Você está na página 1de 13

Hydraulic performance and stability of coarse rockll deposits

D. Hansen PhD, PEng, MCDA, W. Z. Zhao BEng and S. Y. Han MASc


Open-pit coal mining operations often generate enormous
quantities of waste rock. This is dumped into nearby
valleys and the coarsest fraction ends up at the bottom of
the deposit; the nal result is often referred to as a rock
drain. These ow-through rockll drains can be 100 m
high, thousands of metres long, and usually have a stream
owing through their base. When the streamthat supplies
the rock drain oods, the ability of the drain to convey
water is exceeded by its supply, and a large but temporary
pool develops at the upstream end of the dump. More
importantly, the free surface within the rock drain
becomes elevated, as does the point of exit at the
downstream face. Because of historic instances of failures
at the downstream face, the management of these buried
streams and the hydraulic behaviour of the rock drains
through which they ow are of interest. The behaviour of
homogeneous embankments comprised of coarse rockll
has many similarities to the behaviour of long rockll
drains. Methods for assessing the hydraulic performance of
either are presented herein. Denition of the complete
water surface envelope and the tendency toward
unravelling of the downstream face receive primary focus.
It is demonstrated that the hydraulic gradient directly
beneath the seepage face and small variations in its
direction play a far greater role than seepage-face overow
in promoting unravelling failure, and argues that all such
deposits are, in effect, embankment dams that should fall
under the applicable national dam safety regulations.
1. INTRODUCTION
In the Kootenay region of south-eastern British Columbia,
Canada, there are many open-pit mines. It is common for the
waste rock at these mines to eventually cover natural stream(s)
that ow through the site (see Figs 1 to 3). These deposits of waste
rock are sometimes referred to as rockll drains. The
downstream face of a rockll drain that had been built at an
open-pit coal mine failed quite suddenly in the mid-1990s. Its toe
may have moved in direct response to ow from a runoff event
that was occurring at the time. In any case, the stage of the stream
was high and rocks with a diameter of about 2
.
5 m that had been
placed over the toe of the drain to prevent this type of failure
kicked out, without warning. They rolled for hundreds of metres,
all over the mine site. An engineer who was doing eld
investigations of rock drains at the time felt the ground shake
while he was quite some distance from the mine, and thought it
was an earthquake. Fortunately, no one was injured, but the mine
was closed for about a week. The stream owing through the
dump was shallow and only a few metres wide in its ordinary
mid-summer condition.
Rockll drains such as the one shown in Fig. 2 are unlikely to be
overtopped unless the watercourse blocked by the waste rock is a
large one. In such cases, approval for the location of the dump
would probably not be granted in the rst place. Given that these
deposits are built up over time it is conceivable that a relatively
new deposit of waste rock could be overtopped by the
watercourse, and as such, exhibit failure that is like that of an
embankment dam, or of a so-called ow-through rockll dam.
Further, the development of a dangerous situation does not
require actual overtopping. Previous experience indicates that
particles on the downstream face of a deposit of coarse material
that is undergoing signicant through-ow are subject to very
signicant seepage and overow forces, whether they make up
the toe of a rockll drain or of a ow-through rockll dam. This
paper consists of the development a general understanding and
framework for designing, or evaluating the hydraulic behaviour
of, embankment-like deposits of homogeneous coarse rockll.
This design framework is based upon (i) adaptation and synthesis
of accepted seepage and uid mechanics fundamentals, (ii)
detailed modelling efforts, previously described, and (iii) limited
experimental data in a realm that is still the subject of some
research. The foregoing are combined using reasonable extremal
conditions so as to provide the needed design guidance. In doing
so, the relative importance of the destructive forces, as well as
certain research needs, become evident.
2. REGULATORY ISSUES
Section 1.1 on Scope and Applicability of the 1999 Canadian
Dam Association (CDA) Guidelines,
1
currently undergoing a full
review and update, states that
These guidelines apply in general to dams, as dened in section 1.2,
that are at least 2
.
5 m high and which have at least 30 000 m
3
of
reservoir capacity.
Although the phrase full supply level that is associated with
reservoir capacity has little meaning in this context, it is quite
conceivable that a pond 32 m wide by 375 m long (upstream) by
average depth 2
.
5 m could temporarily form on the upstream side
David Hansen
Associate Professor of Civil
Engineering, Dalhousie
University, Halifax, Canada
Wen Zhong Zhao
Research Assistant, Dept of
Civil Engineering, Dalhousie
University, Halifax, Canada
Sang Yoon Han
Research Assistant, Dept of
Civil Engineering, Dalhousie
University, Halifax, Canada
Proceedings of the Institution of
Civil Engineers
Water Management 158
December 2005 Issue WM4
Pages 163175
Paper 13872
Received 27/05/2004
Accepted 20/09/2005
Keywords:
embankments/oodworks/safety
and hazards
Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al. 163
of a waste rock dump (for example), and possibly every spring.
The volume within this pool would ordinarily be small if the
drainage area is small, as is often the case for streams in
mountainous areas. In the UK, Institution of Civil Engineers (ICE)
designates inspect all dams that fall under the Reservoirs Act of
1975.
2
This Act pertains to dams that are at least 15 m high and/
or create reservoirs with a capacity of at least 25 000 m
3
. About
2000 dams in the UK therefore require regular inspection under
the Act. Although it has provisions for non-impounding
reservoirs, it does not have specic provisions for managing
the risk of failure of coarse rockll ow-through structures.
Section 8 of the 1995 CDA Dam Safety Guidelines states the
following requirement for ow-through rockll dams
3
Flow-through rockll dams shall be able to withstand, without local
movement or massive movement of rock particles, the combined
effects of the action of the seepage emerging from the downstream
face, along with any overow.
This section also indicates that ow over the crest of such
structures is not recommended unless the downstream slope is
atter than 1V:1
.
5H. Section 8.8 of the 1999 CDA Guidelines,
1
states
