Professional Documents
Culture Documents
By Dieter Rickenmann1
INTRODUCTION
•Hydr. Engr., Amt fur Gewasserschutz und Wasserbau des Kantons Zurich,
Walchetor, 8090 Zurich, Switzerland.
Note. Discussion open until April 1, 1992. 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 December 13,
1990. This paper is part of the Journal of Hydraulic Engineering, Vol. 117, No. 11,
November, 1991. ©ASCE, ISSN 0733-9429/91/0011-1419/$1.00 + $.15 per page.
Paper No. 26320.
1419
In the past, studies on sediment transport in rivers and flumes were mainly
concerned with bed slopes up to a few percent. It is only recently that
interest on sediment-transport processes in steep channels has grown. Such
studies, including flume slopes in excess of 10%, were performed by Mi-
zuyama (1977), Mizuyama and Shimohigashi (1985), and Smart and Jaeggi
(1983). The transporting fluid used in these flume tests was clear water. In
torrents where debris flows also occur, the influence of large amounts of
suspended fine particles may become important. Accordingly, the objective
of the present study was to examine the effect of an increasing density and
viscosity of the fluid on the flow and the sediment-transport behavior. A
series of steep flume tests was performed using a clay suspension to represent
the fine slurry of a debris flow, and the flow resistance and the sediment-
transport capacity were determined for various clay concentrations. The
experiments of Smart and Jaeggi (1983) [see also Smart (1984)], which were
carried out at the same hydraulic laboratory, served as a reference condition
for the clear-water case.
In the past, two principally different debris-flow theories have been de-
veloped. One of them is the viscoplastic model that was applied to flows
containing a lot of fine material in a viscous slurry. It is based on the
approach by Bingham (1922) to describe a particular type of non-Newtonian
flow. The Bingham model also applies to clay suspension flows as used in
this study and it can be written as:
T = T + T1 (1)
* *(!)
where T = the shear stress in the fluid; iB = the (Bingham) yield stress;
r\B = the Bingham viscosity; and dvldy = the velocity gradient perpendic-
ular to the flow direction.
Considering mainly the interaction between the coarser particles in a
debris flow, Takahashi (1978) proposed an alternative model based on Bag-
nold's (1954) concept of dispersive pressure. Combining these two basic
approaches, more complex models were formulated, for example, Chen
(1988a, 1988b) and Julien and Lan (1991). A detailed discussion on debris-
flow models can be found elsewhere (Bradley 1986; Rickenmann 1990).
Many hyperconcentrated slurries of fine particles suspended in water
behave like a Bingham fluid (Costa 1984; Chen 1986). The flow resistance
of these fluids can be analyzed either by a set of equations in terms of the
Bingham parameters (Johnson 1970; Naik 1983) or by defining an effective
viscosity (xe2 and using conventional Newtonian formulas (Quian et al. 1980;
Wan 1982):
and Culbertson 1964; Bradley 1986) and from laboratory experiments (up-
per-regime tests of Simons et al. 1963; Kikkawa and Fukuoka 1969; Wan
1982; Bradley 1986; Wan and Song 1987). With regard to the effect on bed-
load transport rates qB, an increase in qB due to an increase of the fluid
density is predicted from experimental results by Shields (1936), by Meyer-
Peter and Mueller (1948), and by Luque and van Beek (1976), and from
theoretical considerations by Yalin (1977) and by Woo (1985). However,
Wan and Song's (1987) theory predicts a decrease in the dimensionless bed-
load transport rates Q>B for an increasing fluid density at higher shear stresses.
Low (1989) found that many bed-load transport formulas adequately ac-
count for a change in the solid/fluid density ratio, but that the Smart and
Jaeggi (1983) equation should be modified. Information on the effect of an
increasing viscosity alone on qB in open-channel flow is less conclusive (Hong
et al. 1984; Woo 1985; Lau 1987). Wan (1982) concluded from his experi-
ments that <$>B was smaller in the bentonite suspension flows than in the
corresponding clear-water flows, particularly in the low flow-intensity
region.
