Problem applying symmetry conditions to a FSI analysis

[Raniel Albuquerque]

Hello,

I'm simulating a Fluid-structure interaction of a water jet perforating a soil. In order to reduce computational time, I'm trying to apply symmetry conditions to my problem, which consists in two models, one fluid and another solid. The fluid model doesn't allowes the creation of an axissymetric part, and how the fluid model and the solid model must have a interface of interaction, I can't apply the convencional symmetry. So I'm trying to use the boundary conditions to simulate the symmetry. But my solid model results are strange.

When I use the Symmetry/Antisymmetry/Encastre condition, with U1 = U3 = UR2 = 0, the soil's lateral walls aren't moving down, but the inner region is. I've constrained only the X and Z axis displacements, it should've move down with the inner region.

Any thoughts how I could solve this problem, by means of the correct use of symmetry or changing some elements properties?

 

[FEA way]

Judging by the fact that you wanted to use axisymmetric model and that your geometry looks like a slice of cylindrical part, I think that in this case it would be the best to use cyclic symmetry. It's available in the Interaction module (Create interaction).

 

[Raniel Albuquerque]

Thank you, FEA way. I've tried that strategy, but how my simulation has a fluid, I got this message error:

"Cyclic symmetry model may only be used with static, visco, frequency, steady state dynamics, or heat transfer analyses."

 

[FEA way]

Could you attach a picture showing which part of your model is fluid and which is solid ? You can easily visualize it using Color Code options toolbar. Also is the full model a cylinder as I guessed before ?

Indeed displacement BCs can't be applied to the regions where fluid-solid interaction is defined but other symmetrical regions within the model may have displacement BCs applied. For the fluid part slip wall can simulate symmetry.

 

[Erik Panos Kostson]

Just use quarter symmetry and fix the displacement normal to the symmetry planes in the structural elements and for the cfd ones have 0 normal velocity at the 2 symmetry planes.

 

[Raniel Albuquerque]

Yes, you were right. My analysis consists in two cylindrical models. This is the fluid model:

The model's material is water, there's a water inlet at the center of the model.

The soil's model is at the next picture, it isn't a assembly of the two models, it's the soil model alone.

To co-simulate the two models, I'm using a Fluid-Structure Co-simulation Boundary, at the Interaction menu. Here's the surfaces of both models where ocurrs the interaction:

 

[Erik Panos Kostson]

See my previous response. Also if convergence in cfd or fea domain is hard look on explicit Lagrange or Sph for soil and Sph for water jet. See this

https://m.youtube.com/watch?v=92Psa-bppXk

 

[FEA way]

Is your approach on this topic based on some reference article ? Even if it seems to work properly I would suggest trying with SPH, as Erik said. Water jet eroding soil is hard to model using other methods than SPH. In your case there's a large fluid domain with very small inlet in the middle simulating narrow stream of water. This seems somewhat inefficient. Also it would be much better to include damage model and element deletion for solid domain.