G., Integral methods in the theory of the boundary layer,
Applied Math. and Mechanics U.R.S.S., 6, 453-470 (1941);also N.A.C.A. Tech.
Memo. No. 1070 (1944).
24. MANGLER, W.,Eine Vereinfachung des l’ohlhauscn-Verfahrens zur Berechnung
 MECHANICS OF BOUNDARY LAYER FLOW 7
der laminaren Reibungsschicht, Jahrb. deut. Luftjahrtforsch., 1941, Vol. I,
18-20.
26. PRESTON, J. H., Note on the method of successive approximations for the solution
of the boundary layer equations, Phil. Mag. [I31,452-465 (1941).
26. PRETSCH, Die laminare Reibungsschicht an elliptischen Zylindern und Rotationa-
ellipsoiden bei symmetrischer Umstrbmung, Lujlfahrtforech., 18, 397 (1941).
27. BRAINERD, J. G., and EMMONS, H. W., Effect of variable viscosity on boundary
layers, with a discuasion of drag measurements, J. Applied Mechanics, 9, Al-A6,
(1942).
28. DORODNITZYN, A., Laminar boundary layer in compressible fluids, Compt. rend.
acad. sci. U.R.S.S., 34, 213-219 (1942).
29. LOITSIANSKII, L. G., Approximate method for calculating the laminar boundary
layer on an airfoil, Compt. rend. acad. sci. U.R.S.S., 36,227-232 (1942).
30. LOITSIANSKII, L. G., Laminar boundary layer on a body of revolution, Compt.
r e d . add. sea’. U.R.S.S., 36, 160-168 (1942).
31. KOCHIN,N. E., and LOITSIANSKII, L. G., An approximate method of calculating
the laminar boundary layer, Compt. rend. a d . sci. U.R.S.S., 36, 262-266
(1942).
32. PISKUNOV, N. S., On the problem of flow separation in a viscous fluid, Compt. rend.
a d . en’. U.R.S.S., 37,345 (1942).
33. SCHMIDT, H., and SCHRODER, K., Laminare Greneschichte-Ein kritischer Literatur-
bericht, Lujijahrrtforsch., 19, 65-97 (1942).
34. SCHMIDT, H., and SCHR~DER, K., Die Prandtlsche Greneschichtgleichung als
asymptotische Niiherung der Navier-Stokesschen Differentialgleichungen bei
unbegrenzt wachsender Reynoldscher Kenneahl, Deut. Math., 6,307-322 (1942).
35. SCHERBARTH, K., Greneschichtmessungen hinter einer punktformigen Storung
in laminarer Stromung, Jahrb. deut. Luftfahrtjorsch., 1942, Vol. I, 51-53.
36. STEPANYANTZ, L. G., The calculation of laminar boundary layers around bodies
of revolution, Applied Math. and Mechanics U.R.S.S.,6, 317-326 (1942).
(Russian text, English summary.)
37. VANWIJNGAARDEN, A., Laminar flow in radial direction along a plane surface,
Proc. Koninkl. Nederland. Akad. Wetensch., 46, 269-275 (1942).
38. WEYL,H., On the differential equations of the simplest boundary layer problems,
Ann. Math., 43, 381407 (1942).
39. BASIN,A. M., A new approximate method for the calculation of laminar boundary
layers, Compt. rend. mad. sci. U.R.S.S.,40, 14-i7 (1943).
40. MANQLER,W., Die “iihnlichen” Losungen der Prandtlschen Grenzschicht-
gleichungen, 2.angero. Math. Mech., 28, 241-251 (1943).
41. GBRTLER,H., Verdrhgungswirkung der laminaren Grenzschichten und Druck-
widerstand, Ing. Arch., 14, 286-305 (1944).
42. ACKERET, J., FELDMANN, F., and ROW, N., Untersuchungen an Verdichtungs-
@ t h e nund Greneschichten in schnell bewegten Gasen, Mitt. Inst. Aerodyn.,
E.T.H. Zthich, No. 10 (1946); also N.A.C.A. Tech. Memo. 1113 (1947).
43. OLDROYD, J. G., Calculations concerning theoretical values of boundary layer
thickness and coefficients of friction and of heat transfer for steady two-

-
dimensional laminar flow in an incompressible boundary layer with main
stream velocity U z m or U e“, Phil. Mug., 36, 587-600 (1945).
N

44. LIEPMANN,H., The interaction between boundary layer and shock waves in
transonic flow, J . Aeronaut. Sci.,13, 623-638 (1946).
8 HUGH L. DRYDEN

45. CARRIER, G. F., The boundary lnyer in n corner, Quart. Applied Math., 4, 367
(1947).

111. STABILITY OF LAMINAR BOUNDARY TJAYER FLOW

The circumstances surrounding the breakdown of laminar flow and the

beginning, at least, of the process of transition to turbulent flow are now
fully understood as a result of work completed during the last decade.
It has been assumed from the time of Reynolds that the problem of the
origin of turbulence is associated with or equivalent t o the problem of the
stability of the laminar flow. Reynolds’ own experiments on flow in
pipes showed that at low Reynolds numbers disturbances of any magni-
tude at the entrance were damped and the flow ultimately became steady.
At higher Reynolds numbers the magnitude of the disturbance present
at the entrance was an important factor and the flow remained steady
t o a Reynolds number that increased as the disturbance was reduced.
In 1938 there were two schools of thought as t o the nature of the instabil-
ity. The one, exemplified by the theoretical workers of the Gottingen
school, believed that the laminar flow was unstable for infinitesimal
disturbances. The other, exemplified by experimental workers in
England and America, believed that transition occurred as a result of
external disturbances of finite magnitude.
1. Status of Problem in 1958
The theoretical problem of the stability of two-dimensional laminar
flow was investigated by Heisenberg, Tietjens, Tollmien, and Schlichting
of the Gottingen group during the period 1924-1935. The specific
problem of the boundary layer near a thin flat plate without pressure
gradient was attacked by Tollmien and Schlichting. The problem
was idealized by assuming a layer of constant thickness within which the
velocity distribution was given by an approximation to the Blasius
distribution. Small periodic disturbances were assumed to be present
and their subsequent history was computed. The mathematical discus-
sion is complicated and involves a troublesome joining of solutions in
the neighborhood of the point at which the wave velocity coincides with
the local flow velocity. Tollmien found that unstable waves can exist if
the Reynolds number formed from the free stream velocity and the dis-
placement thickness of the boundary layer is greater t8han120. Schlicht-
ing obtained a value 575 for this critical Reynolds number. A t higher
Reynolds numbers there is a definite locus of nciitrnl disturbances which
are neither amplified nor dnmpcd. For any given Rrynolds number there
are two such neutral disturbnnccs, differing in WLVC Icngth, frequency,
and wave velocity. For intermccliate values of wavv length or frequency
 MECHANICS OF BOUNDARY LAYER FLOW 9

or wave velocity the disturbances are amplified; for values outside these
ranges, larger or smaller, the disturbances are damped.
Experimental workers could find no evidence in 1938 to support the
Tollmien-Schlichting theory. Dryden and his colleagues had studied
the fluctuations in the boundary layer near a plate and had not observed
the unstable wave motion. They found slow fluctuations, which were
interpreted as probable fluctuations in thickness of the layer forced by
disturbances in the free stream. There was no evidence of shear stresses
or effect of the fluctuations on the distribution of mean speed. Transition

