APPARENT FRICTION FACTOR ON THE FLOOD PLAIN-

MAIN CHANNEL INTERFACE OF COMPOUND

CHANNEL SECTIONS
G. C. CHRISTODOULOU
1
and W.R.C. MYERS
2
1
Department of Civil Engineering
National Technical University of Athens
Zografou 15780, Athens, Greece
Tel (+30)-1-7722813, Fax (+30)-1-7722814
e-mail: christod@hydro.civil.ntua.gr
2
School of the Built Environment
University of Ulster at Jordanstown
Newtonabbey, N.Ireland, U.K.
Tel (+44)-1232-366294, Fax (+44)-1232-366816
e-mail: wrc.myers@ulst.ac.uk
ABSTRACT
The paper aims at quantifying the apparent shear on the vertical interface between
main channel and flood plain in symmetrical compound sections. The apparent
shear stress is expressed in terms of an apparent friction factor f
a
and the square of
the velocity difference between subsections. Experimental data from the large scale
Flood Channel Facility in U.K., as well from a small scale facility are analyzed. The
velocities of the subsections are estimated by means of Mannings formula, thus
incorporating any difference in roughness and allowing the applicability of the
method in a predictive sense. The resulting f
a
is found to be well correlated to the
width ratio B/b of the compound section and to the Reynolds number on the flood
plain, Re
f
. These findings confirm that B/b is the most important geometrical
parameter in compound channel flows, in agreement with earlier research, and show
that flow in the least turbulent section, i.e. the flood plain, governs the interaction
with the main channel. A plot similar to the Moody diagram is produced, where B/b
takes the place of the relative roughness of pipe flow. This can be used to evaluate
the apparent shear stress and consequently the discharge of the compound section.
Keywords: Flood plain; compound channel; apparent shear stress; friction factor;
flood hydraulics.
INTRODUCTION
The study of flow in compound channel sections has been the subject of
considerable research in recent years. The practical interest in the problem arises
from the necessity for accurate discharge predictions during flood events and for a
reliable stage-discharge relation for flood control measures and management
schemes. It has long been realized that tranditional hydraulic methods of channel
subdivision are inadequate for discharge calculation due to the significant interaction
between main channel and flood plains. Such interaction leads to loss of
conveyance of the compound section and thus failure to account for it results in
overestimation of the discharge capacity of the channel at a given stage. Many
experimental studies have been carried out addressing various aspects of the
problem, ranging from the boundary shear distribution to the structure of turbulence
in compound section and various methods as well as empirical formulas have been
proposed for discharge calculation (e.g. Prinos and Townsend, 1984; Wormleaton et
al, 1982; Ackers, 1993). A major contribution to the available experimental data
basis was provided by the extensive studies on the large scale Flood Channel
Facility in U.K.
Despite the progress achieved so far, some fundamental issues of the flow in
compound channels remain unresolved. In particular, the resistance characteristics
of compound sections are not yet fully understood to the extent that they can be
readily assessed in engineering applications. Baird and Ervine (1984) tried to
correlate the Darcy-Weisbach friction factor of compound asymmetrical sections to
the overall Reynolds number. Myers (1984) showed, based on dimensional
considerations, that the friction factor depends -in addition to the overall Reynolds
number- on the ratio of Reynolds numbers of the main channel and flood plain
sections. Myers and Brennan (1990) presented detailed results from the Flood
Channel Facility showing that the friction factor varies widely between main channel
and flood plain and also depends on the overall Reynolds number. The obvious
variation of frictional resistance between flood plain and main channal due to the
difference in water depth (and hence relative roughness) and in Reynolds numbers
suggests that seeking a single overall friction factor for the entire compound section
may not be the best way to proceed. An alternative way is to examine the interaction
between the subsections in terms of an apparent shear on their interface. Several
investigations studied the apparent shear force or stress on the (imaginary) interface
between main channel and flood plain, e.g. Knight and Demetriou (1983). Baird and
Ervine (1984), Wormleaton and Merrett (1990) and others have presented empirical
correlations of the apparent shear stress in terms of the velocity difference between
the subsections and geometrical parameters. Christodoulou (1992) proposed, on
physical and dimensional grounds, an expression for the apparent shear stress in
terms of the square of the velocity difference between the subsections and showed
that the respective apparent friction coefficient depends mainly on the width ratio.
This conclusion was based on experimental data mostly from small-scale laboratory
facilities with a certain main channel aspect ratio. The availability of additional data
from larger scale experiments allows further investigation of the dependence of the
apparent friction factor on the geometrical characteristics as well as on the Reynolds
number. The purpose of this paper is to present preliminary results of this on-going
work and the relevant conclusions.
SOME THEORETICAL CONSIDERATIONS
A typical symmetrical compound channel section is shown in Figure 1.
Figure 1. Notation sketch
The apparent shear stress on the (imaginary) vertical interface between main
channel and flood plain may be expressed as

