You are on page 1of 10

SUPERCRITICAL F L O W IN S M O O T H

O P E N CHANNELS
By K a n d u l a V. N. S a r m a 1 a n d P . Syamala 2

ABSTRACT: The effect of Reynolds number and Froude number on the friction
factor is studied for supercritical turbulent flow in smooth rectangular open chan-
nels. The values of Froude numbers for various aspect ratios, beyond which the
flow becomes unstable, are determined using the method of Berlamont and Van-
derstappen and the measured momentum coefficients. For stable supercritical flows,
the friction factor varies with Reynolds number only, as in the case of subcritical
flows. For unstable supercritical flows, the friction factor is found to vary ac-
cording to Rouse's equation.

INTRODUCTION

Formulas developed for pipe friction are, strictly speaking, not applicable
for open channels. But it is common practice to connect the cross-sectional
mean velocity, V, and the mean shear stress T over the entire wetted perim-
eter by formulas similar to those for flows in pipes by replacing the diameter
d0 of the pipe by AR, in which R is the hydraulic radius of the open channel
(Chow 1959; Gonchorov 1970; Henderson 1966; Keulegan 1938; Laksh-
minarayana et al. 1986; Rajaratnam and Muralidhar 1969; Schlichting 1968)
and by adjusting the constants suitably.
While this procedure is satisfactory for stable free-surface flows, it is in-
adequate for unstable free-surface flows. When the Froude number F, ex-
ceeds a certain value F s , at which instability of free surface occurs, the flow
would be nonuniform and unsteady. If the channel is long enough, even-
tually roll waves develop. The value of F s depends on the aspect ratio, rel-
ative roughness, and velocity distribution, and is always greater than unity
(Berlamont and Vanderstappen 1981). The friction f a c t o r / i s influenced by
the Froude number F for flows in which F > F s (Berlamont and Vander-
stappen 1981; Rouse et al. 1963; Rouse 1965; Syamala 1988); hence, the
friction formulas for such flows should include the Froude number, F. In
this paper the results of a study on the variation of / with F are presented
for flows in smooth rectangular open channels.

REVIEW OF LITERATURE

The effect of Froude number on flows in open channels was first men-
tioned by Jegerov (1946). Powell (1949) found experimentally that in the
case of smooth open channels, the friction f a c t o r / ( = 8T/V 2 ) in which p is
the mass density, was influenced by F, for F S: 2.49. He observed no effect
of F o n / f o r F < 1.69.
Rouse et al. (1963) observed that when F was well in excess of unity, it
'Prof., Dept. of Civ. Engrg., Indian Inst, of Sci., Bangalore 560 012, India.
2
Asst. Prof., Dept. of Civ. Engrg., Regional Engrg. Coll., Calicut-1, India.
Note. Discussion open until June 1, 1991. To extend the closing date one month,
a written request must be filed with the ASCE Manager of Journals. The manuscript
for this paper was submitted for review and possible publication on June 22, 1990.
This paper is part of the Journal of Hydraulic Engineering, Vol. 117, No. 1, Jan-
uary, 1991. ©ASCE, ISSN 0733-9429/91/0001-0054/$1.00 + $.15 per page. Paper
No. 25423.

