Professional Documents
Culture Documents
AND R O U G H O P E N CHANNEL F L O W
By M. Salih Kirkgoz 1
Downloaded from ascelibrary.org by Indian Institute of Technology Guwahati on 02/25/20. Copyright ASCE. For personal use only; all rights reserved.
INTRODUCTION
REVIEW
Fully developed turbulent shear layers consist of inner and outer regions
(Cebeci and Smith 1974). In the inner region the mean velocities are gen-
erally controlled by the wall shear stress, wall roughness, distance from the
wall, density, and viscosity of the fluid.
The inner region consists of two distinct parts, namely: the viscous sub-
layer and the fully turbulent part. Experiments by Klebanoff (1954) showed
that the total shear stress at some small distance from the wall is constant
and equals the wall shear T0. Therefore the integration of Newton's law of
viscosity (T = jx du/dz with T = T 0 ), gives the velocity distribution in the
viscous sublayer as
U UifX
- =— (D
«* v
where u* (= V T 0 / P ) = the shear velocity; p = the density; v = the kinematic
viscosity of the fluid; and z = the distance from the solid boundary.
In the fully turbulent part of the inner region, the logarithmic velocity
distribution equation of von Karman-Prandtl (von Karman 1930; Prandtl 1932)
is universally recognized as the logarithmic portion of the "law of the wall."
The equation takes the following form:
U U*Z
— = A In — + B (2)
W* V
in which A = 1/K (K' is the universal von Karman constant, which is in-
dependent of the nature of the wall, whether smooth or rough); and B = a
constant whose value depends on the nature of the wall surface (Schlichting
1968). The constants of Eq. 2 were found to be A = 2.5 and B = 5.5 from
Nikuradse's (1932) experiments for hydraulically "smooth" pipe flow. Keu-
legan (1938) assumed that in the plane normal to the boundary of any open
channel shape, the mean velocity distribution for fully developed turbulent
flow could be expressed by Eq. 2 with Nikuradse's values of constants.
Subsequently, investigations were made to determine the velocity distribu-
tion in the inner turbulent region of boundary layers on a flat wall with zero
pressure gradient. Several authors found through their own experiments or
adopted different values for the constants A and B in Eq. 2—for example,
Klebanoff (1954: 2.44, 4.9), Townsend (1956: 2.44, 7), Huffman and Brad-
shaw (1972: 2.44, 5), and more recently Steffier et al. (1985: 2.5, 5.5), and
Nezu and Rodi (1986: 2.43, 5.29). As may be seen from these findings, A
has a range of variation between 2.43 and 2.5, and B between 4.9 and 7.
It is seen that the constants for "plane" boundaries do not seem to differ
much from those found for channels of circular cross section (Nikuradse
1544
where II = the "profile" parameter (which was given as 0.55 with K = 0.4
and B = 5.1), and w(z/8) is called the "law of the wake" and is given by
2 sin2(-n-z/28) for zero pressure gradient. Eq. 4, in fact, represents the com-
bined effect of velocity control mechanisms that exist in the inner and outer
regions. When Eq. 4 is written at the edge of the boundary layer, and Eq.
4 is then subtracted from it, the following velocity-defect distribution, due
to Coles, can be obtained
um - u i z n ,/ITA
= — In - + - 2 cos2 - (5)
M* K 8 K \28/
Rough Walls
The velocity distribution on rough surfaces is affected by the grading,
shape, and spacing of the surface's roughness elements. In dealing with the
velocity distribution in rough-boundary flows, the most common practice has
been to use Nikuradse's (1933) expression for fully rough turbulent flow in
pipes
Rotta (1962) applied the law of the wall to the inner region of the bound-
ary layer on a rough surface by assuming that the action of the roughness
can be interpreted as being equivalent to a velocity reduction across the vis-
cous sublayer. He gave the velocity distribution as
u 1 udz + Az) AM
— = - In — +B (7)
K* K V U*.
In Eq. 7, AM represents the velocity jump across the viscous sublayer cor-
responding to the Az reference shift from the top of average roughness height,
which gives the location of the hypothetical bed level. By using van Driest's
(1956) expression, Rotta gave the continuous representation of the velocity
distributions in the inner region for different roughness conditions.
The formulas given are the most commonly used ones to represent the
mean velocity distributions in "smooth" and "rough" bedded turbulent flow.