16‘ RCRIT
Fro. 1.-Critical Reynolds number of a sphere in air streams of different turbulence.

occurred intermittently and suddenly with no development of amplified

disturbances. Experiments by the groups at the California Institute of
Technology, Massachusetts Institute of Technology, and Cambridge
University, England, likewise failed to show the amplified disturbances.
Not only did the experimental group obtain this negative result; their
experiments confirmed a theory developed by G. I. Taylor, (84) which
attributed transition to the presence of free stream turbulence. Taylor
assumed that transition occurred as the result of momentary separation
in regions of adverse pressure gradient associated with the free stream
turbulence. The Reynolds number a t transition was found to be a
function of (u’/U)(D/L)&, where u’/U is the intensity of the turbulence
(u’ being the root mean square speed fluctuation, I/ the mean speed), L
is the scale of the turbulence (defined as the integral of the correlation coef-
ficient between longitudinal velocity fluctuations at points distance y
apart in a direction t.ransverse to the flow for 1~ varying from zero t o
infinity), and D is a reference dimension of the body. Figrirrs 1 and 2
10 HUGH L. DRYDEN

show experimental checks of this relation taken from National Advisory

Committee for Aeronautics, Technical Reports 681 and 682.
2. Observation of Tollmien-Schlichting Oscillations
These conflicting theoretical and experimental results have now been
completely reconciled by the work of Schubauer and Skramstad, who first
observed the Tollmien-Schlichting oscillations in 1940 in a wind tunnel a t

“/D

FIG.2.-Location of transition in the boundary layer on an elliptic cylinder in air

streams of different turbulence.

the National Bureau of Standards, for which the turbulence of the free
stream was only 0.02 to 0.04 per cent of the mean speed. Their work is
described in full in the National Bureau of Standards Research Paper 1772,
which is a reprint of the original N.A.C.A. advance confidential report
issued in April 1943, now declassified and issued as N.A.C.A. Wartime
Report W-8. When the free stream turbulence is less than about 0.1 per
cent of the mean speed, transition is preceded by the Tollmien-Schlichting
oscillations, and the theory is beautifully confirmed quantitatively as well
as qualitatively. For ordinary wind tunnels the stream turbulence
usually lies in the range 0.2 to 1.0 per cent. In these wind tunnels the
turbulence is the controlling factor and the mechanism is that of the
Taylor theory.
Fig. 3 shows the first record obtained by Schubauer and Skramstad
 MECHANICS OF BOUNDARY LAYER FLOW 11
in August 1940 of the naturally occurring laminar boundary layer oscil-
lations. After assuring themselves that the oscillations were not effects
of vibration, acoustic phenomena, or other extraneous disturbance, and
that the frequency did in fact agree with the value predicted by the
theory, they devised an ingenious method for proceeding experimentally

4.0 fi

4.5
w
W
LL

FIG.3.-Laminar boundary layer oscillations in the boundary layer of a flat plate.

Distance from surface = 0.025 inch, speed of free stream = 80 ft/sec, time interval
between dots = $a sec.

exactly as Tollmien and Schlichting proceeded mathematically. They

introduced small disturbances of known frequency by the arrangement
shown in Fig. 4. An alternating current of the desired frequency was
sent through a metal ribbon 0.002 inch thick placed in a magnetic field,
causing it t o vibrate about a mean position about 0.006 inch from the
plate, the amplitude being controlled by the magnitude of the current.
The resulting fluctuations in speed in the boundary layer a t various
distances were measured by means of a hot-wire anemometer. The
procedure was repeated for several frequencies and wind speeds. From
these data the damping and amplification coefficients could be determined
12 HUGH L. DRYDEN

for disturbances of various frequencies over the range of Reynolds num-

bers covered, and the frequencies of the neutral oscillations, which are
neither damped nor amplified, could be plotted against Reynolds number.

FIG.4.-Method used by Schubauer and Skramstad to excite controlled disturbances

in the laminar boundary layer of a flat plate.
The wave lengths were independently determined from the distance
through which the hot-wire anemometer had to be moved to return t o the
same phase relationship between the observed wave and the current
through the ribbon. The wave velocities were determined as the product
of frequencies and corresponding wave lengths.
After the experimental work had been completed, C. C. Lin (66)
 MECHANICS O F B O U N D A R Y LAYER FLOW 13

undertook a’revision of the mathematical theory and a clarification of

some features that had been adversely criticized. Lin’s calculation gave
a critical Reynolds number of 420 as compared with Tollmien’s 420 and
Schlichting’s 575, and a slightly different locus for the neutral oscillations.
Figs. 5 and 6 show the nondimensional wave length and frequency loci as
observed by Schubauer and Skramstad compared with the theoretical
results of Schlichting and Lin. The agreement is remarkable. Fig. 7
shows a comparison of the distribution of amplitude of the u-fluctuations

00

F’IG, S.-ExperimentaIIy determined wave lengths of neutral oscillations in the

laminar boundary layer of a flat plate compared with the theoretical curves of Schlich-
ting and Lm.

across the boundary layer for a particular case with the theoretical com-
putations of Schlichting.
These experimental results were confirmed by Liepmann at the
California Institute of Technology for the boundary layer on the convex
side of a curved plate. Schlichting had computed the effect of convex
curvature on the neutral curve and found a slightly stabilizing effect.
Liepmann used the vibrating ribbon technique, acoustic excitation, and
a method employing roughness elements uniformly spaced on the surface
to give a disturbance of known wave length. The results were found to
agree closely with those obtained at the National Bureau of Standards,
the influence of convex curvature being extremely small.
An important factor in these observations is the reduction of the
turbulence of the air stream t o a value as low as practicable. This is
accomplished by the use of damping screens as described by Dryden and
Schubauer in Journal Aeronautical Science, 14, 221-228 (1947). At thc
14 HUGH L. DRYDEN

low levels obtained, the noise of the wind tunnel propeller produces the
predominant speed fluctuation. In such a tunnel it is easy to demonstrate
that sounds of the proper pitch cause an earlier breakdown of the laminar
flow.

FIG.6.-Experimentally determined frequencies of neutral oscillations in the

laminar boundary layer of a fiat plate compared with the theoretical curves of Schlich-
ting and Lin.