54

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit ht
influenced /. They expressed the view that the gravitational influence on
open channel resistance was in some way connected with the free suiface
instability involved in the formation of roll waves. They could usually ob-
serve perceptible roll waves in a long enough flume for runs in which F >
Fs.
Iwagaki (1952) assumed the mixing length to decrease close to the wall
and increase away from the wall due to the effect of free surface. On the
basis of his own data and other Japanese data and the data of Powell (1946),
he concluded that the resistance to flow is influenced by the Froude number
expressed in terms of hydraulic radius R when it is greater than 0.89. How-
ever, the data show large scatter [Fig. 3 of Iwagaki (1954)]. So the authors
are also of the opinion that the gravitational influence on open-channel re-
sistance is associated with free-surface instability and hence can be present
only for F > Fs > 1.
A number of criteria for unstable uniform flow, basically differing only
in the assumptions made regarding the channel geometry, velocity profile
over the depth, and the resistance relationship for uniform turbulent flow
have been developed by the earlier investigators (Brock 1969).
For a wide rectangular open channel with a large aspect ratio, Ar (width-
to-depth ratio), rectangular velocity distribution over the depth and constant
/ , Jeffreys (1925) obtained Fs = 2.0.
For the case of flows in smooth open channel, which is of particular in-
terest in this paper, Rouse (1965) derived for F s , the following equation,
assuming logarithmic velocity distribution and a linear relationship between
/ a n d the momentum correction factor (3 = (/ v2dA/V2A).
1
Fs = — — = = = = = = = = = = = = ^ = ^ = = (1)
V(0.5 + 1.3/1/2 - 0.78/)2 - 0.78/(1 + 0.78/)
Using the Fs value so obtained, / i s related to R, F and Fs as

- ^ = 2 log
r RV?
p\ 2
/3 -0.97 f o r F > Fs (2)

in which R = Reynolds number = 4RV/v and v = kinematic viscosity.


Eq. 2 showed good agreement at least as a first approximation with the
experimental data of Iowa (Rouse 1965).
In the analysis of Escoffier and Boyd (1962) they assumed the resistance
coefficient to be independent of the relative roughness and (3 = 1.0. For a
rectangular open channel, the following equation can be obtained for Fs from
their analysis.
(Ar + 2\
F =L5 (3)
' (—j
in which Ar = B/D where B = width of the channel, and D = depth of
flow.
As per Eq. 3, F, decreases with increasing Ar and is equal to 1.5 for wide
channels with Ar = °°.
For turbulent flows in open channels (Berlamont and Vanderstappen 1981),
55

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit ht
007
0-06 • Author-; (F < F s )
0-05 o Authors(F > F S )
A Tracy & Lester ( F > F S ,'
0-04

0-03

0.02

0-01 —

0.007 _i I i I i I J L I i I i I
8 103 10°

FIG. 1. Variation o f / w i t h R

the value of Fs was determined from two simultaneous equations. These


equations can be simplified in smooth open channels as follows.

- — j - 20C + C2 = 0, (4)

D (2 + M)f
(5)
8 dD 8 8 dV 4/

where C = the ratio of wave velocity of an infinitesimal small-amplitude


wave at F = F„ to V; F = V/VgD; g = acceleration due to gravity; and
M = 1/(1 + 2/Ar).
It may be noted that the friction factor to be used in Eq. 5 is the / for
stable flows.
Knowing the value of (3 and assuming a proper friction law for uniform
flow, Eqs. 4 and 5 can be solved for Fs and C. They computed Fs values
using different friction laws and p values, p values were computed assuming
different velocity distributions over the depth and no lateral variation in the
velocity in all the cases.

PRESENT APPROACH

In an earlier study at the Indian Institute of Science (Lakshminarayana et


al. 1986), the following equation was obtained for/in the case of subcritical
smooth turbulent flows in rectangular open channels
/ = 0.328FT for F < 1.0 (6)
It was observed that F did not influence / and the use of R in computing
R adequately took care of the effect of the channel shape in the friction law
for subcritical smooth turbulent flows in rectangular open channels. Eq. 6
gives/values that are higher than those given by Blasius equation (Schlicht-
56

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit http://w
ing 1968) for pipe flows. Eq. 6 is found to be valid for 1 < F < Ft also
(Fig. 1). So this equation is used in the present study to evaluate F,.
The value of (3 is computed from the velocity distributions measured in
smooth rectangular channels at I.I.Sc. (Lakshminarayana 1980; Syamala 1988)
covering wide ranges of F, R, and Ar.
Substituting Eq. 6 into Eq. 5 and solving the Eqs. 4 and 5, simulta-
neously, one gets for Fs
1 / 5M\ I 5M\2 '
- = P - 2 P ( l + T ) + ( l + T ) (7)

Fs thus depends on |3 and Ar. Fs decreases as the aspect ratio (two-dimen-


sionality) increases.
On the basis of Eq. 7, the data of/collected by the authors for F > Fs
is analyzed.

