A UNIFIED VIEW OF THE LAW OF THE WALL
USING MIXING-LENGTH THEORY
by
V. C. Pate!
Sponsored by
General Hydromechanics Research Program of the Naval Ship Systems Command Naval Ship Research and Development Center
Contract No. N00014-68-A-0198-0002
uHR Report No. 137
Iowa Institute of Hydraulic Research
The University of Iowa
Iowa City, Iowa
April 1972
This document has been approved for public release and sale; its distribution is unlimited.
ABSTRACT
It is shown that, if the well-known mixing-length formula is regarded simply as a relationship between the velocity and the stress distributions in the wall region of a turbulent flow, then a truly universal distribution of mixing length is sufficient to describe the experimentally observed departures of the velocity distribution from the usual law of the wall as a result of severe pressure gradients and
transverse surface curvature. Comparisons have been made with a wide
variety of experimental data to demonstráte the general validity of the mixing-length model in describing the flow close to a smooth wall.
An extension of the relaminarization criterion of Patel and Head, and sorne experimental evidence, suggest that the thick axisymmetric boundary layer on a slender cylinder placed axially in a uniform stream
can not be maintained in a fully turbulent state for values of the Reynolds number, based on friction velocity and cylInder radius, below a certain critical value.
ACGOWLEDGEIVENTS
This report is based upon research conducted under the General Hydromechanics Research Program of the Naval Ship Systems Command, technic-ally administered by the Naval Ship Research and Development Center, under Contract NOO011-68-A-Ol96-OOQ2.
The author gratefully acknowledges the assistance in data
reduction and computation he received from Messers K.C. Chang, A. Nakayama
and R. Damian of the Iowa Institute of Hydraulic Research.
LIST OF COETS
Page
I. INTRODUCTION i
II. RELATION BETWEEN THE VELOCITY AND STRESS DISTRIBUTIONS 3
III. MIXING-LENGTH DISTRIBUTION
IV. STRESS VARIATION MID VELOCITY DISTRIBUTION IN THE WALL
REGION 5
Two-Dimensional Boundary Layer 5
Fully-Developed Flow in a Pipe 8
Thiòk Axisymmetric Boundary Layer 8
Fully-Developed Flow Through Concentric Annuli 11
V. EXPERINTAL VERIFICATION OF TI INFLUENCE OF TRANSVERSE
CURVATURE 13
VI. DISCUSSION 18
VII. CONCLUSIONS 20
REFERENCES 21
LIST OF FIGURES
Page
Definition Sketch, Mixing-Length Theory. 23
The Continuous Mixing-Length Model, Equation (8). 21L
The Influence of Pressure Gradient on the
Law-of-the-Wall. A Plot of Equations
(16)
and (17). 25 Velocity Measurements of Newman (1951) Comparedwith Mixing-Length Theory. 26
Definition Sketch for the Thick Boundary Layer on
a Cylinder. 27
The Influence of Transverse Curvature on the
Law-of-the-Wall; A Plot of Equation (26).
28
Determination of Wall Shear Stress by the Method
of Clauser. 29
Comparison of the Measurements of Richmond with
Equation (26). 30
Comparison of the Measurements of Keshavan with
Equation (26). 31
Comparison of the Measurements of Yu, and
Willmarth and Yang with Equation (26). 32
Comparison of Mixing-Length Theory with the
Measurements of Lawn and Elliott in Concentric Annuli. 33
Figure 1. Figure
2.
Figure3.
Figure 1. Figure 5. Figure6.
Figure7.
Figure8.
Figure9.
Figure lo. Figure 11.iv
LIST OF TABLES
Page Table 1. Summary of Data on Thick Axisymmetric Thrbulent
Boundary Layers 3I
Table ? A Summary of the Concentric Annuli Data of Lawn
A UNIFIED VIEW OF THE LAW OF THE WALL USING MIXING-LENGTH THEORY
I. INTRODUCTION
Eddy-viscosity and mixing-length models daté back to the
earliest days of the study of turbulence. They have received a great
deal of attention over the years both frein research workers who have sought to discredit the concepts involved, or at least to establish their
limitations, and in the opposite camp, from engineers who have recognized
their usefulness as a basis for performing practical shear-flow calculations. Recent experience in turbulent boundary-layer calculations suggests that
these models work considerably better in practice than one ha any right
to expect on physical grounds.
A number of physical interpretations have been placed on the
well known mixing-length formula,
.jTt
in addition to Prandtl's original heuristic one.
The interpretation that appears to have received the widest acceptance in recent years is the one due to Batchelor (1950). According to this, the mixing-length
expression represents the equilibrium between the production and dissipation of
turbulent knietic-enery which is implied when convection and diffusion are
assued negligibly small. Detailed turbulence measurements indicate that such an equilibrium between production and dissipation is approached most
closely in the wall region of turbulent boundary layers. This region is
commonly referred to as the inner layer or the law-of-the-wall region.
Here, the mixing-length expression is used as a relationship
between the Reynolds stress and the mean velocity to study the velocity distributions in the wall region of boundary layers developing in pressure
2
influence of pressure gradients on the law of the wall in boundary layers developing on nominally plane surfaces has been the subject of a
rnber of previous itvestigations. It is nòw generally recoiized that
the usual law of the. wall,
u/U = f(Uy/v),
T T
wh-ich has been found to apply in fully-developed flows i pipes and
channels, as well as in the flat-plate boundary layer, ceases to be valid in the presence of sbstantial streamwise pressure gradients, and that the
departures from the usual law can be predicted using mixing-length theory.
The results of the earlier investigations are re-examined here, and a
parallel analysis is presented for axially-symmetric flows in order to
show that mixing-length theory once again predicts a significant
influence of transverse curvature on the velocity distribution in the
wall region. As far as the flow in this region is concerned., transverse
curvature effects have been observed in two types of experiments: in
the thick axisyinmetric boundary layer developing on a long slender
óylinder placed axially in a uniform stream (there being no streamwise
pressuré gradient in this. case), and in fully-developed turbulent flow in the concentric annulai' space formed by two parallel pipes. The velocity
laws deduced from mixing-length theory a1,e been compared with experimental
data from these sources. The substantial agreement found between the
measurements and theory once again demonstrates the general validity and the usefulness of the mixing-length model.
The results of the analysis presented here and the experimental
data collected elsewhere show that there is a remarkable. similarity
between the influence of a favorable pressure gradient ön the flow in
the wall region of a two-dimensional boundary 1rer and the ifluence of
transverse surface curvature on the flow in the -wall region of an
axisymmetric turbulent flow. One is therefore tempted to apply the
criterion proposed by Patel and Head
(1968)
for the re-laminarizationof a two-dimensionâ.J. turbulent boundary ayer under the influence of a favorable pressure gradient to the case of turbulent flow on a slender cylinder. An extension of this criterion and. rather limited experimental
way
3
evidence suggest that there may be a lower limit to the cylinder Reynolds number, based on friction velocity and the radius of the cylinder, below which the flow in the neighborhood of the cylinder can not be maintained
in a fully turbulent state.
