New Code Workshop : FSFF

**F**ree
**S**urface **F**low with **F**luent

**Wave Propagation and breaking**

Redacted by

ANTIBI Olivier ( mfn01 ) - PERCHAT Alexandre ( mfn09 )

Test case 1 : Simple box

Test case 2 : Simple box with a triangular bump

After having tested the tutorial example we are now ready to compute our own case. The case we would like to simultate is linked to our Industrial Hands-On. It dealed with a wave propagation and moreover with wave breaking. We didn't have enought time to manage to simulate wave propagation in Fluent and we so decided to study water propagation in a air box. Even if the phenomenon is not described with the same equation, the results are quite the same and can well illustrate wave breaking with Fluent. We will analyse two different cases.simulations. A first with a simple box and another with a triangular bump.

To begin with, we used a standard structured mesh thanks to prebfc. The computational domain is 5m lenght and 3m height and the box is 50x30. The mesh is show belown :

It's important to notice that when you adapt a region to patch an initial condition the mesh is refined on the domain. The refined mesh is shown belown :

The boundary conditions are all walls so we must be aware of the problems due to reflexion.

The results below represents the propagation of the free surface at different time steps. In fact it's the air volume fraction. The time step must be chosen cleverly to be enough little to prevent the code form exploding and enough big to make the simulation go faster. We chose a time step of 0.01 and the results are taken each 50 time steps.

As we could imagine, because of reflexion the computation is exploding as shown below :

Simple box with a tringular bump

The mesh was modified in order to add the bump. We chose to increase its length in order to well develop the wave. Its length is now 10m and it's 5m large. We tried to compute severous simulations with unstructured meshes and it doesn't seem to run very well, so we decided to use structured prebfc meshes. The following pictures shows the initial mesh and the mesh refined on the adapted region.

We used the same time step and took the solutions every 10 time steps. The results shows the air volume fraction through the iterations.

The results are in godd agreements in what we can imagine. we can notice a good propagation of the wave and a good behavior to cross the bump but the last pictures highlight a difficulty for the code to compute and obtain clever results concerning wave breaking. This can be explain by the numerical model used. We have many difficulties to install a research code using a particular method ( Sph ) but this method seems more appropriate to simulate this kind of flows.

We tried to simulate a more realistic test with an unstructured mesh. The following meshes illustrate the difference between standard and adapted mesh, after having patch the initial condition.

The following pictures illustrate the initial condition where air volume fraction is set to 1 ( air ) and set to 0 ( water ). In both cases we can notice a zone of disontinuity and this can explain why with the same last parameters and even with smaller time step the computational code can't manage to realise only one iteration.