It is seen, however, that the constants appearing in these formulas have a
wide range of values found by various investigators. The purpose of this
study is to compare the measured turbulent velocity profiles for "smooth"
and "rough" open channel flow with those obtained using the "law of the
wall" and the velocity-defect distributions, which are given in this section.
EXPERIMENTS
1546
CN I N ON
CN en
en
^H • ' CN
-fl-
cn
i u o
60 g o 3 60
•«oo
l£.s 5eo MH
<5 3
co - r j 00
F S
o
CN
IO vi o o •n i n o O O o o
r- en 0 0 0 0 *o cn T f en en T i - • *
o o O en O o
r- 0 0 en l ~ - en eo
O
NO
O
f-
o
e'-
o o O en en o en en en O en
eo r- eo en en en eo en • f l - I— en
•^ > <n > CO
X 3
£•§ ° >< %£ 'c?
e § m <T\ vo IN •n o\ o \r> 0 0 en en m en r- en r~- en en Ti- o en O in en en ON I N ON en r— O eo
O ,3 _ CJ Q C O ° a a EPS oo • * (N m I-~ o \u r~ r^ cn ONo o en CN ON
o • f l - •fl- ON en CN
r~- • f l - eo en o en •en
CN
o •fl- CN
en en - H
x a CN cn cn ^r en r- en en en cn f l - r- en •fl- r -
i
CO ^ 6 u0
en en 0 0 eo en
vn ON o\ r- e n
IN
IN oo r-
•fl-
o
ON
IN IN
en en
o CN
en
NO
CN
ON
eo
a^
I
t i . CO o CN cn en t ~ o ^^ - H CN • f l - o o ^ H CN • f l -
• " CO
- H ^H
^ o • *
•* ON ^-i
^^ oCN •fl- eo -* CN •fl- eo
O 5 X <o X O-H X «" O o o o o o O o o o O o o o o o o o o o o O O o O o o o o o
wi -a co c"^;
•*-• o X Xco "is
""
X .2
T 3 CO * •? g _ >o o 00 o Vf3 D eo r- m en en o IN •fl- 0 0 o 00 o o en en en en • f l - eo en o 1— en o o
• 3 •—
CO CO 2 a °X o O) \ uo ^J-
--< en en oo oo VO O CN en r~ r- en
^< cn CN O cn o r- en en r- 0 0 0 0 oe eo 0 0 o
• M ^ H
en ON O
X
£
3
CO
6 0 CO
^ C
C
3 c
CO CO
T). 5? CO
•fi co"
^ x
cSx
. •*
I •S S ?
-
r^
CN
t~~
00
t-~
o
"*
CN
Tt
o
CN
r~-
o
cn
IN
00
\D
O r-~
en en
en •cn cn CN
IN
l-~
*
in
00
o
oo
O
00
o
IN
CN
en r-
CN CN
en • f l -
NO
IN
en CN CN CN CN en cn en CN cn cn T f
o
o
00
o
o
IN
o
IN
o en
ON
00
r-
00
l—
IN
O
IN
00
00
1—
en
r—
•fl-
IN
r-
en cn T f
en t— en
00 c-- r-
CO •a " a
CJ
a '—'
o 0 0 T i - 0 0 en oo I N ^ 0 0 • f l - en NO en o ON • f l - en o o
CO ?
XI r— o T t V) 0 0 0 0 rt • * 00 en !-> CN NO
O m « j *r> r- r- o o as IN m M- en M- o en O N o 0 0 en en en ON O N ON 1— r- e n en r— eo -H
S CO O l -_ ^i c •o— • • o o »^ r- r- r-
* • *
o r- r- o ON eo ON •fl1 I— o r- t - - - f l - o T f NO en ON 0 0 e n r- - H en CJN
m
CJ O CN en cn cn cn cn NO o CN en e'- CN CN CN en r~- oe CN CN en 0 0 0 0 CM CN en eo
£2 .Co o < •"* in ON
CN
00 CN
CO
3 T3.2 M
. 2 -H" J" cn "
number
ex 5
to O co
Test
X > CO
- cn -ch m eo r~ o - en •* en - en "fl- en - en en - en
(2)
CN 00 ON CN CN CN CN •fl- CN •fl- en
0) K
go* -
a u -
60
X CO ft o O O C*H
O. o
co J 5 CO T3 a
O en en en en en
B *-*to CJ co n CN (N CN CN CN •fl- •fl- •fl- •fl- -fl-
T3
CO
L,
3 g< o o u
is a
§
0 Q . -t-
CO . > > ' — p
J3 J3 X I X I
01 01 01 0/
3 3 3 3
XI XI XI XI XI X! XI XI X! XI
o i : 01 01 01 01 01 01 01 01 0 0
3 3 3 3 3 3 3 3 3 3
XI
00
3
co o
CO . 3 j 5
CO
t- VI(- H H H H f= h h H h 0p0 O f) C) O O C) f ) o f) f) O o f) f) o O n O O O
3 « 00 CO on CO to cn oo 0 0
t/3 00
at OS CA OS a! o; OS OJ OS OS OS OS OS OS OS os OJ OS OS OS
co co T 2
Downloaded from ascelibrary.org by Indian Institute of Technology Guwahati on 02/25/20. Copyright ASCE. For personal use only; all rights reserved.
