Learning Forum

[Webster]

Hi community!

I am using LS Dyna for a FSI simulation. My simulation terminated as the result of an ICFD mesh error. The result comfirmed the large deflection in geometry was attributed to a sudden yet unreasonable oscillation in my fluid pressure (a sudden overshoot).

The problem is not associated with the time step. The time increment is ~ 5E-5, so I am not worrying about this. The velocity inlet - pressure outlet boundary conditions might be the culprit, but I am not sure.

Are there any techniques (or, tricks) I can deploy to stablise the pressure in simulation?  Or, alternatively, do you have any other suspects that I should be careful about?

Thank you.

[Reno Genest]

Hello Binghuan,

Could you share a screenshot of your pressure vs time graph to show to overshoot?

What velocity inlet and pressure outlet boundary conditions are you using?

Are you using a consistent unit system? This is very important. You will find more information here:

https://lsdyna.ansys.com/consistent-units/

Which version of LS-DYNA are you using? Have you tried with the latest R14.1 solver? Make sure to use the double precision MPP solver:

https://user:computer@ftp.lstc.com/user/mpp-dyna/R14.1.0/windows/

Are you using a loose or strong coupling FSI? From our ICFD training course: "Problem: It has been observed that loose coupling between an incompressible fluid
and a solid frequently yields unstable computations. The mass ratio between fluid and structure has a significant influence on the stability
of the system. The bigger the ratio , the more unstable the system. A smaller timestep will cause the instabilities to appear earlier. In order to overcome those effects, it is mandatory to use an iterative procedure also
called strong FSI coupling."

 

So, if you are using a loose FSI coupling, I would suggest you try the strong FSI coupling.

": In order to trigger automatic strong coupling, the LS DYNA implicit solver
must be used for solid mechanics (No further modifications of the CFD input deck is needed). First tip: In case of stability issues, it may be useful to change the Newmark integration
parameters (see *CONTROL_IMPLICIT_DYNAMICS). γ = 0.6 and β = 0.4 for instance
yield a more diffusive but more stable solution. Second tip: Lower the tolerances of the implicit solve (see
*CONTROL_IMPLICIT_SOLUTION. Third tip: Increase the timestep. In some cases, this may help with stability to the
expense of accuracy.".

You will find FSI with ICFD examples here:

https://www.dynaexamples.com/icfd/basics-examples/strong_fsi

https://www.dynaexamples.com/icfd/intermediate-examples/Intermediate_fsi_flap

The last example above has a full tutorial on Youtube:

https://www.youtube.com/watch?v=eLI4p5Ifcds

 

Let me know how it goes.

 

Reno.