13th Australasian Fluid Mechanics Conference

Monash University, melbourne, Australia

13-18 December 1998

255

HYDRAULICS OF RUBBER DAM OVERFLOW : A SIMPLE DESIGN APPROACH

H. CHANSON

Department of Civil Engineering, The University of Queensland, Brisbane QLD 4072, AUSTRALIA


ABSTRACT
Rubber dams have been used for the past 40 years in river
and coastal engineering applications, in Australia and
overseas. Despite the increasing interest for rubber dams,
little information is available on their hydraulic
performances. The overflow characteristics of inflated rubber
dams are re-investigated. New analytical calculations provide
information on the wall pressure distribution and on the
nappe trajectory. The calculations are compared with over
160 new experiments. Altogether the results provide a new
method to predict overflow characteristics and new design
guidelines are derived for the optimum location of the
deflector.

INTRODUCTION
Rubber dams are long tubular-shaped fabrics placed across
channels, streams and weir crest to raise the upstream water
level when inflated (Fig. 1). In open channels, they are
commonly used to raise water levels, to increase water
storage and to prevent chemical dispersion (Table 1). The
interest in inflatable dams is increasing because of the ease of
placement. Such structures can be installed during later
development stages.
The membrane is usually deflated for large overflows. It is
however common practice to allow small spillages over the
inflated dam. During overflows, vibrations might result from
fluid-structure interactions (e.g. OGIHARA and
MARAMATSU 1985, WU and PLAUT 1996), and the
instabilities might damage and destroy the rubber membrane.
Several failures were experienced including in Australia. In
practice, a deflector (i.e. fin) is installed on the downstream
face of the rubber dam to project the nappe away from the
membrane, hence preventing rubber membrane vibrations
(Fig. 2).
Little attention has been paid on the overflow hydraulics.
ANWAR (1967) investigated experimentally small overflow.
Other studies (e.g. SHEPHERD et al. 1969, BINNIE et al.
1973) discussed the fluid-structure interactions of inflated
rubber dams.
In the present paper, the characteristics of rubber dam
overflow are re-investigated. New analytical calculations of
nappe trajectory are presented and the results are compared
with laboratory experiments. New guidelines for the
optimum design of rubber dam deflector are discussed.

NAPPE TRAJECTORY
Considering a fully-inflated rubber dam, the downstream
face of the dam follows closely the shape of a circular
cylinder as observed by ANWAR (1967). For the sake of
simplicity, we shall consider an idealised rubber dam shape
(Fig. 2).



Fig. 1: Sketch of a rubber dam


Year Site Characteristic
s
Manufactur
er
Remarks
(1) (2) (3) (4) (5)
1965 Koomboolo
omba dam,
QLD
L =1 60 m,
D =1.22 &
1.5 m, H
infl
=
0.91 m
Fabridam-
Firestone
No deflector. Water
filled. Placed on
ogee crest. Design
overflow : 2.4 m
2
/s
(inflated), 35 m
2
/s
(deflated).
1967 Proston
weir, QLD
L =1 51 m,
D =1.5 m,
H
infl
=1.4 m
Fabridam-
Firestone
No deflector. water
filled
1983 Val Bird
weir, North
QLD
L =2 82 m,
D =1.9 m,
H
infl
=0.5 m
Fabridam-
Firestone
Water filled.
1996 Lyell dam,
NSW
L =2 40 m,
D =3.5 m,
H
infl
=1.4 m
Bridgestone With deflector. Air
filled
1997 Dumbleton
weir,
Central
QLD
L =2 75 m,
D =2 m
H
infl
=0.7 m
Queensland
Rubber Co.
With deflector. Air
filled.

Table 1: Examples of rubber dams installed in Australia

256

Fig. 2: Rubber dam overflow


Downstream of the crest, the overflowing nappe adheres to
the weir face because the convex wall curvature imposes a
pressure field modification within the nappe inducing a
suction pressure. The resulting Coanda effect acts on the wall
surface in a direction normal to the flow direction.
At the crest, critical flow conditions occur. Downstream of
the crest, the flow depth and velocity derive from the
continuity and motion equations. Assuming an ideal-fluid
flow, it yields :
F = F
crest
2
+ 2 * (1 - cos) (1)
d
R
=

d
crest
R
3/2
*
C
D
k
3/2
*
1
F
(2)
where d is the flow depth, V is the velocity, F =V/ g*R, is
the angular position, F
crest
=V
crest
/ g*R, C
D
is the
discharge coefficient, H
1
is the upstream total head, D is the
dam height, d
crest
is the flow depth at the crest, R is the
radius of curvature and k = d
crest
/d
c
accounts for the
streamline curvature and non-uniform distributions of
pressure and velocity at the crest (Fig. 2).
At the surface of the rubber dam, the wall pressure may be
deduced from the motion equation in the radial direction. At
any position , the dimensionless pressure distribution at the
wall equals :
P
atm
- P
s