II. RELATION BETWE THE VELOCITY AND STRESS DISTRIBUTIONS.
We consider a nominally parállel turbulent shear flow such as
occurs in a pipe or the turbulent
boundary layer. Referring to Figure 1for notation, the total shear stress at a point may be written
T = , (i)
this being the sum of the molecular and Reynolds stresses. If the latter
is related to the mean velocity distribution via the mixing-length theory, equation (i) becomes
au
T =p+p
--- . 2Introducing the usual non-dimensional quantities defined in the following
= U/U y =
Uy/y
=UL/v ,
T = T/Tw and UTwhere T IS the shear stress at the wall, y = O, we have
(j*. .4.)2
(au:) - 1*= O
Solving this quadratic, we obtain
au:
21*ay:
{l
+and since u = O at y = O, this yields
rw (J
-
* U - j0 i + + (1)(5)
This is a r-eltion between the velocity distributior ¿nd the stress distribution. In ordeì to proceed further, we must prescribe the
func-tion ve and also the variafunc-tion with y* of the shear stress
T*.
ÏII..
MIXING-LENGTH .ISThIBUTIONFor the flow iñ the neighborhood of a smooth wall we identify threé regions in the usual way, namely the sublayer, the blending region
and the fully turbulent region. Through the sublayer (r < 5, say) the
Reynolds stress is assumed to be negligible so that
For the fully turbulènt region the molecular stress is neglected in
comparisiön with the Reynolds stress and dimensional reasoning leads co
the familiar linear väiation of mixing length in. the form
= K,
(7)where K isthe von Kármn coistant. In subsequent work K will be
assumed equal. to
o.i8
as suggested by Patel(l965b).
In the blendingregion between the sublayer and the fully-turbulent region., both molecular
and Reynolds stresses are present. Composite mixing-length and
eddy-vis-cosity foulae ihich may bé ued for this region. (5 < y < 50, say) have
been suggested by a number of workers on thç basis Qf empirical correlation
of experimental òbsèrvations. Amongst the most well-known fQrm'llae are
those of Reichardt
(1951),
Deissler(i95),
and varDriest(1956).
Herewe shall use afi expression proposed recently by Landweber and Poreh (to
be published.)! This is ve = cy* { talill(X2y*2
y
(8)
where X is a constant such that the tanh terni approaches
tu4ty
in theregion of
y
= 50.
It can readily be verified that equation(8)
applies thí-oughout the wall region since it gives the correct behavior of
for
small and large y*. A plot of equation
(8)
is shom in Figure 2.In order to obtain formulae for the velocity distribution we need to consider next the variation of the stress through the wall region. The stress distribution will in general dépend upon the geometry
of the flow. In what follows, we shall examine a few cases of interest.
A. Two-Dimensional Boundary Layer. We consider first a
thin two-dimensional boundary layer developing in a pressure gradient.
Using standard notation, the relevant equations may be written
IV. STRESS VARIATION AND VELOCITY DISTRIBUTION IN THE WALL REGION
Du Du
u +v =
DxBy
pdx
pDy
Du + Dv
Dx Dy
O,
where x is measured along the wall and y normal to it. The components
of mean vélocity in these directions are u and y , respectively, and T
is the total shear stress, i.e. the sum of the molecular and Reynolds stresses.
Integration of equation (9) with respect to y gives
= i + py*
+
4-
J(u
+ y
dyo
- v.
where
=
Ur
dx is a non-dimensional pressure-gradient parameter.Equation (u) shows that the stress variation occurs as a result of the
pressure gradient and the flow accelerations, the latter being zero at the wall itself. In the sublayer it is permissible to neglect the flow accelerations so that
= i + (12)
is a good approximation. Further away from the wall, in the fully-turbulent region, say, the acceleration term is not negligible. Experiments indicate that the stress distribution is nearly linear
T* =
lTY
(13)6
where A is a non-dimensional stress-gradient parameter,
T pU y
indepefldent of y* in the wall region iediately outside the blending region.
There are two limiting cases of interest. I-nj a flat-plate boundary layer
it can be shown that the accéleration terms are small so that T* = 3.
throughout the wall region. Notice that this, in conjunction with the
-mixing-length relations of equations (6) and
(7)
lead to the familiarlinear relation,
for the sublayer and the logarithmic law,
i
u = - in y* + B
for the fully turbulent region. These laws can be used in equation (n) to check, retrospectively,, the assumptIon that the acceleration terms are indeed negligible. The second limiting Case is the flow in the
neighborhood of separation. Here again the flow acceleration can be
neglected so that A becomes equal to A. The mixing lengthfomula then
leads tO the well known one-half power law for the velocity distribution as suggested by Stratford (1959).
It will be noted that equation (13) is an empirica], fit to experimental data introduced in order to obtain the velocity distribution
in closed form expressions. This relation was first suggested by
Townsend (1961) and has since been used by a number of authors. In
reality, however, the stress gradient falls continuously through the
wall region from its value equal to the pressure gradient at the wall.
An attempt to represent the stress variation through the blending region,
i.e., one in which A = A near the wall and A = constant further away
T
p
Tfrom it, has been made by McDonald (1969), who suggested a series expansion
for the stress gradient with terms involving a tanh function similar to that in equation (8).
Equations (12) and (13) can now be used in conjunction with the mixing-length model to obtain expressions for the velocity distributions
in the sublayer and the fully-turbulent region. Integration
or
equation(5)
using equations(6)
and (12) leads to the velocity distributionin-in the sublayer
u* =
(15)
In a similar manner, equations (5),
(7)
and (13) give the velocity distribution in the fully turbulent region in the(1 + *)½
i-u* =
when A , end therefore
p
positive for adverse pressure gradients.
T
form
½
+2{(1+Ay*)½_1]
p+B* ,(17)
T
(1 + Ay) + i
pA , are both zero. Note that A and A are
T p T
pressure gradients and negative for favorable
In order to obtain the velocity distribution in the blending region and thereby to find B*, it is of course necessary to integrate
equation (5) using the more complicated mixing-length relation given in
equation (8) and a continuous stress variation model buch as that suggested by McDonald. The integration can be performed numerically and the results
should join smoothly into equations (16) and (17). Such a calculation is
not attempted here but it may be of interest to note that B* is nearly
a linear function of A ,viz
p
B* = B + 3.7 A (18)
(Patel and Head 1968). Using typical values of A and Aequations
(i6)
and (17) are plotted in Figure 3.Equation (17) was first obtained by Townsend (1961) and has been
used, with minor modifications, by Patel (l965a), Mellor (1966) and
McDonald (1969). Experimental verification of this relation is rather scarse.
While measurements of velocity profiles in boundary layers with large adverse
and favorable pressure gradients indicate departures from the usual
loga-rithmic law (equation
15)
in the manner suggested by equation (17), theonly direct comparisons with experiment made so far are those due to
Patel and. Head (1968.) (reproduced here as Figure 1) and to McDònald (1969). in order to check the validity of this relation it is of course necessary to measure the wall shear-stress and the velocity arid shear-stress profiles.
The wall shear-stress has to be measured quite accurately and by a method
which does not rely upon the usual logarithmic law.
where B * is expected to be some function of the sublayer parameter A
B. Fully-Developed Flow in.a Pipe. Here the stress distribution
is exactly linear and. related to the pressure gradient by
2dx
2 d-xwhere. a is pipe radius, r is measured from the pipe axis, while y is
the distance from thê wall. In dimensionless form this may be written
T*=i_
=l+y*=1+½y*
,a*
t
pUa
where a* . It cafl easily be showi that for moderate to high pipe
Reynolds numbers (say, Re > i0), the pressure-gradient parameter is small
and a* is large (t _2/a* = -20 Re7I'8). It is permissible therefore to assume = i in the neighborhood of the wall. This leads then to the
conclusion that the velocity distributions in the wall region in
fully-developed turbulent pipe flow and. flat-plate turbulént boundary layers are identical. ThIs has indeed been demonstrated to be the case by the
experiments of Head and Rechenbérg (1962).