EXPERIMENTS

The experimental setup mainly consisted of a smooth straight rectangular


flume with a length of 15.25 m and a depth of 0.3 m. The experiments were
carried out at a width of 0.305 m and 0.61 m. The maximum depth of flow
was approximately 0.15 m. The flume was connected at the upstream end
to a supply tank with a smooth bellmouth entry. The water was supplied to
the flume through the head tank by a constant speed centrifugal pump ca-
pable of delivering up to 80 lps from the main sump of the laboratory. The
flow from the flume was drained into a tail channel fitted with a calibrated
weir, from where it was led to the main sump.
The flows were nearly uniform on the upstream portion of the channel.
But on the downstream portion of the channel, the flows were uniform only
in certain cases. In other cases the flows were both unsteady and nonuniform
and the free surface was quite rough and uneven. However, it was not pos-
sible to detect the roll waves.
According to Montouri (1963) small disturbances occurring at the up-
stream end of the channel develop into roll waves only after covering a
certain distance and hence the roll waves are not detected in short channels.
This must be the reason in the case of authors' experiments also. When the
authors reduced the depth of flow to about 1 cm in the 61-cm-wide channel,
they could observe an orderly train of roll waves. At this depth R was well
over 10,000 and the flow was still turbulent.
In the case of unsteady nonuniform flows temporal mean depths were used
in computations and the mean shear T was calculated by applying the mo-
mentum equation to a reach. Totally, 67 runs were made in the supercritical
range, details of which are given in Table 1. For runs 1-33 (B — 0.305 m),
velocity distributions were also measured. For runs 34-67 (B = 0.61 m),
velocity distributions were not measured.

FRICTION FACTOR

The values of/ and R obtained for supercritical flow are plotted in Fig.
1. The authors' data are compared with those of Tracy and Lester (1961)
and Eq. 6 of Lakshminarayana et al. (1986).
57

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit
TABLE 1. Values of Friction Factor/and Fs for Different Runs

Run number F K R x 10" 4 F, f


(D (2) (3) (4) (5) (6)
1 1.36 2.05 52.12 3.35 0.0122
2 1.43 4.00 26.92 2.33 0.0144
3 1.43 11.73 6.72 1.765 0.0200
4 1.47 2.03 56.32 3.3 0.0120
5 1.49 6.34 15.68 1.96 0.0164
6 1.50 9.50 9.44 1.80 0.0187
7 1.53 4.83 22.52 2.17 0.0150
8 1.55 7.60 12.88 1.92 0.0173
9 1.56 2.73 44.00 2.82 0.0127
10 1.60 7.40 13.76 1.93 0.0166
11 1.60 2.35 52.52 3.025 0.0118
12 1.65 4.98 21.36 2.20 0.0150
13 2.40 9.14 15.68 1.84 0.0160
14 1.75 4.90 25.44 2.17 0.0146
15 1.78 8.50 12.72 1.86 0.0173
16 1.79 10.00 10.32 1.82 0.0180
17 1.80 3.94 33.76 2.35 0.0136
18 1.80 6.00 20.12 2.04 0.0155
19 1.79 7.00 16.76 1.95 0.0160
20 1.80 8.21 13.56 1.88 0.0174
21 1.85 9.05 12.16 1.84 0.0190
22 2.09 5.10 27.92 2.13 0.0140
23 2.06 7.88 16.28 1.90 0.0180
24 2.09 10.00 12.16 1.82 0.0196
25 2.24 5.10 30.88 2.14 0.0155
26 2.30 7.00 21.28 1.95 0.0172
27 2.35 8.00 18.20 2.08 0.0178
28 2.42 9.84 14.40 1.81 0.0188
29 2.44 9.00 16.28 1.84 0.0183
30 2.55 ' 5.00 36.20 2.15 0.0150
31 2.60 7.00 24.04 1,89 0.0166
32 2.65 10.00 15.20 1.82 0.0186
33 2.65 9.00 17.76 1.84 0.0179
34 2.31 8.00 54.20 1.89 0.0140
35 2.30 9.00 46.04 1.84 0.0145
36 2.29 10.00 40.00 1.80 0.0150
37 2.30 11.00 34.92 1.77 0.0155
38 2.30 12.00 31.44 1.75 0.0158
39 2.28 13.00 28.00 1.73 0.0162
40 2.28 14.00 25.12 1.71 0.0166
41 2.265 15.00 22.72 1.70 0.0170
42 2.65 8.00 61.92 1.89 0.0137
43 2.62 9.00 52.48 1.84 0.0142
44 2.63 10.00 45.60 1.80 0.0146
45 2.61 11.00 39.60 1.77 0.0150
46 2.61 12.00 35.68 1.75 0.0154
47 2.60 13.00 32.72 1.73 0.0157
48 2.59 14.00 29.04 l.yi 0.0161
49 2.60 13.00 26.32 1.70 0.0164