Flow over the crest of such structures is not recommended unless the
downstream slope is designed to limit erosion. Allowance must be
made in the design for the possible accumulation of debris on the
upstreamface of the dam, over time. If the debris is not removed it will
reduce the quantity of ow passing through the embankment and
correspondingly increase the ow over the embankment.
Given the unusual aspect ratio of rock drain deposits, it will be
very unlikely that one would ever be overtopped. On the other
hand, because the hydraulic capacity of even very coarse rockll
is low compared to an open channel, the possibility of the
overtopping of a deposit that only has a short downstream length
is very real.
The schematic in Fig. 4 is greatly exaggerated in the vertical.
Because these deposits are generally built up by end-dumping,
the original upstream and downstream faces tend to be at the
natural angle of repose, with the largest particles at the bottom.
An attempt is often made to buttress the downstream face with
very large particles that are not necessarily placed at the angle of
repose. It can be seen that particles under the seepage face of
height y
exit
are subject to overow, and that particle P
2
is subject
to nearly all the ow being conveyed through the structure, and
as a ow over the particle. If rock waste dumps at an open-pit
mine site can be considered to be ow-through rockll dams
from an engineering perspective, their hydraulic and
geotechnical performance is then governed by CDA Guidelines
section 8 (for example).
This paper focuses on the behaviour of such deposits with regard
to seepage rates, the emergence of this relatively high rate of
seepage on the downstream face, and the question of which rock
particle is the uppermost unstable particle.
Fig. 1. Line Creek, Kootenay Mountains, British Columbia,
Canada, before coverage by millions of tonnes of coarse
material (photo: British Columbia Dept of Energy & Mines)
Fig. 2. Line Creek, Kootenay Mountains, British Columbia,
Canada, partially covered with waste rock; eventual length
2500 m (photo: British Columbia Dept of Energy & Mines)
164 Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al.
3. DESIGN ISSUES
Figure 4 raises three important design questions. First, at what
gross volumetric rate will water be able to move through the
deposit? The ow will be non-Darcy, and although
computationally trivial to estimate, without this rate the next two
questions cannot be addressed. Second, what will the height of
the so-called seepage face be? Third, which particle within the
seepage face (Fig. 4) will be the uppermost unstable particle, P, P
1
or P
2
? Particles P
1
and P
2
occupy the upper-limiting and lower-
limiting positions on the seepage face respectively and have
certain features with respect to the nature of the overow and
seepage to which they are subjected (as will be discussed).
Particle P is located anywhere between these two extremal
locations. The third question speaks directly to the safety
concerns expressed in various dam safety guidelines.
The possibility of a gross rotational failure of the downstream
slope also exists, but is not discussed herein. The rst author has
found evidence that traditional rotational failure methods of
analysis may be ill-suited to massive and/or deep-seated failures
in ow-through rockll structures. A triangle at the toe seems to
kick out by seepage forces in such structures, with the help of the
simultaneous collapse of the material behind it. Garga et al.
4
have
presented a detailed means of analysing this particular failure
mechanism.
Waste dumps at open-pit mine sites are not designed in the
ordinary sense of the word. They are the outcome of the
successive end-dumping (over considerable distances) of very
large amounts of blasted material. The word design is used herein
only because steps can be taken to try to inhibit certain modes of
failure, such as the placement of very large particles on the
downstream face.
4. THEORETICAL CONSIDERATIONS
4.1. Gross volumetric ow rate
The approach that can be taken is analogous to the sizing of an
open channel for uniform ow using Mannings equation.
Hansen et al.
5
considered six non-Darcy ow equations and
discussed the selection issue for these kinds of applications. For
the boundary conditions relevant to the problems considered
herein (all having a free surface and therefore all having
modest hydraulic gradients), most of these equations behave in
a manner similar to that of Wilkins,
6
the relationship favoured
by the mining industry due in part to their quantitative
experience with it (Piteau
7
). It is dimensionally unbalanced,
applicable to cases of nearly fully developed turbulence, and may
be stated as
V
v
Wm
05
i
054
1
where V
v
is void velocity (Q/(nA) (m/s)), n is porosity, W
is Wilkins empirical constant (5
.
243 (L
05
/T)), m is the
hydraulic mean radius (m), i is the hydraulic gradient
(dimensionless), L is length and T is time.
The so-called hydraulic mean radius is a representative pore
diameter that can be estimated using (Garga et al.
8
)
m
e d
r
e
6
2
where e is the void ratio (dimensionless), d is particle diameter (L),
r
e
is particle surface-area efciency (about 1
.
3 for coarse angular
rock (dimensionless)).
Due to the fact that the friction slope and the bed slope will,
in general, have a similar range of values in this application
(the completed dump being about 20 times longer than it
is high), the bed slope can be used with equation (1) to
make a preliminary estimate of the ability of the drain
to convey water.
Expressions like equation (1) are often stated as
i aV
N
v
3
Fig. 3. Composite photo of lower edge of waste dump. Note
man in lower right and distinct vertical gradation (photo: British
Columbia Dept of Energy & Mines)
Phreatic
surface
toe
P
2
P
P
1
y
exit
y
+
x
Fig. 4. Schematic of ow-through rockll drain (vertically
exaggerated, and particle size actually decreases with the height
y above the bed)
Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al. 165
in which case the normal depth may be found using the following
equation (see Appendix B). It is employed by nding that depth
(y
n
) which causes the left-hand side to equal the right-hand side.
A
Q
n
a
S
o
_ _
1=N
4
Q is the design discharge (L
3
/T), A is the cross-sectional area of
the ow associated with normal depth y
n
porous
(L
2
), and S
0
is the
bed slope (dimensionless).
A at parabola is often a reasonable representation of the
in-bank geometry of many streams. Such approximations are
also made because the stream in question may well have been
buried without ever having been surveyed. As shown in
Appendix B, the normal depth in a parabola (y lx
2
b
) with
the ow governed by equation (1) is
y
nporous

