Convergence of CFX field in FSI analysis
 

Hello friends

I do the fluid-structure interaction of the aortic valve and blood. I managed the negative element volume and analysis go ahead. I have only one problem. During every coupled interaction I have no convergence for CFX filed.

*** ANSYS BINARY FILE STATISTICS
BUFFER SIZE USED= 16384
122.938 MB WRITTEN ON ELEMENT SAVED DATA FILE: ANSYS.esav
162.000 MB WRITTEN ON ASSEMBLED MATRIX FILE: ANSYS.full
59.125 MB WRITTEN ON RESULTS FILE: ANSYS.rst

//the end of the structure calaculation - everything is fine.

FOURTH SYNCHRONIZATION POINT OF MASTER CODE:
LOAD TRANSFER AT THE END OF SOLUTION

FIFTH SYNCHRONIZATION POINT OF MASTER CODE:
GET SLAVES LOCAL CONVERGENCE

FIELD SOLUTION CONVERGENCE

CONVERGENCE OF FIELD SOLVER: ANSYS. . . . . TRUE
CONVERGENCE OF FIELD SOLVER: CFX. . . . . FALSE

//not converged ??

SURFACE LOAD TRANSFER CONVERGENCE

CONVERGENCE FOR FIELD 2 ACROSS INTERFACE 0
UX DISP CONVERGENCE. . . . . . 0.70735
UY DISP CONVERGENCE. . . . . . 0.72570
UZ DISP CONVERGENCE. . . . . . 0.70866

CONVERGENCE FOR FIELD 1 ACROSS INTERFACE 1
FX FORCE CONVERGENCE. . . . . 0.68833
FY FORCE CONVERGENCE. . . . . 0.69597
FZ FORCE CONVERGENCE. . . . . 0.68933


FIFTH SYNCHRONIZATION POINT OF MASTER CODE:
SERVE SLAVES THE GLOBAL CONVERGENCE
1

Have you ever seen similar problem ?? Or may you have same idaes what's wrong??

regards

Chris

Hi,

FSI adds another dimension of things to check over a normal static CFD simulation. Have you checked it converges OK without FSI with some geometries typical of what you will get in the simulation? Have you checked the FSI motion updates are small enough that you are resolving the motion properly?

Glenn Horrocks

Glenn

I've checked it but only for the first state of geometry, and the fluid converged. During the simulation I had a problem with negative volume element then I decrease the substep. Now the deformation is close to 0.1 mm and I think the FSI motion is small enough. May I should decrease the time of the substep again ? I must admit that I have a quite poor mesh.