cally), h is the water depth, b/h is the flow aspect ratio, F (= vVgh) is
the Froude number, V is the average velocity of flow, R (= 4VR/v) is the
Reynolds number, and R is the hydraulic radius. The tests shown in Table
Downloaded from ascelibrary.org by Indian Institute of Technology Guwahati on 02/25/20. Copyright ASCE. For personal use only; all rights reserved.
The results of the tests described in Table 1 are presented separately for
"smooth" and "rough" surfaces. A number of symbols used for the flow,
and the present assumption for the hypothetical bed in uniform roughness,
along with Rotta's model (Rotta 1962), are shown in Fig. 2. Fig. 3 gives
the measured velocity distributions for "smooth," and the four different "rough"
surfaces. It is apparent from this figure that the nondimensional velocity
distributions are consistently, albeit gradually, diverging from near unifor-
mity as the average uniform roughness height increases.
V u
v-
The calculated values of the shear velocities using Eq. 8 are given in Col.
(8)
10 of Table 1. These values seem to compare well with those given by Nezu
and Rodi 1986, who found values between 4.4 and 39.9 mm/s for the sim-
ilar range of Reynolds numbers. It is clearly seen from the velocity distri-
bution of test 1 in Fig. 4 that the viscous sublayer is quite large for R =
5,220, indicating a (laminar-turbulent) transitional flow regime. A somewhat
similar feature may also be detected from test 2 in the same figure. The
thickness of the viscous sublayer tends to decrease with increasing Reynolds
number, from about 2 mm down to approximately 0.5 mm.
"Law-of-the-Wall" Distribution
With known values of shear velocities, the "law-of-the-wall" distributions
are calculated and given in Fig. 5. Disregarding the data for test 2 (R =
8,770), the overall data in the turbulent part of inner region are best rep-
resented by
1548
(<•) (b)
U UifZ
— = 2.44 In — + 5.5 (9)
w# v
In Eq. 9, the value 2.44 in the first term confirms that in these tests the von
Karman universal constant K has a value of 0 . 4 1 . From Figs. 5 and 6 it may
be suggested that for the range of Reynolds number conditions tested Eq. 9
is valid in the range 50-80 ^ u*z/v :£ 200-600. Fig. 6 shows only two
test data that correspond to the velocity distributions of flows with "low"
and "high" Reynolds numbers. The value of u*z/v marking the measured
velocity distribution from Eq. 9 at lower and upper limits depends on the
Reynolds number in that it increases with increasing Reynolds number. Hinze
(1975) suggested u*z/v = 500 to 1,000 for the upper bound. The lower and
upper limits for the fully turbulent inner region correspond to the following
proportions of the boundary layer thickness: 0.14-0.05 < z/8 < 0.6-0.5.
These values are higher than that which is normally quoted in the literature
(Cebeci and Smith 1974).
In the lower part of Fig. 5, the velocity distributions seem to conform
well with Eq. 1, showing a viscous sublayer for u*z/v ^ 5-12, which again
increases with an increase of Reynolds number. These values correspond to
the viscous sublayer thickness of 1-1.5% of the whole boundary layer thick-
ness (decreasing with increasing Reynolds number).
Velocity-Defect Distribution
The velocity defect distributions of the mean velocities are given in Fig.
7, wherein Eq. 3 and Eq. 5 are also included. The data seem to conform
reasonably well for z / 8 & 0.03. While the prediction of the velocity defect
distribution by Eq. 3 is reasonable, Coles's expression, Eq. 5, falls com-
pletely outside the data points. Graf (1984) argued that the second term in
Coles's wake law must have another constant in order to fit the channel data.