Q
WS
054
0
3

l
p
4 n m
05
_ _2=3
5
where l is an empirical coefcient describing the parabolic
shape.
Since the phreatic surface is analogous to an M2 water surface
prole for pure open-channel ow(Dake
9
), the depth found using
either equation (4) or (5) is asymptotically approached in the
upstream direction. Given that the energy gradient at the
upstream end of the deposit will, in general, be somewhat greater
than S
0
for M2-analogue conditions, the results of applying
equations (4) or (5) represent upper limits on entry depth. For
ow-through rockll dams the typical distance upstream of the
point of emergence is not nearly enough to allow the porous
media y
n
to be realised.
4.2. Exit height of point-of-emergence of seepage face
Applying Wilkins
6
equation (1) to the idea that the exit gradient
that acts beneath the seepage face and within the zone of the toe
may be adequately approximated by the angle u of the toe
(Taylor
10
) yields that the exit height y
exit
for a rectangular
cross-section is (and since Q VA nV
v
y
exit
L)
y
exit

Q
n LWm
05
(tanu)
054

q(cot u)
054
n Wm
05
6
It may be expected, however, that the true angle representative
of the emergent seepage eld is less than u, that this effective
ow angle u
ff
varies with the relative exit height, and that
it approaches u as the relative exit height increases, as in
Fig. 5. For non-Darcy ow in a crushed limestone, Hansen
11
found that
u
ff
u
141
y
exit
H
017 r
2
08 7
where H is height of the dam, u is angle at the downstream toe
and u
ff
is angle of the emergent ow eld within the toe.
The above equation was found to work best if y
exit
, 0
.
5H.
Substituting u
ff
from equation (7) into equation (6) and
rearranging
q n Wm
05
y
exit
tan 141u
y
exit
H
017u
_ _ _ _
054
8
Fig. 6 shows how equation (8) performed for a series of observed
exit heights measured in the hydraulics laboratory of the
University of Ottawa.
11
(The non-Darcy ow parameters were
found fromcolumn tests on the specic rock material in question,
with the result being a power function with parameters similar to
those arising from Wilkins equation.) The model embankments
in question were indeed much more like embankments in their
aspect ratios (height H divided by length in direction of ow)
than rock drains.
Another approach to nding the exit height is to use the analogy
with critical depth y
c
in a truly open channel.
12
The exit height at
a freefall for a rectangular cross-section lled with a porous
medium having a vertical downstream face is
y
cporous
k
1
aq
2
g cos
2
b
0
n
2
_ _
1=3
9
where a is kinetic energy correction coefcient and b
0
the angle
of the stream bed above the horizontal (cos
2
b
0
often close to 1).
Similarly, the exit height from a vertical face for a buried channel
having a parabolic cross-section is (Appendix A)
y
cporous
k
2
27alQ
2
32 g cos
2
b
0
n
2
_ _
1=4
10
where k
1
and k
2
are empirical adjustment factors.
High phreatic surface
High Q
At high flow
approaches q q
ff
Foundation
toe
Low phreatic surface
At low flow
<< q q
ff
Foundation
toe
q
q
ff
q
q
ff
Low Q
Fig. 5. Illustration of concept of angle of emergent ow-eld, u
ff
Computed
Observed
12 0
9 0
6 0
3 0
0 0
D
i
s
c
h
a
r
g
e
:
/
s
L
Exit height: cm
0 0 5 0 10 0 15 0 20 0 25 0 30 0 35 0 40 0
1V:1H
1V:2H
1V:3H
Fig. 6. Performance of exit-height-based rating curve for three
model dams
166 Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al.
Experiments indicate that depths computed using (9) with k
1
1
are lower than the actual exit height, y
exit
. Bari
13
reported values
of k
1
of between 2
.
5 and 3 for deposits having a vertical
downstream face. Further, the existence of rockll downstream
of the exit height cross-section might be expected to contribute
somewhat to a correction factor. By comparing equations (9) and
(8), the variation of the correction factor k
1
over a range of
relative depths was inferred from the exit-height rating curves
presented in Fig. 6. It is not presently known whether the pattern
of corrections shown in Fig. 7 would also be applicable to k
2
but
such information would be useful, especially in the rockll drain
context (because the buried cross-section must often be
approximated). However, this is clear evidence that the exit
height for dams with sloping downstream faces can be estimated
using the critical depth analogy, but with a correction factor that
reaches a limiting value of between 3 and 5, and which depends
on the slope of the downstream face. It must be noted that if the
imposed depth on the downstream face is greater than that which
would occur in a freefall situation (as might be computed by
equation 10), the imposed depth will govern.
Johnson
14
and Kluber and Breth
15
both considered the exit
height of the partially completed Hellhole Dam in California,
USA, just before it failed in December 1964. Table 1 presents the
basic information found in these articles.
If the assumptions in Table 2 are made then application of
equations (2) and (8) yields q 4
.
6 m
2
/s, which corresponds to
Q 705
.
2 m
3
/s. This is slightly larger than the reported ow at
incipient failure of Q 679
.
6 m
3
/s.
The most upstream depth is connected to the exit height (or to the
imposed tail-water level if the latter is greater) by a curve known
as the phreatic surface. It is of interest partly because it represents
the initial spatial condition for the numerical modelling of any
unsteady (wave-like) water-surface proles travelling through
the deposit. Such perturbations are of hydrologic cause and begin
at the upstream end of the deposit. The modelling of these
translatory waves is outside the scope of this paper. The
behaviour of such temporary waves within coarse rockll has
been described by Stephenson
12
and is the subject of on-going
research by the authors. The steady-state phreatic surface
through a series of stream cross-sections of arbitrary individual
geometry can be modelled quite accurately using dedicated
algorithms.
16
Alternatively, for channels having cross-sectional
geometries that can be described by a single power function, the
phreatic surface can be approximated by
u
1
y
sN2s
u
2
y
sN1
u
3
x 11
where x is distance from the point of emergence (L), y is height of
phreatic surface above the bed (L), and u
1,2,3
are parameter groups
which include cross-sectional geometry descriptors, a level of
turbulence index, the ow rate, porosity, and so on (see
Appendix C).
Comparison of equation (11) with post-correction for typical
bed-slope drops (S
0
less than 5%) against ne-scale numerical
solutions of the unit width gradually varied water-surface
equation showed that the error in the prole was acceptably
small (only centimetres) as long as the length of the prole was
less than about 50 m. The error was found to become more
signicant (metres) if the longitudinal length was increased to
hundreds of metres. Equation (11) is an approximate but direct
approach for getting the initial condition needed for unsteady
ow calculations. Given the uncertainties in the hydraulics of the
downstream face and indeed of the coarse porous media itself, the
fact that equation (11) gives only an approximate water surface
prole if correction is made for the bed slope after its application
is not considered fatal to the goals considered herein. If the role of
bed slope is important and/or the structure is long, an adaptation
of the standard step method can be used to compute the water
surface prole, as laid out by Table 7 (Appendix C), the
uncertainties in which are described by Hansen and Bari.
17
Figure 8 presents the phreatic surface so computed for Hellhole
Dam on the Rubicon River (local bed slope not reported). As
can be seen, although the agreement between computed and
observed exit heights (from equations 2 and 8) is very good
(computed/reported 0
.
98), the agreement between the
inferred upstream depth using equation (11) with zero bed
slope and the reported upstream depth at failure is moderate
Slope of downstream toe 1V:1H 1V:2H 1V:3H
5 0
4 0
3 0
2 0
1 0
0 0
I
n
f
e
r
r
e
d
k
1
Exit height/total height of dam
0 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8
Fig. 7. Variation in k
1
over a range of relative downstream
depths (and sloped downstream face)
Parameter Reported value
Nominal size of crushed rock 127 mm (5 inches)
Angle of toe 37
.
68 (1V:1
.
3H)
Height

of dam 67 m (220 ft)


