N 62 57585

CASE FILE
I .
-L

UATIONAL ADVISORY COi9iIITEE FOR ARCNAUTICS

No. 585

ATOAL D; jURY COiMI1 ILL FOR AERONAJT1CS

1724 S1IE1, LW.,

WASwNGT0 tJ.C.

VELOCITY DISTRIBUTION IN THE BOUNDARY LAYER

OF A SUBMERED PLATE

By M. Hansen

Fro: Abhand1uren atis dern Aerodynanischen Institut

an dei' Technisohen Hochschule Aachen, No. 8, 1928

Washington
OrccT, 1i3C'
 NATIONAL ADVISORY COvlld1ITTEE FOR RONAUTICS.

TECHNICAL iiIEMORANDUM NO. 585.

VELOCITY DISTRIBUTION IN THE BOUNDARY LAYER

OF A SUBMERGED PLATE.*

By M. Hansen.

This report deals with the measurement of the velocity dis-

tribution of the air in the vicinity of a plate placed parallel

to the air flow.

I. Notation, Definitions and Theoretical Results

p, density of the air,

.L, viscosity of the air,

i,, kinematic viscosity coefficient,

distance of test point from leading edge of plate,

y, distande of test point from plate,

l, length of plate, 50 cm (19.69 in.),

b, width of plate, 38 cm (14.96 in.),

U, air velocity outside of boundary layer,

u, component of boundary layer air velocity parallel to

plate,

- 6, thickness of boundary layer,

To shearing stress on plate,

*"Die Geschwindigkeitsverteilung in der Grenzschicht an einer
eingetauchten Platte." From Abhandlungen aus dem Aerodynami-
schen Institut an der Technischen Hochschule Aachen, No. 8 1928,
pp. 31-45.
N.A.C.A. Technical Memorandum No. 585 2

Rx = J-2 characteristic coefficients.

U

-u
1)

L. Prandtl and H. Blasius succeeded in determining theoret-

ically the velocity distribution for the case of a laminar flow.*

Their results may be briefly summarized as follows.

Let x and y be the coordinates, respectively, parallel

and perpendicular to the plate, and U the velocity of the un-

disturbed air flow. Tae velocity curves are similar for differ-

ent values of x, the distance at which the same velocity oc-

curs, increasing with the square root of x. The velocity

obeys the law:

u=Uf( ) (1)

For comparison with experimental results, it is convenient

to introduce a boundary_layer thi•ckness, although no exact

value can be assigned to it, because the velocity u changes

asyrnptoticallr to U. For example, y may reDresent the boun-

dary-layer thickness for which the function £ differs from

unity by a certain fraction (say l). If the function f in

equation () is approximated by a parabola, which changes asymp-

totically into the line u = U, then
*H. Blasius, Zeitschrift für Mathematik und Physik, Vol. 56, No.
1, 1908.
N.A.C.A. Technical Memorandum No. 585 3

y = 5.5 (2)

is then obtained for the transition point. This value of y

will be taken for comparison with the experimental results.

The surface friction is represented by T0

= U (--.
According to the Blasius theory, it is

1
0.332 (3)

Various observations led Prandtl to surmise that the lami-

nar flow changes, with increasing Reynolds Number, to a turbu-

lent flow, similar to the flow through a tube. Since the ele-

mentary law of turbulent friction is not yet known, the veloc-

ity distriition cannot be theoretically determined for this

case. Nevertheless, if the velocity distribution experimentally

determined for tubes and the likewise experimentally determined

law of resistance be transferred to the case of the boundary

layer, the thickness of the boundary layer can be determined

with the aid of the momentum theorem even for this case. This

calculatlon was made Dy Prandtl and Von Karman.* The thickness

of the boundary layer was found to be

8 = 0.370 Rx1 (4)

The shearing stress was

1/4
= 0.0225 .p U2

*Zeitsch:ift fir angewandte Mathematik und iecharik, Vol. I,

1321, p. 233-298.
N.A.C.A. Technical Memorandum N. 585 4

The fundamental velocity distribution for

O<y<ô is u=U(- andfor y>5 is uU.