9. Relation of Tollmien-Schlichting Waves lo Transition

The Tollmien-Schlichting waves do not in themselves constitute
turbulent flow. The regular waves grow in amplitude, become distorted,
then exhibit intermittent bursts of high frequency fluctuations. At
some distance from the surface the bursts usually appear in the low veloc-
ity part) of the cycle and near the surface there is occasional evidence of
intermittent separation. It is believed but not definitely proved that
 MECHANICS O F BOUNDARY LAYER FLOW 16
intermittent separation produces dynamically unstable vortex sheets,
which roll up into small eddies.
Thus the two originally conflicting pictures coalesce. If the free
stream turbulence is greater than 0.2 per cent, intermittent separation

/----
_---

0 8 1.2 I6 I 3 2 4 i

FIG.7.-Experimental and theoretical distribution of amplitude across the boundary

layer for two particular neutral oscillations computed by Schlichting.
Solid curvea are theoretical, broken curves with points are experimental.

occurs as soon aa the Pohlhausen parameter ( P / p U ) ( d p / d z ) reaches the

appropriate value, 6 being the thickness of the boundary layer, I( the
viscosity of the fluid, U the free stream speed, and dp/& the pressure
gradient. For lower turbulence this step is preceded by the appearance
of amplified waves. At the higher turbulence the intensity and scale
of the turbulence determine transition; at the lower turbulence the spec-
trum of the disturbances originally present is of great importance.
10 HUGH L. DRYDEN

Whether these disturbances are small free stream turbulence, noise, or

surface roughness, they are selectively amplified until large inusoidal
oscillations are present. In free flight noise may well be the most impor-
tant factor.
4. Effect of Curvature
Liepmann completed an extensive study of the effect of curvature on
transition. The boundary layer on convex surfaces, at least up t o values
of momentum thickness equal t o 0.001 time the radius of curvature,
exhibits the same Tollmien-Schlichting instability as the flat plate, and
the effect of curvature is so small as to lie within the precision of measure-
ment. Transition on the upper surface of an airfoil will therefore not be
noticeably influenced by the curvature.
The experimental work on concave surfaces was described by Liep-
mann in N.A.C.A. Advance ConJidential Report No. 4528 (now declassified
and available as N .A.C.A . Wartime Report W-87).When the momentum
thickness 0 is greater than 0.0005 times the radius of curvature T , the
laminar flow is dynamically unstable, owing to three-dimensional dis-
turbances as studied theoretically by Gortler (50). In a stream of the
lowest turbulence the Gortler parameter Re d v r , where Re is the Reyn-
olds number based on momentum thickness, is equal t o 9.0. This value
decreases to about 6.0 at a turbulence level of 0.003 (value of u’/U).
If the value of e/r is less than 0.00005 the flow is unstable because of
Tollmien-Schlichting waves like the flat plate and the convex surface,
and transition occurs at a constant Reynolds number whose value depends
on the free stream turbulence. For values of O/r between 0.00005 and
0.0005 there appears to be a more or less continuous change from transi-
tion due to Gortler vortices to transition caused by Tollmien-Schlichting
waves.
6. E$ect of Pressure Gradient
The theory of the instability of a laminar boundary layer subjected
t o a falling or rising pressure was investigated by Pretsch (68) at Gottingen
and by Schlichting and Ulrich (62) at Braunschweig.
Pretsch considered the velocity distributions corresponding to the
“similar” solutions for a free stream velocity U proportional to a power
law of the distance x, i.e., U = kxm. A positive value of m corresponds
t o accelerated flow (falling pressure) and a negative value to decelerated
flow (rising pressure). These solutions had previously been computed
and tabulated by Hartree (4)in terms of a parameter p connected with
m by the relation m = p / ( 2 - p). For positive values of m(O < p < 2)
 MECHANICS O F BOUNDARY LAYER FLOW 17
Pretsch found it possibIe to approximate Hartree’s tabulated values by
the expression

and for m negative or zero (-0.198 < @ < 0) by the expression

= + (1 - ): sin
u- I
[(! - 1) sin-1 - 24,

FIG.&-Wave lengths of neutral oscillations for flows with pressure gradient &B
computed by Pretsch.

where the quantities n, a, u,, and s were selected to give the best possible
approximations to Hartree’s values. The stability of laminar flows
represented by these approximations to Hartree’s solutions was investi-
gated, but for convenience the distributions were identified in terms of
Hartree’s parameter p.
The locus of the neutral oscillations for various values of p are shown
in Fig. 8 as a plot of the nondimensional wavelengths against Reynolds
number. The usual nondimensional parameter for the pressure gradient
is Pohlhausen’s parameter X = U’62/u. Pohlhausen’s four-term approxi-
mate solution for U = kxm is X = constant where X is the function of ,8
shown in Fig. 9. Fig. 10 shows the critical Reynolds number, based on
displacement thickness, as a function of X. While the values of m, p,
and X are useful in giving a rough idea of the velocity profiles considered,
18 HUGH L. DRYDEN

FIQ.9.-Relation between Hartree parameter p and the Pohlhaueen parameter X.

Re
FIG.lO.--Effect of pressure gradient on stability of laminar flow.
 MECHANICS OF BOUNDARY LAYER FLOW 19

the stability limits actually apply to the approximations previously

referred to, for which the constants are tabulated in Pretsch’s paper.
In a second paper Pretsch (61) computed the damping and amplifi-
cation of disturbances other than the neutral one for the same pressure
distributions. The maximum observed amplification coefficient as a
function of X is shown in Fig. 11. The values are always larger for

x
FIQ. ll.-Mauimum amplification coefficient in flows with pressure gradient as com-
puted by Pretsch.
decelerating flow than for accelerating flow and become very large as X
approaches large negative values corresponding to separation of the flow.
At about the same time as Pretsch’s investigation a similar study was
carried out by Schlichting and Ulrich (62). They studied the velocity
distributions given by a six-term power series of the Pohlhausen type,
choosing as the additional boundary conditions the vanishing of dsU/dya
a t the wall and a t the outer boundary. The resulting parameter X P 6
characterizing a given pressure gradient differs somewhat from the param-
eter h for the four-term series approximation. The relation between
values giving the same momentum thickness is shown in Fig. 12. The
accuracy of the stability computations was studied by computing the
criticalReynoldsnumber for the case of zero pressure gradient. The four-
20 EUQH L. DBYDEN

FIQ. 12.-Relation between Pohlhausen parameters for six-term and four-term

approximations giving same momentum thickness.