Reservoir depth

at failure 44
.
8 m (147 ft)
Exit height

at failure 29 m (95 ft)


Width

of ow at failure 152
.
4 m (500 ft)
Flow-through at failure 679
.
6 m
3
/s (24 000 ft
3
/s)

Above foundation (elevation 1295


.
4 m)

Approximately trapezoidal (Johnson


14
), average from Kluber and
Breth
15
Table 1. Hellhole Dam characteristics and failure data
Parameter
Assumed
value Notes
Porosity n 0
.
4 Material reported as being dumped,
this n reasonable
r
e
1
.
3 Material reported as crushed rock,
this r
e
reasonable
Table 2. Assumptions about porous media of Hellhole Dam
Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al. 167
(computed/reported 0
.
92). The discrepancy may be due
to uncertainties and/or assumptions made about the
porous medium, the complexity of the full geometry of
this partially completed dam (an incomplete second tip is shown
in Johnson
14
) and the possibility that a surge
was taking place at the time that failure conditions were
being recorded.
4.3. Uppermost unstable particle
The uppermost unstable particle resides somewhere within the
seepage face. There are three primary forces relevant to this
particle: the overow hydraulic forces, the seepage force and
the submerged weight of the particle (see Fig. 9).
The destabilising hydraulic forces, F
hyd
, associated with the
overow acting on a single particle that is located within the
seepage face include the shear force, the drag force, and the
hydrodynamic lift. The coefcient C
D
in the following standard
expression includes the effects of shear
F
hyd
C
D
rA
U
2
2
12
where C
D
is the drag coefcient (dimensionless, see Fig. 10), A is
the area of forward projection of object (L
2
), U the velocity of
uniform innite ow eld impinging upon object (L/T) and r is
the uid density (1000 kg/m
3
in this case).
The hydrodynamic drag force acting in the direction of the ow
associated with half of case (a) (Fig. 11) will obviously be greater
than the force associated with case (b), which in turn will be
greater than the drag force for case (c). If it can be shown that the
contribution of the force associated with half of case (a) is small
compared to another particle-destabilisation effect considered
herein, it might be concluded that this is an acceptable method of
estimating the contribution of overow to particle destabilisation
within a seepage face, especially for design purposes. Fig. 11(d)
depicts a situation closer to the realities under analysis. It too
must be associated with less individual particle drag than cases
(a) or (b), but its relationship to case (c) is less clear. The absence
of the no-slip condition (a condition that is nearly universal in
initiation-of-motion analyses found in standard literature)
should, however, lessen the drag below that of case (c). This is
because the separation zones downstream of each particle have
relief from negative pressures via the voids in the bed (the
downstream face of the dam), thus reducing the pressure
differential across a given particle.
It is difcult to nd clear guidance on hydrodynamic lift effects.
Some researchers have reported a negative lift (that is, particle
pushed into the bed) as a result of investigations on the initiation
of motion problem (Watters and Rao
19
), while others have
reported a positive lift (Stevens and Simons
20
). Hydrodynamic lift
on a rough and irregular semi-sphere would probably not be
continuously positive (outward) in this context. Further, one
would expect that the presence of the many voids in the plane of
the bed would tend to prevent the occurrence of negative
pressures in the boundary-layer separation zones because
drawing water out of these voids would relieve such pressures. No
effort was therefore made to quantify possible hydrodynamic lift
effects (positive or negative).
The force required to hold a sphere in a free stream is highly
variable, but for characteristic dimensions between 0
.
1 and
1
.
0 m, and velocities between 0
.
5 and 1 m/s, the Reynolds
number varies from 5 10
3
to 1 10
6
, and the drag coefcient
varies as shown in Fig. 10.
The approach taken herein for assessing the possibility of the
initiation of motion of one particle is based on consideration of
how the various forces result in destabilising (or stabilising)
moments (see Fig. 12).
Computed phreatic surface
Reported
* above elevation 1295 4m
y
exit
80
70
60
50
40
30
20
10
0
H
e
i
g
h
t
*
:
m
Distance upstream: m
0 50 100 150 200
Fig. 8. Cross-section of Hellhole Dam, reported by Johnson,
14
with water surface envelope (S
0
assumed 0)
0 6
0 5
0 4
0 3
0 2
0 1
0 0
C
D
Re ( / ) Ud n
0 200 000 400 000 600 000 800 000 1 000 000
Fig. 10. Drag coefcient for sphere, Reynolds number between
5 10
3
and 1 10
6
(adapted from Giles
18
)
F
hyd
W
sp
F
s
spatially vaired
overflow
Emergent seepage
Non-Darcy seepage
q toe
Fig. 9. Forces acting on a single particle located on the
downstream face
168 Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al.
If it is assumed that half of the particle in question is completely
exposed, the length of the destabilising moment arm could not be
more than d/2. For overow occurring essentially parallel to the
downstream face, an estimate of the destabilising moment is then
M
destab1

d
2
F
hyd
13
A moment-arm length of d=2 in Eq. (13) is conservative from a
design point of view.
The seepage face of a ow-through rockll structure is an
example of spatially varied ow under high relative-roughness
conditions, making it difcult to relate discharge and depth.
Further, the intersection of the rockll drain with the valley side
walls will often be parabolic, meaning that the available channel
width will decrease with decreasing elevation. If the downstream
slope is very at and the ow in the receiving channel is
subcritical, the depth at the toe could be legitimately estimated
using the normal depth of the receiving channel. This will rarely
be the case because the downstream face will be hydraulically
steep, resulting in spatially varied ow that is supercritical over
most of the length concerned,
21
the conditions for which have
been described by Li.
22
The presence of a hydraulic jump where
the supercritical ow meets the sub-critically-owing stream is
itself a safety concern with respect to the stability of the particles
at the toe.
It is suggested that estimation of U may be approached by
assuming that it varies from zero at the point of emergence at the
top of the seepage face up to a conservatively high value at the
toe. Table 3 presents extreme outcomes for F
hyd
.
Regardless, F
hyd
is evidently not nearly as important as F
seep
. The
destabilising seepage force is given by
F
seep
8
P
(1 e)g
w
i 14
where 8
P
is the volume of the particle in question (L
3
), g
w
is unit weight of water (F/L
3
) and i is hydraulic gradient
(dimensionless).
The authors are not aware of any previous statement of equation
(14), but it is based on the accepted fact that the force exerted on a
unit bulk volume of porous medium is the product of the unit
Isolated sphere,
infinite flow field
Isolated hemisphere
within boundary layer
Bed of hemi-spheres
with overflow
Spheres under seepage-face
accretion
(a) (b)
(c) (d)
Fig. 11. Four drag-force scenarios for spherical particles
Exit angle
Mag. of vector
x
50
45
40
35
30
25
20
15
10
5
0
A
n
g
l
e
:
x
Downslope distance/Length of seepage face: %
0 20 40 60 80 100
0
.
9
0
.
8
0
.
7
0
.
6
0
.
5
0
.
4
0
.
3
0
.
2
0
.
1
0
M
a
g
n
i
t
u
d
e

