Physical and Numerical Modeling of Rough Ramps and Slides
A new European Water Framework Directive forces the review and possible transformation of existing weirs, bank mounts, stream straightenings and river mouths into more natural conditions. A state-of-the-art solution to naturalize large water surface steps is the arrangement of so-called rough ramps and slides. According to German DIN 19661/2 and DIN 4047/5, ramps range from 1:3 to 1:10, and slides range from 1:10 to 1:30. Several natural obstacles are arranged in the flow area to get a nature-orientated solution which provides more climb possibilities for fishes.
To analyze the effects of boulders on the water surface levels and flow velocities, as well as the resistant coefficients of the boulders a physical model was built at the University of Wuppertal’s Hydraulic Laboratory. State-of-the-art measurement techniques were used to give information about flow characteristics and forces on the boulders. The physical model allowed variations of boulder arrangements and discharges. All model runs were compared with 3-D simulations using FLOW-3D. The animation to the left shows one possible solution for a natural rough slide.
The physical model was scaled 1:15 and included two 2 cm deep scours (0.3 m in prototype) upstream and downstream of the ramp (Fig. 1). A simple weir overflow marked the end of the model. A fine-grain layer represented the river bed’s roughness. The arranged boulders were sized 6 x 6 x 6 cm in the model. Six ultrasonic sensors were used to measure the water surface elevation in a 5 x 5 cm grid. The positioning of the sensors was done with ISEL multiphase motors.
Additionally, vertically arranged load-cells measured the forces on the boulders in the flow direction. Altogether, more than 100 model runs are carried out in the physical model. Four varying discharges were combined with 20 boulder configurations. A photo of the physical model can be found in Fig. 2, and an example result for water depths in Fig 3.
Figure 1: Schematic plot of the physical/numerical model.
Figure 2: Example photo of model run ML32.
The numerical model can be identically compared with the physical model. The model is build within FLOW-3D with an external STL-file for the geometry. 1,086,010 cells, while 781,790 cells are active, are chosen in a 1 cm grid. The output interval is 1 s; the initial time step is 0.001 s and the maximum time step 0.01 s. The river bed roughness with 2 mm can be compared with the physical model’s roughness. After approximately t = 40 s a quasi steady flow situation is achieved. As turbulence model the RNG-model is used. The simulation time depends on flow characteristics and is between 2 and 8 h.
Figure 3 (left): shows the water depth in the physical model. Figure 4 (right): shows the water depth in the numerical model.
Figure 4 shows an example for water depths in the numerical model. All model runs were done in dimensions corresponding to the physical model as well as in prototype dimensions to detect possible scaling effects. Comparisons of both sets of runs show identically results, hence, no scaling effects can be assumed. The FLOW-3D simulations were run are calculated on a Windows 7 system with 4GB RAM and an Intel Core Quad 3,0 GHz machine with four parallel processes.
Figure 5: Comparison of water depths, numeric minus laboratory.
Post-processing was done with MATLAB to create comparable graphic outputs. All model runs showed very good results for water surface elevations (see Fig. 5) and flow velocities. Small uncertainties were found, but these are believed to be caused by laboratory measurement techniques and local high turbulent flow structures. Also, the larger measurement grid in the physical model generates some deviations. Finally, FLOW-3D allows a fast, efficient and cost-effective variation of boulder configurations to create a continuative measurement program within the numerical model.
This material was contributed by Dr.-Ing. Mario Oertel of University of Wuppertal's Hydraulic Engineering Section in Wuppertal, Germany, a runner-up of the 2nd Flow Science 30th Anniversary Simulation Contest. For visualization as well as post-processing of the simulation data, Matlab was used.