Beruflich Dokumente
Kultur Dokumente
Maarten G. Kleinhans
Universiteit Utrecht, Fac. Geosciences, Dept. Physical Geography, Utrecht, The Netherlands
ABSTRACT: For the prediction of river bed destabilisation and fractional sediment transport of mixtures, we
aim to solve two problems that are poorly understood. First, the flow and pressure fluctuations surrounding both
the embedded and exposed particles must be parameterised for hydraulically smooth to rough flow. Second, an
adequate relation between particle size and particle exposure should be based on the particle size distribution and
the (water-worked) bed structure. We use a recently developed force balance model for the threshold of motion
of uniform sediments incorporating the effects of particle exposure, pressure fluctuations into the bed, very
shallow flow and bed slope. The flow module is extended to non-uniform roughness of sediment mixtures. Our
extended model predicts the critical Shields values of arbitrary mixtures directly as function of exposure and no
longer needs empirical hiding-exposure relations. Several empirical and geometrical relations between particle
size and exposure were tested. The results are compared to extensive datasets from the literature of incipient
fractional transport rates. The modelled hiding-exposure relations are very sensitive to the relation between
particle size and exposure, which differ for unimodal, skewed and bimodal mixtures. This is explained by the
pore structure of these sediments. The existing relations fail particularly for the smaller particles in bimodal and
skewed distributions. These small particles percolate through the pores so their exposure or embedding strongly
depends on the fractional content and pore structure, in agreement with empirical data. We are working on a
universal relation for exposure containing particle size distribution, pore structure and water-working. The model
reproduces data of uniform sediments well for the entire physically possible range of particle exposures and for
hydraulically rough to nearly smooth conditions. Trends in existing data for mixtures are also reproduced but
depend strongly on exposures that were not measured.
INTRODUCTION
REVIEW
The classical empirical study by A.F. Shields demonstrated that incipient motion of sands and gravels of
all densities occurs in a narrow range of a nondimensional ratio of the sediment entraining and detraining
forces:
where c = critical shear stress, s and are the densities of the sediment and water, g = gravitational acceleration and D = particle size. This critical Shields
541
10
Soulsby
Zanke
Wiberg Smith
this paper
data
10
10
10
10
10
grain size D (mm)
10
*cr model
10
10
10
*cr data
for Di /Dm > 0.3; for Di /Dm < 0.3Ashida and Michiue
(1971) proposed a correction: ci = 0.07, wherein
Di = fractional and Dm = geometric mean particle
size. The Egiazaroff model cannot account for variations in the particle size distribution, however, which
has large effects on the effective hiding function (Van
der Zwaard 1974; Shvidchenko et al., 2001). In fact,
the exposure of the fine mode of bimodal sediment
strongly depends on its relative abundance. Starting
from a gravel bed with few fines deeply embedded, adding fine sediment will increasingly fill the
pores of the coarse sediment until the coarse sediment is entirely covered while the fine sediment is fully
exposed. At the same time the decreasing roughness
of the bed affects the mean shear stress and the turbulence characteristics (Van der zwaard 1974; Sambrook
smith and Nichols, 2005). So far no model is available
incorporating these effects, but the problem has relevance for downstream fining and for aquatic ecology,
e.g. spawning of salmon.
Kleinhans and Van Rijn (2002); Zanke (2003);
Hofland (2005) and many earlier workers showed that
near-bed turbulence increases the mobility. The largest
shear stresses in the turbulent flow field entrain the
sediment. Hence the original Shields curve refers to
generalised sediment motion while individual particle motion is attained at smaller Shields values.
vollmer et al., (2002) measured and parameterised
turbulence-induced pressure fluctuations in the porous
bed. vollmer and Kleinhans (2007) showed that the
pressure fluctuations cause lift forces on the particles
that contribute to initial motion even if particles are
flush embedded and not exposed to the attacking nearbed flow velocity. Their model performed well to the
entire range of possible particle exposure: from deeply
embedded to nearly fully exposed. Verification on a
dataset for a large range of particle sizes (Buffington
542
MODEL DESCRIPTION
Model for uniform sediment
543
where n = percentile of size fraction. For better understanding of these empirical relations we also study
three relations based on simplified bed structures
(Fig. 2).
The pivot angles for unimodal and bimodal sediment demonstrate the effect of the presence of fines
in the bed (Figs 3, 4). The exposures of large grains
Di > 2Dm (Figs 3, 4) hardly differ between the various
models, except for the same centre model (see Fig. 2)
which predicts half the typical numbers. The modelled
exposures for the smaller grains Di < 2Dm , on the other
hand, are very different (see inset in Figs). The exposure is the largest for small grains on Dm and the
smallest for eq. 8. There is a considerable difference
between the Kir and Buf equations. For large grains
both resemble the same centre result, illustrating that
the large grains sink about halfway into the bed rather
than be fully exposed (the latter would lead to overpassing with extremely small critical shear stresses).
