Upload
herbert-french
View
219
Download
0
Tags:
Embed Size (px)
Citation preview
Incorporating sediment-transport capabilities to
DSM2
Kaveh Zamani and Fabián A. BombardelliDepartment of Civil & Environmental Engineering
University of California, Davis
Technical Advisory Committee, Department of Water Resources,
January 13, 2010
2
Background
AdvectionGo with the flow
Advection Dispersion Erosion Deposition Source/Sink Tributaries
Erosion /Entrainment + and –
Sources and Sinks
DispersionSpreading out
Deposition/Settling
Progress to Date: Single Channel
Next step: Complete single channel model
Next step: Extend model to a channel network
Background Sediment Transport Math Treatment Num. Method Tests Questions
3
Modes of sediment transport
1-Bed LoadMainly empirical formulasLagrangian solution for each particle
2-Suspended LoadAdvection-Dispersion-Sink/Source
3-Wash Load There is a third mode of sediment
transport called wash load, whereby very fine particles are transported downstream with very little interaction with the bed sediments.
During floods, the wash load is deposited in the floodplains (usually ignored in numerical simulations).
Background Sediment Transport Math Treatment Num. Method Tests Questions
4
Modes of sediment transport
Sediment transport as bed-load in rivers
Source: Prof. Dietrich’s website
Background Sediment Transport Math Treatment Num. Method Tests Questions
5
Modes of sediment transport in the Delta
We find the transport of sediment as bed-load and in suspension. Not much information exists about wash load.
In large portions of the Delta, the sediment can be cohesive.
The USGS has numerous stations in which sediment in suspension is monitored via optical backscatter sensors (OBS).
Background Sediment Transport Math Treatment Num. Method Tests Questions
6
Modes of sediment transport in the Delta
Source: USGS website
Bed-forms at Garcia Bend in Winter 2000.
Background Sediment Transport Math Treatment Num. Method Tests Questions
7
Mathematical treatment of the problem
Tracking individual particles is not feasible for a system of the size of the Delta. Then, we need to use the continuum approach.
Source: Abad et al., 2007
Background Sediment Transport Math Treatment Num. Method Tests Questions
8
Mathematical treatment of problem: Sediment in
suspension
SSCqDE
s
CKA
ss
CQ
t
CALL
ss
ss /
Sediment transport in suspension:
A: cross-sectional wetted area (m2)
sC: volumetric cross-sectional-averaged concentration of sediment in suspension (-)
Q: flow discharge (m3/s)
sK: dispersion coefficient (m2/s)
E Dand : entrainment rate of sediment into suspensionand deposition rate of sediment per unit width, respectively (m2/s)
Lq LCand: lateral discharge (m2/s), and concentration (-), respectively
SS / : non-point sources/sinks (m2/s)
Background Sediment Transport Math Treatment Num. Method Tests Questions
9
Mathematical treatment of the problem: bed-load transport
Sediment transport as bed-load:
bq: bed-load solid discharge per unit width (m2/s)
sR
pd: sediment particle diameter (m)
: specific gravity (-)
stressshearexcessfdgR
q
P
b 3
g: acceleration of gravity (m/s2)
The equation for sediment in suspension comes from the integration in the cross section up to zb.
Background Sediment Transport Math Treatment Num. Method Tests Questions
10
Mathematical treatment of the problem: Entrainment and Deposition
: sediment concentration at the bottom (-)
BwEE ssBwCD ssl
slC
Recent developments in sediment transport refer to several activelayers, which could be incorporated in a second stage of themodel development.