FIQ. 13.-Wavelengths of neutral oscillations for flows with pressure gradient as

computed by Schlichting and Ulrich.
 MECHANICS OF BOUNDARY LAYER FLOW 21

term approximation gives 1150, the six-term approximation 645, as

compared with Pretsch’s value of 680 and Schlichting’s “exact” value
575. The results obtained for several values of X are shown in Fig. 13 in
the form of loci of the neutral disturbances. The critical ReynoIds num-
ber is plotted against X in Fig. 14, in which Pretsch’s result is also shown.
The accuracy of the computed values is open to some question since
the values obtained for zero pressure gradient differ from Lin’s more
accurate value 420, which agrees with the experimentally determined

Rc

FIG.14.-Comparison of results of Pretsch and those of Sohlichting and Ulrioh for

effect of pressure gradient on stability of laminar flow.

value of Schubauer and Skramstad. Nevertheless the trend and general

order of magnitude are probably correct. The effect of pressure gradient
is very large; for -3 < X < 3 the critical Reynolds number is approxi-
mately proportional to eo.6A. It is easy t o understand why the observed
critical Reynolds number for airfoils is so large; the effect of the accelera-
tion of the flow over the upper surface is to increase greatly the critical
Reynolds number.
In both papers computations were made of the parameter X at various
points along the surface of airfoils, and the point for which the critical
Reynolds number corresponding to the prevailing value of X was first
reached was interpreted as the most forward possible position of the
transition point. The experiments available indicated that transition
did not occur until some distance further back even in the fairly turbulent
wind tunnel available. It was recognized that the amplification of the
disturbance must be considered.
22 HUGH L. DRYDEN

I n view of the good experimental check of the theoretical computations

for the case of zero pressure gradient, it would seem worth while to
devote considerable effort to refining the theoretical stability computa-
tions for flows with pressure gradient and to carrying on additional
experimental investigation. Schubauer and Skramstad made only a few
exploratory measurements.
6. Efect of Compressibility
A theoretical investigation of the stability of the laminar boundary
layer in a compressible fluid has been carried out by Lees and Lin (68),
although in their first report the investigation did not reach the stage of
numerical computations, It was found that for Reynolds numbers of
the order of those encountered in most aerodynamic problems the tem-
perature disturbances have only a negligible effect on those particular
solutions that depend primarily on the viscosity. The chief physical
mechanism of the instability is essentially the same for compressible flow
as for incompressible flow.
It was found that when the solid boundary is heated, the boundary
layer flow is destabilized by the change in the distribution of the product
of density and vorticity, but stabilized by the increase in kinematic
viscosity near the boundary. When the boundary is cooled, the reverse
is true. The detailed results are given in N.A.C.A. Tech. Note No.
1360 by Lester Lees, entitled The Stability of the Laminar Boundary Layer
in a CompressibEe Fluid, The first effect was found to predominate.
BaMiography on Stability of Laminar Boundaqj Layer Flow and Transition
46. CLAUSER, M., and CLAUBEB, F., The effect of curvature on the transition from
laminar to turbulent boundary layer, N.A.C.A. Tech. Note No. 613 (1937).
47. ROSENBROOK, G., Instabilitiit der Gleitschicht im schwach divergenten Kanal,
2. ongew. Math. Mech., 17, 8-24 (1937).
48. SILVERSTEIN, A., and BECKER, J. V., Determination of boundary layer transition
on three symmetrical airfoils in the N.A.C.A.full-scale wind tunnel, N.A.C.A.
Tech. Rept. No. 637 (1938).
49. VONDOENHOFF, A. E., Preliminary investigation of boundary layer transition
dong a flat plate with adverse pressure gradient, N.A.G.A. Tech. Note No.639
(1938).
50. G~RTLER, H., tfber eine dreidimensionale Instabilitiit laminarer Grenzschichten
an konkaven Wiinden, Nachr. Ges. Wiss. Gistlingen, Fachgruppe I (N.F.), 2,
1-20 (1940).
51. DRYDIN,H. L., Turbulence and the boundary layer, J . Aeronaut. Sei., 6, 85- 100
(1939).
52. FAOB,A., Fluid motion transition from laminar to turbulent flow in a boundary
layer, Phys. SOC.Repts. on Progress in Physical 6, 270-279 (1939).
53. TOWNEND, H. C . H., Note on boundary layer transition, Brit. A.R.C., Repts. and
Memo. No. 1873 (1939).
 MECHANICS OF BOUNDARY LAYER FLOW 23
54. FAQE,A,, Experiments on the breakdown of laminar flow, J . Aeronaut. Sci., 7 ,
513-517 (1940).
55. LANQER, R. E., On the stability of the laminar flow of a viscous fluid, Bull. Am.
Math. Soc., 46,257 (1940).
56. SCHLICETINQ, H., Uber die theoretische Berechnung der kritischen Reynoldsschen
Zahl einer Reibungsschicht in beschleunigter und verzogerter Stromung, Jahrb.
deut. Luftfahrtjorach., 1940,Vol. I, 97.
57. FAQE,A,, and PRESTON, J. H., Experiments on transition from laminar to turbu-
lent flow in the boundary layer, Proc. Roy. SOC.(London), A, 178, 201-227
(1941).
58. PRETSCH, J., Die Stabilitiit einer ebenen Laminarstromung bei Druckgefiille und
Druckanstieg, Jahrb. deut. Luftfahrtjorech., 1941, Vol. I, 58-75.
59. FAQE,A., Transition in the boundary layer caused by turbulence, Brit. A.R.C.,
Repta. and Memo. No. 1896 (1942).
60. PRANDTL, L., Bericht Iiber neuere Untersuchungen uber das Verhalten der
laminaren Reibungsschicht, insbesondere den laminar-turbulenten Umschlag,
Mitt. deut. Akad. Luftfahrtforsch., 2, 141 (1942).
61. PRETSCH, J., Die Anfachung instabiler StBrungen in einer laminaren Reibungs-
schicht, Jahrb. deut. Luftjahrtforsch., 1942,Vol. I, 54-71.
62. SCHLICETINB, H., and ULRICH, A., Zur Berechnung des Umschlages laminar/tur-
bulent, Jahrb. deut. wvahrtjorech., 1944,Vol. I, 8-35.
63. BATCHELOR, G. K., The laminar flow characteristics of three related airfoils,
Australia, Council Sci. Id. Research Div. Aeronaut., Rept. No. A20 (1943).
64. CHARTERS, A. C., Transition between laminar and turbulent flow by transverse
contamination, N.A.C.A. Tech. Note No.891 (1943).
65. PILLOW,B. A., A review of hydrodynamic stability and its bearing on transition
to turbulent flow in the boundary layer, Australia, Council Sci. I d . Research,
Div. Aeronaut., Rept. NO.A36 (1945).
66. LIN, C. C.. On the stability of two-dimensional parallel flows, Quart. Applied
Math. 8 (No. 2), 117-142 (July 1945); 8 (No. 3), 218-234 (Oct. 1945); 9 (No. 4),
277-301 (Jan. 1946).
67. LOFTIN,L. K., JR., Effects of specific types of surface roughness on boundary-
layer transition, N.A.C.A. Wartime Rept. L-48 (1946).
68. LEES,L., and LIN, C. C., Investigation of the stability of the laminar boundary
layer in a compressible fluid, N.A.C.A. Tech. Note No. 1116 (1946).
(See also references under Boundary Layer Suotion.)