a
=1/2c
fa
V
2
=1/8f
a
V
2
(1)
where is the fluid density
V is the difference of mean velocity in the two subsections
f
a
is an apparent friction factor analogous to the familiar Darcy-Weisbach
friction factor (f
a
=4c
fa
)
Assuming uniform flow and considering the balance of forces along the flow direction
in the main channel leads to:

a
.2y+
m
P
m
=gS
o
A
m
(2)
where S
o
is the bottom slope
A
m
, P
m
are the area and wetted perimeter (excluding the interface) of the main
channel, respectively

m
is the average shear stress on the main channel boundary
If
a
can be estimated, eq. (2) may be used to evaluate the main channel discharge
for a given flow depth, whereas the flood plain discharge could also be obtained by a
similar procedure (Radojkovic and Djordjevic, 1985; Wormleaton and Merrett, 1990).
The apparent friction factor f
a
should in principle depend on the geometry of the
cross section and on some Reynolds number(s) as well as on the boundary
roughness(ess). For a smooth typical section as shown in Fig.1 dimensional
considerations lead to an expression of the form
f
a
=(B/b, y/h, b/h, s
m
, s
f
, Re
m
, Re
f
) (3)
where Re
m
and Re
f
are the main channel and flood plain Reynolds numbers,
respectively, defined as:
Re
m
=4R
m
V
m
/ Re
f
=4R
f
V
f
/
where R
m
, R
f
are the respective hydraulic radii
ANALYSIS OF RESULTS
To explore the dependence of f
a
on the parameters shown in eq. (3) the
experimental results from series 1,2,3,7,8,10 conducted at the FCF facility were
analysed (HR. Wallingford, 1992). The boundaries were smooth in all cases expect
for the flood plains in Series 7. In addition, and in order to detect in particular the
Reynolds number dependence, data of small-scale experiments presented by Knight
and Demetriou (1983) were also considered. The analysis was based on the
reported value of
a
on the vertical interface, calculated from the measured data; the
velocity difference between subsections was obtained not from the measured values
but from the velocities estimated by the Manning formula for each subsection,
assuming n=0.010 for the smooth boundary (in accordance with the single channel
value reported by Myers and Brennan, 1990) and n=0.020 for the rough boundary.
This approach was preferred to allow the evaluation of f
a
in a predictive sense, i.e.
without the need for measuring the velocities of the subsections. Table I shows the
main geometrical and flow characteristics of the data sets used.
Figure 2. Variation of apparent friction factor with relative depth and width ratio in
the Flood Channel Facility experiments
Figure 2 shows the dependence of f
a
on the relative depth for various width ratios
and section geometries. It is seen that f
a
remains nearly constant for 0.3<y/h<1.0, its
value depending clearly on B/b, as noted earlier by Christodoulou (1992).
Remarkably, the results of Series 2 and 7 essentially coincide despite the different
roughness, implying that the effect of roughness can be successfully incorporated in
the value of the mean velocities estimated by proper selection of Mannings n. It is
also seen that for small y/h the apparent friction factor increases appreciably in all
cases suggesting a Reynolds number influence for low overbank flows. After several
trials it was found that f
a
can be best correlated to the flood plain Reynolds number
Re
f
, as shown in Figure 3. This Figure includes data from both large scale (HR
Wallingford, 1992) and small scale (Knight and Demetriou, 1983) experiments.
Despite the inevitable experimental scatter, it is clear that f
a
tends to a constant
value depending on B/b for large Re