58

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit http://ww
TABLE 1. {Continued)
(1) (2) (3) (4) (5) (6)
50 2.00 7.00 55.64 1.95 0.0140
51 2.02 8.00 47.48 1.89 0.0144
52 2.02 9.00 40.44 1.84 0.0150
53 2.00 10.00 43.00 1.82 0.0154
54 2.00 11.00 30.40 1.77 0.0159
55 2.00 12.00 27.36 1.75 0.0163
56 1.98 13.00 24.36 1.73 0.0167
57 2.01 14.00 22.16 1.71 0.0171
58 2.01 15.00 19.76 1.70 0.0178
59 1.80 5.00 76.20 2.15 0.0110
60 1.80 6.00 60.96 2.03 0.0115
61 1.80 7.00 50.00 1.95 0.0124
62 1.80 8.00 42.00 1.89 0.0125
63 1.80 9.00 36.00 1.84 0.0135
64 1.81 10.00 31.60 1.80 0.0156
65 1.76 12.00 24.08 1.75 0.0175
66 1.79 14.00 19.72 1.71 0.0185
67 1.78 15.00 17.92 1.70 0.0190

Fig. 1 shows that while some points fall on the curve given by Eq. 6,
others show a substantial shift. The points that fall on the line were found
to be for F < Fs. The remaining larger set of data, including those of Tracy
and Lester, pertained to F > Fs. There are, however, nine data points of
Tracy and Lester for 1 < F < Fs that are not shown in Fig. 1. These points
have fallen midway between the two sets of data.

DETERMINATION OF FS

Fs can be determined for a given Ar from Eq. 7 if the value of p is known.


Usually (3 is either assumed as 1 or its value is calculated from an empirical
equation wherein it is linearly connected to the friction factor/. The value
of p can also be theoretically derived for given velocity profiles. In the case
of flow through open channels that are not wide, the shapes of velocity
profiles are not simple and vary along the width (Sarma et al. 1983; Syamala
1988). So in the present study the authors used the average of measured
values of p from 33 runs (Table 1) of Syamala (1988) and 44 runs of Laksh-
minarayana (1980).
Lakshminarayana (1980) measured velocity distributions for 44 runs in the
subcritical flow range. Syamala (1988) measured velocity distributions for
33 runs for supercritical Froude numbers. The range of variables covered by
Syamala in these experiments (runs 1-33) are given in Table 1. Experi-
mental range of Lakshminarayana is given in Lakshminarayana (1980) and
Lakshminarayana et al. (1986). The value of p varied from 1.015 to 1.03,
with a mean value of 1.022 for these 77 runs (Syamala 1988). This range
is in agreement with the range of 1.01-1.03 mentioned by Watts et al; (1967).
So the mean value of 1.022 was substituted in Eq. 7 for p to determine F s .
Fig. 2 shows the variation of Fs for smooth rectangular open channels.
The figure shows that Fs decreases for increasing Ar values.
59