o
f

h
y
d
.
g
r
a
d
i
e
n
t

v
e
c
t
o
r
Fig. 13. Behaviour of surcial exit gradient for model dam
with u 458, 0% corresponds to exit point, 100%
corresponds to toe
Parameter Low High
U (m/s) 0
.
1 3
.
0
r
e
1 10
4
6 10
6
C
D
0
.
4

0
.
2 ?
d (m) 0
.
1 2
.
0
A (full circle) (m
2
) 0
.
004 1
.
571
Nominal F
hyd
(N) 0
.
016 2827
.
4
Half of F
hyd
(N) 0
.
008 1413
.
7
Moment-arm (m) 0
.
025 (d/4) 1 (d/2)
M
destab1
(Nm) 0
.
0002 1413
.
7

see Figure 10
Table 3. Range of possible effects of ow over seepage face
Point of
rotation
toe angle Stream-bed or foundation of dam
Moment-arm is sin (both cases)
2
x d
F
seep
, particle P
F
seep
, particle P
1
x
15
90
30
60
d
2
q
Fig. 12. Idealisation of particle position on downstream face
Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al. 169
weight of the uid moving through that medium and the
applicable hydraulic gradient (such as Holtz and Kovacs
23
).
The term 8(1 e) represents the bulk volume occupied by a
particle within the porous medium in question. In this
case, the particle is actually at the edge of said media and
must be of a representative size, from a design point of
view. Given that in this context there will be a ow
separation zone on the downstream side of each such
particle (a separation that would not exist if it were
completely enclosed by more particles on its downstream
side), it may be that the force given by equation (14) is
an underestimate of the true force. The applicable
hydraulic gradient can be estimated via formal numerical
modelling of the non-Darcy ow eld but in and near the toe
would not usually exceed a value of about 1
.
0. If a precise pattern
of surcial heads is needed so as to obtain more precise
exit gradients and their associated directions within a toe
having a particular geometric conguration, said numerical
modelling is indeed indicated. Given that the modelling of
such non-Darcy seepage is not widely accessible as a
software-based undertaking, it is of interest to note that the
outcomes, as ow-nets, do not appear to be very different
from conventional ow-net outcomes in most cases.
24
The seepage path can be expected to act at the angle of the
downstream face at the uppermost point of emergence on the
seepage face, but theoretically cannot exceed 458.
25
For this case,
the seepage force is therefore at its highest (unity) but if
u 458 the moment-arm length is zero, therefore tending to
negate the effect of the seepage force at this location. The
intermediate situation occurs when the seepage force acts at
an angle j (measured back from the plane of the downstream
face, see Fig. 12). Fig. 13 is a typical outcome of the detailed
modelling of the complete variation in hydraulic head within an
embankment undergoing non-Darcy ow.
11
This result is very typical; the surcial exit-gradient magnitude
approaches unity as its direction becomes more horizontal.
Table 4 indicates the typical limits of action for this seepage
vector. It also includes the possibility of an undamped surge
arriving at the downstream face.
The moment arm for the seepage vector is
L
d
2
sin j 15
where j is the angle for the seepage vector, measured below the
plane of the downstream face (see Fig. 12). The destabilising
moment is then
M
destab2
F
seep
d
2
sinj 16
and the maximum value of j is u.
With regard to the submerged weight of particle, for the
arrangement shown in Fig. 12 this stabilising force will act
using a moment arm of length
L
d
2
cos u 17
The stabilising moment is then
M
stab
W
sp
d
2
cos u 18a
where, if the particle is a sphere of diameter d, its submerged
weight is
W
sp
(g
p
g
w
)8
P
(g
p
g
w
)
pd
3
6
18b
The term g
p
2g
w
is about 16 kN/m
3
.
Table 5 presents extremal outcomes for this force. The
destabilising effect of the seepage force is evidently about 36 to
378 times larger than the overow effect, for particles on the exit
face in this engineering context. This is non-intuitive.
Given that the arrangement of particles on the downstream face
is quite erratic, it will be both credible and conservative to
assume that no constraint is provided by particles in the same
plane as the particle in question. This absence-of-constraint
assumption is also appropriate from a safety perspective because
a constrained particle could be struck by an unconstrained
particle that has become dislodged; this self-promoting process is
sometimes referred to as unravelling. The overall factor of safety
(FS) of the particle of interest is then
FS
stab. moment

destab. moments

W
sp
d=2 cos u
05F
hyd
d=2 F
seep
d=2 sin j
or: 19
FS
W
sp
cos u
05F
hyd
F
seep
sin j
Parameter
Top of
seepage face At the toe
Exit hydraulic gradient, i 0
.
500
.
75 0
.
751
.
0

n (e) (dimensionless) 0
.
33 (0
.
493) 0
.
43 (0
.
754)
d (m) 0
.
10 2
.
0
j 158

458
F
seep
(N) 5
.
7 71,937
M
destab
2
(Nm) 0
.
1 50,867

Normally about 0
.
75

Zero degrees results in a moment of zero Nm


Table 5. Approximate upper limits on seepage force and
associated destabilising moment
Location Direction Magnitude
Top of seepage face

If coincident

with
downstream face, j 0
1
.
0
Top of seepage face If not coincident with
downstream face, j % 0
0
.
5
At the toe

Horizontal, j u of toe 1
.
0

Particle P
1
at seepage breakout point, u 458

As may happen if a severe surge arrives relatively undamped.