ws : settling velocity
Background Sediment Transport Math Treatment Num. Method Tests Questions
11
Numerical Method: Operator Splitting
2nd order accurate Strang type splitting algorithm RDcucc xxxt
Background Sediment Transport Math Treatment Num. Method Tests Questions
],[),,(),()121
***
nnnnt tttxtcxtcuccx
],[),,(),()2 121
*******
nnnnt tttxtcxtcRDccx
],[),,(),()3 121
*****
nnnnt tttxtcxtcuccx
12
Numerical Method: Diffusion
2nd order, implicit )()(
x
CAK
xt
AC ss
s
13
11
1
11
31
1
212
1
212
1
212
11
212
)()()()(ni
ni
ni
n
isn
isn
isni
n
is
C
C
C
AKx
tAK
x
tAK
x
tAAK
x
t
11
121
21
211
212
)()()()()()()()()1(
)(
nis
n
isnis
n
isnis
n
isnis
n
isnis CAKCAKCAKCAK
x
tAC
Background Sediment Transport Math Treatment Num. Method Tests Questions
)(2
: 2110
12 xOs
x
CCconditionboundaryNeumann n
nS
nS
knownisCconditionboundaryDirichlet ns1:
13
Numerical Method: Advection
)(2
)22/12/1
2/12/1
ni
ni
ni
ni DS
A
tCC
`12
1
22)1
2/12/1
ni
ni
nnin
i
ni C
A
Q
x
tC
t
t
Cx
x
CCC i
SSDSSx
CAK
xx
QC
t
AC ss
ss //)()()(
Background Sediment Transport Math Treatment Num. Method Tests Questions
2nd order explicit
fluxLimitedCD
xCDxx
CC
niited
niited
ni
)(
)(
lim
lim
)()(2
)31
21
21
21
21
1
ni
ni
n
i
n
ini
ni CSCS
t
x
QCQCtACAC
x
t n
n+½
n+1
i-1 i+1i
14
Numerical Method: Flux limiter
Gibbs phenomenon Admissible limiters region for 2nd order schemes (Sweby
1984)
van Leer limiter (1974)
)1974(1
1
1
Leervanr
rr
CC
CCr
ii
iii
Background Sediment Transport Math Treatment Num. Method Tests Questions
15
Test: Mesh Convergence
Norms as measure of functions
Background Sediment Transport Math Treatment Num. Method Tests Questions
ivvL max
n
v
n
v
L
n
ji
ij
1
1
n
v
L
n
ji
i
21
2
2
(The most restrictive norm)
(The less restrictive norm )
16
Test: Mesh Convergence II
Background Sediment Transport Math Treatment Num. Method Tests Questions
`Table 1: Errors and Norm of Errors in Different Mesh Sizes for Θ=0.5
Nu
m o
f volu
mes
L1
Rate
of L
1 c
han
ge
L2
Rate
of L
2 c
han
ge
L
∞
Rate
of L
∞ ch
an
ge
Dd
t/dx²
Sta
bility
252.030E-
031.094E+0
08.407E-
075.355E+0
03.622E-
043.844E+0
03.858E-01 O.K
501.855E-
031.025E+0
01.570E-
075.797E+0
09.422E-
054.090E+0
01.482E+0
0O.K
1001.809E-
031.006E+0
02.708E-
085.760E+0
02.304E-
054.072E+0
06.050E+0
0O.K
2001.798E-
031.001E+0
04.701E-
095.845E+0
05.657E-
064.151E+0
02.445E+0
1O.K
4001.796E-
031.000E+0
08.044E-
106.301E+0
01.363E-
064.611E+0
09.827E+0
1O.K
8001.795E-
031.993E+0
01.277E-
106.708E+0
02.956E-
073.385E+0
03.941E+0
2O.K
16009.008E-
041.002E+0
01.903E-
111.518E+0
08.732E-
081.000E+0
01.578E+0
3O.K
32008.991E-
04X
1.254E-11
X8.732E-
08X
6.317E+03
O.K
Second order
17
Tests: Analytical Solutions I (Diffusion)
Background Sediment Transport Math Treatment Num. Method Tests Questions
Case D_D
0
0.02
0.04
0.06
0.08
0.1
0.12
-20 -15 -10 -5 0 5 10 15 20
X
Co
nce
ntr
atio
n
t=3.2 exact t=6.4 exact
t=9.6 exact t=12.8 exact
t=16 exact t=19.2 exact
t=25.6 exact t=32 exact
t=38.4 exact 3.2 model
6.4 model 9.6 model
12.8 model 16 model
19.2 model 25.6 model
32 model 38.4 model
2
2
x
cD
t
c
0,0
0,)0,(..
:
x
xxxCCI
toSubject
0),(:.. txCxCB
Dt
etxC
Dt
x
exact 4),(
4
2
18
Tests: Analytical Solutions II (Diffusion)
2
2
x
cD
t
c
Background Sediment Transport Math Treatment Num. Method Tests Questions
Dirichlettcx
NeumanntDx
cxCB
2),1(:)1(
2exp05.0sin22:)1.0(..
2
tDxxtxcexact
2
2exp5.0cos42),(
2
cos42)0,(:..