II. Researches of Burgers and Zjnon

In 1923-1925, J. M. Burgers and B. G. van der Hegge Zijnen,

in the Aerodynoica1 Institute of the Deift Technical High

School, investigated the velocity distribution in the boundary

layer for both kinds of flow and particularly in the region of

transition from the laminar to the turbulent state.* In the

laminar region they found very good agreement between the theo-
retici nd exp eriment results, as regards the thickness of

the boundary layer. The discrepancies were greater, however,

as regards the shearing stresses, whether determined frbm the

velocity gradient on the plate, or from the momentum theorem

with the help o± the formula

= d
p 1 u (U - u) d y (6)

Thepoint of transition from the laminar to the turbulent region

could. be quite accurately determined by plotting the measured

boundary-layer thiekness as a function of the coordinate x.

The resulting curve has a sharp bend at a certain value of Rx.

This value of R may he designated as the ucritical character-
istic (K'mnzl) On the other hand., a Ucharacteristicit can -
be anAd Lo th buria.:,r layer_itself, by expressing_it in
of th vc1octy Distrhution in the Boundary Lay-
er Along a Plane Surface,u Deift, 1924-.
N. A. C • A • Technical Memor anthim No. 565 5

the form R 5 = TJS/i,. R, and R are theoretically connected by

the expression R = 5.5 which was also approximately

confirmed by experiment. The measurements showed that R var-

ies between 1650 and 3500 and is therefore considerably greater

than in tubes. In the turbulent region the measurements showed

that the velocity varied. anproximately as /r. The curve of

the boundary-layer thickness does not correspond to the formula

5 = 0.37 X Rx_h/5, because in its derivation, it was assumed

that the turbulent flow begins at x = o, though, in reality,

the laminar, flow continues up to the critical value of R.

Zijnen extended the formula by introducing a parameter x 0 and

putti±ig
/
5 = 0.37 (x - x 0 ) ( 7)

in which x 0 de p ends uarticularly on the shape of the leading

edge of the plate (sharp or rounded). The experimental results

can be represented very well by this modified formula.

III. Cbect 0±' Recent Experiments - Laboratory Equipment

My own experiments were intended, on 'the one hand, to ac-

count for the discrepancies 'in the above-mentioned experiments

and, on the other hand, to determine the velocity distribution

on rough plates. The means for these experiments were placed

at my disposal by the otgerneinschaft der Deutschen Wissen-

schaft (Emergency Fund for the Promotion of German Science).

The experiments were nerformed in the small wind tunnel

N.A.O.A. Technical Memorandum No. 585 6

belonging to the Aerodynamic Institute of the Aachen Technical

High School (Figs. i-2). The diameter of the entrance cone is
30 cm (11.8 in.). The length of the free jet between the entrance

and exit cones is about 2.5 m (8.2 ft. ). The measurements were

riade in this free jet where the static pressure was constant,

which was essential for the method of measurement used.

The wind velocity U was accurately controlled by en-larg-

ing or diirishing the exit opening a by the use of plates.
It was not feasible to regulate the wind velocity by means of

a slide valve at a point b of the enclosed portion of the wind

tunnel, because this disturbed the air flow too much. This

method Was used, however, to produce artificial turbulence.

Figure 3 shows the p lates used in the tests, to•ether with their

degrees of roughness. The plates were successively mounted in

a wooden frame on the test stand. Underneath the plate and

mounted on a block of concrete, there was a lathe bed which suo-

ported the measuring instrument. This arrangement made it pos-

sible to move the measuring instrument either parallel or per-

pendicular to the plate. x was read on a scale parallel to
the plate. The perpendicular distance y from the surface of
the plate was measured to within 0.01 mm (.0004 in.) by a slide

gauge mounted on the same support. The p ressure Was determined

by means of two alcohol pressure gauges. The temperature and

pressure Were read frequently during the tests.

Though the Delft measurements were made with a hot-wire

N.A.C.A. TechnicaJ. Memorandum No. 585

instrument, I decided to use small Pitot tubes, which are cam-

paratively easy to make by drawing out and bending the ends of

very small glass tubes. The smallest tube drawn by me had an

outside dieneter of 0.135 mm (0.0053 in.). They worked so slow-

ly, however, that the readings would have taken too much time.