Particle P
2
Table 4. Ranges for hydraulic gradient vector for seepage force
calculations (based on numerical analyses of steep dams
presented in Townsend et al.
24
)
170 Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al.
5. OUTCOMES
Figure 14 presents the variation in the FS for the input data
presented in Table 6. The insensitivity of FS to the overow
hydraulic forces as indicated by this gure is consistent with the
results of Tables 4 and 5.
If the angle of action of the seepage force is taken to be
horizontal, j 458 and this j maximises the moment-arm
length. Further, if F
hyd
0, equation (19) reduces to
FS
g
P
g
w
(1 e)g
w
i
20
Fig. 15 was generated using equation (20). The authors are not
aware of data from full-scale dams or large-particle ume studies
that would help to verify Figs 14 or 15. Hansens model
11
dam
tests with d
50
25 mm showed that initiation of motion occurs
part-way down the seepage face (for example, a particle P in
Fig. 5, and one that is nearer to P
2
than to P
1
). Hansen observed
that the top of the zone of initiation of motion started at between
one-third and two-thirds of the way down the seepage face, with
the latter associated with the atter toe slopes. Assessment of
initiation of particle motion is difcult because it is known to
have an inherently random component to its behaviour (for
example, Neill and Yalin
26
). Equations 19 and 20 indicate (i) that
if the surcial exit gradient approaches unity and simultaneously
acts nearly horizontally (both of which are more true as one
approaches the toe itself), individual particles on the surface of
the downstream face become unstable, and that (ii) prima facie,
rockll deposits with high porosity are more unstable under these
same conditions. If n 0
.
4, SG 2
.
65, and i
extreme
1
.
0,
equation (20) gives FS 0
.
99. Any value of FS only just above
unity should also be considered unsafe because of the turbulent
nature of the ow.
It must also be pointed out that the true effect of increasing
porosity is more complex because it will be associated with
different boundary conditions, through equation (1), via m in
equation (2). For example, although Fig. 15 indicates that
n 0
.
35 would be safer than n 0
.
43, the former would result in
a seepage face that is 1
.
16 times greater in vertical extent, in
accordance with equation (6).
6. CONCLUSIONS
Rockll drains should be regulated by national dam safety
guidelines. With regard to the three questions posed at the onset
as to how such embankments perform, the following can be said.
(i) The ow is non-Darcy, and the associated overall
volumetric ow rate can be readily estimated using Wilkins
equation
6
or similar. If the normal depth occurs at the
upstream point of entry (which is conceivable in the case of
rock drains because they are thousands of metres long), an
upper limit on this depth may be estimated using equation
(5).
(ii) The height, y
exit
, of the highest point of emergence on the
seepage face can be estimated using the concept of the
angle of the emergent ow eld. This height is generally at
least three times greater than height arising from the porous
media analogue for the critical depth in an open channel.
The elevation of intermediate water levels can be
determined by modelling,
16
or approximately by equation
(11) with correction for the drop in bed elevation (as long as
the structure is not more than about 50 m in upstream
downstream length).
(iii) It is not denitively clear which particle within the seepage
face of a full-scale dam will always be the uppermost
unstable one, but it will generally not be the particles at the
limiting locations (P
1
or P
2
). It will be the particle under the
seepage-face accretion that has the greatest destabilising
moment that is induced by the exiting seepage, and this
tends to be in the lower half of the toe. In this regard, the
direction of the exit seepage is very important. If precise
exit-gradient directions are needed for a given toe, formal
numerical modelling is indicated. In lieu of this, the
conservative assumption can be made that the surcial
seepage vector acts horizontally with magnitude of unity.
i =1 00, =5 x
i =0 92, =15 x
i =0 83, =30 x
i =0 75, =45 x
9 0
8 0
7 0
6 0
5 0
4 0
3 0
2 0
1 0
0 0
F
S
U: m/s
0 0 0 5 1 0 1 5 2 0
Fig. 14. Variation in FS of one particle, d 1
.
5 m, on
downstream face under a range of seepage-face conditions
2 0
1 8
1 6
1 4
1 2
1 0
0 8
F
S
Exit hydraulic gradient, i
0 5 0 6 0 7 0 8 0 9 1 0 1 1 1 2
n=0 35
n
n
n
n
=0 37
=0 39
=0 41
=0 43

Fig. 15. Effect of porosity and exit gradient on FS of particles on


downstreamface with zerohydraulic overowforces and j 458
Parameter Value
Angle at toe of dam, u 458
Particle diameter, d 1
.
5 m
Porosity 0
.
4
Particle specic gravity 2
.
65
Table 6. Input data for calculation of moment-based Factor of
Safety of one particle
Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al. 171
7. APPENDICES
7.1. Appendix A. Normal depth for open-channel
non-Darcy ow
The relationship between depth and cross-sectional area for a
parabolic section is shown in Fig. 16.
If uniform ow is occurring i S
0
and y y
n
. Non-Darcy ow
equations are commonly expressed in the form i aV
v
N
. Then
V
v

i
a
_ _
1=N
with V
v

Q
nA
yields:
Q
nA

S
0
a
_ _
1=N
so
A
Q
n
a
S
0
_ _
1=N
21
where the desired y
n
is within A, and the form of A against y
depends on the cross-sectional geometry of the stream. For a
channel boundary that is dened by a parabola y
b
lx
b
2
with l
being a small number for stream beds that are rather wide and
at. For parabolas it is also true that
A
4
3

l
p y
3=2
22
Wilkins equation
6
may be written Q AWnm
05
i
054
, and if
used here as the analogue of the Manning equation for uniform
ow, applying the above to a parabola gives
y
nporous

Q
WS
054
o
3

l
p
4nm
05
_ _2=3
23
7.2. Appendix B. Estimation of the exit height by
critical depth analogy
An M2 water surface prole, in an ordinary open channel and a
Dupuit parabola for unconned ow through a porous medium
are analogues of one another.
9,12
It is usually very undesirable to have any seepage face on the
downstream slope of a conventional embankment dam. However,
one will be always present on the downstream slope of a
ow-through rockll drain or dam, with the height of that
seepage face being mainly a function of the porosity and the
imposed rate of discharge. The depth at the brink-end of an M2
prole is theoretically associated with a condition of minimum
specic energy, known as the critical depth (denoted y
c
). The
specic energy of ow in a porous medium is
E y cos
2
b
0
a
V
2
v
2g
24
where a is the kinetic energy correction factor and b
0
is the
angle of the stream bed above the horizontal. For typical open
channels (no porous media) a % 1
.
05 and cos
2
b
0
% 1. Allowing
in both cases values of unity for the moment, and using
V
v
V/n, V Q/A, and A ry
s
E y
Q
2
2gn
2
r
2
y
2s
25
Taking dE/dy and setting the result equal to zero to nd the
minimum energy condition (at which y y
c
)
0 1
Q
2
s
gn
2
r
2
y
2s1
cporous
26
y
cporous

