Learning Forum

[Mark Jordan]

Hello,

 

I am modelling a steady state two-way fluid-structure interaction of a foil air bearing. In this model, I have an air domain developed in cfx which generates a pressure distribution due to the rotation of a shaft which deforms a compliant top foil which is modelled in a static structural analysis. Shown in the image below is the air domain in green and surrounding compliant top foil which spans 350deg of the air domain.

 

 

I have followed best practice system coupling steps by modelling each analysis (mechanical/cfx) independently and also as a one-way fluid-structure interaction prior to attempting two way simulation, each of which produced the expected results.

 

When performing the fully coupled simulation, a fatal error of negative element volume occurs terminating the solution without results. Usually, this happens after 3 to 6 coupled iterations. The Skewness of both cfx and mechanical meshes are below 0.5 so I believe mesh quality is acceptable. In the image below, step control was set to a minimum iteration of 3 and max of 6. It reached the 6th iteration and then failed without results due to negative volume error.

 

 

I have also referred to both cfx and system coupling user guides, which have not helped resolve this issue. I tried ramping the data transfer in system coupling and this has not helped either. Could someone please advise how I can fix this error and move forward with my simulation?

[Federico Alzamora Previtali]

Hello Mark, 

Negative volume cells is typically due to unreasonable displacement received by Fluent or unreasonable forces sent to Mechanical. Assuming your mesh is of good quality, since you are following Best Practices recommendations from our user guides, it is a good idead to monitor the pressure/force convergence and displacement convergence for those FSI surfaces within a time step to see which of the above is causing problems.

Another thing to look at are the Solver Controls for Mesh Displacement equations. In the Convergence Criteria, you may try to decrease the Residual Target and and Increase the Max Coeff. Loops. 

I hope this helps.