Tube No. 13, with an outside diameter of 0,35 mm (0.0138

in.) and. an inside diameter of 0.21 mm (00083 in.), was the

most satisfactory and was used for most of the measurements.

Figure 4 shows the calibratia curves of this tube. The re-

suits obtained with it were then compared with those obtained

with an ordinary Pitot tube. If discrepancies were found, they

could be mostly eliminated by grinding the open end of the tube.

The measurements could be made only in a velocity region of 4

to 38 rn/s (13 to 118 ft./sec.), since the tubes vibrated

strongly at higher speeds.

In order to determine the distance between a tube and the

plate, I had to find the position at which the tube touched the

plate. This position was easily determined by observing when

the tiD of the tube and its image came together (Fig. 5).

A sufficiently accurate adjustment of the plate was effect-

ed by mounting it perpendicular to the plane of the Pitot tube

with the aid of a carpenter's square. Aiiy more accurate adjust-

ment did. not appear necessary, since measurements on both sides

of the plate showed that the results wore hardly affected by a

slight obliqueness of the plate.

N.A.0.A. Technical Memorandum No. 585 8

IV. Results

a) Lwninar Flow

Figure 6 shows the results for the thin dural plate No. 1.

It is seen that the measured velocity distribution agrees very

well with the theoretically calculated velocity. The velocity

ratio is plotted as a function of the quantity and
shows that, within the accuracy of the measurements, the sec-
tions very well obey the law of similarity.

In Figure 7 the thickness of the boundary layer is plotted

as a function of the quantity d likewise agrees well
with the theory.

For the thicker plate No. 4 (Figs. 8-9), the first sections,

with small values of x, show systematic deviations which dis-

appear for thicker sections.

Figure 10 gives a verification of the impulse balance by

comparing the shearing stress, as determined from the velocity

gradient, with the value obtained from the momentum by means of

formula (6). As shown in the figure, the two values obtained

from my ex p eriments agree very well, while there is a discrep-

ancy in the values obtained from the measurements by Burgers.

In my experiments there are differences at the leading edge of

the plate. Hence it may be inferred that the slight discrepan-

cies in my measurements are due -to the influence of the leading

edge, while there must be some other reason for the discrepan-
N.A.C.A. Technical Memoranthun No. 585 9

des in the measurements made by Burgers.

In order to ascertain the effect of the finite thickness

of the plate, especially the difference between the rounded and

sharpened plates, I tried first to calculate the p otential flow

and then the curve of the boundary-layer thickness for such

plates. The potential flow was calculated by the so-called in-

verse method, by assuming source distributions which, when super-

posed on the parallel flow, yielded streamline shapes similar

to the plates used by me. The rounded section can be generated

by pointlike sources, while distributed sources must al w ays be

adopted if the sharp section is to be substituted.

Figure 11 shows the velocity curve calculated from the

source distribution. In addition to the velocity curve, its

first and second differential quotients are given according to

the length of the arc, because these quantities are necessary

to determine the thickness of the boundary layer. This thick-

ness was calculated by A. Pohihausents method (Abh. Acrodyn.

Inst. Aachen, No. i). Figure 12 shows the course of the quantity

z = 5 2 /i, for both the cases represented in Figure 11. With the
rounded plate (pointlike source) the thickness of the boundary

layer increases very rapidly arid separation soon occurs. With

the sharp-edged plate (linear source distribution), on the other

hand, the deviation from the Blasius case (infinitely thin plate)

is not very great. This result is in accord with the observed

fact that, even at a short distance from the leadtng edge of the
N.A.C.A. Technical Memorandum No. 585 10

rounded plate, the boundary layer becomes turbulent, while with

the sharp-edged plate the laminar flow persists much longer.

As regards the experiments o± Burgers and Zijnen, this cal-

culation could not ex plain the great discrepancy between the

theoretical and. experimental results as found, e.g. , in the

velocity gradient on the wall of the tunnel. The calculation

might be utilized at most to ex p lain the S-shaped velocity

curve found by the above-mentioned investigators. The boundary-

l ayer theory, in fact, gives such curves, provided a pressure

increase or velocity decrease (u = -< 0 in Fig. 11) is

present.

