Sediment Transport Model: 7.2.1 Bed Shear Stresses
Sediment Transport Model: 7.2.1 Bed Shear Stresses
Sediment Transport Model: 7.2.1 Bed Shear Stresses
7.1 Introduction
This chapter describes the sediment transport model implemented in COHE-
RENS. There are six sections. In the first one, a discussion is given of phys-
ical parameters and processes (bed shear stress, molecular viscosity, waves)
which are of importance for the sediment and the influence of sediments
on the physics through density gradients. In the next section a description
is given of the basic sediment parameters (Shield parameter, fall velocity,
critical shear stress), related to sediment transport. This is followed by an
overview of the bed and total load equations available in COHERENS. Then,
the suspended sediment transport module is presented. Finally, numerical
techniques, specific for the sediment module, are described.
311
312 CHAPTER 7. SEDIMENT TRANSPORT MODEL
for the 3-D, respectively 2-D case. In the equations, u∗b is the wall shear
(friction) velocity1 , ub the bottom current, |ū| the magnitude of the depth
mean current and Cdb the bottom drag coefficient given by
h κ i2 h κ i2
Cdb = or Cdb = (7.2)
ln zb /z0 ln H/z0 ) − 1
for the 3-D, respectively 2-D case. Here, κ is the Von Karman constant, zb
height of the first velocity node above the sea bed, z0 the roughness height
and H the total water depth.
In the discussions of the load formulae, given in the sections below, other
formulations, obtained from engineering practice, are mentioned and given
here for completeness:
g
Cdb = Chezy (7.3)
C2
gM 2
Cdb = 1/3 Manning (7.4)
H
g
Cdb = White-Colebrook (7.5)
[18 log (2H/5z0 )]2
The coefficients C and M are denoted as the Chezy, respectively Manning
coefficients.
In case of a rough bed, the roughness length z0 can be calculated using
the following relation:
ν kb kb
z0 = 0.11 + ≈ (7.6)
u∗ 30 30
Here, kb is the roughness height from Nikuradse, ν the kinematic viscosity
and u∗ the friction velocity.
Different kinds of roughness height are considered:
• Physical or “form” roughness kb representing the heights of the elements
composing the bottom roughness. The corresponding physical bottom
stress is used as boundary condition for the momentum flux at the
bottom.
• “Skin” roughness ks usually related to the median particle size. Skin
stress is used in the sediment module to calculate the amount of sedi-
ment material resuspended from the sea bed, once the stress exceeds a
critical value (see below).
1
Since the friction velocity always refers to the bottom in this chapter, the subscript b
will be omitted in the following.
7.2. PHYSICAL ASPECTS 313
Aw = Hs /2 , ωw = 2π/Tw (7.7)
The near-bottom wave orbital velocity determined from linear wave theory
is given by
Aw ωw
Uw = (7.8)
sinh kw H
with the wavenumber kw obtained from the dispersion relation
The latter equation is solved for kw using the approximate formula of Hunt
(1979) s
f (α)
kw = ωw (7.10)
gH
with
−1
f (α) = α + 1.0 + 0.652α + 0.466422α2 + 0.0864α4 + 0.0675α5 (7.11)
314 CHAPTER 7. SEDIMENT TRANSPORT MODEL
and
ωw2 H
α= (7.12)
g
With this option, it is assumed that the near bed wave period is equal to the
mean wave period, such that the near bed wave excursion amplitude can be
written as
Uw
Ab = (7.13)
ωw
The combined effect of currents and waves on the bed shear stress is calcu-
lated using the method described in Soulsby (1997). Two types of bottom
stress are defined
• The mean value over the wave cycle which should be used for currents
" 3.2 #
τw
τm = τc 1 + 1.2 (7.14)
τc + τw
• The maximum value during the wave cycle used in the criterion for
resuspension p
τmax = τm 2 + τ 2 + 2τ τ cos φ (7.15)
w m w w
1
τw = fw Uw2 (7.16)
2
Two formulations are used for the wave friction
• Swart (1974)
Here ρ is the density of the mixture, ρw the density of the fluid (including the
effects of temperature, pressure and salinity), cn the volume concentration
of fraction n, N the number of sediment fractions and ρs,n the particle den-
sity for sediment fraction n. A stable vertical sediment stratification leads to
damping of turbulence and affects the settling velocity of the sediment. How-
ever, care must be taken when applying this in combination with hindered
settling (Section 7.3.3.2), because hindered settling models already account
for the increased buoyancy of the mixture. Thus using a changed equation
of state in combination with hindered settling will result in too low settling
velocities.