There is not yet much information available on sediment transport in
laminar flow. In their experiments, both Wan and Song (1987) and Bradley
(1986) observed an increase in the total bed-material discharge; in Bradley's
study the transport rates in laminar flow were about an order of magnitude
higher than in the corresponding turbulent flow (with the same fluid dis-
charge). Pipe flow tests reported by Thomas (1979b) may suggest that the
bed-load transport rates decrease in laminar flow, assuming that bed load
is the predominant mode of transport at a velocity slightly above depositional
conditions. Negative lift forces for particles near the bed (Coleman 1967;
Davies and Samad 1978) and theoretical considerations by Bagnold (1956)
also seem to indicate that bed-load transport should be smaller in laminar
flow than in turbulent flow.
EXPERIMENTS
7% and 20%, and the fluid flow rate Q between 10 and 30 1/s. Experiments
were performed at five different clay concentration levels; the corresponding
fluid properties are summarized in Table 1. The measured flow parameters
are given in Table 2 for the experiments without sediment transport (case
A). The fixed rough bed in these experiments had the same grain-size
characteristics as the bed material used in the sediment. The data of clay
suspension flows with bed-load transport (case B) is shown in Table 3. The
experimental results labeled H z O were taken from Smart and Jaeggi (1983).
(See section on bed-load transport for distinction between case Bl and case
B2 experiments.
The Bingham parameters iH and r\B were measured with a rotational
viscometer, and the effective viscosity |x(>2 was determined according to (2).
The salt tracer technique was used to measure the mean fluid velocity; at
the highest concentrations this method was not reliable because of insuffi-
cient mixing of the salt solution in the clay suspension (see also remarks in
Tables 2 and 3). The flow depth was inferred from a flow surface level
determined by ultrasonic distance measuring devices and information about
the bed level. In the case B tests, the boundary between nonmoving grains
of the bed and the transport layer was detected by electric conductivity
readings at different levels across the flume width, since no visual obser-
vation was possible in the muddy suspension. All measurement methods
are described in more detail in Rickenmann (1990).
In the analysis of both the case A and the case B experiments, a correction
for sidewall friction effects was applied. A procedure used by Smart and
Jaeggi (1983) was adapted; the method is based on the approach by Einstein
(1934) who postulated equal velocities in flow sections with different rough-
ness. Thus the reduced flow depth, hr, and the reduced flow rate per unit
width, qn were determined from the measured flow depth, H, and the
1422
FLOW RESISTANCE
$!•*»{$+* <3>
where ks = the equivalent sand roughness; (3 = a coefficient depending to
some extent on the relative flow depth; and A is constant in the hydraulically
fully turbulent region where no viscous effects are present (Yalin 1977). In
Fig. 2, the data is plotted in terms of A' = [1/V/ — 2 • \og(hrld9Q)] and
RJ, and R2 = v*dp/\ie2 is the particle Reynolds number, with v* = (gh,.S)0-5
and d = dgo. It is seen that viscous effects only become important for
R2 :£ 10. Above this critical value, A' is approximately constant with an
average of 2.4; from a comparison with (3) it can be concluded that a value
of ks = 0.76 • d90 would best describe the clay suspension experiments in
the turbulent region.
In the experiments with bed-load transport (case B), no flows in the
laminar region were observed. The analysis of the flow resistance data shows
only a very slight increase of the friction factor / for some of the C2 and
C3 tests [Fig. 1(b)]. This minor change is associated with an increase in bed-
load concentration and thus, flow depth, and is not a direct effect of altered
fluid properties.
In further analysis, a search was made for a prediction equation for the
fluid velocity that would apply to both the bed-load transport experiments
in the fully turbulent clay suspension flows (for which R* > 10, case Bl,
see also section on bed-load transport) and the bed-load transport tests of
Smart and Jaeggi (1983).
Jaeggi (1983) proposed a modified flow resistance equation for a sediment
transporting flow over a movable bed at smaller relative depths hldgo, be-
cause many conventional formulas tend to overpredict the velocities in the
range 5 < h/d9n < 20. His equation is given as:
1423
Q V H Tfl •<\B P
a S (l/s) (m) (cm) (N/m2) (ops) (g/cm3
Downloaded from ascelibrary.org by "Indian Institute of Technology, Guwahati" on 03/06/24. Copyright ASCE. For personal use only; all rights reserved.