The reason why Burgers and Zijnen found a considerable de-

viation from the theory of Blasius, while my measurements show

a much better agreement, is due, in my opinion, to the fact that

the Delft experiments were performed. in a closed. wind tunnel,

while mine were made in a free air stream. In a closed. wind

tunnel, a decrease in pressure or an increase in velocity, cor-

responding to the frictional resistance of the tunnel walls

and of the plate itself, must take place outside the boundary

l ayer . The air flowing between the tunnel wall and the test

plate is in a similar state to that in the first part of a tube.

The resulting deviations from the Blasius theory can be approx-

imately calculated., since Zijnen himself established. the increase

in velocIty along the test plate, though for another purpose

(Thesis, Report No. 6, Delft, 1924, pp. 39-42). He expressed. the

N.A.C.A. Technical Memorandum No. 585 11

mean velocity increment by U. The factor denotes

the velocity increment for the entire length 7. s a fraction

of the velocity U. For example, Zijnen found that B = about

0.06 for U = 8 rn/s (26 ft./sec.). I have investigated the

effect of the velocity increment on the thickness of the bounda-

ry layer according to K. Pohihausen' s method and found smaller

values for the thickness than the Blasius theory would lead one

to expect. In fact, the Deift measurements show a deviation

of the some nature. The velocity gradients on the plate can be

compared. for the two cases. For this purpose, in the calcula-

tion of the boundary-layer thickness, I introduced Zijnen's

experimentally found values for 5 and the above value for

U' = into Pohlhausen's differential equation for the boun-

dary layer. The result is shown in Figure 13. The plain lines

represent the velocity gradient according to the Blasius method,

while the dash lines represent the same according to my own

method, with allowance for the velocity gradient in the tunnel.

The agreement is very good.up to the last value, which corre-

sponds to a velocity curve at the distance x= 62.5 cm (24.6

in.) from the leading edge of the plate. With this velocity

curve the transition to the turbulent condition has evidently

taken place already, so that agreement can no longer be expected.

The abov&ntioned calculation also shows that the boundary-

layer formation is very sensitive to relatively slight local

variations in the velocity of the air flow along the plate so

N.A.C.A. Technical Memorandum No. 585 12

that such measurements must really be linked with a very, accurate

control of the static pressure. Thus, even an undulating or
wavy shape of the plate may affect the formation of the laminar
boundary layer.

b) Transition from Laminar to Turbulent Flow

assigned a critical characteristic value Rô = 5.5 f,

We have already remarked that the boundary layer can be
at
which the transition from the laminar to the turbulent condition
takes place. I have attempted to determine R by two methods.
The transition is evidnced, on the one hand, by a sudden growth
of the boundary layer and, on the other hand, by the development
of a shearing stress. Zijnen determined R by plotting the
boundary-layer thickness. In a similar manner (Fig. 14) I first
plotted the nondimensional quantity and then also the
quantity - , -1---_---_ as a function of the characteristic
U312 (, x1'2
value. Both quantities must be constant for a purely laminar
boundary layer.
Despite the scattering of the test poiits, a bend is quite
clearly indicated at R = about 3100. This result agrees with
the Delft measurements. If this value of R is compared with
the critical characteristic of tubes or rings, it appears very
high. If it is considered, for examp le, that the thickness of
the boundary layer should logically be taken as the hydraulicS
radius in our case, R 5 would have a value about six times as
N.A.C.A. Technical Memoandum No. 585 13

large as the corresponding value for tubes. It should be noted,

however, that, even for tubes and rings, the value of the crit-

ical characteristic can be greatly increased by a steady inflow.