The numerical methods, described in Section 5.3.13 for discretising the baro-
clinic gradient are easily extended to include sediment stratification.
Implementation
The following switches are available:
iopt sed tau Selects type of roughness height zs used for sediment transport
1: set equal to the form roughness used in the hydrodynamics
2: user-defined uniform value
3: spatially non-uniform value supplied by the user
iopt waves Disables/enables wave effects and selects type of input data wave
data, for use in the sediment transport models.
7.3. SEDIMENT PROPERTIES 317
0.3
+ 0.055 1 − e−0.02d∗
θcr = (7.32)
1 + 1.2d∗
This equation should only be used in combination with the bed load and
total load equations from Wu et al. (2000). In COHERENS, the critical shear
stress is calculated for each fraction separately at each horizontal location,
using the local values of the kinematic viscosity and the mixture density.
The sediment transport model in COHERENS has an option available for
the user to manually set the critical shear stress. The user can either set
the critical shear stress to a uniform value throughout the whole domain,
or to a spatially varying value. Be aware that the value given should be
the kinematic critical shear stress defined as the square of the critical shear
velocity u2∗,cr = τcr /ρ.
7.3. SEDIMENT PROPERTIES 319
A hiding and exposure factor can be implemented for the critical shear stress
τcr or the critical Shields parameter θcr to account for the change in sediment
transport when different fractions are present. In general, the correction
factor will increase the critical value for the smaller, hidden fractions and
will reduce the critical threshold for motion for the coarser, exposed, frac-
tions. Several formulations can be found in literature, usually as a function
ξ(dn /d50 ).
Two methods are available in COHERENS for hiding and exposure. The
first one uses the formulation of Wu et al. (2000) based upon a stochastic
relation between size and gradation of bed materials and the hidden and
exposed probabilities. The probability of particles dm in front of particles dn
can be assumed to be the fraction of particles dm in the bed material, pbm .
The total hidden and exposed probabilities of particles dn can be described
by:
N
X dm
phn = pbm (7.34)
m=1
dn + dm
N
X dn
pen = pbm (7.35)
m=1
dn + dm
where d50 is the median grain size diameter, which, in case multiple size
fractions are present, means that particles with sizes less than d50 account
for 50% of the total mass.
320 CHAPTER 7. SEDIMENT TRANSPORT MODEL
1/m m
s
2/m 3 1/m
ν ν 1 A 4 d∗ 1 A
ws = Rep = + − (7.39)
d d 4 B 3B 2 B
(1 − cf )(1 − c)
ws = ws,0 (7.44)
1 + 2.5cf
Here ws,0 is the settling velocity of a single mud-floc, cf = c/cgel , cgel is the
gelling concentration (the mass concentration at which the mud flocs form a
322 CHAPTER 7. SEDIMENT TRANSPORT MODEL
dp 3−nf
cgel = ρs (7.45)
df
ws 1 + aG
= (7.46)
ws,r 1 + bG2
p
Here ws,r is a reference velocity, G the shear rate, defined as G = ε/ν
and a, b empirical constants. In order to estimate the shear rate if the
turbulent dissipation ε is not known (e.g. when an algebraic turbulence
model is used), G can be obtained assuming equilibrium between production
and dissipation of turbulent kinetic energy (4.204) which gives 2
r
νT M 2 − λT N 2
G= (7.47)
ν
where νT , λT are the turbulent diffusion coefficients for respectively mo-
mentum and density and M , N the shear and buoyancy frequencies (see
Section 4.4 for details).