For RerÍiolds numbers less than about i0 we may epect depä.rtures
from the usual inner-law equations (li-L) and (15). Suc1 departures have
been observed by Patel and Héad (1969) and may be attributêd to the influence
of eithér the radius of curvature (a*) or the increasingly favorable
pressure gradient ( ), the two effects being inseparable in this case.
p
The influence of transverse crvature on the velocity distribution
is considered in the two examplès that follow.
C. Thick Axisetric BOUrIda±Y Layer. Here we shall consider
the turbulent boundary layer develoing on a long cylinder of constant
radius a placed adally in an unifôm stream of velocity U If the
cylinder is long and slender enotigh the b,oündary layer will become thick
in comparison with the rad:ius of the c1inder. We may then expect the flow to be influeuiced by transverse curvature effects. Since the pressure
gradient is absent, the relevant boundary layer equations are
Bu
= )
1 ,tr. 203x ay
ray
p9
and (ur) +
(vr) =0
(21)the notation being shown in Figure 5. By ana1or with the flat-plate
boundary layer, e again neglect the acceleration terms in equation
(20) to obtain the stress distribution in the neighborhood of the wall. (Note that this assumption can be verified, retrospectively, just as
in the case of the flat-plate boundary layer.) Thus,the stress distribution
is given by a T = T
-vr
or non-dimensionally by = a*+y* a* (22)This expression is valid throughout the wall region, i.e. in the sub-layer, the blending region, as well as in the fully-turbulent region.
By following a procedure similar to that outlined for the two-dimensional boundary layer one obtains the velocity distribution in the sublayer in the form
u = a* £n (i + y*/a*) and that in the fully-turbulent region in form
1 (
ti +
U
- -
14*'
y,a,,
-- K
(i +
y*/a*)+ i J
r
where B* is somewhat similar to 3* in that it is expected to be a
function of a* such that it reduces to the constant B in equation (is)
when a* = - , i.e. for a thin boundary layer.
We may note here several points of interest concerning the velocity distribution in the wall region of a thick axisymmetric boundary layer. First, we see that for sufficiently small values of a* significant variations in stress occur through the wall region as a consequence of the flow geometry. The stress continues to fall monotonicälly from its value at the wall. This is to be compared with the situation in a
two-dimensional boundary layer developing in a favorable pressure gradient < O , < o) where the stress falls from its value at the wall as
p
Tlo
It isnot surprising therefore to find that equations (17) and (214) show similar departures, from the usual logarithc law obtaining in
a fiat-plate boundary layer (compäre the velocity distributions in
Figure 3 for t < O with those in Figure
6).
Secondly, vemy comparethe velocity distribution in the fully-turbulent region, equation (214),.
obtained via mixing-length arguments with the one proposed earlier by Rao
(1967)
from purely heuristic considerations, namely= - 2n {a*Zn (i + y*/a*)} + B (25)
where B is the usual additive constant of equation (15). Rao obtained
this expression simply by replacing y* by the quantity a*9..n (i + y*/a*)
in equation (15). This substitution of variables was justified by Rao on the basis of the òbservation that a similar substitution in the usual sublayer law, equation (i14), gives the correct sublayer law, equation (23), for axisymmetric flow. Be: that as it may, it is indeed surprising to
find that, if the expected departure o B from B. is ignored, the velocity
distributions given by equations (214) and (25) are in excellent agreement with each other over a wide range .of variables. This is easily demonstrated by comparing
(i + y*/aI)½
-with { 2n (i y*/a*) }
(i + y*/a*)½ + i
j
to show that the disagreement is of the order of 1 percent for .y* = a'
and less than 8 percent for
y
= 6a*,
the latter value of y* being expectedto be well outside the range of applicabi.ity of the
wall
region analysis.The,two laws may however disagree or account of the fact that the values.
of Br* in the present formula may be considerably different from B.
In order to obtaIn the velocity distrIbitioi that is continuous
through the sublayer, the biendng region, as well as he fully-turbulent
region it is of course necessary to integrate equation
(5)
using thecontinuous mixing-length model of equation
(8)
and the stress di stribut ion(26)
given by equation (22). Thus, we have y* * = ) a
)]2
;. J&r
J { 1 + [1 + 141(21*2 tarxh(X2y*2) a*+y* 2(a**
This expréssion has been integrated numerically for a range of values of a* using K = o.1l8 and X = v'/63, and the resulting velocity distributions are shown in Figure 6. The results of these calculations can be compared
with equation (21f) for alues of y* jying in the fully-turbulent region
to obtain a suitable expression for the variation of B* with a*.
We shall compare the velocity laws obtained in this section with available experimental data from thick axisymmetric boundary layers
in a subsequent seôtion following some discussion of yet another class of flows which has received considerable attention from previous investi-gators.
D. Fully-Devèloped Flow Through Concentric Annuli. Here we
consider fully-developed turbulent flow in the annular space formed by two concentric cylinders of radii a. and a, subscripts i and o denoting
inner and outer cylinders, respectively. This flow has been investigated
experimentally and theoretically by many workers over a number of years. Among the most compréhensive studies are those due to Rothfus, Monrad
and Senecal
(1950),
Owen(1951),
Knudsen and Katz(1950),
Brighton andJones
(l96I),
Quarmby(1967),
Levy(1967),
and Lawn and Elliott(1971).
One of the major observations of these studies is that the velocitydistribution near the inner wall does not conform with the usual
logarith-mic law of equation (15). The profiles on the outer wall are, however,
adequately correlated on the basis of this equation. The disagreement
with the usual inner law on the inner surface has generally been attributed
to the influence of transverse curvature. Some authors have indeed used
eddy-viscosity and mixing-length arguments to demonstrate that the observed departures are to be expected. In what follows we shall again apply the mixing-length analysis to analyse the velocity distribution in the region close to the inner and outer walls using the appropriate stréss variation. while this will undoubtedly duplicate some of the previous efforts it
will serve several useful purposes. First of all, it will enable us to
obtain a unified view of the success of mixing-length analyses in predicting the velocity distributions in turbulent flows close to a solId wall regardless of the nature of the flow farther away from the wall. Secondly, it will show that the flow in the wall regions of (i) a
two-dimensional boundary layer developing under a. favorable pressure-gradient, a thick axisyimnetric boundary layer on a long slender cylinder, and
12
closely similar dynaznícs insofar as the velòôity and shear-stress dis-tributiöns are' concerned. ina1ly, unlike the previois studies, we shall
Obtain the velocity distribution not only in the
fully-turbulent
regionbut
also
in the sublayeränd
the blending regions.If x is measured in the direction of flow
and
r is the radialdistance measured from the common
axis
of the inner and outer cylinders,it isreadily shown that the stress distribution in the annulus is given by
and
a.
a.
a.