In supporting this view, Nezu and Rodi (1986) found that for R > 105 the
profile parameter II in Eq. 5 was constant at a value of approximately 0.2,
which was considerably smaller than the Coles's value of 0.55. They also
reported that the II-values found by Steffler et al. (1985) were even lower
(varying between 0.08 and 0.15). In Fig. 7, Coles's velocity-defect law is
drawn with values K = 0.41 and II = 0 . 1 . As may be seen from the figure,
Coles's expression with the aforementioned values follows the experimental
1549
-«irr-
i o o,
O n N 5 " (A
eg m in t^ •- O
o © a> e e
3
-*t-
<Se
I
tJ
o O
0 0 CO
o Q o>
V
e(3
CM CO to r^
«
—
~^S"
-«r IS
1
E *
oC> <*>
o -3o <n
o t^
o
o in c- o en
fDMON N
. - N S Cs OJ
a
»o &
° i% » • » j l &
o
s ^ - ^e8><i
f
ra
e
o
to o> o> to o
r- CO (M O) O ._
CM ro in co <- £
0)
E
V e ol %
o 4 c
o
(3
-. 9 ° o -oi-3
assess
to oo ;£ io MT csj
o in.oo
CN e r - rei S-
e
o
NtO
1550
points reasonably well. This leads to the conclusion that in open channel
flow the profile parameter II can be below the 0.55 value given by Coles.
ap >
4. in
& in
N
+
•o
1
In m
c
_J
<olL
lb X
II
i *
I
_ c
^fc ? x o
^
1 <di
M» * «>
St® < 10
•\ 5
0 *( X
< 35
X
0)
a u
X N 1
X
xe
St.|jj- K. 1
Ss.
f'S*" H*
"B»
X*
/
'^Gc'l / <3
X
\
\
= |d»
1552
G
\\
•
\
\
v
\
\
• \
V
0E
0Vg
rV
\
•
TT
A_.
Nl
_J
c
+
-J-
m
in
—i\
\
=V
!' •
»l^
\
^
1553
\
0
\ o
\
0
1*1' ^
!
3
6
5
CD
CO
o
o
o
o
w
•a
ra
Downloaded from ascelibrary.org by Indian Institute of Technology Guwahati on 02/25/20. Copyright ASCE. For personal use only; all rights reserved.
making the water depth, in effect, h + Az; and (2) the fictitious flow through
the depth Az is in laminar condition (this is identical with Rotta's assump-
tion). The values of Az have been determined from the experimental velocity
profiles given in Fig. 8 for the near-wall region. As may be seen from the
figure, Az is obtained by extending the velocity profiles below the lowest
velocity reading AM to intercept the u = 0 line. The experimental values of
Az have been found to lie between 0.25 mm and 0.75 mm. With known
values of AM and Az, shear velocities can be calculated using Eq. 8 (see
Table 1). It may be noticed that the size of roughness elements is reflected
in the values of u* in that the value of M* increases with increasing roughness
size. From Fig. 8 the following points seem to be significant: (1) Az, which
represents the thickness of the fictitious viscous sublayer, is small compared
to the thickness of the smooth surface sublayer given in Fig. 4 (this fact was
also reported by Rotta 1950); and (2) Az is more sensitive to Reynolds num-
ber than the range of values of the uniform roughness height used in this
study.