IV. BOUNDARY
LAYERSUCTION
During the war German investigators studied the effect of removing
a part of the boundary layer air by suction through a homogeneous
porous surface. At a considerable distance from the leading edge of a
plate in a flow of uniform free stream velocity, the velocity distribution
is given by u/V = 1 - t r v a u / r where u is the velocity at distance y from
the surface, U is the free stream velocity, V ois the velocity of the air
sucked through the surface, and v is the kinematic viscosity. The
boundary layer displacement thickness is constant and equal ‘to 1~ / V O ( .
Pretsch (77) obtained a critical Reynolds number of 55,200 and Bussmann
24 HUQH L. DRYDEN

and Miinr (76) 70,000 as compared with Schlichting's value of 575 for
the plate without suction and Pretsch's values of 10,000 to 20,000 for
the plate without suction but with a large favorable pressure gradient.
The required suction velocity V Ois of the order of 0.000014 time the free
stream speed. The maximum amplification parameter BiS*/U in the
unstable region is 4.65 X lo-' for the flow with suction as compared with
3.45 X lo-* for the flow without suction. These results are extremely
important as indicating a possible method. of greatly increasing the
stability of laminar flow and thus delaying transition.
Some experimental work has been carn'ed out in Switzerland and
England with attention given to the practical aspect of reducing drag.
J. Ackeret (74,83) and his colleagues at the Aerodynamic Institute of
the Federal Institute of Technology at Ziirich, by removing a part of
the boundary layer air by suction through a small number of slots, were
in fact able to partially restore a turbulent boundary layer to the laminar
condition. By careful attention to the detailed design of the internal
ducting of the suction system they obtained extremely low drag
coefficients.
Mr. E. F. Relf (82) has given a general account of work carried out
in England during the war on boundary layer suction as applied to the
maintenance of laminar flow and reduction of drag of rather thick airfoils.
A single slot was used on each surface of an airfoil especially designed t o
suit the suction, an idea due t o A. A. Griffith and followed up by S. Gold-
stein and by Lighthill on the theoretical side and presumably by Relf
and his collaborators on the experimental side. Reference is also made
to the use of distributed suction acting through a porous surface with
theoretical work by J. H. Preston and a flight test by F. G. Miles. No
experimental results are given and no references to detailed reports.

Bibliographg on Boundary Layer Suction

B. S., Sucking off the boundary Isyer, Aeroplane, 62,9&100 (1937).
69. SEIENSTONE,
70. GERBER,A., Untersuchungen iiber Greneschichtabsaugung, Mitt. Inst. Aerodyn.,
E.T.H. Ztirich, No. 6 (1938).
71. RICHARDSON, E. C., Manipulating the boundary layer, J . Roy. Aeronaut. Soc.,
42,816-831 (1938).
72. MILES,F. G., Sucking away boundary layers, Flight, 38, 180-184 (1939).
73. WESKE,J. R., Reduction of skin friction on flat plate through boundary layer
control, J . Aeronaut. Sci., 6, 289-291 (1939).
74. RAS, M., and ACKERET,J., On prevention of boundary layer turbulence through
removal by suction, Heh. Phy8. Acta, 14, 232 (1941).
75. SCHRENK, 0., Grenzschichtabsrtugung,Luftw., 7,409-414 (1940) ; also N .A.C.A.
Tech. M e m . No. 874 (1941).
76. BUISMANN, K., and MUNZ,H., Die Stabilitgt der laminaren Reibungsschicht mit
,
Absaugung, Jahrb. deul. Luftfahrtfursch. 1949, Vol. I, 36-39.
 MECHANICS OF BOUNDARY LAYER FLOW 25
J., Umschlagbeginn und Absaugung, Juhrb. deut. Luftfuhrlforsch.,
77. PBETSCH,
lD42, Vol. I, 1-7.
78. SCHLICHTINQ, H., Berechnung der laminaren Grenzschicht mit Absaugung,
Luftfahrlforsch.,19, 179 (1942).
79. SCHLICHTINQ, H., Die Grenzschicht an der ebenen Platte mit Absaugung und
Ausblasen, Lujtfuhrtforech., 19, 293-301 (1942).
80. SCHLICHTINQ, H., and BUSSMANN, K., Exakte Losungen fiir die laminare Grenz-
echicht mit Absaugung und Ausblasen, Schl.iften deut. Akud. Luftfahrtforsch.,
7B, No. 2 (1943).
81. FREEMAN,H. B., Boundary layer tests in Langley propeller research tunnel,
N.A.C.A . Tech. Note No. 1007 (1946).
82. RELF,E. F., Recent aerodynamic developments, J. Roy. Aeronaut. Soc., 60,
421-449 (1946).
83. PFENNWQER, W., Untersuchungen uber Reibungsverminderungenan Tragfltigeln,
insbesondere mit Hilfe von Grenzschichtabsaugung, Mitt. Inst. Aerodyn.,
E.T.H.ZIirich, No. 18 (1946).
V. TUR~ULENT FLOW IN BOUNDARY LAYERS
There have been no notable advances in the theory of fully developed
turbulent motion during the last decade, although the great advances in
experimental techniques for studying the turbulent velocity fluctuations
lead one to expect the early appearance of that long-awaited new idea
that will permit further progress. In the period 1934-1938 Taylor
developed his statistical theory of turbulence, which was so fruitful in
treating the problem of isotropic turbulence. Von K4rm4n extended the
theory, clothed it in more elegant mathematical form, and attempted,
with incomplete success, to treat the problem of shear flow. The con-
cepts and language of the statistical theory have been found most useful
in guiding the experimental work.
At the Fifth International Congress for Applied Mechanics in 1938
von K4rm4n discussed some measurements of turbulent velocity fluctu-
ations in a channel by Wattendorf and by Reichardt and suggested the
existence of a kind of statistical similarity between the fluctuations a t
different points over the region not too close to the wall or to the channel
center. He pointed out the difficulty of reconciling Reichardt’s measure-
ments of the plane stress tensor with the picture of momentum transfer
by a mixing process involving only a scalar “mixing length.” I n the
discussions Tollmien and Prandtl suggested that the turbulent fluctu-
ations might consist of two components, one derivable from a harmonic
function and the other satisfying an equation of the heat-conduction
type, i e . , a nondiffusive and a diffusive component or viscosity inde-
pendent and viscosity-dependent type.
1. Development of Empirical Methods of the Buri-Gruschwitz Type
Since engineers have to make computations relating to turbulent
skin friction and turbulent boundary layers, there has been some
26 HUGH L. DRYDEN

development of empirical methods. These are of the "one-dimensional "

type involving the application of the von K&rm&nintegral equation for
the momentum and the use of empirical relations obtained from a few
experimental studies of flow in convergent and divergent passages.
While the earliest procedure was suggested by Buri, the method proposed
by Gruschwitz has been more widely used and apparently was used in
substantially its original form by German engineers throughout the war.
In the United States a modified procedure suggested by von Doenhoff
and Tetervin (86) has been found useful.
In common with Buri and Gruschwitz, von Doenhoff and Tetervin