Q
2
s
gn
2
r
2
_ _
1=2s1
27
For a rectangle, r b (the width) and s 1 and Q/b q,
in which case
y
cporous

q
2
gn
2
_ _
1=3
28
According to Henderson
27
the actual depth right at a brink for an
open channel (no porous media) is about 5/7 times the nominally
computed value
y
cchannel

5
7
y
c
29
Equation (29) is consistent with the fact that (i) the nominally
computed critical depth (for example) always occurs upstream of
the brink, at a distance of between three and four times the
nominal y
c
, and that (ii) the M2 curve (or Dupuit parabola)
decreases in depth toward the brink. If we accept the Henderson
27
value and transfer it to the porous media setting, then equation
(29) becomes
y
cporous

5
7
q
2
gn
2
_ _
1=3
30
A
x
b
y
Fig. 16. Denitions for parabolic stream cross-section
M2
Open
channel
Brink
Nominal
y
c
Dupuit parabola
Brink
y
exit
Fig. 17. M2 water surface prole and Dupuit parabola
172 Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al.
where y
cporous
is the nominal exit height of freefall seepage from
a vertical rectangular face of porous media (L) (see right-hand
side of Fig. 17). Generalising
y
cporous
k
1
aq
2
g cos
2
b
0
n
2
_ _
1=3
31
Further analysis of Hansens
11
data on sloped downstream
embankment faces undergoing non-Darcy ow in rectangular
cross-sections showed that the empirical factor k
1
increases with
the relative exit height to a limiting value of between 3 and 5, as
the u of the toe approaches u
ff
(Fig. 13), contrary to the brink
behaviour of pure open channels. Generalising from equations
(27) and (31):
y
cporous
k
aQ
2
s
g cos
2
b
0
n
2
r
2
_ _
1=2s1
32
For a parabola with area function
A
4
3

l
p y
3=2
33
y
cporous
k
2
27alQ
2
32g cos
2
b
0
n
2
_ _
1=4
34
Research is needed on the behaviour of the adjustment factor k
2
in equation (34). The coefcient a is probably very close to unity
because a relatively uniform velocity distribution is likely in
Dupuit ows.
7.3. Appendix C. Derivation of analytic expression for
phreatic surface
A well known power-function form of non-Darcy ow equation
is that due to Wilkins
6
V
v
Wm
05
i
054
35
which can be stated as
i aV
N
v
1 N 2 36a
According to Wilkins N 1
.
85 and
a
1
Wm
05
_ _
N
36b
If the ow area of the stream geometry can be described by
A ry
s
, knowing that Q nV
v
r(y
s
) and using equation (36a)
gives
i a
Q
nry
s
_ _
N
37
and
Q nry
s
i
a
_ _
1=N
38
It is a denition that
dE
dx
i 39
Under the Dupuit assumptions and if the bed slope is zero, E may
be evaluated at the surface (y), in which case equation (39) can be
evaluated using equation (37)
dE
dx
a
Q
nry
s
_ _
N
40
If the energy is dened as
E y
V
2
v
2g
41
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Station:
m
Guessed
Z: m
Inferred
y: m
Bed
z: m
A: m
2
V
v
: m/s V
v
2
/2g: m H: m i Average
i
Dx: m h
f
: m H: m DH: m
0 28
.
9560 28
.
9560 0 2885
.
16 0
.
6327 0
.
0214 28
.
977 1
.
41592 28
.
977 0
1
.
3162 1
.
0 1
.
31623
1
.
0 30
.
2755 30
.
2755 0 2914
.
76 0
.
5829 0
.
0182 30
.
294 1
.
21652 30
.
294 2
.
04 10
29
0
.
7112 1
.
0 0
.
7112
2
.
0 31
.
0022 31
.
0022 0 3045
.
21 0
.
2232 0
.
0027 31
.
005 0
.
20595 31
.
005 1
.
38 10
210
(1) distance upstream, measured relative to beginning of seepage face (in this case of Hellhole Dam, this point of emergence is 37
.
643 m
upstream from toe of dam)
(2) elevation of waters surface at station in column (1); increasing in magnitude for case of M2 analogue
(3) local depth (relative to stream bed), and as required by assumed elevation Z in column (2)
(4) elevation of stream bed, column (2) column (3)
(5) cross-sectional area implied by column (3), in this case trapezoidal with unequal side slopes
(6) Q/(nA)
(7) column (6)
2
/2g
(8) column (2) column (6), elevation of energy grade line above datum
(9) energy gradient from i aV
v
N
; a estimated using equation (36b), (i often initially unrealistically high)
(10) average energy gradient, using values at limits of sub-reach under consideration
(11) distance between stations bounding the sub-reach
(12) column (10) x column (11)
(13) H value upstreamcolumn (12), assumed equal to column (7) for row 1
(14) column (8) column (13)
Column (2) is adjusted until column (14) is nearly zero, then column (2) at the next station is similarly adjusted, and so on. Columns (2), (4),
(8) and (13) are elevations above an arbitrary datum, such as mean sea level.
Table 7. Standard step method algorithm; ow through coarse porous media, M2 analogue
Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al. 173
substitution of equation (38) into equation (41) yields
E y
Q
2
2gn
2
r
2
y
2s
42
Differentiating
dE
dx

dy
dx

Q
2
s
gn
2
r
2
y
2s1
dy
dx
using equation (40)
a
Q
nry
s
_ _
N

dy
dx

Q
2
gn
2
r
2
y
2s1
dy
dx
where the negative sign is introduced because ow moves in the
direction of declining specic energy. Rearranging
1
1
a
dy
dx
Q
2N
s
gn
2N
r
2N
y
sN2s1

n
N
r
N
Q
N
y
sN
_ _
call
a
Q
2N
s
gn
2N
r
2N
and b
nr
Q
_ _
N
separating variables
dx
1
a
ay
sN2s1
by
sN
dy
Integrating
x
_
1
a
ay
sN2s1
by
sN
dy

1
a
a
y
sN2s
(sN 2s)
b
y
sN1
sN 1
_ _
C
x
1
a
a
1
2s sN
y
sN2s
b
1
sN 1
y
sN1
_ _
C
at x 0, y y
exit
so
C
1
a
a
1
2s sN
y
sN2s
exit
b
1
sN 1
y
sN1
exit
_ _
x
1
a
a
1
2s sN
y
sN2s
exit
b
1
sN 1
y
sN1
exit
_ _