In the case of the sharp-edged. plate, we have only a compara-

tively steady flow. If the flow is disturbed, e.g., by the sep-

aration of the boundary layer on a rounded leading edge, the

transition takes place at a lower characteristic value. Also

when the air strear itself, for example, due to strong throt-

tling, was more turbulent than under normal conditions, I found

that the reversal occurred sooner. The lower limit of R5 has

not yet been established. Figures 15-16 give examples of ve-

locity distribution in the transition region.

c-i) Turbulent Flow over a Smooth Plate

Since the preliminary experiments showed. that the transi-

tion to the turbulent flow is accelerated by rounding the lead-

ing edge of the plate, a rounded strip of wood was applied to

the front edge of the glass plate. With this arrangement a tur-

bulent flow had already developed at x 10 cm (3.94 in.). The

tests covered avelocity range of 16 to 36 rn/s (52-118 ft./se.c.).

Between x = 10 cm (3.94 in.) and 50 cm (19.69 in.), measure-

ments were made on the plate section at intervals of 5 cm (1.97

in.). The corresponding shearing stresses on the wall were de-

termined from formula (6). No direct determination of the shear-

ing stresses on the wall could be made with the help of the ye-
N.A.C.A. Technical Memorandum No. 585 14

locity gradient, because the measurement with the Pitot tube

did not give sufficiently accuiate results close to the wall.

The results of the measurements are shown in Figure 17, both the

distance from the wall and the velocity being plotted logarith.

mically. It is obvious that, disregarding the immediate -oroxim--

ity of the wall (y < 1 mm), the velocity distribution can be

quite well approximated by a potential law.

In Figure 18, the velocity ratio u/U is likewise loga-

rithmically represented a.s a function of the ratio y /5 . The

values of 5 were derived from the points of intersection of

the sloping lines in 'igure 17 with the horizontal lines in u/U.

In other words, the lines corresponding to the different values

of x were brought into coincidence by shifting horizontally.

Figure 18 shows that the slope of the lines, i.e., the exponnt

of the potential expression depends somewhat on the velocity.

Ifweput
n
= c(" (c = proportionality factor) (8)
U \ 5,!

the exponent n then increases with the velocity. For example:

n = 0.183 for U = 20 m (65.6 ft.);

n = 0.196 U 28 (91.9 U );

n = 0.198 U
U = 36 '(118.0 ).

The exponents obtained for the smooth plate by this method are
all higher than the exponent n = 1/7 0.143 derived by Prandtl
and VOfl Karman from the law of resistance. This is due in'.part

to the fact that the test points y < 1 mm were disregarded in

N.A.C.A. Technical Memorandum No. 585 15

in the evaluation.
If the velocity distribution is approximated by a poten-

tial law, the integral in formula (6) can then be evaluated.

If the boundary-layer thickness is introduced in the above-

mentioned sense, it is found that

a p U2 (9)

in which the quantity a is connected with the exponent n by

the expression
a- 1 1
n+1 2n+l

I have usually determined the shearing stress with the aid of

formula (9).
If it is assumed that, between the shearing stress, the

distance from the tunnel wall and the velocity, there exists a

definite relation governed only by the physical constants, den-

sity and viscosity, this relation can then be expressed in the

form
= ±, (10)
pU 2 \P .1.

This equation is the usual mathematical expression of the Prandtl

dimensional consideration, which leads to the previously men-

tioned. formula for the turbulent velocity distribution. The

quotity uy/, can be regarded as a kind, of characteristic coef-

ficient for the point y.. If the velocity distribution is

approximated by a potential formula, equation (10) is changed to

N.A.C.A. Technical MernorandumNo. 585 16

____ =
(11)
Pu2 \1 I
The relation between the ex p onents n and. m is evidently ex-

pressed by the formula m = - is a coefficient

which cji be determined from the measurements.

Theodor von KaTh'ian, on the basis of the Blasius law of re-

sistance, arrived at the values: = 0.0225, m = 0.25,

n = 0.143. 1 have comp ared all the data obtained for the smooth

plate with the von Karman theory (Fig. 19). It is seen that

the ?OifltS for small values of uy/i, , in the vicinity of

u y/, 20O0, scatter about the theoretical curve, which a Sys-.
tematic deviation occurs for greater values. Geater values

than a = 0.143 are likewise obtained by disregarding the val-

ues in the immediate vicinity of the tunnel wall, as already

mentioned.

c-2) Turbulent Flow along Rough Plates

Af ter the measurements on the smooth plate, I determined

the velocity distribution on an undulating plate and on two

rough plates. It was found that the interpolation with the po-

tential formula gave good results even in this case. The data

were worked out in the same manner as for the smooth plate

(Figs. 20-22).