A similar empirical approach to flocculation, which considers the influence
of the sediment concentration is given in Van Rijn (2007b)
ws α
= 4 + log10 2c/cgel = φf loc (7.48)
ws,r
that Van Rijn (2007b) suggests α = dsand /d50 − 1, with dsand the diameter of
the transition from sand to silt (62 µm).
It is also possible to combine the effects of these two models in COHE-
RENS. In this case, the settling velocity is multipled by the two corrections
factors on the right of (7.46) and (7.48).
Implementation
The methods in this section are selected with the following switches
iopt sed taucr Selects type of method for the critical shear stress
1: user-defined value for each fraction
2: Brownlie (1981) equation (7.31)
3: Soulsby & Whitehouse (1997) equation (7.32)
4: Wu et al. (2000) equation (7.33)
iopt sed hiding Type of hiding/exposure factor for the critical shear stress
0: hiding disabled
1: Wu et al. (2000) equation (7.36)
2: Ashida & Michiue (1972) equation (7.37)
iopt sed ws Type of method for the settling velocity
1: user-defined value for each fraction
2: Camenen (2007) formulation (7.39) for sand
3: Camenen (2007) formulation (7.39) for mud
4: Stokes formula (7.40)
5: Soulsby (1997) formula (7.41)
6: Zhang & Xie (1993) equation (7.42)
iopt sed hindset Formulation for hindered settling
0: hindered settling disabled
1: Richardson & Zaki (1954) equation (7.43)
2: Winterwerp & van Kesteren (2004) formula (7.44)
iopt sed floc Type of flocculation factor for the settling velocity
0: flocculation effect disabled
1: Van Leussen (1994) equation (7.46)
2: Van Rijn (2007b) equation (7.48)
3: combination of the two previous methods
324 CHAPTER 7. SEDIMENT TRANSPORT MODEL
where qb is the bed load solid discharge per unit width in m2 /s.
The Meyer-Peter-Muller and the Einstein equations are two examples
using dimensionless parameters for relating solid discharge to the shear stress
at the bed. However, the solid discharge rates of bed load, suspended load
and total load in this numerical model are given as the dimensional quantities
expressing the volumetric discharge per unit flow-perpendicular width (in
m2 /s).
The bed load transport is assumed to occur below the roughness level z0 ,
i.e. below the lowest σ-layer. The bed load transport vector is decomposed
in the same manner as the horizontal velocity vector. The bed load transport
magnitude is computed as the vector sum of transport components along the
ξ1 and ξ2 directions, assuming local orthogonality of the computational grid:
The ξ1 and ξ2 components are computed at the U-, respectively V-nodes grid
points. This avoids that interpolations of the velocity components to e.g. the
C-points need to be taken prior to the determination of sediment transport
rates.
cl 2 πd2
FD = ρU (7.53)
2 p 4
The coefficient cl can be seen as a correction of the drag force for lifting, Up
is the particle migration velocity.
The friction force experienced by the particles can be expressed as:
πd3
Ff = ρg(s − 1) β (7.54)
6
where β is a dynamic friction coefficient.
When both forces are in equilibrium the following is obtained:
Up h p i
= α 1 − θ0 /θ (7.55)
u∗
with θ > θcr and fd the dynamic friction coefficient of a grain, assumed to
be equal to 0.51.
This relation was proven to be accurate for fine to medium sands, with
experiments using sand with diameters of 190, 270 and 930 µm.
The following steps are taken to calculate the suspended load and bed
load transport. Firstly, the settling velocity of fraction n is calculated with
formula (7.42) from Zhang & Xie (1993). Then the hiding and exposure
factors phn and pen are computed as defined in Section 7.3.2.1.