I
assumed that the boundary layer at any section could be described by
the momentum thickness 8 = (1 -); );( dy and by a single
parameter that fixes the shape of the velocity distribution curve. Buri
had used as the parameter the quantity (U'6!U)(U8/v)i. Gruschwitz
had used 1 - (u/U)2v,eas primary parameter and the ratio of displace-
ment thickness [1 (1 -); &]to momentum thickness as secondary
parameter. This ratio will be denoted H . Von Doenhoff and Tetervin
selected H as an appropriate shape parameter. They give a compre-
hensive summary and analysis of available experimental data, plotting
u/U versus H for a series of constant values of y/8. The available data
gave a reasonably close approximation to a single family of curves. The
value of H for separation was found to lie between 1.8 and 2.6.
The particular velocity distribution curve corresponding to the single
parameter adopted to describe it is of course dependent on the external
forces acting on the boundary layer, that is, on the shear stress at the wall
and the pressure gradient. Yon Doenhoff and Tetervin adopt the ratio
of the nondimensional pressure gradient to the shearing stress coefficient
at the wall as a parameter characteristic of the external forces acting
on the boundary layer. Calling the shearing stress at the wall rr, this
parameter is 2pU U59/rw in their notation * It is then assumed
that this parameter determines not H itself, i.e., the particular velocity
distribution curve of the one-parameter family, but the rate of change of
dH
H with x. Thus 0 - is plotted against 2pUU'8/~,,rw being computed
ax
from the Squire-Young formula (88)

pUa
TW
[
= 5.890 log,, (4.075 -2I):
 MECHANICS OF BOUNDARY LAYER FLOW 27

The experimental data are fairly well represented by the empirical

equati.on

The momentum equation gives the relation

Elimination of rP gives two simultaneous first-order differential equa-

tions, which can be solved by a step-by-step calculation to give B and H.
Kalikhman (91) describes a still different empirical approach based
also on the momentum relation and the Squire-Young formula.
All such methods rest on empirical assumptions and become more and
more unsatisfactory as separation is approached. They do not con-
tribute to the construction of a rational theory founded on knowledge of
the underlying physical phenomena.

2. Developments PToceeding from Reynolds Theory of Turbulent Stresses

The fundamental equations of turbulent flow of an incompressible
fluid developed by Reynolds are identical with the Navier-Stokes equa-
tions except for the addition of the six components of the turbulent
stress tensor, i e . , -pu'2, -pvf2, - P W ' ~ , - - p W , -pDIW', and -p=.
With these six additional unknown functions, the number of unknowns
exceeds the number of equations and there arq insufficient data for a
solution. All of the past procedures assume that the turbulent stress
tensor at a point is determined by the mean flow in the neighborhood of
the point. In the most general form it is assumed that the stress tensor
is equal to the product of the deformation tensor by a mixing length
tensor. The most widely used theory of Prandtl assumes the mixing
length tensor to reduce to a scalar mixing length, that is, one number
instead of six. Prandtl related this mixing length to the geometry of
the flow field (distance from the wall) by a kind of dimensional reasoning
but von K4rmBn related it to the derivatives of the mean velocity. As
will be seen later, the available experimental data suggest re-examination
of the assumption that the turbulent fluctuations at a point are deter-
mined by the mean flow in the neighborhood of that point, that is,
whether the entire field must be considered, leading perhaps to an integral
equation for the relation between the fluctuations at a point and the
mean flow.
28 HUGH L. DRYDEN

A recent paper by Nevzgljadov (94) is of interest in this connection.

He writes down the Reynolds equations, the continuity equation, and the
energy equation for the turbulent velocity fluctuations. The unknown
functions are the mean velocity, the mean pressure, the turbulent pres-
+ +
sure ?r = i p ( u ’ 2 u’* w ’ ~ ) , the turbulent stress tensor, the current
density of the kinetic energy of the turbulent fluctuations, the energy
flux transferred by the pressure fluctuations, and the viscous dissipation.
It is proposed t o select the mean velocity, the mean pressure, and the
turbulent pressure as independent fundamental quantities and to assume
that the other unknowns can be related to these fundamental quahtities
by (‘equations of state.”
Nevzgljadov discusses the question whether the shearing stress a t a
given point can be expressed in terms of the values of the fundamental
quantities in the immediate neighborhood of the point. His theory
differs from that of Prandtl in assuming that the turbulent shearing
stress may depend on the turbulent pressure as well as on the mean
motion. The relation proposed, that is, that the shearing stress is
proportional to the product of the turbulent pressure by the mean velocity
gradient, is not supported by the experiments to be described; the experi-
ments in fact show a more direct relation between shear stress and turbu-
lent pressure than between shear stress and local mean velocity gradient.

9. National Bureau of Standards Experiments

The only experimental work known during recent .years on the turbu-
lent stress tensor in a turbulent boundary layer is that carried out by the
National Bureau of Standards for the National Advisory Committee for
Aeronautics and reported by Dryden at the Sixth International Congress
for Applied Mechanics. The paper will appear in the proceedings of the
Congress and also as N.A.C.A. Tech. Note No. 1168. It is a progress
report on a long range study of the mechanics of a separating turbulent
boundary layer.
The experiments were made in the boundary layer along a partition of
airfoil-like section extending in a diametral plane across the 10-foot wind
tunnel of the National Bureau of Standards. The partition was 27.9 f t
long and had a maximum thickness of 2 ft. The leading edge radius was
1 in. and the leading edge was joined tangentially to cylindrical surfaces
of 23-ft radius forming the nose. The downstream portion of the par-
tition was flat on one side and in the form of a cylinder of 31-ft radius on
the other. By various tricks two-dimensional flow was obtained over the
central region with separation occurring along a line 25.7f t from the lead-
ing edge. The variation of the velocity just outside the boundary layer is
 MECHANICS OF BOUNDARY LAYER FLOW 29

shown in Fig. 15. The maximum velocity occurs at 17 ft from the

leading edge, and was about 161 ft/sec in the measurements t o be

FIQ.15.-Velocity and pressure distribution just outside the turbulent boundary

layer studied a t National Bureau of Standards.

FIG.16.-Velocity distribution within turbulent boundary layer studied a t National

Bureau of Standards.
described. Its exact value was adjusted to be proportional t o the vis-
cosity of the air, thus maintaining a constant Reynolds number,
A t the 17-ft position the boundary layer was about. 2.3 in. thick,
30 HUGH L. DRYDEN

corresponding to that which would have prevailed on a flat plate 14.3 f t

long with fully turbulent boundary layer and no pressure gradient, the
equivalent flat-plate Reynolds number being about 14 X 10'. The free
stream turbulence was about 0.5 per cent and the boundary layer flow

Y,FEET

FIG.17.-Displacement thickness, 8 *, momentum thickness, el and parameter,

H = b*/B, for NBS turbulent boundary layer.

wm turbulent from the leading edge as desired because of the unsym-

metrical form of the partition.
The experimental program has been directed principally at the meas-
urement of the turbulent fluctuations and of the turbulent stress tensor
by means of hot-wire anemometers and contemplates the development
and application of such additional hot-wire measurements as will throw
light on the underlying phenomena. The usual traverses of mean speed
were also made. Unfortunately the hot-wire methods are not adaptable
to the simultaneous measurements of a number of quantities at a large
 MECHANICS O F BOFNDARY LAYER FLOW 31
number of points in the flow field. The measurements extend over
months and are not suitable for analysis of differential changes from
point to point with high accuracy.
The distribution of mean speed is plotted in the manner suggested by
von Doenhoff and Tetervin (86) in Fig. 16, and the displacement thick-
ness, momentum thickness, and their ratio, the parameter H , are shown

