Learning Forum

[Abhemanyu Palaniswamy Chandrasekaran]

I have set up a FSI simulation using the S-ALE keywords. I am replicating the method used in the following paper to find the added mass of a floating body.

https://doi.org/10.1016/j.marstruc.2016.05.005

With a coarse mesh of 1m size for the ALE elements, the simulation runs fine with the timesteps remaining almost consistent throughout the entire simulation.

But when the S-ALE mesh is made with progressive spacing, such that at the free surface of air and water, the element size reduces to 0.4m, the timesteps drop from the initial calculated timestep. How to identify the cause of this problem and resolve it to have a consistent timestep throughout the simulation?

I am currently using the default settings PFAC=0.1 for the coupling.

[Ian Do]

Hello Abhemanyu ,

I don't think I can answer this question as to the reason for dt drop.

When the ALE mesh resolution is changed, this affects many things in the computational domain. It could mean the mesh can now pick up high gradients (P, vel, rho, ...) in the flow field, or in the FSI force gradients(?), or advection of smaller chunks of ALE mat (?), etc. dt limits are computed based on many criteria, such as contact, FSI, advection, CFL, mesh size+sound speed, ... Then the smallest dt is used. So you can see that there isn't a clean way to pin down the reason why when dx changes what may happen to dt.  

 

Ian Do