1424
Q V H Tfl 1)B P
a S (l/s) (m) (cm) (N/m2) (cps) (g/cm3)
Downloaded from ascelibrary.org by "Indian Institute of Technology, Guwahati" on 03/06/24. Copyright ASCE. For personal use only; all rights reserved.
12.27ft
2 1 exp In (4)
7 =" L - U^V M9o
where the coefficients (^ and ax depend on the grain-size distribution and
the packing and the shape of the bed material. For their steep channel bed-
load transport tests, Smart and Jaeggi (1983) determined o^ = 0.05 and Bt
= 15. Using the same coefficients, the performance of (4) is shown in Fig.
3 for the steep flume data sets of Smart and Jaeggi and of the writer. It is
observed that (4) tends to slightly overpredict the velocities of the clay
suspension tests. It may be noted that (4) was developed from experimental
data with flatter slopes; for large relative flow depths and small slopes, the
correction factor in square brackets of (4) approaches unity and the formula
becomes similar to a Nikuradse-type equation.
Based on the concept of dispersive stresses between the moving grains,
Takahashi (1978) proposed an equation for the mean velocity of a quasi-
steadily moving debris-flow front. Replacing concentration and density terms
together with grain-shearing coefficients by one single parameter A*, the
following expression is obtained (Takahashi 1978):
1425
Q V H O fl f« T}B P
a s (l/s) (m/s) (cm) (kg/s) (N/m2) (cps) (g/cm3)
Downloaded from ascelibrary.org by "Indian Institute of Technology, Guwahati" on 03/06/24. Copyright ASCE. For personal use only; all rights reserved.
1426
U
(gdf ^5 = A*(sin a)" 2
(5)
3L
gd3
where U = the mean velocity of the moving grains; q = the fluid discharge
per unit flume with; and a = the bed slope angle. Takahashi plotted his
experimental results in terms of the nondimensional velocity [given by the
left-hand side of (5)] versus the bed slope tan a. As predicted by (5), there
is only a slight dependence on tan a in his figure. From his diagram (Tak-
ahashi 1978), it can be determined that A* ~ 1.3. Using this experimental
value for the parameter A*, (5) can be transformed into:
(sin af-2q°-6g0-2
U = 1.3 (6)
d0A
Interestingly a regression analysis with the two steep flume data sets dis-
cussed previously results in an empirical relation very similar to (6). In fact,
Fig. 4 shows good agreement for the steep flume data between measured
and calculated fluid velocities; putting sin a ~ tan a = S, d = d9Q and U
= V, (6) becomes:
6
'0.2„0.6 n 0.2
13S^q°- g
V 0 (7)
d'
«Q<)
1427
\ 11 f S = 20°/c1
0.40 ^ . 96 J-~\
Downloaded from ascelibrary.org by "Indian Institute of Technology, Guwahati" on 03/06/24. Copyright ASCE. For personal use only; all rights reserved.
\ \ ]S=15%f.
0.30
f
" Re ->!
1-
\
0.20
f
0.10
\
i .l^*--^
\I
% V
fS = 5%] ^
"
i Sl~A
0.08
\ ^ * s X, ^
•>. N <\»\ \ * \ ~^
0.06 \ \ N , \ -tr~ X
* «<\ r
\ \V \ \
• »
0.04 N \
0.03 \
102 10 3 10" 10 5 10s
Re2
C, [%]
0.80 96 1 S = I5%[ | | |
\
\ '/ |
0.60
x \ |s-m%l\ nL5^20%]
0.40 \ -r
^ \ \L-
\ n• "—
I
^r-
V \ \ \ ^ V V
0.30 kv
0.20
\\
A
4.
X +
^ \ -N \ \
^\
\
\
^ \
\ xa
f
0.10
\
V
1
s?
*
\
^ \ | S = 5°/ V
•u ft
• *
0.08 2
10 3
\ 10 4 105 „ 106
10
Re 2
FIG. 1. Flow Resistance Data of (a) Clay Suspension Flows over Fixed Rough
Bed; and (b) Experiments with Bed Load Transport. Data Points with Equal Slope
are Connected by Straight Lines
BED-LOAD TRANSPORT
In the case of bed-load transport in the clay suspension (case B), each
experiment was performed with the same preset slope and flow rate as for
the clear-water tests by Smart and Jaeggi (1983), which served as a reference
condition. The ratio of the corresponding bed-load transport rates per unit
width, qB/qB,Hicn w a s found to increase with increasing clay concentration
of the suspension (and thus, fluid density) up to a limiting value, above
which transport rates decreased strongly. This change in bed-load transport
capacity can also be plotted against the particle Reynolds number R* (Fig.