For the small grains, Kir sets the grains only slightly
deeper than the same centre, while Buf puts the grains
much deeper: down to the same floor function. The
result will be a much decreased mobility for the fines
544
70
70
Buffington
Kirchner
same floor
same centre
on Dm
50 1
40
30
60
Exposure E = p/Dm ()
Exposure E=p/Dm ()
60
20 1
10
10
-4
10
10
D (m)
10
10
150
100
50
10
10
D (m)
10
10
20 1
10
10
-4
10
10
D (m)
10
10
RESULTS
Exposure models
Buffington/Kirchner
same floor
same centre
on Dm
size distrib pi
150
100
50
0
5
10
4.1
30
200
Buffington/Kirchner
same floor
same centre
on Dm
size distrib p
40
-10
5
10
200
0
5
10
50 1
0
-10
5
10
Buffington
Kirchner
same floor
same centre
on Dm
10
10
D (m)
10
10
545
0.5
10
narrow
wider
wide
0.4
1
10
pi
cr,i
0.3
0.2
10
0.1
10
10
10
10
10
10
0
3
10
10
cr,i
10
fine skew
coarse skew
bimodal
0.4
0.3
pi
10
0.5
Buffington
Kirchner
same floor
same centre
on Dm
Egiazaroff
Ashida Michue
10
10
0.2
0.1
3
10
10
10
10
D (m)
10
10
0
3
10
10
D (m)
DISCUSSION
1982). The reason is that the new model can predict incipient motion for various exposures above and
below the bed with appropriate, independently derived
drag and lift coefficients. For individual grains this
approach is well verified (Fig.1) (Vollmer et al., 2002;
vollmer and Kleinhans, 2007). Semi-empirical relations for the velocity profile over rough beds, for the
effect of low submergence (h/Dm ) in shallow flow
and for turbulent pressure fluctuations over and in the
bed can be improved or replaced in the future pending further work on double-averaging of the Reynolds
equations (Nikora et al., 2002).
For mixtures, a description of the exposures and
their related pivot angles of individual grain sizes is
necessary. The effects of this bed structure on the
beginning of motion are very large. The advance of
insight by the model presented in this paper is therefore not that the empirical hiding functions can now
be replaced, but in pointing out the missing of knowledge on the bed structure of mixtures. In the future
the model may predict hiding functions for different
mixtures on a physical basis rather than an empirical
as has been common practice for half a century.
Simple bed structures were shown to be inadequate
as empirical results clearly transitioned between two
546
10
narrow
wider
wide
Egiazaroff
1
cr,i
10
10
10
10
10
fine skew
coarse skew
bimodal
Egiazaroff
1
cr,i
10
10
10
10
D (m)
6
Figure 7. Nondimensional reference shear stress of the
Shvidchenko et al. (2001) data and predicted by the model
with Kir.
types of simple structures. Two empirical bed structures were shown to give very different results while
their authors already warned that these functions may
only be valid for the specific beds from which they
were derived (Buffington et al., 1992).
This dependence of exposure functions on the grain
size-distribution and water-workedness of the bed is
entirely natural and must be investigated in experiments and nature. The physical basis for this dependence is illustrated by reference to three hypothetical
types of sediment for which the model will be suitable
after an appropriate exposure function has been constructed based on grain size-distribution and porosity
considerations. We are presently attempting to find an
exposure function that incorporates these three types:
Type I. A few fines settling from suspension on a
gravel bed will percolate deeply into the bed, which is
then defined as clast-supported. The negative exposure
of these grains will be too low for entrainment even by
in-bed pressure fluctuations.
Type II. Fine sediment larger than the pore size of the
clast-supported bed will not percolate into the bed but
remain near the surface. The exposures of these grains
will be large enough for entrainment. Water-working
CONCLUSIONS
ACKNOWLEDGEMENTS
MGK is supported by the Netherlands Earth and Life
sciences Foundation (ALW) with financial aid from
the Netherlands Organisation for Scientific Research
(grant ALW-VENI-863.04.016). A. Shvidchenko is
gratefully acknowledged for discussion and providing
his data and MGK thanks Gary Parker for discussion.
547
REFERENCES
Ashida, K. and M. Michiue (1971). An investigation of river
bed degradation downstream of a dam. In Proc. of 14th
IAHR Congress, Volume C30, Paris, pp. 19. IAHR.
Buffington, J., W. Dietrich, and J. Kirchner (1992). Friction angle measurements on a naturally formed gravel
streambed: implications for critical boundary shear stress.
Water Resources Research 28(2), 411425.
Buffington, J. and D. Montgomery (1997). A systematic analysis of eight decades of incipient motion studies, with
special reference to gravel-bedded rivers. Water Resources
Research 33(8), 19932029.
Colebrook, C. F. and C. M. White (1937). Turbulent flow
in pipes with particular reference to the transition regio
between the smooth and rough pipe laws. J. Inst. Civ. Eng.
7, 9, 99118, 318400.
Egiazaroff, I. (1965). Calculation of nonuniform sediment
concentrations. J. of the Hydraulics Division, ASCE
91(HY4), 225248.
Hofland, B. (2005). Rock and Roll turbulence-induced damage to granular bed protections, Published PhDthesis.
Delft, The Netherlands: Department of Civil Engineering
and Geosciences, TU-Delft.
Kirchner, J., W. Dietrich, F. Iseya, and H. Ikeda (1990). The
variability of critical shear stress, friction angle and grain
protrusion in water-worked sediments. Sedimentology 37,
647672.
Kleinhans, M. (2005). Upstream sediment input effects on
experimental dune trough scour in sediment mixtures. J.
of Geophysical Research 110, F04S06.
Kleinhans, M., A. Wilbers, A. D. Swaaf, and J. Van den Berg
(2002). Sediment supply-limited bedforms in sand gravel
bed rivers. J. of Sedimentary Research 72(5), 629640.
Kleinhans, M. G. and L. C. Van Rijn (2002). Stochastic prediction of sediment transport in sand-gravel bed rivers. J.
of Hydraulic Engineering 128(4), 412425.
Lisle, T. and S. Hilton (1999). Fine bed material in pools of
natural gravel bed channels. Water Resources Research
35(4), 12911304.
Nikora, V., D. Goring, and A. Ross (2002). The structure and
dynamics of the thin near-bed layer in a complex marine
environment: a case study in Beatrix Bay, New Zealand.
Estuarine, Coastal and Shelf Science 54, 915926.
548