:
xxxcCI
toSubjected
Cell number
19
Tests: Advection
Background Sediment Transport Math Treatment Num. Method Tests Questions
Flux limiter
Numerical diffusion
20
Tests: A-D-R (to be included)
Background Sediment Transport Math Treatment Num. Method Tests Questions
Dt
ec
Dt
utxt
4
4
2
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.11
-16 -8 0 8 16 24 32 40 48 56 64 72 80 88 96 104 112X
C
Source: McDonald, 2007
21
Tests: Experimental Data I (to be done)
Comparison with experimental lab dataNewton (1951).
Background Sediment Transport Math Treatment Num. Method Tests Questions
Source: Wu and Vieira, 2002
22
Tests: Experimental Data II (to be done)
Comparison with experimental lab dataCui et al. (1995/6).
Background Sediment Transport Math Treatment Num. Method Tests Questions
Source: Wu and Vieira, 2002
23
Question: Physical Bed roughness
Ks = Ks’( skin friction )+ Ks”( form drag )1-Methods based on bed-forms and grain-related parameters such as bed-
form length, height, steepness and bed-material size:K’s min= 0.01 m
K’s = 3 d90 for θ <1 (lower regime)
K’s = 3 θ d90 for θ >1 (upper regime)
Is there information on d90,d50, and available for the
Delta? 2-Methods based on integral parameters such as mean
depth, mean velocity and bed material size
Background Sediment Transport Math Treatment Num. Method Tests Questions
)](7001[50, crcs dk
parametermobilitygds
b))(( 50
24
Question: Physical – Distribution in Junctions
Background Sediment Transport Math Treatment Num. Method Tests Questions
)( 21444
333
333
3 SSQKQK
QKS
nn
n
Source: Mike11, DHI
• Do you know of any studies of sediment transport at junctions in the Delta or in other systems?
25
Question: Physical Treatment of Bed-load
1- Solve an advection-diffusion-reaction equation for suspended load, and use empirical formula for computing bed-load
2- Solve an advection-diffusion-reaction equation for both the bed-load and the suspended load, following the proposal by Greimann et al. (2008):
eyxtt S
y
ChfD
yx
ChfD
xy
hCV
x
hCV
t
hC
)()(
)sin()cos(
Se= Erosion source termf = Transport load parameter, fraction of suspended load to total load h = Flow depth a = Angle of sediment transportb = Ratio of sediment velocity to flow velocityVt = Total flow velocity
Background Sediment Transport Math Treatment Num. Method Tests Questions
26
Question: Physical - Entrainment & Deposition
1-In the Delta, is there any tested method for representation of settling velocity (or deposition instead) for cohesive sediment particles beyond Krone’s (1962)
work? 2-In the Delta, is there any tested expression for entrainment of cohesive
sediment particles beyond the work by Krone (1962)?
3-To reduce the number of variables for description of cohesive sediment, which are the most important variables for the Delta? Ws = Ws (Salinity, Concentration, ...)
4-In the Delta, is there any tested method for representation of settling velocity
(or deposition instead) for cohesive sediment particles? i) Ws= constant, ii) Ws,m= Ws(1- ac)b , iii) Ws,m= Kcm
Background Sediment Transport Math Treatment Num. Method Tests Questions
cr
crbE
27
Question: Numerical
1-Do you know of any reliable second order methods for updating boundary conditions for advection-diffusion-reaction equations with operator splitting?
2-What order, in the splitting procedure, should we solve the advection-diffusion-reaction equations? Our initial thought is to solve advection first, then reaction and finally diffusion. Advection is always the dominant term and it should come first.
Background Sediment Transport Math Treatment Num. Method Tests Questions
28
Questions: User need/request
What kinds of analytical tests and comparisons to data (field and laboratory) would you like to see in the STM code?
What units for sediment/constituent concentration you would like to see in DSM2-STM? Volume per volume or mass per volume?
Is it desirable for STM to have a feature that allows the user to select the numerical scheme to be used to solve the advection part?
Initial non-cohesive implementation has: 1) Garcia and Parker (1991); 2) van Rijn (1984); 3) Smith and
McLean (1977); 4) Zyserman and Fredsoe (1994) Is there any other formulation you prefer to have in DSM2-STM?
Background Sediment Transport Math Treatment Num. Method Tests Questions