5), which represents the effect of a changing fluid viscosity; here RJ is
determined with the mean grain size dm. It can be concluded from Fig. 5
1428
ft
"++ «
t-
[» * *
+ i+ * *** A A<£ .A
Y
r i 0
T
1
Y 1
1
r 1
1
_|~Re 2 * = 10[]Re 2 * = 5a •
1
1
0 1 2 3 4
log (Re2*)
FIG. 2. Flow Resistance Data of Clay Suspension Flows over Fixed Rough Bed,
Shown in Terms of Parameters A' = [l/(Vf - 2 • log(hJdmi)] and K
that the strong decrease in transport rates occurs for values of R* below
about 10-15.
For open-channel flow, the thickness of the laminar (viscous) sublayer,
k, is given as (Yalin 1977): X = 11.6Wu* = 11.6(T|/p)/w*, where v is the
kinematic and T| is the dynamic (Newtonian) viscosity. This implies that in
a flow with R* ~ 12, the laminar sublayer is of the same magnitude as the
equivalent sand roughness ks. It is assumed that the relationship for X is
also valid in the case of the clay suspension, using the effective viscosity
|xe2. With regard to the critical value of R* shown in Fig. 5, this implies
that the thickness of the laminar sublayer is of the order of the grain size
of the transported bed material. Thus it appears that once the flow around
the grains becomes laminar, the bed-load transport capacity starts to de-
crease, for otherwise equal flow conditions.
Interestingly, also for the clay suspension flows over fixed rough bed (case
A), a limiting value of RJ = 10 was found below which viscous effects
become important. It is therefore concluded that in the bed-load transport
experiments, the decrease in qB for RJ <~ 10 (case B2) is due to the
increasing viscosity while viscous effects are negligible in the region R* > ~
10 (case Bl).
Vmeas.[m/S]
FIG. 3. Comparison between Measured Fluid Velocities, V„„m, and those Calcu-
lated with (4), V„k, for Two Steep Flume Data Sets
-^fePM-fc) ; <»
where d30 = the grain size at which 30% by weight of the material is finer;
6„ = the critical Shields stress at initiation of motion; and 9„, = hrjn • SI
[{s — 1) • dm] — the dimensionless bed shear stress determined with the
mixture flow depth hrm (including the space occupied by the transported
solids but corrected for sidewall friction) and the mean grain size d,„. It is
noted that (8) was developed in order to predict also the bed-load transport
rates of the Meyer-Peter and Mueller (1948) experiments covering a slope
range of 0.04% < S < 2% [data given by Smart and Jaeggi (1983)].
The data of the clay-suspension experiments (case Bl) was analyzed to-
gether with the Smart and Jaeggi tests, and a regression calculation yielded
the following transport equation, valid for the slope range 5% < X < 20%:
(9)
' • - < r r ^ (£)"<''"'-^
where qcr = the critical discharge at initiation of motion. The correlation
coefficient squared is r2 = 0.95 and the standard error is SE = 20% of the
1430
X A
X ./
Downloaded from ascelibrary.org by "Indian Institute of Technology, Guwahati" on 03/06/24. Copyright ASCE. For personal use only; all rights reserved.
