FSI simulation - negative rho error and pressure diffusion through type 18 master part

Ingeniorator
Ingeniorator New Altair Community Member
edited May 2023 in Community Q&A

Hello everyone, I have some questions about FSI modelling. When using LAW51 with a LAW5 JWL material, LAW4 hydraulic JC and LAW6 with ideal gas EOS (all Eulerian), I sometimes have models abort due to a negative rho error in some element. What exactly causes this? The cutoff pressure for the LAW6 material is set to -1e-30.

On other occasions using a similar setup with a Type 18 contact between the air/explosive (slave) components and a solid part (master) being affected by the pressure wave, I noticed that some of the pressure wave seems to travel through the solid part. Even though the wave is influenced when it hits the part, some of it seems to diffuse through it. I did not use any artificial minimum time step such as DT/NODA/CST to speed up the computation in some models and it looked like the same behaviour. The contact stiffness and gap factors were defined as described in the manuals

Related to this: how does a larger time step (i.e. set by DT/NODA/CST) influence the Type 18 contact behaviour? Would a time step larger than the one controlled by DT/ALE contribute to this artificial diffusion? Also, how are the master and slave mesh sizes correlated - should they have a similar size or should/can the master size be (much) larger? Is it better to define the master entity as component, set or surface, or is there no difference?

Tagged:

Answers

  • Mathis
    Mathis
    Altair Employee
    edited May 2023

    Hi Ingeniorator,

    -1- /DT/NODA/CST has only effect on lagrangian parts. So it shouldn't have any effect on fluid or on coupling interface type18. It will just be side effect from the different behavior of the structure due to added mass. 

    -2- If you want to speed-up computation, you may want to increase a little bit the scale factor in the card /DT/ALE. By default, the value is 0.5. You can try to increase this up to 0.7. Be aware that it may cause some instabilities. 

    -3- If you have some diffusion through the interface, you can check your gap and stiffness values wrt the documentation. If you still have some diffusion you may try to increase stiffness value. It may however lead to some reduced displacement of the structure. 

    -4- Finally, depending to the complexity of your structure and how much it will move forwards inside the fluid. You may switch to ALE and tied the structure to the ALE domain through a type2  interface. There will no longer be any diffusion. 

    I hope it helps.

    Regards,

    Mathis

Categories