w
*g*R
=
d
R
*

F
2
- cos *

1 +
d
2*R
(3)
where P
s
is the absolute pressure at the wall, P
atm
is the
atmospheric pressure,
w
is the water density. Equation (3)
predicts an increasing suction pressure (P
atm
- P
s
) down the
rubber membrane as the flow is accelerated.
The nappe adherence on the wall might lead to flow
instability at the base of the nappe (i.e. next to the separation
position), pressure fluctuations on the downstream face of the
dam and vibrations of the flexible membrane. Nappe
adherence instabilities may be eliminated by deflecting the
nappe off the rubber wall (Fig. 2).
At take-off the flow properties (d
o
, V
o
,
o
) are deduced from
the Bernoulli and continuity equations, neglecting energy
losses and assuming that the effects of the developing
boundary layer are small (i.e. Eq. (1) and (2)). Usually the
deflected nappe angle
o
at take-off is smaller than the
deflector angle , and it may be estimated in first
approximation as :

= tanh

h
d
o
*
(4)
where h is the deflector height measured normal to the wall
and d
o
is the nappe thickness at take-off (Fig. 2).
The trajectory equations of a ventilated nappe are :
x
R
=
V
o
g*R
* cos(
def
-
o
) *
g * t
2
R
+
x
o
R
(5)
y
R
= -
1
2
*
g *t
2
R
-
V
o
g*R
* sin(
def
-
o
) *
g * t
2
R

+
y
o
R
(6)
where x is the horizontal direction, y is the vertical direction
positive upwards, t is the time, x
o
and y
o
are the co-ordinates
of the deflector edge,
def
is the angular position of the
deflector.
Equations (1) to (6) may be combined to predict the nappe
trajectory from the crest down to the nappe impact.


COMPARISON WITH EXPERIMENTAL DATA
Experimental setup
Laboratory experiments were conducted to validate
qualitatively and quantitatively the basic flow patterns. The
inflated rubber dams were idealised with circular concrete
cylinders (Fig. 3). Two rectangular channels were used. Each
flume is supplied with recirculating water supplied by a
constant head tank, the discharge being measured by 90-
degree V-notch weirs.
Several configurations were tested : i.e., three cylinder sizes
(R =0.052, 0.0755, 0.117 m), several deflector sizes and
positions (0 h 0.11 m, 0
def
90 deg.). All cylindrical
weirs had a smooth varnish surface and the deflected napped
were ventilated by sidewall splitters. The flow depths were
measured on the centreline using point gauges connected to a
Mitutoyo
TM
digimatic caliper (Ref. No. 500-171). The error
on the flow depth was less than 0.1-mm and the error on the
longitudinal position was less than 0.5 mm.
The hydraulic characteristics of the crest were investigated
separately and results were presented in CHANSON and
MONTES (1997). The present experimental study was
focused on the nappe trajectories and altogether over 160
new experiments were performed.


257


(A) Flow from right to left : q =0.0048 m
2
/s, d
1
=0.178 m,
R =0.117 m,
def
=45 deg., h =0.004 m


(B) Nappe re-attachment : q =0.006 m
2
/s, W =0.25 m, R =
0.117 m,
def
=25 degrees, h =0.026 m
Fig. 3: Overflow above an ideal rubber dam with deflector


Experimental results
Visual observations (e.g. Fig. 3) showed satisfactory
operation of the rectangular deflectors. Upstream of the fin,
the nappe accelerates until it is projected upwards. Once
deflected, the flow becomes a free-falling nappe. The jet
trajectory depends critically upon the location
def
of the
deflector. Further the trajectory expands with increasing
deflector height h for the same weir, flow conditions and
deflector location
def
.
Note that the deflector induced a backwater effect upstream
for
def
=0 (h/R >0, all flow rates) and
def
=25 degrees
(h/R 0.22, all flow rates) only. For these geometries,
critical flow conditions occurred at the deflector rather than
at the crest.
Careful measurements of the upper nappe and recirculation
pool were performed and compared with ideal-fluid flow
calculations (e.g. Fig. 4).
Overall reasonable agreement was found between
calculations and data, but for the smaller cylinder size (i.e. R
=0.0524 m). For R =0.0524 m, the deflected nappe tended
to attach to the fin, leading to shorter jet trajectories than
predicted in some tests.





0
0.5
1
1.5
2
2.5
-1 -0.5 0 0.5 1 1.5 2 2.5
Cylinder
Data
Calculations
Run C3F125
Deflector
y/R
x/R
Q =5 L/s
Pool free-surface

Fig. 4: Nappe trajectories - Comparison between
experimental data and ideal-fluid flow calculations

def
=25 deg, q =0.020 m
2
/s, R =0.0755, d
1
=0.2012 m, h
=0.0039 m, D =0.1544 m, d
crest
=0.03305 m


Comparison with air jet studies
Several researchers investigated the Coanda effect with air
jets flowing past circular cylinders. FEKETE (1963)
performed very careful pressure measurements with various
cylinder diameters (1.2E+5 < *V*R/ < 7.4E+5) while
SARPKAYA (1968) investigated the deflection of air jets
blowing past circular quadrants (2.45E+5 < *V*R/ <
4.0E+5). A re-analysis of their data (Fig. 5) indicates that the
dimensionless pressure distributions on the cylinder wall
were close to ideal-fluid flow calculations :

P
atm
- P
s
*g*R
= F
2
*
d
R
Air flow (3b)
where V, F and d are defined in terms of the flow properties
at the jet nozzle, and and are the fluid density and
viscosity respectively. Equation (3b) is compared favourably
with the original data of FEKETE (1963) and SARPKAYA
(1968) on Figure 5.