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit h
(a) B = 30.5 cm

Slope
.1 in 51-2

3-5 (b) = 61 0 cm

—^£~_ A in 5L2
3.0
Vs
-_____^1JJT65

2-5
_____Jin88.5
F.F S
2-0 ' - ^ /firTrio"

1-5

1-0 i i i i i 1 1 1 1
10 12 H 16 18 20
Ar

FIG. 2. Variation of F, and F (for Different Slopes) with Ar

The variation of F with Ar for a given width and slope of the channel can
be computed using Eq. 6. Fig. 2 also shows such a variation for the two
widths of 0.305 m and 0.61 m for some slopes used in the experiments. It
can be seen from Fig. 2 that F is greater than Fs for a given slope only
beyond a certain Ar and that the free surface is always stable for flows on
a slope of 1 in 135 in the 0.305-m-wide channel. The nature of F and Fs
curves suggest that for a given width of the channel there is a limiting slope
such that for the slopes flatter than this limiting slope the flows are always
stable.

60

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit http://ww
EFFECT OF F

Fig. 1 shows that with the onset of instability, the values of the friction
factor/ suddenly rise above those for stable flows. For F > Fs the following
equation fits the data of writers and Tracy and Lester (1961) well.
/ = 0.368FT0 25 (8)
Eq. 8 implies that F has no further effect on/once F > Fs.
For the same data, Eq. 2 of Rouse (1965) is fitted in Fig. 3 and the fit
is good. Eq. 2 implies that F influences / once F > F s .
Eqs. 8 and 2 are apparently contradictory. This contradiction can be re-
solved, once it is realized that, when the variation of log (F/F s ) 2/3 is small
for the range of experimental data, that term gets replaced by its average
value in the friction equation. That F/F s can approximately be constant for
certain combinations of Ar and bed slopes in the experiments, can be under-
stood from Fig. 2. This feature was also explained by Berlamont and Van-
derstappen (1981). They showed that the ratio of the rise i n / value to /
becomes nearly a constant for a certain range of F/F s ratios. For this range,
/ appears to depend only on R through F > F^.

CONCLUSIONS

1. As the flow becomes unstable the friction factor / becomes consistently


higher than that for stable flows.

9.5

8-5

7-5
A Tracy & Lester
— Rouse (Eq.2)
6.5 _i I i i * • i

9-5

8-5
J_
VF
7-5

6.5
.10* 2 4 6 8 10 5
2/3
R^T/(F/FS)

FIG. 3. Variation of 1/v?with RVf/(F/F s ) 2 / 3

61

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit
2. When F > F,, F influences / according to Eq. 2.
3- For a certain range of F / F J ; the influence of f is io increase/"by a constant
amount and in that range, / i s given by Eq. 8.