1
a
a
1
2s sN
y
sN2s
b
1
sN 1
y
sN1
_ _
43a
where
a
Q
2N
s
gn
2N
r
2N
and b
nr
Q
_ _
N
43b
Rearranging (C-9a)
a
a
1
2s sN
y
sN2s

b
a
1
sN 1
y
sN1

1
a
a
1
2s sN
y
sN2s
exit
b
1
sN 1
y
sN1
exit
_ _
x
which is an equation of the form
u
1
y
sN2s
u
2
y
sN1
u
3
x 44
where
u
1

a
a
1
2s sN
u
2

b
a
1
sN 1
u
3

1
a
a
1
2s sN
y
sN2s
exit
b
1
sN 1
y
sN1
exit
_ _
N 1
.
85 for nearly fully developed turbulence and the parabolic
case for r and s is given by equation (22).
Equation (44) cannot handle fully developed turbulence (N equal
to exactly 2
.
0). Downstream is dened as positive and x is the
distance upstream from the point of emergence of the seepage
face (so x 0 at y y
exit
). The result is a water surface prole
that is an M2-analogue (that is, increasing depth in the upstream
direction, relative to the bed) for ow through a rockll structure
with zero bed slope. Equation (44) only gives an approximate
phreatic surface (above a given datum) if allowance is made for
the slope of the bed after the fact. If the structure is more than
about 50 m long, a numerical method, such as that presented in
Table 7, should be used instead.
REFERENCES
1. CANADIAN DAM ASSOCIATION. CDA Dam Safety Guidelines.
CDA, Alberta, 1999.
2. Reservoirs Act 1975: Elizabeth II. Chapter 23. Her Majestys
Stationery Ofce, 1975.
3. CANADIAN DAM ASSOCIATION. CDA Dam Safety Guidelines.
CDA, Alberta, 1995.
4. GARGA V. K., HANSEN D. and TOWNSEND D. R. Mechanisms of
massive failure for owthrough rockll embankments.
Canadian Geotechnical Journal, 1995, 32, No. 6, 927938.
5. HANSEN D., GARGA V. K. and TOWNSEND D. R. Selection and
application of a 1-D non-Darcy ow equation for 2-D ow
through rockll embankments. Canadian Geotechnical
Journal, 1995, 32, No. 2, 223232.
6. WILKINS J. K. Flow of water through rockll and its
application to the design of dams. Proceedings of the 2nd
Australia-NZ Conference on Soil Mechanics and Foundation
Engineering, 1956, Canterbury University College,
Christchurch, New Zealand, pp. 141149.
7. PITEAU ENGINEERING (KOMEX INTERNATIONAL LTD). Mined rock
and overburden piles, rock drain research program. Prepared
for the BC Dept of Energy & Mines with the assistance of
Manalta Coal and Canmet, Calgary, Alberta, 1999.
8. GARGA V. K., TOWNSEND D. R. and HANSEN D. A method for
determining the surface area of quarried rocks. ASTM
Geotechnical Testing Journal, 1991, 14, No. 1, 3545.
9. DAKE J. M. K. Essentials of Engineering Hydraulics.
MacMillan, London, 1972.
10. TAYLOR D. M. Fundamentals of Soil Mechanics. John Wiley &
Sons, New York, 1948.
11. HANSEN D. The Behaviour of Flowthrough Rockll Dams. PhD
thesis, University of Ottawa, Canada, 1992.
12. STEPHENSON D. Rockll in Hydraulic Engineering. Elsevier
Scientic, Amsterdam, 1979.
13. BARI R. The Hydraulics of Buried Streams. MASc Thesis,
Department of Civil Engineering, Technical University of
Nova Scotia, Canada, 1997.
14. JOHNSON H. Flow through rockll dam. ASCE Journal of
Soil Mechanics & Foundations Division, 1971, 97,
No. SM2, 329340.
174 Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al.
15. KLUBER T. and BRETH H. Discussion of ow through rockll
dam. ASCE Journal of Soil Mechanics & Foundations
Division, 1971, 97, No. SM11, 16091612.
16. BARI R. and HANSEN D. Application of gradually-
varied ow proles to simulate buried streams. IAHR
Journal of Hydraulic Research, 2002, 40, No. 5,
674683.
17. HANSEN D. and BARI R. Uncertainty in the water surface
prole of a buried stream owing under coarse material.
ASCE Journal of Hydraulic Engineering, 2002, 128, No. 8,
761773.
18. GILES R. V. Fluid Mechanics and Hydraulics. McGraw-Hill,
New York, 1977.
19. WATTERS F. H. and RAO M. V. P. Stabilization of Channels
with Gravel. MS thesis, Colorado State University, Fort
Collins, Colorado, 1967.
20. STEVEN M. A. and SIMONS D. B. Stability analysis for coarse
granular material on slopes. In River Mechanics, Vol. I.
H. W. Shen, Fort Collins, Colorado, 1971.
21. SHARP B. B. and JAMES J. P. Spatially varied ow at the toe of
a rock-ll slope. Proceedings of the First Australasian
Conference on Hydraulics and Fluid Mechanics, Perth,
Australia, 1963.
22. LI W. Open channels with non-uniform discharge.
Transactions of the American Society of Civil Engineers,
1954, 120, 255274.
23. HOLTZ R. D. and KOVACS W. D. An Introduction to
Geotechnical Engineering. Prentice-Hall Co., Englewood
Cliffs, NJ, 1982.
24. TOWNSEND D. R., GARGA V. K. and HANSEN D. Finite difference
modelling of the variation in piezometric head within a
rockll embankment. Canadian Journal of Civil Engineering,
1991, 18, No. 2, 254263.
25. PARKIN A. K. Rockll dams with inbuilt spillways Ihydraulic
characteristics. Bulletin 6, University of Melbourne
and Water Research Foundation of Australia,
Melbourne, 1963.
26. NEILL C. R. and YALIN M. Quantitative denition of the
beginning of bed movement. ASCE Journal of the Hydraulics
Division, 1969, 95, No. HY1, 585588.
27. HENDERSON F. M. Open Channel Flow. MacMillan, New York,
1966.
What do you think?
To comment on this paper, please email up to 500 words to the editor at journals@ice.org.uk
Proceedings journals rely entirely on contributions sent in by civil engineers and related professionals, academics and students. Papers
should be 20005000 words long, with adequate illustrations and references. Please visit www.thomastelford.com/journals for author
guidelines and further details.
Water Management 158 Issue WM4 Hydraulic performance and stability of coarse rockll deposits Hansen et al. 175

Você também pode gostar