= __:_ + . _2:. (-r- -T rdxT
'a. r' w.. V. 1i
or by i = Tr
2dxt
a
r
w w o o0
where T
and T
are the wall shear stresses at the inner and outeri o
cylinders, respectively. If we now measure distances y and y0 froni the
inner and outèr walls, and introduce non-dimensional quantities as before,
these expressions beöoine
T*
where the subscripts i and o indicate that.the variables have been made
non-dimensional by using the shear velocity UT = or
U
IT/
. Equations (29) and (30) show immediately. that in thiscase the stress variat-iöns occur not
only
as a result of transverecurvature but also due to the favorable pressure
gradient
driving theflow.
The relative importance of the two .nfluences will obviously
depend upon the parameters a.*, a0*, 4
and
,
or alternatively,
upon the radius ratio a./a0
andthe flow Reynolds nunber.
(
i
i½
pi1-
½a.*
a* (2
y.* y*
(29)(30)
a*+y.*
a*
°
y*
)y*
=a*
y0* {13
The velocity distribution in the wall regions of the outer and inner cylinders can now be found by again integrating equation (5)
making use of the stress distributions given above and the mixing-length expressions of Sectionill. The velocity laws 'obtained by performing the integration numerically will be compared with experimental results in the next section. It is, however, interesting to note here that in
situations where the pressure gradient terms can be neglected in equations (29) and (30), the velocity laws can be obtained in closed f rm for the
sublayer and the fully-turbulent regions. The laws for the innêr wall
are then identical with, those for the thick axisymmetric boud' y layer given previously.
V. PERIMENTAL VIFICATION 0F TIE. INFLUENCE OF TRANSVERSE CURVATURE.
We have already mentioned that the influence of pressure
gradi-ents on the law of the wall in the case of thi two-dimensional boundary
layers has been studied by a number of workers. The expressions obtained
from mix±ng-length analysis (equations 16 and 17) have been found to give
substantial agreement vith experimental data. In this' section we shall
therefore restrict our attention to the influence of transverse, curvature. in order to verify the expressions för the velocity distribution obtained in the previous section we shall rely upon data from two types of
experiments: measurements in thick a.xisymmetric boundary layers and
those ïn fully-developed turbulent flow in concentric annuli. Both flows
have been investigated experimentally over a wide range of parameters. Considering the thick axisynimetric boundary layer first, we find that this flow has been examined experimentally by Richmoiid (1957), Yu (1958), Yasu.hara (1959), Keshavan '(1969), and WilJ.marth and Yang(1970),
and theoretically by Landveber (19I9) and Cébeci (1970). The vide range
of variables encountered in these experiments can be' seen from the data summary given in Table 1. It should be emphasized that this table does not contain al]. the measurements that are available; many more measurements vere examined during the course of this investigation and Table 1 was
l4
ICeshavari ánd Yu measured velocity profiles at severa]. streamwise statiQns for each cylinder diameter, we have listed only the measurements at
the more downstream stations so as to emphasize the influence of
trans-verse ciirvature.
It wifl be clear that in order to compare the measured velocity
distributions with the expressions of the previous section it is necesary
to determine, independently, the wall shear stress. In Table 1 we have
listed the values of the wall shear-stress coefficient, Cf quoted by
the originators. The values quoted by Richmond for his measurements
on the smaller cylinder are thought to be in error. In the experiments
of ICeshavan the wall shear stress was deduced from the measured streainwise
momentum-thickness development by applying the momentum-integral equation..
As coimriented upon by a number of previous investigators, this method is
not altogether reliable. In view of this, much of Keshavan's subsequent
analysis of his data must be regarded with caution. In situations where
such doubts existed concerning the correct value of the wall shear stress,
we followed an alternative method to determine Cf. This is the method due
to Clauser
(1956).
In this method, a chart is made by plotting the usualflat-plate law of the wall in a parametric form, the parameter being Cf.
The measured velocity profile is then superimposed on this chart and the value of Cf read d-irectly by examining the measurements close to the wall where the law of the wall is expected to apply. rpical "Clauser plots"
for two profiles measured by Richmond and one by Keshavan are shown in Figure 7. This method of determining Cf applies strictly to flows in which the usual flat-plate law of the wall is expected to be valid.
Nevertheless, since the departure from this law in the present case is
gradual, being s,n1l in the sublayer and. the blending regions, the method
gives a good approximation to the actual wall shear stress provided due
emphasis is placed only on the experimental points lying in the sublayer and. blending regions. Where such measurements are not available, as for exainple in the profile of Keshavan shown in Figure
7,
a reasonableextrapolation of the data points can still be made to estimate the value
of C.
The valuès of the skin-friction coefficient determined by thedifferences betweén these and the values quoted by Richmond and Keshavan. For the experiments of these two authors we have also tabu-lated the
values obtained theoretically by Cebeci (1970). In this recent paper,
Cebeci obtained numerical solutions to the eq.uatiòns of a thick
axisym-metric turbulent boundary layer on a slender cylinder using an eddy-viscosity model. The skin-friction coefficient was found to -depend primarily upon R, the Rernoìds number based on the cylinder radius, and only marginally
upon R, the Reynolds number based on the streamwise distance. The values
of Cf attributed to Cebeci's theory in Table 1 were obtained from Figure 13
in his paper after estimating the value of R in each case. It will be
seen that the theoretical values of Cebeci are in excellent agreement with those obtained by. the method of Clauser for the experiments of
Keshavan. For the smaller cylinder of Richmond, however, the theory of Cebeci predicts values of Cf which are much higher than those indicated
by the Çlauser plots. A possible reason for this discrepancy will be
discussed later on.
The value of thé parameter a*, which is a measure of the transverse curvature as it áffécts the flow in the wall region, was obtained for each case by using the quoted value of Ra and the value of
Cf indicated in Table 1, since a* = R/Cf/2 . Notice that the available
experiments cover a very wide range of this parameter, the smallest value
of a* being about 6 in the experiment of Richmond. Using the known value
of a*, the integral in equation (26) was evaluated numerically to obtain
the variation of u* with y* as predicted by mixing-length theory. The
results of these computations were compared with the measured velocity distributions. Ty-pical results, covering a representative ránge of values
of a*, are shown in Figures 8, 9, and 10. These, and similar comparisons
not included here, show several features of interest. First of all, it is
seen that the experiments indicate a very significant influence of trans-verse wall curvature ön the velocity distribution in the wall region of the boundary layer. The departures of the velocity distribution from the usual
flat-plate law of the wall are qualitatively similar to those observed in
thth plane-surface boundary layers developing under favorable pressure
gradients. The curvature parameter a* adequately describes the influence
pi
T = w. J-a. 2. i - a.* a J. Oi6
ixing-lént]:i theory appears to predict the velocity distribution. in the
wail region (y* of the order öf' 2a*) quité well for values of a* greater
than about 30. For a* less than this vaitlé (as in the Ra
= 93.8
end. 253experiments of Richmond), there are signfiòant discrepancies between
the meàsured profiles and mbdng-length predictions While these may, in
part, be due to the difficulty of obtaining thé correct value of the wall
shear stress, rather gross changes in C are required to reconcile
mi,dng-length theory ìith experiment, and such óhanges are not indicated by the C1ausr plots. A alternative explanation must therefore be offered for
the failure of the mixing-length theory, and also for thé observation that
the experiientai values of Cf are considerably lower than the theoretical
values of Cebeci, for a* less than about 30. Such an explanation is
attempted in the next sectiön.