APPENDIX I. REFERENCES

Berlamont, J., and Vanderstappen, N. (1981). "Unstable turbulent flow in open


channels." J. Hydr. Div., ASCE, 107(4), 427-449.
Brock, R. R. (1969). "Development of roll wave trains in open channels." / . Hydr.
Div., ASCE, 95(4), 1401-1427.
Chow, V. T. (1959). Open channel hydraulics. McGraw-Hill Book Co., Inc., New
York, N.Y.
Escoffier, F. F., and Boyd, M. M. (1962). "Stability aspects of flow in open chan-
nels." / . Hydr. Div., ASCE, Vol. 88(6), 145-166.
Gonchorov, V. N. (1970). Dynamics of channel flow • 2nd Ed., Israel Program for
Scientific Translations, Jerusalem, Israel.
Henderson, F. M. (1966). Open channel flow. MacMillan Co., New York, N.Y.
Iwagaki, Y. (1952). "On the laws of resistance to turbulent flow in open smooth
channels." Proc. 4th Japan Nat. Congress for Appl. Mech., 245-250.
Jeffreys, H. (1925). "The flow of water in an inclined channel of rectangular sec-
tion." Philosophical Magazine and Journal of Sci., London, United Kingdom,
793-807.
Jegerov, S. A. (1940). "Turbulente Uberwellenstromung." Wasserkraft und Was-
serwirtschaft, 35, 55-57.
Keulegan, G. H. (1938). "Laws of turbulent flow in open channels." Research Paper
RP 1151, Nat. Bureau of Standards, U.S. Dept. of Commerce, Washington, D.C.,
Vol. 21, 707-741.
Lakshminarayana, P. (1980). "Velocity and shear distribution in smooth rectangular
open channels with small aspect ratios," thesis presented to the Indian Institute of
Science, at Bangalore, India, in partial fulfillment of the requirements for the de-
gree of Doctor of Philosophy.
Lakshminarayana, P., Sarma, K. V. N., and Lakshmana Rao, N. S. (1986). "Shear
in smooth rectangular open channels." J. Inst. Eng. India Part CI, 66(5), 204-
213. - -
Montouri, C. (1963). Discussion of "Stability aspects of flow in open channels," by
Escoffier and Boyd. J. Hydr. Div., ASCE, 89(4), 264-273.
Powell, R. W. (1946). "Flow in a channel of definite roughness." Trans., ASCE,
New York, N.Y., 531-554.
Rajaratnam, N., and Muralidhar, D. (1969). "Boundary shear stress distribution in
rectangular open channels." Houille Blanche, 6, 603-609.
Rouse, H. (1965). "Critical analysis of open channel resistance." / . Hydr. Div.,
ASCE, 91(7), 1-25.
Rouse, H., Koloseus, H. J., and Davidian, J. (1963). "The role of the Froude num-
ber in open channel resistance." J. Hydraul. Res., 1(1), 14-19.
Sarma, K. V. N., Lakshminarayana, P., and Lakshmana Rao, N. S. (1983). "Ve-
locity distribution in smooth rectangular open channels." J. Hydr. Engrg., ASCE,
109(2), 270-289.
Schlichting, H. (1968). Boundary layer theory. 6th Ed., McGraw-Hill Book Co.,
New York, N.Y.
Syamala, P. (1988). "Velocity, shear and friction factor studies in smooth rectangular
channels of small aspect ratios for supercritical flows," thesis presented to the
Indian Institute of Science, at Bangalore, India, in partial fulfillment of the re-
quirements for the degree of Doctor of Philosophy.
Tracy, H. J., and Lester, C. M. (1961). "Resistance coefficients and velocity dis-
tribution—Smooth rectangular channel." Water Supply Paper 1592-A, U.S. Geo-
logical Survey, 1-18.
Vedernikov, V. V. (1946). "Characteristic features of a liquid flow in open channel."
Trans., U.S.S.R. Academy of Sci., Vol. 52, 207-10.

62

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit http:
Watts., F. J., Simons, D. M., and Richardson, E. V. (1967). "Variations of a and
values in a liiicu open channel. J. Hydr. Div., Ai>Cli, yj(o), 2 1 7 - 2 J 4 .

APPENDIX II. NOTATION

TVie following symbols are used in this paper:

A = area of cross section of flow;


Ar = aspect ratio (B/D);
B = width of rectangular open channel;
C = critical dimensionless wave velocity;
D = depth of flow;
d0 = pipe diameter;
F = Froude number ( V / V g D ) ;
Fs = Froude number beyond which flow becomes unstable;
/ = Darcy-Weisbach friction factor;
g = acceleration due to gravity;
M = 1/(1 + 2/Ar);
R = hydraulic radius;
R = Reynolds number (4VR/v);
V = cross-sectional mean velocity;
(3 = momentum correction factor = / v2dA/V2A;
p = mass density; and
v = kinematic viscosity.

63

Downloaded 25 Nov 2010 to 144.122.101.171. Redistribution subject to ASCE license or copyright. Visit http

You might also like