Furthermore, the resistance and the shearing stress were

determined from the impulse integral. For comparison I deter-

mined the integral first on the basis of the test points, but
N.A.C.A. Technical Memorandum No. 585 17

then also with the aid. of formula (9). The deviations were

greater for the rougher plate, apparently because the interpola-

tion with the aid of the potential law does not hold good up

to values of u = U or y = 5, but the velocity-distribution

curve is sharper. In this case the exact integral was naturally

resorted. to for the calculation of the shearing stress.

The next step was to determine to what extent the shearing

stress can be represented by formula (ii). Figures 23 and 24

show the result. The quantity To/pu 2 is plotted as the ordi-

nate and uy/i, as the abscissa, both in the logarithmic scale.

It is obvious that the absolute value of the exponent m in-

creases, oi the one hand, with increasing roughness and, on

the other hand, with increasing velocity.

For all the data, the coefficient varies between 0.03

and 0.6, the exponent m between -0.3 and -0.5 and correspond-

ingly the exponent n between 0.186 and 0.325. I have now made

the noteworthy observation that the values of , as a function

of n in the logarithmic scale, lie on a straight line.

In my opinion, the fundaenta1 importance of the results

resides in the fact that there is apparently a definite rela-

tion between the exponent of the velocity distribution and the

coefficient . The latter corresponds to the coefficient which

is assumed to be constant in the theory for the ideal case of

the smooth plate. For rough rjlates, however, both the quantity

and the exponent m are no longer constants, but are affected

N.A.C.A. Technical Men'iorandum No. 585 18

by the velocity as well as by th roughness. It may be surmised.

that both depend on the nondimensionsJ. quantity uK/ p , in which

K is the mean roughne.s. The general resistance formula

would then have the form

m
1) /
To = i1 _i (12)
Pu2 \1)1

It is still uncertain whether this equation represents only an

interpolation formula or a definite law.

Translation by
National Advisory Committee
for Aeronautics.
1.LC.A.Technica1 Memoranthim No.585 Pigs. 1,2,5

Pig. 1

PIg. 2
IT.A.C.A. Tochnical iioi:orancium ITo. .585 Figs.3,4

No. P_•____ T1LI.d

1 heot dira1 2 5 r Sharp
(smooth) ____________
2 Glass (smooth) 9.0 mm r-
_iiii_ Rounded
3 U II
4.5 mm ____ Sharp
Turbulence
4 - 6.0 1E S1'iarp
1tr1_
- 5 Gloss ci1a-
- 2Jii Rounded.

I)__ -Sharp

•. Q3ess :____E Rounded.

Longth of plates, I = 50 cm
Sriiooth
Tindulat ing
Roughness I
Roughness II

1 OC
liiiI 111 lfi
Fig. 3

--
1 - - -- --
a, Result with large
pitot tube.

H
8C b, Difference in results.
0
0
H

c,Result with tube ITo, 13

0
before grinding and after
grinding the open end.

C)
F-i
* -- -f-..4- d, The rcsult with large
and small tubes agree.
LI)
w
r')
F-i
e, To difference

L)
-- -- _ L -- L

t , mm,
Fig. 4 Calibration cunre of
tube No. 13
N.A.C.A. Technical Morandan No. 585 Pigs. 6,8,918,19
__-

T , frt'-
7 7,

I-

___ 0

/ _J.X-3cm
--___
--.-

- -'--
. • -
0
' I
i.-

8/csius
---f

I
IJ-8rn,'s

-
-'3

7
7?,5
75,0

-
--- L5_.