The critical shear stress (taking into account hiding and exposure) for
fraction n is calculated:
0.6
phn
τcr,n = 0.03(ρs − ρ)gdn (7.62)
pen
The bed load sediment transport is then determined with:
2.2
τb
Φb,n = 0.0053fn max − 1, 0 (7.63)
τcr,n
whereby the hiding/exposure factor ξn for sediment class n, is always
calculated with the method of Wu et al. (2000), as given by (7.34)–(7.36).
and Φbc = Φbn = 0 if θmax ≤ θcr . θm is the value of θ averaged over the wave
cycle, θw , is the amplitude of the oscillatory component of θ and φ the angle
between current and wave propagation direction. The maximal value of the
Shields parameter θmax over the wave cycle is calculated as follows
p
θmax = θm 2 + θ 2 + 2θ θ cos2 φ (7.65)
w m w w
where θc is the current related Shields parameter. Note that (7.66) and (7.65)
are the dimensionless versions of (7.14)–(7.15).
From the equations it can be seen that when the wave direction is per-
pendicular or along the current direction, the component at right angle to
the current direction is zero. No wave asymmetry is taken into account.
In the absence of waves, one has Φbc,2 = Φbn = 0 so that
√
Φb = Φc = 12 θ max (θ − θcr,n , 0) (7.67)
in which
where φ is the angle between wave propagation and current direction and Uw
the near-bed wave orbital velocity given by (7.8).
The instanteneous near-bed flow velocity magnitude due to combined
waves and currents is determined as the magnitude of the vector sum of the
current and wave components
h 2 i1/2
Uδ,cw (t) = Uw sin ϕ + u2cδ + 2ucδ Uw cos φ sin ϕ (7.79)
Z 2π
1 uc,δ + Uw sin ϕ cos φ
q̄bc = qb (t) dϕ
2π 0 Uδ,cw (t)
Z 2π
1 Uw sin ϕ sin φ
q̄bn = qb (t) dϕ (7.80)
2π 0 Uδ,cw (t)
• a change in the bed load transport rate and direction beyond incipient
motion.
332 CHAPTER 7. SEDIMENT TRANSPORT MODEL
In COHERENS, the second and third effect can be modeled using the
switch iopt sed slope. The second effect was discussed in Section 7.3.2. For
the third effect, a correction for the transport rate vector magnitude has to
be combined with a correction on the transport direction, i.e. the correction
has a longitudinal component parallel to the current vector and transverse
component, normal to the current.
The longitudinal ∂h/∂c and transverse ∂h/∂n bed slopes are given by
∂h 1 ∂h 1 ∂h
= cos φc + sin φc (7.81)
∂c h1 ∂ξ1 h2 ∂ξ2
∂h 1 ∂h 1 ∂h
= cos φc − sin φc (7.82)
∂n h2 ∂ξ2 h1 ∂ξ1
where φc is the angle between the ξ1 - and current directions.
The slope effects are modeled using the relation of Koch & Flokstra
(1981). They developed the following relation for a correction factor αs on
the bed load transport rate
∂h
αs = 1 − βs (7.83)
∂c
with βs = 1.3. They included also the effect of the slope transverse to the
flow by the change in direction of δs degrees:
∂h
tan δs = βs (7.84)
∂n
Note that these corrections are only applied for bed load equations and not
for total load.
The bed load formulae in the previous subsections were derived along the
current direction and an additional transverse component in the presence of
waves. The components along the coordinate directions are calculated by
the following transformations
qb1 = αs qbc cos(φc + δs ) − qbn sin(φc + δs )
qb2 = αs qbc sin(φc + δs ) + qbn cos(φc + δs ) (7.85)
where
Implementation
The bed load formulae in this section are selected with the following switches
iopt sed slope Disables (0) or enables (1) bed slope efects using the Koch &
Flokstra (1981) formulation.
with d = d35 , fn the fraction of size class n in the bed and |ū| the magnitude
of the depth averaged current.
