Abstract:
|
The main objective of this work is to analyse the two different turbulence
methodologies, LES and RANS, using three numerical OpenFOAM simulations based on
the Finite Volume Method. Also, the non-hydrostatic effects produced by the highly
curved surface are analysed. In order to assess the 3D model, two dam-break
experiments are used to calibrate (Kleefsman et al., 2005) and validate (Fraccarollo
and Toro, 1995) the model, respectively.
LES methodology is simulated using Smagorinsky submodel while in RANS the RNG kepsilon
is used. To simulate the water surface, the Volume of Fluid and the interFoam
solver are used.
A computer with 32 GB of RAM memory and Intel Core i7-6700k processor is used to
simulate the experiments. The computational time to simulate the second experiment
with 1 cm mesh is around 2 days.
In conclusion both models simulate quite well the water height, whereas the pressure
results are less precise. Even though, Smagorinsky model gets better results in the first
experiment. In the second experiment, Smagorinsky model shows that it is able to
follow the variability in time of the variables under study; however, there are parts in
which the model underestimates the values while in others it overestimates the
values. Regarding to the comparison, done in this experiment, between the pressure
head and the water height, they are not equal when there is a highly curved water
surface. |