Some of the variations of AM and Az are presented nondimensionally in
Figs. 9 - 1 2 . Fig. 9 shows that the ratio Au/um remains practically between
0.3 and 0.4 throughout the experimental range. It has average values of
0.32, 0.33, 0.35, and 0.37 for the rough surfaces, and an overall average
of 0.34. Fig. 10 gives the variation of AW/M* (or M#AZ/V) with u*k/v. The
best-fit expression for the data presented in this figure is
AM {u*k\
— = 3.50 (10)
«* \ V /
Fig. 11 shows the variation of the relative shift Az/k with u*k/v. The equa-
1554
' •
A
•
•
V
0
*0 A
G
•
A )
f (I A
0.2
0
I " ^ = 0.34 o Rough 1
A Rough 2
yi, „
7 Rough 4
12X104
4VR
Re
1555
U.AZ
V
v3-5^)
A Rough 2
a Rough 3
V Rough 4
100
v O
••
^ = 3.25
/ k sf
Az
//
A^ y
k
V
0 Rough 1 ^U
A Rough 2
V Rou 3h « \J'
7
Y
FiQ. 11. Variation of &z/k with ujc/v
1556
^ = 0.0,1 ( | )
^
(i1 T1^
1.( xK^
\ ^ 2.(
A
5.3
// «.A a ^
. / ,,-P 96 B \
U.' 2.1-
8 6 v
A 33^
o Rough 1
A Rough 2 •V/.bB 5.2X
11
• Rough 3 ?!?.,
V Rough 4 7.5 N
11.4
0.01 0.1 0.4
k/ I
"Law-of-the-Wall" Distribution
Fig. 13 shows the "law-of-the-wall" distribution of the mean velocities on
the "rough" surfaces. The distribution is quite different from that of the
"smooth" wall (see Fig. 5). Compared to the smooth wall results, the values
of u/u* are much lower for "rough" surfaces. Rotta (1962), using his the-
oretical model, gave the velocity profiles that varied depending on u*k/v
for sand-grain roughnesses. Such a systematic distinction, however, does not
seem to exist in Fig. 13. As may be seen from the figure, the point where
the law-of-the-wall distribution becomes applicable (with K = 0.41) moves
to a higher value, about u*z/v = 100, in comparison to the "smooth wall"
case. This point was mentioned also by Hinze (1975). In the fully turbulent
part of the inner region (that is, between u*z/v = 1 0 0 and 400), the data
seem to follow the expression
u u*z
— = 2.44 In 0.8 (13)
K* V
Velocity-Defect Distribution
The velocity-defect distribution of the data for the four "rough" surfaces
considered is shown in Fig. 14. As may be seen from the figure, the data
are so scattered that it is difficult to draw a single line to fit the data, even
for an individual rough surface. This is, perhaps, due to the fact that the
outer region of the turbulent boundary layer is influenced by the wall rough-
ness that affects the flow with increasing roughness height and Reynolds
number. Consequently, the velocity distribution becomes increasingly non-
uniform, with the considerable scatter that is seen in Fig. 3.
The velocity distribution for "rough" surfaces is not given in terms of z/
k because such presentations exhibit a separate tendency for each roughness.
It is known that the velocity distribution, especially in the inner region, is
essentially independent of the channel aspect-ratio (Kirkgoz and Darici 1986;
1557
gh 3
gh 4
— CM
V cn cn
• o o o o
\ cr cr cr cc
\ 1 0 0 <] D >
\ E
\\ \ b "6 5 "VIODN
c
'd V,
\ t& ° \ D
t FPKH
\ '&
\ ^
>m
w Lfl
00
\ d in
•Q
I
\ % N1
\ c
\ —1
3
\ O
\ CM'
K
\ 0^ gfiiH 11
\ o 0
\
\
*.-.N 0 6 » h
a. /
\l'* / 3
\ / *\>
L\ H)1- /
% <* / (A
^ 5
3
^ m
• ^
-4
;\ \
1 \ N|
a>
it
v
%• \ §
\ ,3>c \
\ \ «3
\ w I
\ ^Ch;
\ ^w
>
\
\
\
<\
\
\
\
\
1558
D
va o
V a i
° 8° 7 'D
A
' D
57
^
L
3' o
A D V
o V
V *> °[
A Aw
A
O
• 0s
° A
D
AO a
V
( hi B v D Va
v A V
P
O • AO
A
\ []
O A A° V {:* av
T
O A V O „ v
8 a
Sarma et al. 1983). Accordingly, the results of the present study, which
covers a wide range of aspect ratios (see Table 1), apply to channels that
are frequently encountered in civil engineering practice.
CONCLUSIONS
1559
tribution is quite reasonable for the particular roughnesses used in this study. On
the other hand, the velocity-defect distribution shows considerable scatter.
ACKNOWLEDGMENTS
APPENDIX I. REFERENCES
1560
b = width of channel;
"90 = particle diameter so that 90% of particles are smaller;
9 = gravitational acceleration;
h = depth of flow;
k = average height of roughness elements;
R = hydraulic radius;
u = mean flow velocity;
Au = mean velocity measured at z = Az;
um = maximum mean velocity in distribution (i.e., u at z = 8);
u* = shear velocity;
V = average flow velocity;
z = distance above bed;
Az = reference shift;
8 = thickness of boundary layer;
K = von Karman constant;
M- = dynamic viscosity;
V = kinematic viscosity;
p = density; and
T = shear stress.
1561