Considering next the measürements in fully-developed turbulent flow in concentric annuli we find that there should be no difficulty in
this case concerning the wail shear stresses on the inner and outer surfaces.
since these are determinéd by thé streamwise pressure gradient (dp/dx) and
the location of the zero stress point (r r) in the annulus. Using
momentulii considerations t is easily verified that
2 - a 2 r 2a. dx 1 T
-a. w _2. + a T o i (31) (33) and.a 2r
2 T 2a dx z. (3?) o oElimination of r between these equations leads to a useful. relationship
between the dimensionlesS essure-gradi-eflt and curvature parameters
17
It will be recalled that the stress distribution in the annulus can be found from equations (29) and (30) once the values of the parameters
a , and are known.
i o
p
As we have already mentioned., the cöncentric annulus flow has
been investigated by a large number of workers. In the earlier
measure-ments the assumption has been made that the zero stress point coincides
with the positioji of maximum velcity. Although all experimenters agree
on the qualitative features of the flow insofar as the behávior of the
-friction factor and veIôci-y distribution is concerned, there is' much cOntroversy concerning the exact location of the zero stress point,' and
its dependence on thé radius ratio and Reynolds number. This controversy
of courSe carries over to the determination of the wall shear stresses
and consequently to the applicability of the law of the wall. In view of
this, ve have chosento analyze the most recent experiments of Lawn an
Elliott
(1971)
where, instead of relying upon the location of the maximumvelocity point (Brighton and Jones, 19614) or upon Preston tube measurements
(Quarmby,
1967),
the zero stress point is determined directly from hot-wire measurements. The authors demonstrated the accuracy of- their stressmeasurements by comparing them with equations (27) and (28). Fortunately,
these experiments cover a range of radius ratios and Reynolds numbers that not only duplicates previous experiments but also makes a meaningful
comparison with mixing-length theory possible.
From the values of the wall shear stresses given by Lawn and
Elliott the curvature 'and pressure-gradient parameters were first determined
using the equations given above. These are summarized in Table 2 tO indicate
the range of variables encöuntered in the experiments. It will be seen that
the pressure gradient parameter' for the outer wall is larger than that for
the inner wall but, for the onditions of these experiments, the valués on
both. surfaces are considerably less than those for which significant departures and Po =
Pl
T w. J-3/2 (314) T w o18
from the usual law of the wall have been observed in thin plane-surface
boundary layers (Figure 3). This suggests that the departures observed
in the annulus experiments are primari1 due to the influence of
trans-verse curvature. The. curvature parameter a* is of course smaller for the inner surface indicating that the curvatuz'e effects v1111 first become
prominant Í'or the flow near the inner walL This is in accordance with
expérirnentál óbservations.
The values of the paraméters a* and shown in Table 2 were
used in equation (29) to find the st±ess variation, and. the velocity
distribution was computed from equation
(5)
using the continuousmixing-length model of eqjiation (8). Tyieál comparisons with experimental data
are shown in igure 11. Once again it is seen that mixing-length theory
predicts the departures of the velocity distribution from the usual law
of the wll with acceptable accuracy. It is in fact rather surprising to
find that the theory agrees with experimental data right up to the
loca-tion of the zero-stress point in the annulus. This is somewhat similar
to the often Úiade observation that, with minor changes in the constants,
the well-known logarithmic law, equation (15), can be made to agree with
the vèlocity distribution in fully-developed pipe flow right up to the center of the pipe.
VI. DISCUSSION
Attention has already been drawn, to the rather remarkable similarity between thé influence of a favorable pressure gradient and
that of transverse wall curvature insofar as the stress and velocity
distributions in the wall region are concerned. As has been demonstrated
by Patel and Head
(1968),
there is a definite limit to the favorablepressure gradient that can be imposed on a turbulent boundary layer
with-out destroying the essential equilibrii. that ed.sts between the pròduction
and dissipation of turbulent kinetic enerr in the wall region, and thereby
provoking relaminariZatiOn. For favörable pressure gradients larger than
this limiting value, the flow in the boundary layer can notbe maintained
in a fülly turbulent state. From their experiments, Patel and Head
19
criterion for the onset of this reverse transition process. In view of
the observed similarity between the influence of favorable pressure gradients and transverse wall curvature, we inquire whether there exists a similar
limiting value of the curvature parameter a below which the boundary layer
on a slender cylinder, and also the. flow closéto the inner wall in
concentric annuli, can not be maintained in a fully turbulent state. If
we compute the stress gradient at the edge of the blending region for the former case from equation (22) and set it equal to -0.009, we find that
the corresponding value of a* is about 28. If the similarity between
favorable pressure-gradient and transverse-curvature effects can be extended in this manner, the result shows that the flow in the neighborhood of
slender cylinders with values of a* less than about 28 frust be regarded
as transitional and not fully turbulent. Mixing-length theory will obviously
fail to apply in such cases. This appears to explain the large discrepancy observed earlier between the measurements of Richmond for a* = 6.0 and
13.85
(Figure 8) and the velocity distributions predicted by mixing-lengththeory. Furthermore, if such an interpretation is accepted, it is not surprising to find, as commented upon by White (1970), that the wall shear stressés measured at these low ialues of a* are considerably lower than those calculated from theories, such as that of Cebeci (1970), which assume
the existence of fully turbulent flow. The limiting value of a* for fully
turbulent flow on slender cylinders postulated here needs to be confirmed. by direct experimentation.
The present analysis of the influence of transverse curvature on the law of the wall also sheds somelight on the controversy concerning the use of Preston tubes to measure the wall shear stress on the inner wall of
a concentric annulus. Experiments as well as midng-1ength theory have shown that the usual universal law of the wall obtaining in flat-plate
boundary layers and fully-developed pipe flow does not apply in the boundary layer on slender cylinders and also on the inner wall of concentric annülus
flow. This suggests that it is not permissible in these cases to use the
Preston tube method of determining the wall shear stress. Since, however,
the departures from the universal laws are small in the sublayer and the blending regions, it shouldbe possible to obtain the correct value of the wall shear stress by making measurements with several Preston túbes of different diameters and then extrapolating the results to a hypothetical tube of zero diameter.
20
VII. CONCLUSIONS
It has been shown that, if the wéll-known mixing-length formula is viewed simply as a relationship between the velocity and the stress
distributions in the wall region of a turbulent flow, then a truly universal
distribution of mixing-length in the form 9.*(y*), used in conjunc.tiön with
the appropriate stress variation, is sufficient to determine the velocity distribution. The experimentally observed influence of streamwise pressure gradients on plane-surface boundary layers, and. of transverse wall curvature on thiòk axisymmetric boundary layers and fully developed flow in concentric
annuli, is then predicted with acceptable accuracy. It is' not necessary, as
has been attempted by a number of previous workers, particularly with regard to curvature effects (see, for example, Ktiudsen and Katz or Keshavan), to
correlate the experimental data on the basis of variations of the. constants
K and B appearing in the usüal logarithmic law, equation (15).
The general success of the mixing-length relation shown here may
be regarded as a demonstration of the universal nature of wall turhulence.
as postulated in recent turbulent kinetic-enerr theories. Furthermore, the
present work lends support to the often made claim, particularly by workers
developing boundary-layer calculation methods, that mixing-length and
eddy-viscosity models work considerably better in practice than one has aiy right to expect from Prandtl's original physical model so well described in testbooks.