Y, INCHES

FIG.18.-Typical distributions of turbulent shearing atress coefficient in NB8 turbu-

lent boundary layer.

in Fig. 17. The agreement with previous results is good, although

there are systematic differences slightly greater than the experimental
dispersion.
Typical distributions of the turbulent shearing stress are shown in
Fig. 18 and a contour map for the whole region studied is shown in Fig. 19.
The measurements were made by the procedure devised by Skramstad
and described by Dryden a t the Fifth International Congress for Applied
Mechanics. The viscous shearing stresses are wholly negligible in the
regions studied, because of the large Reynolds number, being less than
3 per cent of the turbulent shearing stress at 0.1 in. from the surface.
The laminar sublayer is never approached in any of the measurements.
The stress coefficient used in Fig. 19 is based on the velocity at 17
ft as a constant reference velocity and hence the contours are those of
equal absolute values of the stress. In the region of adverse pressure
gradient, that begins at 17 f t the maximum value of the shearing stress
gradually shifts from the wall on proceeding downstream until at separa-
32 HUGH L. DRYDEN

tion it is located near the middle of the boundary layer. The maximum
value slowly increases downstream.
In addition to the shearing stress, the root-mean-square values of the
three components of the turbulent fluctuations were also measured, u'

c
0

parallel t o the mean flow, v' perpendicular to the mean flow and to the
surface, and w' perpendicular to the mean flow but parallel to the surface.
Contours of their distributions are shown in Figs. 20, 21, and 22. The
values are made nondimensipnal by dividing by the reference velocity
just outside the boundary layer at 17 ft from the nose, but since the same
reference value is used throughout, the contours are also contours of
equal absolute values of u', v', and w'. The mean velocity just outside
 MECHANICS O F BOUNDARY LAYER FLOW 33

20 21
X. Feet

FIQ.20.--Contpur map of u' for NBS turbulent boundary layer.

X , Feel

FIQ.2l.--Contour map of v' for NBS turbulent boundary layer.

X. Feel

FIG.22.-Contour map of w' for NBS turbulent boundary layer.

34 HUGH L. DRYDEN

the boundary layer at each section varies from section to section as given
in Fig. 15. The turbulence is three-dimensional even though the mean
flow is two-dimensional. The turbulence is strongly nonisotropic, v'
being considerably less than u' and w' being intermediate in value.
Isotropy is reached in the free stream.
From these data the complete turbulent stress tensor can be computed
and in particular the angle a between the direction of the principal axis
60

50

40

a 20
~

0 .4
-i .6 LO

FIQ.23.-Angle (I of the principal axis of the turbulent stress tensor to the mean
direction of flow.

of the stress tensor and the mean flow. Typical results are shown in Fig.
23 along with Reichardt's results for two-dimensional flow under pressure
between two plates as given by von K&rm&p. These results confirm the
difficulty pointed out by von K&rm&n,namely, that the principal axis
of dilatation is at 45" to the mean flow as compared with 10" to 30" for
the principal axis of the turbulent shearing stress. There are directions
for which there is a shearing stress but no rate of shear of the mean flow,
and vice versa. This result cannot be reconciled with the concept of a
scalar mixing length,
As further indication challenging the correctness of the assumption
that there is a close relation between the turbulent fluctuations and the
mean flow in the immediate neighborhood, consider Fig. 24,which shows
the stress coefficient plotted against the mean velocity gradient. Over
extended regions the veIocity gradient is nearly constant; yet the shearing
stress changes by a factor of 4 or 5.
 MECHANICS OF BOUNDARY LAYER FLOW 35
In order to test Nevfigljadov's assumption that the shearing stress T
+ + d 2 )dy
is proportional to (d2 v ' ~ *J the ratio of r to (u" + v'z + to'*)
was plotted against duldy. The results are shown in Fig. 25. For the

8 du
""i;-
FIO.!&.-Relation between &awing atwea coefficient and mean velocity gradient.

FIG. 25.-Ratio of shearing etress 7 to mean energy of turbulence as a function of the

mean velocity gradient.
17$- and 20-ft positions the ratio is almost independent of du/dy except
in the outer part of the layer where T falls to aero and the turbulence
decreases to that of the free stream. Perhaps if the free stream were
sufficiently free from turbulence, both quantities would fall to wro
together, retaining a constant ratio. Nevzgljadov's assumption is not a
36 HUGH L. DRYDEN

good approximation but confirmation is given to von K4rmin’s assump-

tion of statistical similarity between the fluctuations a t different points in
the same cross section. However, for stations further downstream as the
boundary layer proceeds further against the adverse pressure gradient,
there is considerable variation in the ratio of T to (d2 v f 2+ +

19 20 21 22 23 24 25 26
X , Feat

FIQ. 26.-Contour map of uu correlation coefficient for NBS turbulent boundary

layer.

X , Feet

FIG.27.--Contour map of u’/u for NBS turbulent boundary layer.

Contours of the uv correlation coefficient are shown in Fig. 26.

Comparison of Figs. 19, 20, 21, 22 and 26 shows that the turbulent stress
near the wall falls regularly and smoothly as separation is approached
and that the decrease is caused by decreasing uv correlation and not by
any marked diminution in u’ or u’. The laminar sublayer apparently
serves only as a medium for transferring the stress from the fluid to the
wall and does not play any fundamental r61e in the phenomenon of turbu-
lent separation.
 :
MECHANICS OF BOUNDARY LAYER FLOW 37
Although the absolute values of u', v', and w' near the wall decrease
slowly as separation is approached, they become continually increasing
fractions of the mean speed because of the decreasing mean speed. At
separation the u' fluctuation near the wall exceeds 40 per cent of the
local mean speed as seen in Fig. 27. Hence there are probably extended
areas of intermittent reverse flow just upstream from the mean separation
point, since it is well known that momentary fluctuations of three times
the root mean square value are quite frequent.

4" a4

0.2

7
o? -0.2
00

I ae -
y. kh.

FIG.28.-Correlation between velocity fluctuations a t two points in the same cross

section of the NBS turbulent boundary layer a t z = 20 ft.
At left, one point is fixed a t g = 1.56 inches, a t right one point is fixed a t y = 3.62
inches.
A few measurements have been made of the correlation between u
fluctuations a t two points in the same cross section of the boundary layer
as function of location of one of the points and their distance apart.
The cross section selected was a t 20 ft from the leading edge. Two
typical curves are shown in Fig. 28. The surprising result is that there
is a positive correlation between the fluctuation a t any point in the
boundary layer and that at a point in the middle of the boundary layer,
and a negative correlation between the fluctuations at a point just outside
the boundary layer and that a t the middle of the boundary layer. The
spatial extent of the disturbance at any instant is large compared with
the boundary layer thickness. Here again the mixing length concept
seems wholly inadequate.
The negative correlation may be interpreted as the effect of the outer
rough "edge" of the boundary layer on the external flow, an effect sug-
38 HUQH L. DRYDEN