2.5 4 " / *
ft
A * / *
A &
2.0 A /
A / X
«
A^ Oy/ x
1.5
* "
£
Uniform
1.0 Data set
grain size Mixture -
J *X
Smart/Jaggi A x
Rickenmann *
0.5 I
0.5 1.0 1.5 2.0 2.5 3.0
Vmeas.[m/S]
FIG. 4. Comparison between Measured Fluid Velocities, Vmem, and those Calcu-
lated with (7), V,.„,c, for Steep Flume Data
qB
qB,H20
3.2
-i 1 1
c, [%]
2.8 T 21.6-22.0
"-••-J
ft 15.6-17.8
+ 12.2-14.9
2.4
« 8.6-9.7
s V-':" < * 4.4-5.8
2.0
+ 0
1
0
ft <5
1.6 *
* **
1.2
"f >
+ " * + /
>
0.8
T *
n J5
1.0 10.0 100.0 1000.0
Re 2 *
FIG. 5. Ratio of Bed Load Transport Rate Measured in Clay Suspension to Value
Obtained in Clear Water, qBlqBM20, as Function of Grain Reynolds Number R*
1431
l°g(qB,caic.[m3/s m])
,«*/
4m
f
I'rf
o
e
/i*
s°
i*
• !&
J
43: 1
s>
Uniform
' , y< i
Data set grain size
Mixture
i
/ I
Smart/Jaggi * *
Rickenmann *
o
Mtiller
C I r~
X
1 1
-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1
3
l°g(qB,meas.[m /s m])
FIG. 6. Comparison between Measured Bed Load Transport Rates, <7B,mcas., and
those Predicted by (11), qB,mlc_, for Three Data Sets Covering Slope Range 0.1%
< = 5 < = 20%
1433
steep slopes. These findings are consistent with the aforementioned con-
clusion that in the form of (8) and (9) the exponent of the slope factor
should be different in the steep and low slope region. To distinguish between
the two regions, a limiting slope of about 10% may be assumed above which
the space occupied by moving grains is significant with respect to the total
flow depth.
DISCUSSION
sion (not supported by turbulence but due to a very small settling velocity).
It is difficult to predict what the combined effect will be on the bed-load
transport capacity in laminar open-channel flow. Experimental evidence
(Bradley 1986; Wan and Song 1987) suggests that the total transport capacity
may be considerably higher in laminar flow.
Three mechanisms were proposed by Takahashi (1981) for the initiation
of a debris flow: (1) A landslide from a hill slope may transform into a
debris flow; (2) at sufficiently steep slopes ( > ~ 27%) and with large bed
shear stresses, material deposited in a channel bed may be mobilized and
eventually form a debris flow; and (3) accumulation of debris material from
the side slopes can cause a temporary blockage of material to be built up
in the channel, and a debris flow may be initiated when this material starts
to move.
It was noted by Davies (1988) that a temporary blockage of large grains
can occur in a uniform flow of a grain-fluid mixture in a steep, narrow
channel. He further pointed out that the development of roll waves may
cause the pulsing of a debris flow; the critical Froude number for the oc-
currence of roll waves is much smaller in laminar than in turbulent flow.
Based on field data, Davies showed that the limiting conditions developed
for clear-water flows also apply generally for debris-flow conditions. Re-
cently, Savage (1989) demonstrated theoretically that the critical Froude
number for the occurrence of roll waves decreases with increasing cohesion
(Bingham yield stress), viscosity and particle interaction.
In the light of the experimental results of this study, the pronounced
decrease of the bed-load transport rate in macroviscous flow (i.e., having
R2 s 10) may also be seen as an element enhancing the instability of a
uniform flow, and thus offer a possible explanation for the pulsing behavior
of debris flows. In the experiments, macroviscous flow occurred at clay
concentrations Cf above about 17%. Scaling considerations suggest that a
higher critical value of Cf can be expected in the field (Rickenmann 1990).
In the range of fine-material concentrations Cf between 20% and 50%, the
Bingham yield stress TB increases strongly with Cf (Chen 1986; O'Brien and
Julien 1986). According to field data compiled by Davies (1988), the tran-
sition to macroviscous flow may be expected to occur in this concentration
range. For a given flow rate in a torrent, a local input of fine material might
thus cause a sudden decrease in bed-load transport capacity (at R* — 10),
causing a part of the grains to be deposited. If such events are frequent and/
or large enough, the deposited material could form a temporary dam. In
contrast to direct deposition of material from the side slopes, a partial or
full blockage to the flow might therefore also be caused by a change in flow
properties.
CONCLUSIONS
APPENDIX I. REFERENCES
of Philosophy.
Chen, C. (1986). "Chinese concepts of modeling hyperconcentrated streamflow and
debris flow." Proc. 3rd Int. Symp. on River Sedimentation, Univ. of Mississippi,
1647-1657.