21
REFERENCES
Batchelor, G.K. 1950 "Note on Free Turbulent Flows, with Special Reference to the Two-Dimensional Wake," J. Aero. Sci., 17, 14141.
Brighton, J.A. and Jones, J.B. 19614 "Fully Developed Turbulent Flow in
Annuli," J. Basic E'ig., Trans. ASME, Ser. D,
86, 835.
Cebeci, T. 1970 "Laminar and Turbulent Incompressible Boundary Layers on Slender Bodies of Revolution in Axial Flow," J. Basic E'ig., Trans.
ASZtfE, Ser. D,
92, 5145.
Clauser, F.H. 1956 "The Turbuleit Boundary Layer," Advances in Applied
Mechanics, 14, 1.
Deissler, R.G. 19514 "Analysis of Turbulent Heat Transfer, Mass Transfer and Friction in Smooth Tubes at High Prandtl and Schmidt Numbers," NACA Tech. Rep. 1210.
van Driest, E.R.
1956
"On Turbulent Flow Near a Wall," J. Aero. Sci.,23, 1007
and1036.
Head, M.R. and Rechenberg, I.
1962
"The Preston Tube as a Means of Measuring Skin Friction," J. Fluid Mech.,114, 1.
Keshavan, N.R.
1969
"Axisyiimetric Incompressible Turbulent Boundary Layers in Zero Pressure Gradient Flow," M.Sc. Thesis, Indian Institute of Science, Bangalore, India.Knudsen., J.G. and Katz, D.L.
1950
"Velocity Profiles in Annuli," Proceedingsof
Midwestern Conference on Fluid Mechanics, 1950.Landweber, L. 19149 "Effect of Transverse Curvature on Frictional Resistance," David Taylor Model Basin Report 689.
Lawn, C.J. and Elliott, C.J.
1971
"Fully Developed Turbulent Flow ThroughConcentriò Annuli," Central Electricity Generating Board, Berkeley,
England, Report RD/B/N1878.
Levy, S. 1967 "Turbulent Flow in an Annulus," J. Heat Transfer, Trans. ASME,
89, 25.
McDonald, H.
1969
"The Effect of Pressure Gradient on the Law of the Wall in Turbulent Flow," J. Fluid Meöh.,35, 311.
Mellor, G.L.
1966
"The Effects of Pressure Gradients on Turbulent Boundary Layers," J. Fluid Mech.,24, 255.
Newman, B.G.
1951
"Some Contributions to the Study of the Turbulent Boundary Layer," Aust. Dept. Supply, Report ACA-53.Owen, W.M.
1951
"Experimental Studj of Water Flow in Annular Pipes," Proc. AScE, 77, Sep. No.88.
22
Patel, V.C. 1965a "Contributions to the Studr of Turbulent Boundary Layers," Ph.D. Thesis, Cambridge University, England.
Patel, V.C. 1965b "Calibration of the Preston Tube and. Limitations on It's
Use in Pressure Gradients," J.
Fluid
Mech.,23, 185.
Patel, VC. and head, M.R.
1968
"Reversion of Turbulent to Laminar Fiov,"J.
Fluid
Mech.,314, 371.
Patel, V.C.. and Head, M.R. 1969 Velocity Profiles in Fully Mech.,
38, 181.
Quarrnby, A. 1967 "An Experimental Study of Turbulent Flow
Annuli,"
mt.
J. Mech. Sci.3 j, 205.Rao, G.N.V.
1967
"The Law of the Wall in a Thick Axisy etric Turbulent Boundary Layer," J. Basic Eng., Trans. ASME, Ser. D,89L, 237.
Réichardt, H.
1951
"Voi1stndige Darsteiling der TurbulentenGeschwindig-keitsverteilung in Glatten Leitungen," Z.A.M.M., 31,
208.
Rich±ond, R.L.
1957
"Experimental Investigation of Thick Axially Symmetric Boundary Layers on Cylinders at Subsonic and HypersQnic Speeds,"Ph.D. Thesis, California Institute of Technolor, Pasadena.
RothÍ'û.s, R.R., Monrad, C.C. and Senecal, V.E.
1950
"Velocity Distributionand Fluid Friction in Smooth Concentric Annuli,"
md and Eng
Chem,
142, 2511.
Stratford, B..S. 1959 "The Prediction of Separation of the Turbuleirt Boundary Layer," J.
Fluid
Mech.,5, 1.
Townsend, A.A.
1961
"Equilibrium Layers and Wall Turbulence," J.Fluid
Mach., 11, 97.
White, F.M.
1970
"Written Discussion of the Paper of Cebeci,1970," J. Basic
Eng., Trans. ASME.,
92, 550.
Willmarth, W.W. and Yang, C.S.
1970
"Wall-Pressure Fluctuati.ons BêneathTurbulent Boundary Layers on a Flat ].ate and a Cylin4er," J.
Fluid
Ñech., 141, 147.
Yasühara, M.
1959
"Experiments of Axisymzaetriç Boundary Layers Along aCylinder in Incompressible Flow," Trans. Jqpan Soc. Aerospace Sci.,
?L,
33.
Yu, Y.S.
1958
"Effect öf Transverse Curture on Turbu1et Boundary Layer Characteristics," J. $hip Research, 3, 33."Some Observations on Skin Friction and Developed Pipe and. Channel Flows," J.
Fluid
23
L
*
50
40
30
21450
y*
lOO
45
40
35 30 25 20 15 IO 5 o 25-
0.05 0.025 0.00500
o -1 -0.018 -0.009-2 -0.04
-0.02- -3 -0.08
-0.04 i I I 11111lp
T 0.25 I I 111111 EQUATION (I?) EQUATION (16) I I I II lilt
I I I Ii Ill
I -i liii
4 0.50 3 0.10 2 0.05 I 0.01 tor
102Figure 3. The Influence of Pressure Gradient on the Law-of-the-Wall.. A Plòt of
50 45 40 35 30 25 20 15 Io - Equation (16) Equation (14) Equation ('7)
I
Station G F D Equation (15) X X ¿ z X Xo-o
I. I l0 26 102 io3Figure )4 Velocity Measurements of Newman (1951) Compared with I"lixingLength Theory.
8 .0094 .0045
D .0279 .0141
F .1127 .075
Ue
y
w A a y 27Figure 5. Definition Sketch for the Thick Boundary Layer on a Cylinder.
30
III
y1t Figure6.
The Influence of Transverse Curvature on the Law-o'-the--Wall; A Plot of Equation (26).
i I I
ut
I I i iill
I0o
IO
102
UeY
1/
Figure 7. Determination of Wall Shear
u* 30 20 I0 I0
ye
102 Figure 8.30 20 I0
l0
io
yR
102
Author(s) Vu WIIImorth & Yortg
Figure IO. Comparison of the
Measurements of Yu, and Willmarth
and
I I I I
I uil
i i ii uil
i I I iliti
io 102Figure 11. Comparison of Mixing-Length Theory with the
Measuremen-ts of
Lawn and Efliott in Concentric Annuli.
30 c/a I Re xi5 I
Ill
o Theory X 0.088 0.362 120 I 0088 0.577 $80 2 o 0.088 1.200 335 3 + 0.176 0.94$ 567 4 20 5 II,.