This transport formula was derived by uniting bed load and suspended
load transport relations in one function through a transition in the dimen-
sionless d∗ parameter. A revised set of values for the parameters mw and
7.5. TOTAL LOAD 335
Cw which are used in the current implementation, have been derived after
experiments in 1990:
nw = 1 − 0.243 ln d˜∗
mw = 1.67 + 6.83/d˜∗
ln Cw = 2.79 ln d˜∗ − 0.426(ln d˜∗ )2 − 7.97
Aw = 0.14 + 0.23/d˜∗1/2 (7.91)
3 Ucw (t)
qt = 40fn ws dn θcw (7.92)
|Ucw (t)|
in which ws is the particle fall velocity, dn the grain size of class n, with
fraction fn , θcw the instantaneous Shield parameter of dimensionless shear
stress due to currents and waves and Ucw the current due to current and
waves, whose magnitude is given by (7.79).
The Shields parameter is computed here with the velocity due waves and
currents and with a combined friction factor for waves and currents:
fcw |Ucw (t)|2
θcw (t) = (7.93)
2(s − 1)gd50
with
where fc is the friction factor due to currents and fw the wave-related friction
factor given by (7.18). The wave boundary layer thickness δ is the same as
the one for the Van Rijn (2007a) bed load formula.
The mean total load is obtained by decomposing the instantaneous sedi-
ment total load vector into components, along and transverse to the current
direction, and integration over the wave period using (7.80) with qb replaced
by qt . The components along the coordinate axes are obtained using (7.85)
without slope effects (i.e. αs = 1, βs = 0, δs = 0).
The integration over the wave period is executed in the code with the
numerical integration method of Gaussian quadrature. For details about the
applied method see Section 7.7.3. The number of data points needed in this
method to reach a stable solution is optimised for the Madsen & Grant (1976)
method. The fact that the transport is determined with the fifth power of the
Shields number makes that the oscillating signal of the wave-current velocity
results in a very peaked instantaneous transport. The consequence is that a
relatively high number of points is needed to capture the sharp peak in the
transport during maximal near-bed velocity. The optimal number for this
method was determined at 15 points per wave period, see Figure 7.1.
The procedure is repeated for each class of sediment grain size defined by
the user.
7.5.4 Wu et al (2000)
The calculation procedure for bed load transport is described in Section 7.4.5.
Adding the following equation for suspended load to the bed load formula
results in the total sediment transport due currents only:
1.74
−4
p τb |ū|
qs,n = 0.262.10 (s − 1)gd3 −1 (7.98)
τcr,n ws,n
Here, the settling velocity is calculated with equation (7.42) of Zhang &
Xie (1993) and the critical shear stress using relation (7.33) from Wu et al.
(2000).
Figure 7.1: Convergence of Gauss-Legendre quadrature for wave period integration of instantaneous sediment trans-
port. Different peak orbital velocities and wave-current angles are considered.
338 CHAPTER 7. SEDIMENT TRANSPORT MODEL
computed following the approximate formula, but with the fully integrated
instantaneous oscillating transport, as described in Section 7.4.7. The sus-
pended load, however, is computed using the explicit approximation formula
of Van Rijn (2003).
The suspended sediment transport due to currents and waves is calculated
by
qs = qsc + qsw (7.99)
which is the sum of the current-related suspended transport (with wave agita-
tion), directed in the main current direction and the wave-related suspended
transport in the wave propagation direction (associated with wave asymme-
try). For simplicity, it is assumed in the following that there is only one grain
size class in the sediment bed. In case of multiple fractions, the procedures
are repeated for all classes, and the transports are multiplied with each class’
fraction in bed.
2
log(H/30z0 )
µc = fce /fc
= (7.108)
log(H/1.5d50 )
2 2
ln(90δw /ka ) ln(30H/ks ) − 1
αcw = (7.109)
ln(90δw /ks ) ln(30H/ka ) − 1
αcw,max = 1 (7.110)
with δw the wave boundary layer thickness and ka the apparent roughness.
For definitions, see Section 7.4.7. Note that the skin roughness height is set
to 1.5d50 which is the actual median grain size in the bed at each location
from the particle size distribution in the highest bed layer.
e
The effective wave-related bed-shear τb,w is in turn determined by:
e
τb,w = µw τb,w (7.111)
In the 2003 version of the van Rijn model, µw is given by 0.125(1.5−Hs /H)2 .