@_tty E
I

________-
, ,7 " a
3'
Pig. 6
Pig. 8

a
fJO2&36m/3
L--T-

I.t _L-

i-LJ
01

-'h--
H-

'1

-4

/1
____________
P ______- _________ _____
'V
7
P/o/e2

1E Srnoêfh
ffIe
- 7Z. - 0,798
U-iOm/s.
ii -0,796 " U-28 -
• i

7rbq:
qs,1 42
Tur'bIent j
0,3
18 45 4f a? ?6 491
NIA.O.A. Tocimical ::omorand.uni No. 585 Figs. 7,10

-( -1
__ I It
.1 .5 1.0

Fig. 7 VU--

40
- pJu(L1u)d ry, k€f /m Iansn

rO')
7( H0
•- 30
b
H biD

C,
- ',)

10 10°

p(U-)dy Burer
_L I i -- I
0 10 20 30
Fig. 10 x,cm
 N.A.O.A. Technical Memorandum No. 585 Figs. 11,12

1000
800 4000
Ti, cm/s TJjU1 andi Tilt fo t ie 3000
p(in: -Uike source.
400 'ii 2000 S

it.I.I
77Z?
0
-1000
TJ tt , 1/s cm

-5000

800
A
U, cm/s
500 U,U and U as a
function of x.
Ut, s - source
0
1J', 1/s cm
-500
0 2 '4 6 8
Fig. 11 x= arc 1eiith in ci
1
Pbint-1i4 sourc.
(separatlion)
Solution
- / Ut0 for source
C2 I __ 'V
arrangement.

--_____ -
• _ _
I

I I I
1 2 3
1.0 x=arc len2'tn in nri

U'0, __--

.5

N I-.

0 5 10 15
Fig. 12 x=arc ngth in cm
N.A.O.A.,Technical ivIemoranure No. 585 . Figs. 13,14

0 - __________-
x=10 cm x20 cm
Burgers Bur:.ers
8 g,' Burgers.

x=30 cm x=40 cm x=82.5 cm

Fip'. 13
From boundary-
.16 l a yer thick-

ness 6
14
-
i -
12 _ _

10

+ # -- - -
_+++++ +r' +.
- - .-- - - From shearing
I (0 4 6tr3s

0
-
---
C\7
-..--
0

1
•r-I
2 I ___
__
a, 10001
0± -
V)

I
p.J

a b
2 oJI C
4
R
6 8
in lO'r
10 12

Fig. 14
N.A.Ci.. Technical Momorandun No. 585 Figs. 15,16

Ca I CD
CC

C)
CD C) 0
c H xx 4-' C\
A
U H IL
C'- - I::
If)
H
)' )'__
0
0
LI ç
C'
b
L1 (O
LI
'Cc

x 0H x
C'
C'
X
'C

o x 'C
C'
'C
H C'----%
0
H
L0-

1:: H
CL

(\:l 0'-
LI
LU 0IC) 0 0 10 0 LI) 0
H H C') H H
51w 'U S/ILl U

CD (0
tO H
CO
x C)
4-' (1)
cC3 H
H CD
(1 \ 4-, ('C?
"C' ct

C'
Kk H
A k'X
C

\ \ )1
C',1

\ C' xx-

x \

C' 'C'

C'
_____ _____

-_____
I-

___ 00
I "C

'—'C

t_i_iiIt[ liii 4th liii

LU 10

s/rn 'U s/rn C

 .A.C.A. Tcthuic&1 osrt Jo. 586 Pigs. 17,20,21,2223,24
-
_____________
!!I!1
- 7

Jo
4?
H1ITf
4'

75

10
- --
_____ _____

as

as

0,7

46

45
Pig. 20
0,1,1
u=f(y) - - - - -

__ = -
0,6

as

a_.uiiii
75 42 4.5 4 (5 lö 47 44&7

IC

Fig. 21"' --

Pig. 22

a = smooth
b=wy
a = rougbfles2 I
d = rotghxese II

"-
A' -ROIQIt---
— b-- ------ - -
/ait-2Yrn/s nv'-4-
V5
6- 11 '

_______
---i
- -
we/hg: 16m/s. in
-- z" " "
'.
/frVn
2
.
7 2 3
I
4'
Ii
.5 70
I
20 .7 4'OSO 700 2) i0044
Pig. 23