f
and increases for lower Re
f
towards a limiting
line on the (Re
f
, f
a
) plane independent of B/b, in a way similar to the classical Moody
diagram for flow in pipes. The width ratio B/b is seen to play the role of the relative
roughness
s
/D of pipe flow; even the variation of the ultimate value of f
a
(for large
Re
f
) with B/b is nearly logarithmic.
Figure 3. Dependence of the apparent friction factor on the flood plain Reynolds
number and the width ratio (HRW=HR Wallingford, 1992; KD=Knight & Demetriou,
1983; see Table I)
Reference
B
(m)
b
(m)
h
(m)
Sm Sf b/h B/b
Q
(l/s)
y/h
Re
(10
4
)
Ref(10
4
)
HR Wallingford (1992)
- Series 1 10 1.50 0.15 1 0 10 6.67 208-1014 0.060-0.67 6.8-32.6 0.41-22
- Series 2 6.3 1.50 0.15 1 1 10 4.2 212-1114 0.043-0.92 11.8-63.4 0.26-42
- Series 3 3.3 1.50 0.15 1 1 10 2.2 225-835 0.053-1.0 22.9-75.6 0.34-36.6
- Series 7* 6.3 1.50 0.15 1 1 10 4.2 216-543 0.04-1.01 12.7-31.5 0.12-25
- Series 8 6.0 1.50 0.15 0 1 10 4.0 185-1103 0.053-1.0 9.5-57.4 0.32-43
- Series 10 6.6 1.50 0.15 2 1 10 4.4 237-1092 0.053-0.86 12-54.7 0.35-33.8
Knight , Demetriou (1983)
- Series 1 0.608 0.152 0.076 0 0 2 4.0 4.9-29.4 0.12-1.02 2.5-12.7 0.53-10.7
- Series 2 0.456 0.152 0.076 0 0 2 3.0 5.0-23.4 0.15-0.96 3.1-12.3 0.68-9.0
- Series 3 0.304 0.152 0.076 0 0 2 2.0 5.2-17.1 0.12-0.97 4.3-11.2 0.49-5.7
* Rough flood plain
Table I. Main geometrical and flow characteristics of experiments analyzed
CONCLUSIONS
Analysis of available data from large scale and small scale experimental facilities
revealed that the apparent friction factor f
a
on the vertical interface between main
channel and flood plain in symmetrical compound channels depends primarily on the
relative width ratio B/b and the flood plain Reynolds number Re
f
. The plot of f
a
vs Re
f
is reminiscent of the Moody diagram, but instead of the relative roughness, the
ratio B/b governs the value of f
a
for large Re
f
. Use of the diagram allows a
straighforward prediction of the apparent shear stress and consequently of the
discharge distribution and overall discharge in a compound channel at a given
stage. Work proceeds towards testing the success of the discharge prediction by
comparison to available data.
REFERENCES
1. P. Ackers (1993), Flow Formulae for Straight Two-Stage Channels. J. Hydr.
Res., 31, 4, 509-531.
2. J.I. Baird and D.A. Ervine (1984), Resistance to Flow in Channels with Overbank
Flood Plain Flow. Proc. 1
st
Intern Conf. on Hydraulic Design in Water Resources,
Southampton, Springer-Verlag, 4.137-4.150.
3. G.C.Christodoulou (1992), Apparent Shear Stress in Smooth Compound
Channels. Water Res. Management, 6, 235-247.
4. HR Wallingford (1992), SERC Flood Channel Facility, Experimental Data-Phase
A, Vol.1, Report SR 314.
5. D.A. Knight and J.D. Demetriou (1983), Flood Plain and Main Channel Flow
Interaction. J.Hydr.Engrg, ASCE, 109,8,1073-1092.
6. W.R.C. Myers (1984), Frictional Resistance in Channels with Flood Plains. Proc.
1
st
Intern. Conf. on Hydraulic Design in Water Resources, Southampton,
Springer-Verlag, 4.73-4.87.
7. W.R.C. Myers and E.K. Brennan (1990), Flow Resistance in Compound
Channels, J.Hydr.Res., 28,2,141-155.
8. P. Prinos and R.D. Townsend, (1984), Comparison of Methods for Predicting
Discharge in Compound Open Channels. Adv. Water Res., 7, 180-187.
9. M. Radojkovic and S. Djordjevic (1985), Computation of Discharge Distribution in
Compound Channels. Proc. 21
st
IAHR Congress, Melbourne, 3, 367-371.
10. P.R. Wormleaton and D.J. Merrett (1990), An Improved Method of Calculation for
Steady Uniform Flow in Prismatic Main Channel/Floodplain Sections.
J.Hydr.Res., 28,2,157-174.