This parameter will be different in the 2007 version described in the next sec-
tion. The bed shear stresses due to waves and due to currents are determined
using the same friction factors as in the bed load transport model, see Sec-
tion 7.4.7.
terms of onshore and offshore directed peak velocities. These velocities have
been computed using the method of Grasmeijer & Van Rijn (1998). The
wave propagation directed suspended transport is given by:
with
Due to the relation adopted for turbulence damping due to high sediment
concentrations, the modified Rouse number in the approximative formulation
for current-related suspended load (Zc in equation (7.104)) takes a different
form
ws
Zc = (7.113)
βc κu∗ φd
Here, φd is the turbulence damping function used, originally varying with
concentration, but here evaluated for the reference concentration
Implementation
The type of total load formulation is selected with the following switches
ws ∂z h3 ∂s
ũs = − = −ws (7.117)
h1 ∂ ξ˜1 s h1 ∂ξ1 z
ws ∂z h3 ∂s
ṽs = − = −ws (7.118)
h2 ∂ ξ˜2 s h2 ∂ξ2 z
case (see below), are modeled by source terms in the depth averaged sediment
concentration equation
∂c̄ 1 ∂ ∂
+ h2 c̄ū + h1 c̄v̄ =
∂t h1 h2 ∂ξ1 ∂ξ2
1 ∂ h2 ∂c̄ ∂ h1 ∂c̄ E−D
D̄H + D̄H + (7.119)
h1 h2 ∂ξ1 h1 ∂ξ1 ∂ξ2 h2 ∂ξ2 H
For details see Section 4.3.2.
Garcia & Parker (1991) showed that the flux can be expressed as an non-
dimensional flux E∗ ≡ E/ws and that in situations not too far from equili-
brium, this non-dimensional entrainment is equal to the near bed reference
concentration (i.e. E∗ = ca ). Therefore one can write:
∂cn
− DV (a) = fn ws,n (a)ca,n (7.123)
∂z a
Here ca,n is the near-bed reference (equilibrium) sediment concentration at
a reference height an for sediment size class n with fractional amount fn at
the bed.
In COHERENS, the near bed reference concentration for sand is calculated
either by the method of Smith & McLean (1977) or by the one in Van Rijn
(1984a), which can be selected with the switch iopt sed bbc.
The near-bed boundary condition of Smith & McLean (1977) is given for
each fraction n by:
Tn
ca,n = 0.0024cmax
1 + 0.0024Tn
a = ks + 26.3(θn − θcr,n )dn
ks = 30z0 (7.124)
dn Tn1.5
ca,n = 0.015 (7.125)
an d0.3
∗n
with d∗ defined by (7.29). The reference level a (either half the size of the
dunes or the roughness length scale ks ) is limited to be between 0.01H and
0.1H.
It may occur that the reference level an is not located within the bottom
grid cell where the bottom boundary condition is applied. The methods used
in COHERENS to overcome this problem are described in Section 7.7.1. The
influence of waves on the erosion of sediment is included in this equation only
by the change in the bed shear stress due to the near bed wave motion using
the equation of Soulsby (1997).
7.6. SUSPENDED SEDIMENT TRANSPORT 345
H
En − Dn = (c̄e,n − c̄n ) (7.127)
Te,n
is given by
R
ceq H − z̃ a
= (7.128)
ca z̃ H − a
where a is the reference level, H the total water depth, z̃ = z + h the
height above the sea bed, ca the reference concentration, R = ws /βκu∗ the
Rouse number, ws the particle fall velocity, u∗ the skin shear velocity, β the
inverse of the Prandtl-Schmidt number (see Section 7.6.4) and κ the von
Karman constant. The exponent expresses the relative importance of the
suspended load, where it is equal to about 5 for bed load only and decreases
to unity for suspended load up to the surface (Van Rijn, 1993). The reference
concentration is defined in Section 7.6.3.1.
Another way to calculate the depth averaged sediment concentrations is
by using one of the fomulae for suspended transport, described in Section 7.5.