F I0 ____-Table 1.
Summary of Data on Thick Axisymmetric Turbulent Boundary
Layers
*jndjcates the value of Cf used
to find a* Author(s) a inch x feet R a Cf Author Cf Clauser Plot C. Cebeci Theory a* Richmond .012 16.0 93.8 .0059 .0082* .0i46 6.00 (1957) .012 16.0 253 .0052 .0060* .0101 13.85 .5 10.0 40,200 .0029 .0030* 1555 Yu(1958) 1.0 8.0 15,250 .00318 .0036* 646 1.0 8.0 30,7140 .00282 .0030* 1192 1.0 7.0 145,060 .002614* .0026 1638 Yasuhara .3914 3.28 21,800 .003147* .0035 908 (1959) Keshavan .0625 1.29. 1425 .0068 .0085 .0088* 28.2 (1969) .0625 1.29 825 .0022 .0075 .0072* 49.6 .125 1.29 1320 .0059 .0065 .00614 714.5 .125 1.667 16140 .00147 .006o .0056* 87.6 .25 i.66 39140 .0055 .00514 .0050* 197.0 Wiflmarth & 1.5 24.0 70,200 .00276* .0028 2600 Yang (1970) 1.5 214.0 115,Q00 .00219* . .0022 3800 1.5 24.0 1314,000 .00230* .0023 14550
Table 2.
A Suninary of the Concentric Annuli Data of Lawn
and Elliott
(1971)
Radius Ratio
aia
Reynolds Number U Ti
ft/sec U T oft/sec
U T oa.*
i
-1
p.
i
a *
0 o UT..088
.362
0.914 0.714.7872
120
.00101461069
.0021414.577
1.1421.12
.7887
180
.000700
i6i14 .00114271.20
2.6142.08
.7879
335.000375
3000
.000767
2.01
14.303.39
.78814 5146.000230
14892 .00014692.37
5.01
3.95
.78814636
.000198
5702
.00014014.176
.14961.26
1.06
.81413 320.001003
1530
.00i684
.640
1.59
1.33
.8365
1404.000787
1920
.00i345
.91412.23
1.87
.8386
567.000563
2700
.000955.
1.38
3.11
2.61
.8392
791 .000140143765
.000684
2.19
4.71
3.95
.8386
1198
.000267
5700
.0001i53
.396
.350
1.16
1.08
.9310
662
.001792
1558
.002221.
.613
1.90
1.77
.9316
1085
.00109142550
.001353
.854
2.5142.37
.9331
11452.000819
31420.001008
1.24
3.5143.30
.9322
2022
.000588
14760.000726
1.61
14.364.05
.9289
21492 .0001475 5850.000593.
Sttirì ( ('1 sst (i(.It Ion
DOCUMENT CONTROL DATA-R&D
1V ,!a j li, Itjon of ti 1fr, l,,IdI of .ths trio t anti ind,'x ill,! i,nnotfthi(.flflIu.'t be ttI Ii'r,d UVh,StU i
v,,t r,.
- ...,- j.,.. tu UNA TI N G AC TI Vt t Y (('..rporate author)
Institute of Hydraulic Research The University of Iowa
¿a. REPON 15CC UIl Y . SI T( .'tI ON
Unclassified
2h. GROUP
.1 Ni PORT TITLE -.
"A Unified View of the Law of the Wall Using Mixing-Length Theory"
4 DESCRIPTIVE NOTES (Type of report and nclusivc ¿atc) IfliR Report No. l37
S AU THOR(S) (First name, middle initial, last name)
-V.C. Patel
6 REPORT DATE
-April 1912
70. TOTAL NO. OF PAGES
4l
7h. NO. O NEFS
31
86. CONTRACTOR GRANT NO. -
--NOOOl-68-A-Ol96-OOO2
b. PROJECT NO.
e. d.
98. ORIGINATOR'S REPORT NUMBE.R(5)
iiim
Report No. 1319h. OTÑER REPORT NO(S) (Any other numbers thot may be assigned
thla report)
IO DISTRIBUTION STA'MENT
This document has been approved for public release and sale; its distribution is unlimited.
II SUPPLEMENTARY NOTES - - 12. SPÖNSORING MILITARY ACTIVITY
Naval Ship Research and Development Center
13, AØSTRACT
It is shown that, if the well-known mixing-length formula is regarded simply as a relationship between the velocity and the stress distri-butions in the wall region of a turbulent flow, then a truly univeral d-istri-bution of mixing length is sufficient to describe the experimentally òbserved
departures of the velocity distribution from the usual law of the wall as a
re-sult of severe pressure gradients and transverse surface curvature. Comparisons
have been made with a wide variety of experimental data to demonstrate the general validity of the mixing-length model in describing the flow close to a smooth wall.
An extension of the relaininarization criterion of Patel and Head, and some xperimental evidence, suggest that the thick axisyetric boundary layer on a slender cylinder placed axially in a uniform stream cannot be majntained in
a fully turbulent state for values of the Reynolds number, based on friction velocity and cylinder radius, below a certain critical value.
D
Uncias sifi ed.
'UtiIV ('1.ui t iet ton
FORM 4
i473 (BACK
.4
--KEY WORDS
-LINK A LINK U . -:
ROLE WT ROLE WT hQL f.
Thrbu.ient boundary 3 ayer s
Law of the
a1l
Mixing-length theöry Pressure gradient Transverse curvature Reiam-inarizatiOfl i-r
1Q Commanding Officer and Director
Naval Ship Research and Development Center, Attn: Code
041
(39)
Department of the NavyWashington, D.C. 20007
Attn: Code
513
(i)
2 Officer-in-Charge
Annapolis Division Naval Ship Research and
Development Center
Annapolis, Maryland
21402
Attn: Library
6 Commander
Naval Ship Systems Command Department of the Navy
Washington, C.C.
20360
Attn: Code037
Code205?
CodePMS81/525
Code03412
20 DirectorDefense Documentation Center 5010 Duke Street
Alexandria, Virginia
22314
3 Chief of Naval Research
Department of the Navy
Washington, D.C.
20360
Attn: Code
438
(2)
Code 4h (i)
1 Commanding Officer
Office of Naval Research Branch Office
495 Summer Street
Boston, Massachusetts
02210
1. Coîrnnnding Officer
Office of Naval Research Branch Office
219 S. Dearborn Street
Chicago, Illinois
60604
DISTRIBUTION LIST FOR REPORTS PREPARED UI'IDER T1 GENERAL HYDROCHANICS RESEARCH PROGRAM
1 Office of Naval Research
Resident Representative 207 West 24th Street
New York, New York 10011
Commanding Officer
Office of Naval Research Branch Office
1030
East Green StreetPasadena, California
91101
1 Commanding Officer
Office of Naval Research Branch Office
1076 Mission Street
San Francisco, California
94103
3 Commanding Officer
Office of Naval Research Branch Office
Box
39,
FPO, New York 09510i Dr. F. H. Todd
Office of Naval Research Branch Office
Box 39,
FF0, New York 095103 Conur&ander
Naval Ship Engineering Center Department of the Navy
Washington, D.C.
20360
Attn: Code 6120 Code 6136
Code
6140
Special Projects Office Department of the Navy
Washington, D.C.
20360
Attn: Dr. John Craven
Code NSP-OO].
