Learning Forum

[Marcin Nowak]

Hi, I am modeling FSI problem in Fluent-only - using intrinsic solver. 
I encounter a problem with the negative cell volume, but in the solid body zone. The dynamic mesh (smoothing) in the fluid region works well.
Below picture presenting mesh cross-section through a solid domain, clearly shows, that deformation in absorbed mainly in the first cell layer, closest to the fluid.

I did not find any option to tune, for the solid body mesh motion. Also, I do not set any options for the solid body dynamic mesh (only for fluid) in the setup workflow.
Are there any options to change a dynamic mesh settings for the FEM solid body mesh?
Or, what else possibly could I try to mitigate the problem? Now I solve a model with reduced timestep size but I really dont think this would help as the time step is already small.

[Ravi Sastry]

Hi User,

Hope you are follwing the Fluent user-guide settings (link shared below). 

27.2. Setting Up an Intrinsic Fluid-Structure Interaction (FSI) Simulation (ansys.com)

If the defaults aren't working can check on some of the below

 

Regards

[Marcin Nowak]

I tried:
define → models → structure → controls → enhanced-strain? ->yes
define/models/structure/controls/numerical-damping-factor? -> I set value of 0.5
define → models → structure → controls → unsteady-damping-rayleigh? ->yes
These options, have no effect on the mesh motion. However “enhanced strain” option gave slightly higher deformations (diameter was higher by 2.4%). 

The timestep size does not have impact on the results or the mesh motion – I checked two smaller time step sizes.

I was able to mitigate somehow a problem, by setting a mesh bias factor – having less dense solid mesh at the fluid-solid interface region, at the beginning of the computation. But this does not solve my problem completely as I was not able to perform full transient calculations (but I was able to get further in time). 

Also I dont want to use “bias factor” as the results are quite highly dependent on this (I checked it). By the way, I was performing a mesh sensitivity analysis for the solid body mesh, by using different number of mesh layers (from 2 to 8) and the solution in the Intrinsic FSI Fluent solver, is very different for these meshes (stabilizes at 6 layers). Whereas for the Mechanical-only (applying constant pressure inside a pipe) with multiple substeps, the solution was not dependent on the number of mesh layers.

To sum up, I would like to find an option in Fluent, which will let me to regulate the solid body mesh motion, to have more uniform mesh motion in the solid body domain (now the nodes moves primarily in the layers that are closer to the fluid). Please let me know whether there are such options.

 

I have also another idea. Check the real-scale deformation field below, for the Mechanical-only simulation. It can be observed, that there are present negative cell volumes, however this does not affect the solution process, also I dont receive any error or warning. I think that in the FEM calculations the cell volumes do not matter. So maybe there is an option to turn-off the calculation abortion in Fluent, when the negative cell volume in the solid mesh is present?

 

 

Regards

[Ravi Sastry]

The solid body motion is result of FE solution and comes automatically. Cannot control it in any manner. The point is that the FE solution should not give you any non-physical results. Again, the problem is maybe not where one may be thinking and could be elsewhere. The only mechanical solution will not show the problem because it does not have a full coupled mesh and therefore negative volume at the interface would not be reported. One way to also check is the FE solution in intrinsic FSI correct is to perform as with mechanical 1-Way FSi solution and to compare such results with Mechanical.

Also on other suggetions is to check if the material properties are realistic - a real material should not behave like this. As solid nodes position is coming from intrinsic solver, this check may help.