This case is a simulation of sediment deposition in a sand trap. The results for both velocities and sediment concentrations are compared with results from a physical model study and farily good agreement was found (Olsen and Skoglund, 1995).
Here are some pictures from the physical model study.
View from the top looking in the downstream direction.
View inside the expansion, looking in the downstream direction.
Looking at the inflow section with the sediment feed. Looking in the downstream direction.
Grid seen from above.
Three-dimensional view of the sand trap, where the bed, the water surface and a longitudinal section along the centerline is shown. The color shows the concentration. The bed is shown with blue color.
Three-dimensional view of the sand trap, where the bed, the water surface and a longitudinal section along the centerline is shown. The color shows the water velocity.
Longitudinal profile velocity vectors at the entrance region. The colors show the magnitude of the velocity. Red is maximum value and blue is minimum value.
Longitudinal profile along the centerline of the geometry. The colors show the magnitude of the sediment concentration. Red is maximum value and blue is minimum value.
Comparision of calculated and measured sediment concentrations in vertical profiles along the centerline of the geometry. Lines are calculated concentrations and crosses are measured values.
koordina file
koomin file
control file for water flow computation
control file for sediment flow computation
The current "koordina" file was made with a varying length of the cells in the main area of the sand trap. If making a new "koordina" file today, a better solution would probably be to use constant cell lengths. However, the results would probably not change much.
Two "control" files are enclosed. It is technically possible to compute the water flow and sediment transport with one file, but it is usually more convenient of doing this in two steps. Often, one will want to do parameter tests for the sediment computaiton, for example vary the inflow concentration, grain size distribution etc. Then it is impractical to compute the water flow over again for each sediment computation. Using two files, the water flow field is read from the "result" file before the sediments are computed.
The SSIIM program originally computed the sediment concentration with a steady solver. The main problem was then how to compute the changes in the grain size distribution at the bed. For the current case, this is not so important, as the pick-up rate of sediments at the main part of the sand trap is zero, and then the grain size distribution does not affect the concentration. However, in the entrance region algorithms has to be used to ensure that there is no deposition in the entrance channel. After having worked with unsteady algorithms for the last years, these are used in the present setup. To prevent erosion in the geometry, the "koomin" file is used to specify the level of the non-erodable bed. This is identical to the "koordina" file, as the bottom of the model was made of wood/ concrete and there were no sediments at the bed initially.
The control file for the water flow computation is fairly straightforward. The file presented here also contain a number of data sets used in the sediment computations. Actually, only the F2, G1, G3, G7, W1, W2 and the K 1-5 data sets are used. Note the use of the K5 data set to specify block-correction to improve the convergence speed. Also, the relaxation factors on the K3 data set has been lowered slightly compared with the default values. This also improves the convergence speed.
The control.s file contains the F 11 data set, with a negative Shields coefficient. This means that the Shields curve will be used. The time step specified on the F 33 data set is fairly long, but this will reduce the total computaitonal time. The F 37 2 data set is used to specify a time-dependent sediment computation. The integer 2 on the data set specifies that a formula for the pick-up rate is used instead of specifying a concentration at the bed cell. The use of F 37 2 instead of F 37 1 gives better sediment continuity. The F 68 2 data set specifies that the water flow should not be recomputed for each time step. This saves computational time. It is then assumed that the magnitude of the sediment deposition is so small that the bed changes will not affect the water flow field. If the F 68 data set is removed, a time-dependent computation of the bed elevation changes is done.
The results in form of fluxes in and out of the geometry are given in the "boogie" file:
35 Sedim. continuity: In, Out, Susp, Bedch., Bedmove, ContDef., MoveDef: (qm)
Dt: 0.018868 0.002288 0.000000 0.016559 0.000000 -0.000022 0.016559
Sum: 0.679245 0.080085 0.005669 0.593007 -0.000000 -0.000484 0.593007
Grain size breakdown:
Size: Inflow Outflow LayerActive LayerInac. Suspended Defect
1 0.084906 0.000001 0.000090 0.084914 0.000014 -0.000114
2 0.084906 0.000048 0.000478 0.084047 0.000133 0.000200
3 0.084906 0.002150 0.001616 0.080354 0.000618 0.000168
4 0.084906 0.037844 0.002145 0.042859 0.002069 -0.000012
The trap efficiency for each size is computed from the numbers giving the inflow and outflow fluxes.
Looking at a longitudinal profile of the water velocity vectors, it is possible to estimate the length of the recirculation zone. This length is overpredicted compared with observations from the physical model. Lowering one of the parameters in the k-epsilon model gave better results for the length of the recirculation zone. This also gave better correspondence with the measured sediment concentrations, as shown in the figure above. However, using this water flow field gave fairly similar trap efficiencies as using a flow field based on the standard values in the k-epsilon model.
koosurf file
timei file
control file for the grid generation
control file for water flow computation
control file for sediment flow computation
interpol.u file
interpol.c file
verify.u file
verify.c file
This page was last updated: 7 March 2019