Nappe re-attachment
The agreement between data (air jets and water nappes) and
calculations suggests that the ideal-fluid flow calculations
may be used as a design tool to predict the optimum deflector
position and size and to calculate the corresponding nappe
trajectory.
The purpose of the fin being to deflect the nappe away from
the wall, nappe re-attachment downstream of the deflector is
not acceptable. In laboratory, nappe re-attachment was
observed for
def
=0 (all flow rates) and for
def
=25
degrees at low flow rates (e.g. Fig. 3B). No nappe re-
attachment was observed for
def
= 30 to 90 degrees.
However, with
def
=90 degrees, little nappe deflection was
observed and the nappe impact was located not far away
from the rubber dam toe.
258
Overall maximum nappe deflection was observed for 30

def
60 degrees. This result is very close to calculations
(CHANSON 1996).


0.00
0.20
0.40
0.60
0.80
1.00
1.20
0 20 40 60 80 100 120 140
FEKETE
SARPKAYA
Ideal-fluid

(degrees)
(Patm-Ps)/(ro*g*d*F2)

Fig. 5: Adherence pressure distribution (P
atm
-
P
s
)/(*g*d*F
2
) versus - Comparison with air jet data



SUMMARY AND CONCLUSION
Overflow of rubber dams is characterised by an accelerating
nappe which adheres to the membrane wall. Hydrodynamic
and fluid-structure instabilities may develop and must be
avoided. Current design techniques include the installation of
a fin an the upper quadrant of the rubber dam, to deflect the
flow away from the flexible membrane wall.
Analytical calculations have been developed to predict the
nappe trajectory above rubber dams, including the free-
falling jet. The results have been compared successfully with
experimental data. The ideal-fluid flow calculations may be
used by designers to predict the nappe trajectories and to
prevent nappe re-attachment situations. The new laboratory
experiments and calculations (CHANSON 1996) have been
used also to show that the optimum location of the deflector
is 30
def
60 degrees.
A major limitation of the present work is the assumption of
steady flow and boundary conditions. Rubber dams are made
of flexible membranes which respond to hydrodynamic
fluctuations (e.g. wall turbulence). The shape of the rubber
dam is also a function of the upstream head and this was
ignored.

ACKNOWLEDGMENTS
The author acknowledges the assistance of his former
students R. McCONAGHY, A. SWINCER, E. SHAW and T.
SIAW.

REFERENCES
ANWAR, H.O. (1967). "Inflatable dams." Jl of Hyd. Div.,
ASCE, Vol. 93, No. HY3, pp. 99-119.
BINNIE, G.M., THOMAS, A.R., and GWYTHER, J .R.
(1973). "Inflatable Weir used during Construction of Mangla
Dam." Proc. Instn. Civ. Eng., Part 1, Vol. 54, Nov., pp. 625-
639. Discussion : 1974, Part 1, Vol. 56, pp. 189-194.
CHANSON, H. (1996). "Some Hydraulic Aspects during
Overflow above Inflatable Flexible Membrane Dam." Report
CH47/96, Dept. of Civil Engineering, University of
Queensland, Australia, May, 60 pages.
CHANSON, H., and MONTES, J .S. (1997). "Overflow
Characteristics of Cylindrical Weirs." Research Report No.
CE154, Dept. of Civil Engineering, University of
Queensland, Australia, 96 pages.
FEKETE, G.I. (1963). "Coanda Flow of a Two-
Dimensional Wall J et on the Outside of a Circular Cylinder."
Report No. 63-11, Dept. of Mech. Eng., McGill University,
Canada.
OGIHARA, K., and MARAMATSU, T. (1985). "Rubber
dam : Causes of Oscillation of Rubber Dams and
Countermeasures." Proc. 21st IAHR Congress, Melbourne,
Australia, pp. 600-604.
SARPKAYA, T. (1968). "The Deflection of Plane
Turbulent J ets by Convex Walls." US Govt Research Report
No. AD 673249, Naval Postgraduate School, Monterey CA,
USA.
SHEPHERD, E.M., McKAY, F.A., and HODGENS, V.T.
(1969). "The Fabridam Extension on Koombooloomba Dam
of the Tully Falls Hydro-Electric Power Project." Jl Instn. of
Eng., Australia, Vol. 41, pp. 1-7.
WU, P.H., and PLAUT, R.H. (1996). "Analysis of the
Vibrations of Inflatable Dams Under Overflow Conditions."
Thin-Walled Structures, Vol. 26, No. 4, pp. 241-259.