SOL103 Normal Modes Analysis Using MFLUID

[mashst24]

I'm attempting to perform a normal modes analysis on a cylindrical aluminum tank filled with water. I am interested in determining the mode shapes and modal mass participation of the coupled structural/fluid response. I am attempting to model the water using a Fluid Region (fluid), but when I run the analysis I get a fatal error. My model is a simple cylindrical tank modeled with plane elements, that runs fine with the fluid region disabled. To apply the fluid region, I select the top, bottom, and 2 half-cylinder faces of the tank, uncheck the box for "Z Free Surface" because the analysis I am running is for a full tank, and specify the water mass density. Is there something in this process that I am missing which is causing the fatal error? Any help would be much appreciated; I haven't been able to find much information or tutorials regarding using Fluid Regions. Here is the fatal error that I get in the error log:

** SYSTEM FATAL MESSAGE 3034 (LNNHERR)
INTERNAL FAILURE IN THE LANCZOS PROCEDURE:
M-ORTHOGONAL QR PROCEDURE FAILED TO CONVERGE. PROBABLE CAUSE:
MASS MATRIX IS INDEFINITE (MODES) OR STIFFNESS MATRIX IS INDEFINITE (BUCKLING).
*** SYTEM FATAL MESSAGE 7340 (LNNHERR)
WARNING REPORTED BY SUBROUTINE LNNDRVD (IER= 728)
USER INFORMATION: GRAM-SCHMIDT DID NOT CONVERGE
*** SYSTEM FATAL MESSAGE 7340 (LNNRIGL)
TRIDIAGONAL QL DID NOT CONVERGE
THIS RUN IS TERMINATED. PARAM, BAILOUR, -1 MAY BE USED TO CONTINUE THE RUN

 

[jbrackin]

What version of Femap are you using? In Femap Help, see the section in Connections on Fluid Regions for descriptions of the inputs and setup.
I just executed the problem you described in v11.1.1 with no errors.

 

[mashst24]

I am running v11.0.0. I took a closer look at the help topic for MFLUID, and this part confused me:

"There is a Check box which enables you to turn the Z Free Surface OFF completely, so no value is written out to Nastran for the ZFS field. When the Z Free Surface is completely OFF and the XY Plane and YZ Plane fields are NOT set to Antisymmetry, Nastran will treat the MFLUID card as a special form of “external fluid”."

My takeaway from that is that turning the Z Free Surface off is not the proper way to model an internal fluid. I am able to get my analysis to run when I specify a Z Free Surface at the top of the tank, but I didn't think that was the correct method in my case since the tank is full and there are no free surfaces. For my case with a cylindrical tank, should I select just the bottom surface and the cylindrical sides, or should I select the top surface as well?

Thanks for your help!

 

[jbrackin]

Can you post your model?