1 Commanding Officer (Tech. Lib.)
U. S. Naval Air Development Center
i Commanding Officer and Director
Naval Applied Science Laboratory Flushing & Washington Avenues
Brooklyn, New York 11251
i Director
Naval Research Laboratory
Washington, D.C. 20390
Attn: Code 2027
i Commanding Office
Navy Uñderwater Weapons Research and Engineering Station
Newport, Rhode Island 028140
1 Commander
Boston Naval Shipyard
Boston, Massachusetts 02129
Ättn: Technical Library
1 Commander
Charleston Naval Shipyard Naval Base
Charleston, South Caròlina 291408
Attn: Code 2145b14 Technical Library
1 Commander
Long Beach Nava]. Shipyard
Long Beach, California 90802
Attn: Technical Library
i Commander
Norfolk Naval Shipyard
Portsmouth, Virginia 23709
Attn: Technical Library
i Commander
Pearl Harbor Naval Shipyard Box 1400, Fleet Post Office
San Francisco, California 96610
1 Commander
Philadelphia Naval Shipyard
Philadelphia, Pennsylvania 19112
Attnt Code 240
i Commander
Portsmouth Nava]. Shipyard
Portsmouth, New Hampshire 03801
Attn: Technical Library
-2-i Commander
Puget Souñd Navál Shipyard
Bremerton, Washington 983114
Attn: Engineering Library Code 245.6
1 NASA Scientific and Technical
Information Facility P.O. Box 33
College Park, Maryland 207140
i Library of Congress
Science and Technolor Division
Washingtoñ, D.C. 205140
i U. S. Coast Guard
1300 E Street N. W.
Washington, D.C. 20591
Attn: Division of Merchant Marine
.1 University of Bridgeport
Bridgeport, Connecticut 06602
Attn: Prof. Earl Uram
Mechanical Engr. Department
14 Naval Architecture Department
College of Engineering University of California Berkeley, California 914720 Attn: Librarian Prof. J. R. Paulling Prof. J. V. Wehausen Dr. H. A. Schade
2 California Institute of Techno1or
Pasadena, California 91109
Attn: Dr. A. J. Acosta Dr. T. Y. Wu
1 Cornell University
Graduate School of Aerospace Engr
Ithaca, New York 114850
Attn: Prof. W. R. Sears i The University of Iowa
Iowa City, Iowa 522140
Attn: Dr. Hunter Rouse
2 The State University of Iowa
Iowa Institute of Hydraulic Researòh
Iowa City, Iowa 522140
Attn: Dr. L. Landweber. (1)
Dr. J. Kennedy (i)
(J.)
14 Massachusetts Institute of Technology
Department of Naval, Architecture and Marine Engineering
Cambridge, Massachusetts 02139
Attn: Dr. A. H. Keil, Roòm.5-226 (1)
Prof. P. Mandel, Room 5-325 (1)
Prof. J. R. Kerwin, Room 5-23 (i)
Prof. M. Abkowitz (i)
U. S. Merchant Marine Academy
Kings Point, L.I., N. Y. 110214
Attn: Capt. L. S. McCready,. Head Department of Engineering
3 University of Mich.gan
Department of Naval Architecture and Marine Engineering
Ann Arbor, Michigan 148i014
Attn: Dr. R. F. Ogilvie Dr. F. Michelsen Prof. H. Benford 2 U. S. Naval Academy Annapolis, Maryland Attn: Library
Prof. Br.uäe Johnson
i U. S. Naval Postgraduate School
Monterey, California 939140
Attn: Library
1 New York University
University Heights
Bronx, New York 101453
Attn: Prof. W. J. Pierson, Jr.
2 The Pennsylvania State University
Ordnance Research Laboratory
University Park, Pennsylvania 16801
Attn: Director (i)
Dr. G. Wislicenus (i)
2 Scripps Institution of Oceanography
University of California La Jolla, California .92038 Attn: J. Pollock
M. Silverman
3 Stevens Institute of Technology
Davidson Laboratory 711 Hudson Street
Hoboken, New Jersey 07030
Attn: Dr. J. Breslin
-'3-(1)
(3)
i University of Washington
Applied Physics Laboratory 1013 N.E. 140th St±eet
Seattle, Washington 98105
Attn: Director
2 Webb Institute, of Naval Architecture Crescent Beach Road
Glen Clove, L.I., New York 115142
Attn: Prof. E. V. Lewis (i)
Prof. L. W. Ward (i)
i Worcester Polytechnic Institute
Alden Research Laboratories
Worcester, Massachusetts 01609
Attn: Director
1 Aeroj et-General Corporation 1100 W. Hoflyvale Street
Azusa, California 91702
Attn: Mr. J. Levy.
Bldg. i60,. Dept. 14223
:1 Bethlehem Steel Corporation
Central Technical Division Sparrows Point Yard
Sparrows Point,, Maryland 21219 Attn: Mr. A. D. Haff
Technical Manager Bethièhem Steel Corporation 25 Broadway
New York, New York 100014
Attn: Mr. H. de Luce Electric Boat Division
General Dynamics Corporation
Groton, Connecticut 063140
Attn: Mr. V. T. Boatwright, Jr.
1 Esso International
15 West 51st Street
New York, New York 10019
Attn: Mr. R. J. Taylor, Manager R & D, Tanker Dept.
1 Gibbs and Cox, Inc.
2]. West. Street
New Yórk, New York i0006
Attn: Technical Information Control Section
i Grumman Aircraft Engineering Corp.
Bethpage, L.I., N. Y. 11714
i Lockheed Missiles & Space Co. P.O. BOx 501t
Sunnyvale, California 9088
Attn: Dr. J. W. Cuthbert, Facility i Dept. 5_01, Bldg. l50
1 Newport News Shipbulding and
Dry Dek Company l0l Washington Avenue
Newport News, Virg-inia 23607 Attn: Technical Library Dept.
1 Oceanics, incorporated
Technical Industrial Park
Plainview, L.I., N. Y. 11803
Attn: Dr. Paul Kaplan
1 Robert Taggart, Inc.
3930 Walnut Street
Fairfax, Virginia 22030
Attn: Mr. R. Taggart
i Sperry Gyroscope Company
Great Neck, L.I., N. Y.
UO20
Attn: Mr. D. Price G-2
i Sperry-Piedmont Company
Charlottesville, Virginia 22901
Attn: Mr. T. Noble.
i Society of Ñavai Architects and.
Marine Engineers Trinity Place
New York, New York i0006
1 Sun Shipbuilding and Dry Dock Co.
Chester, Pennsylvania 18013
Attn: Mr. F. L. Pavilk Chief Naval Architect
i TRG/A Division of Control Data Corp.
535 Broad Hollow Road (Route 110)
Melville, L.I., N. Y. i17l6
3. Woods HOle Oceanographic Institute
Woods Hole, Massachusetts 025143
Attn: Reference Room
2 Commander
San Francisco Bay Naval Shipyard Vallejo, California
Attn: Tecbìiical Library, Code 130L7 Code 250
i Commandant (E)
U. S. Cöast Guard (Sta 5-2) 1300 E Street N.W.
Washington, DC. 20591
2 Hydronautics, Incorporated
Pindell School Road Howard County
Laurel, Maryland 20810
Attn: Mr. P. Eisenberg Mr. M. P. Tulin