The depth averaged equilibrium concentration is then given by
qs,e
c̄e = (7.129)
|U|
Here qs,e is the equilibrium suspended sediment transport and |U| the
magnitude of the depth-integrated current (in m2 /s).
The adaptation time scale Te is given by:
H
Te = T∗ (7.130)
ws
Here, T∗ is a non-dimensional correction factor, which depends on the shape
of the sediment concentration profile, parametrized by the Rouse number R
defined below. For COHERENS, this relation was obtained by performing
simulations with COHERENS in water column (1DV) mode and determining
the adaptation time scale from the data (see Figure 7.2). Through this data
an eight order polynomial was fitted:
T∗ = p1 R8 + p2 R7 + p3 R6 + p4 R5 + p5 R4 + p6 R3 + p7 R2 + p8 R + p9 (7.131)
with:
R = min(wsb /u∗ , 1.14) (7.132)
and
p1 = 67.342 p2 = −321.0 p3 = 614.14
p4 = −592.75 p5 = 292.15 p6 = −62.141
p7 = 3.667 p8 = −2.2571 p9 = 0.97978 (7.133)
and wsb the settling velocity at the bottom.
7.6. SUSPENDED SEDIMENT TRANSPORT 347
1
Data 1DV simulation
0.9 8th degree polynomial
0.8
0.7
0.6
T ws/h
0.5
0.4
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1
ws/u*
For cohesive sediments, the erosion and deposition rates in 2-D are obtained
in similar way as for sand, using an equilibrium time schale and a depth
averaged concentration (Section 7.6.3.3). For cohesive sediment, the only way
in which the depth averaged equilibrium concentration can be calculated is by
averaging the Rouse profile (equation 7.128). However, a near-bed reference
concentration and reference location are needed to obtain the depth averaged
concentrations. The near bed reference concentration is obtained from the
near bed erosion rate E using
E
ca = (7.134)
ws (a)
The near bed reference location a is, in the 2-D case, an extra tunable
parameter height c cst. This is not very different from the usual approach
to use a threshold of deposition as an extra tunable parameter for simu-
lating cohesive sediment transport. However, the present approach has the
advantage that unphysical results that may result from the threshold of de-
position method (such as the unlimited erosion leading to unphysically high
sediment concentrations) are avoided. The same time schale is used as for
the calculation of depth averaged sand transport.
348 CHAPTER 7. SEDIMENT TRANSPORT MODEL
DV = βνT (7.135)
Implementation
The following switches are used for the transport of suspended sediment:
iopt sed mode Type of mode for applying the sediment transport model
1: bedload transport only computed by a formula, which is determined
by iopt sed bedeq
2: suspended load transport only (computed with the advection-diffusion
equation)
3: bedload and suspended transport (i.e. option 1 and 2 together)
4: total load transport computed with a formula, which is determined
by iopt sed toteq
iopt sed nodim The number of dimensions used in the sediment formula-
tion.
2: depth averaged transport3
3: 3-D sediment transport.
3
Note that iopt sed nodim is always set to 2 if iopt grid nodim = 2.
350 CHAPTER 7. SEDIMENT TRANSPORT MODEL
ones for a scalar quantity, as described in Section 5.5. However, there are
some differences, mainly related to the modeling of erosion and deposition,
which are discussed here.
7.7.1 Erosion-deposition
7.7.1.1 Three-dimensional sediment transport
In three-dimensional simulations, erosion and deposition come into play through
the near-bed boundary condition. For mud, the erosion rate is given directly
by the equation of Partheniades (1965). For sand, the erosion E is calcu-
lated from the near bed reference concentration ca at a height a above the
bed using E = ws ca . However, the height a does not normally coincide with
the used computational grid. Hence, a transformation has to be made, such
that the erosion is calculated at a location coinciding with the computational
grid.