Chen, C. (1988a). "Generalized viscoplastic modeling of debris flow." /. Hydr.
Engrg., ASCE, 114(3), 237-258.
Chen, C. (1988b). "Generalized solutions for viscoplastic debris flow." /. Hydr.
Engrg., ASCE, 114(3), 259-282.
Coleman, N. L. (1967). "A theoretical and experimental study of drag and lift forces
acting on a sphere resting on a hypothetical stream bed." Proc. 12th IAHR Con-
gress, Fort Collins, Vol. 3, 184-192.
Costa, J. E. (1984). "Physical geomorphology of debris flows." Developments and
applications of geomorphology Costa and Fleisher, eds., Springer-Verlag, Berlin,
Germany, 268-317.
Davies, T. R. H. (1988). "Debris flow surges—A laboratory investigation." Mitt.
Nr. 96 der Versuchsanstalt fur Wasserbau, Hydrologie und Glaziologie, ETH
Zurich.
Davies, T. R. H., andSamad, M. F. A. (1978). "Fluid dynamic lift on a bed particle."
/. Hydr. Div., ASCE, 104(8), 1171-1182.
Einstein, H. A. (1934). "Der hydraulische oder Profilradius." Schweiz. Bauzeitting, 8.
Hong, R.-J., Karim, M. R., and Kennedy, J. F. (1984). "Low-temperature effects
on flow in sand-bed streams." J. Hydr. Engrg., ASCE, 110(2), 109-125.
Jaeggi, M. N. R. (1983). "Alternierende Kiesbanke," Diss. ETH Nr. 7208, Ziirich
Mitt. Nr. 62 der Versuchsanstalt fur Wasserbau, Hydrologie und Glaziologie der
ETH Zurich.
Johnson, A. M. (1970). Physical processes in geology. Freeman Cooper and Co.,
San Francisco, Calif.
Julien, P. Y., and Lan, Y. (1991). "Rheology of hyperconcentrations." /. Hydr.
Engrg., ASCE, 117(3), 346-353.
Kikkawa, W., and Fukuoka, S. (1969). "The characteristics of flow with wash load."
Proc. 13th IAHR Congress, Kyoto, Japan, Vol. 2, 233-240.
Lau, Y. L. (1987). Discussion of "Low temperature effects on flow in sand-bed
streams," by Hong et al. (1984). J. Hydr. Engrg., ASCE, 113(1), 111-115.
Low, H. S. (1989). "Effect of sediment density on bed-load transport." /. Hydr.
Engrg., ASCE, 115(1), 124-137.
Luque, R. F., and van Beek, R. (1976). "Erosion and transport of bed-load sedi-
ment."/. Hydr. Res., 14(2), 127-144.
Meyer-Peter, E., and Mueller, R. (1948). "Formulas for bed-load transport." Proc.
2nd meeting IAHSR, Stockholm, Sweden, 39-64.
Mizuyama, T. (1977). "Bedload transport in steep channels." thesis presented to
Kyoto Univ., at Kyoto, Japan, in partial fulfillment of the requirements for the
degree of Doctor of Philosophy.
Mizuyama, T. (1981). "An intermediate phenomenon between debris flow and bed
load transport." Proc. Symp. on Erosion and Sed. Transp., IAHS Publication No.
132, Int. Association of Hydrol. Sci., 212-224.
Mizuyama, T. and Shimohigashi, H. (1985). "Influence of fine sediment concentra-
tion on sediment transport rates." Jpn. Civ. Engrg. J., 27(1), 46-49 (in Japanese).
Naik, B. (1983). "Mechanics of mudflow treated as the flow of a Bingham fluid,"
thesis presented to Washington State Univ., at in partial fulfillment of the re-
quirements for the degree of Doctor of Philosophy.
O'Brien, J. S., and Julien, P. Y. (1986). "Rheology of non-Newtonian fine sediment
mixtures." Proc. Speciality Conference on Advances in Aerodynamics, Fluid Me-
chanics, and Hydraulics, ASCE, New York, N.Y., 989-996.
Quian, Y., Yang, W., Zhao W., Cheng X., Zhang L., and Xu W. (1980). "Basic
1437
APPENDIX ! l . NOTATION
1438