gested by 0.I. Taylor in private discussion. If the disturbance is pre-

dominantly a slowing up of the boundary layer at any moment, the
boundary layer will thicken by reason of the continuity of the flow. This
presents a bump to the outer potential flow, and the outer flow will speed
up in passing over the bump. Thus the two effects will be in opposite
phase and there will be a negative correlation.
I n terms of the kinetic theory analogy or the mixing length descrip-
tion, these results show that the “mean free path,” mixing length, or
scale of the turbulent processes is large compared with the thickness of
the boundary layer. Considerable masses of fluid move as more or less
coherent units. The process cannot be smoothed by averaging over a
small volume because it is not possible to choose dimensions small com-
pared with the boundary layer thickness and at the same time large
compared with a single fluid element. The mixing-length idea and even
the idea that the turbulent fluctuations and turbulent shear stress are
directly related to the mean speed at a point and its derivatives at that
point must be abandoned. Shall the flow then be regarded as a mean
flow that merely transports and distorts large eddies superposed on the
flow, these eddies being of varying she and intensity? Or shall the flow
be regarded as a mean flow on which traveling wave systems are super-
posed, each wave extending across the entire boundary layer and traveling
at a speed determined by the geometry of the whole boundary layer, the
speed outside the boundary layer, and the physical properties of the
fluid? It is hoped such questions will be answered by further experiment.
The rapidly developing theory of random functions (96) may possibly
form the mathematical framework of an improved theory of turbulence.
However it is necessary to separate the random processes from the non-
random processes. It is not yet fully clear what the random elements
are in turbulent flows. The experimental results described suggest that
the ideas of Tollmien and Prandtl that the measured fluctuations include
both random and non-random elements are correct, but as yet there is
no known procedure either experimental or theoretical for separating
them.
Bibliography
on
Turbulent Flow in Boundary Layers
84.TAYLOR,G. I., Statistical theory of turbulenoe, V. Effect of turbulence on
boundary layers, Proc. Roy. Soc. (London)A , 166, 307-317 (1936).
85. TAYLOR,G. I., Flow in pipes and between parallel planes, Proc. Roy. Soe. (Lon-
don) A , 169, 496-506 (1937).
86. VONDOENHOFF, N., Determination of general relations
A. E., and TETERVIN,
 MECHANICS OF BOUNDARY LAYER FLOW 39
for the behavior of turbulent boundary layers, N.A.C.A. Wartime Report,
L-382 (1943).
87. GOLDSTEIN, S.,The similarity theory of turbulence, and flow between parallel
plates and through pipes, Proe. Roy. Soc. (London) A , 169, 473-496 (1937).
88. SQUIRE, H.B., and YOUNQ, A. D., The calculation of the profile drag of aerofoils,
Brit. A.R.C., Repts. an& Memo.No. 1838 (1938).
89. JACOBS, W., Umformung eines turbulenten Geschwindigkeitsprofs, 2. angdw.
Math. Mech., 19, 87-100 (1939);also N.A.C.A. Tech. Memo. No. 961 (1940).
90. SASAKI, T.,On the theory of turbulent boundary layer on a flat plate, Aeronaut.
Research Lab. Rept., Tokyo Imp. Uniu., 16 (No. 8), 483-492 (1941). (Japanese
text, English summary.)
91. KALIKHMAN, L. E.,A new method for calculating the turbulent boundary layer
and determining the separation point, Compt. Rend. Akad. Nauk U.S.S.R.
Leningrad, 38, 180-185 (1943).
92. KEHL,A., Untersuchungen iiber konvergente und divergente turbulente Rei-
bungsachichten, Zng. Arch., 13, 293-329 (1943).
93. MOHR,E.,Laminare und turbulente Stromungen mit besonderer Beriiktsichtig-
ung des Einflusses von festen Wlinden, 2.Physik, 121, 301-350 (1943).
94. NEVZQLJADOV,
' V.,A phenomenological theory of turbulence, J. Phys. (U.S.S.R.),
9, 235-243 (1945).
95. TETERVIN, N.,Approximate formulas for the computation of turbulent boundary
layer momentum thicknesses in compressible flows, N.A.C.A. Wartime Rept.
L-119 (1946).
96. BASS, J., Lea fonctiona alhtoires e t leur interpretation mkanique, Remre sci.,
83, 3-20 (1945).
Bibliography
of
Related Papers
Drag Calculations
97. F h Q E , A., Profile and skin friction airfoil drags, Brit. A .R.C., Repts. and Memo.
No. 1862 (1938).
98, PIERCY, N. A. V., PRESTON, J. H., and WHITEHEAD, L. G., Approximate pre-
diction of skin friction and lift, Phil. Mag. [7],26, 791-815 (1938).
99. SQUIRE, H.B.,and YOUNQ,A. D., The calculation of the profile drag of airfoils,
Brit. A.R.C., Repts. and Memo. No. 1838 (1938).
100. PRETSCH, J., Zur theoretischen Berechnung des Profilwiderstandes, Jahr. deut.
Luftfahrtjorsch., 1938,Vol. I, 60;also N.A.C.A. Tech. Memo. No.1009 (1942).
101. YOUNQ, A. D., The calculation of the total and skin friction drags of bodies of
revolution a t 0' incidence, Brit. A.R.C., Repts. and Memo. No. 1874 (1939).
102. TETERVIN, N.,A method for the rapid estimation of turbulent boundary-layer
thicknesses for calculating profile drag, N.A.C.A. Wartime Rept. L-16 (1944).
103. JACOBS, E. N.,and VON DOENHOFP, A. E., Formulas for use in boundary-layer
calculations on low-drag wings, N.A.C.A. Wartime Rept. L-319 (1941).
104. SCHULTZ-GRUNOW, F., Neues Reibungswiderstandsgeseta fur glatte Platten,
Luftlahrlforsch., 17, 239-246 (1940); also N.A.C.A. Tech. Memo. No. 986
(1941).
105. FALKNER, V. M.,A new law for calculating drag. The resistance of a smooth
flat plate with turbulent boundary layer, Aircraft Eng., 16,65-69 (1943).
40 HUGH L. DRYDEN

106. NITZBERO,G. E., A concise theoretical method of profile-drag calculations,

N.A.C.A. ACR No. 4BO6 (1944). (Nowdeclassified.)
Flight Tests on Boundarg Layers
107. JONES,B. N., Flight experiments on boundary layera, J . Aeronaut. Sci., 6,
81-101 (1938); also Engineering, 146, 397 (1938); also Aircraft Eng., 10,
136-141 (1938).
108. STEPHENS, A. V., and HASLAM, J. A. V., Flight experiments on boundary layer
transition in relation to profile drag, Brit. A.R.C., Repts. and Memo. No. 1800
(1938).
109. RUYPH, L. R., Js.,and SCHAIRER, R., Boundary layer and wake survey measure-
ments in flight and in the wind tunnel, J . Aeronaut. Sci., 7, 426433 (1940).
110. YOUNG, A. D., and Momus, D. E., Note on flight tests on the effect of slip-
stream on boundary layer flow,Brit. A .R.C. Repts. and Memo. No.1967 (1939).