In COHERENS, three methods are available to do this, which can be set
with the switch iopt sed bbc type. The first two options are based on the
method used in the EFDC model. In this method, the flux is calculated
based on an averaged of either the lowest cell (iopt sed bbc type = 1) or the
cell in which a lies (iopt sed bbc type =2). At a given reference level a, the
net upward flux is given by
Herein, the following expression is used for the non-equilibrium near bed
sediment profile (derived assuming a linear varying eddy viscosity near the
bed)
a R F
0
c = ca − (7.146)
z ws
Combining this expression with equation (7.144) gives
Z zn R
ca a
c̄eq = dz (7.147)
∆zn zn−1 z
aR c a 1−R 1−R
c̄eq = zn − zn−1 (7.148)
∆zn (1 − R)
and for R = 1
aca
c̄eq = [ln zn − ln zn−1 ] (7.149)
∆zn
The range of values for the Rouse number R are limited by the theo-
retical maximum of 2.5, since no suspension can exist when the settling
velocity is larger than the shear velocity. When the method of cell-averaged
reference concentration described above is used while the reference level is
located within the bottom cell, the integrated profile (equations (7.148)-
(7.149)) starts from the bottom of the near-bed cell, at level z0 . Since the
Rouse profile is not valid that close to the bed, another lower limit for inte-
gration is used, with the minimum set at the Nikuradze roughness ks = 30z0 .
The level z0 , in turn, is the level at which the assumed logarithmic velocity
profile becomes equal to zero.
This method with iopt sed bbc type =2 can give very accurate results,
provided that a high resolution is used near the bed. However, when the
resolution is relatively low (and thus the cells near the bed are relatively
large), the averaging of the profile leads to an overestimation of the concen-
trations in the lowest cells and therefore in sediment concentrations that are
too high. Therefore, this method is not recommended for practical applica-
tions. Instead, another method should be used by setting iopt sed bbc type
= 3. In this method, the Rouse profile (equation (7.128)) is used to calculate
the sediment concentration from the given boundary condition at the first
C-node above the bed.
The deposition flux is just an advection term from the lowest cell. The
type discretisation of this term at the bed is determined by the switch
iopt scal depos. The options are
7.7. NUMERICAL METHODS 353
H n+1
ce − θv cn+1 − (1 − θv )cn
E−D = (7.152)
Te
In case of 3-D simulations, erosion is taken explicitly whereas deposition is
integrated semi-implicitly with the implicity factor θa which is the same as
the one used for the vertical advective term in scalar transport equations (see
Section 5.5.3).
Letting
∂h
β∗ = βs = tan δs (7.154)
∂n
one has
1
cos δs = cos arctan β∗ = p
1 + β∗2
354 CHAPTER 7. SEDIMENT TRANSPORT MODEL
β∗
sin δs = sin arctan β∗ = p (7.155)
1 + β∗2
so that
cos φc − β∗ sin φc
cos(φc + δs ) = p
1 + β∗2
sin φc + β∗ cos φc
sin(φc + δs ) = p (7.156)
1 + β∗2
with ϕi = π(1 + xi ).
7.7. NUMERICAL METHODS 355
a∗ R 0 φ −φ
Z
R
c̄e = ca e e − 1 dφ
1 − a∗ ln a∗
n
ca a∗ R X R
' − ln a∗ wi eφi e−φi − 1 (7.161)
2 1 − a∗ i=1
with φi = 21 (1 − xi ) ln a∗ .
• P
The total volume of negative concentrations is determined: M− =
ijk cijk (cijk < 0)∆Vijk where ∆Vijk = h1;ij h2;ij h3;ijk is the volume
of a grid cell.
Implementation
Numerical methods, described in this section, are selected with the following
switches:
356 CHAPTER 7. SEDIMENT TRANSPORT MODEL
iopt sed bbc type Selects the method to transpose the near bed boundary
condition to the computational grid. It is strongly recommended not to
change the default value of 3.
1: EFDC method applied to lowest cell (not recommended)
2: EFDC method applied to the first the cell above (not recommended)
3: Using the Rouse profile
iopt sed filter Disables (0) or enables (1) the application of the Bartnicki filter
to prevent the occurrence of negative concentrations .