cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Showing results for 
Search instead for 
Did you mean: 

Wildfire 2 FEA analysis failing in Wildfire 3

Highlighted

Wildfire 2 FEA analysis failing in Wildfire 3

I have an FEA of a somewhat complex casting that was run in WF 2.0 about 1.5 years ago. I was revisiting the analysis and opened the file in WF3.0. I meshed the part and attempted to run a static analysis with the same constraints and loads from the WF 2.0 file in order to attempt to duplicate the results obtained in WF 2.0. The part does not fail in the meshing process, but WF 3.0 does create approximately 2000 additional elements than WF 2.0 did (~36,000 vs ~38,000). The static analysis fails soon after it begins stating there is bad geometry and to consider adding seed points in the failed area before re-running the analysis. The mesh settings between WF 2.0 and WF 3.0 are the same, as is the accuracy setting. I have resolved/fixed the first failed area and rerun the analysis only to get another message with another geometry issue. I do have the concern that I may only be seeing the tip of the iceburg, as it were, in regards to geometry failures. Has anyone else run into this problem? Is there something that has changed in WF 3.0 that makes it more sensitive in the calculation portion of an analysis but not in the meshing portion? Is there any type of work around short of going through and resolving the geometry issues step by step? Since the failures don't show up until the analysis, finding where the bad gepmetry is with a "dbid number" is cumbersome.
Thanks in advance for your help.

David W. Ambrose, P.E.
Principal Engineer
Compaction

Wacker Corporation
N92 W15000 Anthony Avenue
Menomonee Falls, WI 53051-1504
USA


1 REPLY 1

Wildfire 2 FEA analysis failing in Wildfire 3

David,

You must have had noticed that What's new in WF3 lists "meshing algorithm improvements", which means the internal meshing procedure (though being pretty much a derivative of WF2 routine) is not precisely the same as it was in WF2, hence seeing a different mesh in WF3 is normal. (though I wouldn't expect this to happen on relatively simple geometry).

Now, the reality is that those "improvements" may in fact cause unanticipated side effects. In other words, WF3 may now mesh parts that WF2 didn't, but it may also, rather unexpectedly, fail on parts that WF2 was able to handle. Pro/M developers were improving/debugging their meshing most likely using their internal library of parts as well as failing parts/geometries they were getting from bug reports from the field (That's the way software development and testing normally works). Well, they didn't have your particular part to test their improvements, did they?

If I were you, I'd simply stick with WF2 for that particular analysis. You should also report your trouble to PTC, so you might see your parts running OK in, say, WF4, or even next build of WF3.

Regarding the "tip of an iceberg", you're actually upon one, and it is called "repeatability of results", and it's the one that FEA vendors have been struggling with, have no solution for, and, in my honest opinion, will never solve, for it's more educational problem than software problem. But I don't really want to start discussion on that...

Regards,
Yuri


David Ambrose <david.ambrose@am.wackergroup.com> wrote:
I have an FEA of a somewhat complex casting that was run in WF 2.0 about 1.5 years ago. I was revisiting the analysis and opened the file in WF3.0. I meshed the part and attempted to run a static analysis with the same constraints and loads from the WF 2.0 file in order to attempt to duplicate the results obtained in WF 2.0. The part does not fail in the meshing process, but WF 3.0 does create approximately 2000 additional elements than WF 2.0 did (~36,000 vs ~38,000). The static analysis fails soon after it begins stating there is bad geometry and to consider adding seed points in the failed area before re-running the analysis. The mesh settings between WF 2.0 and WF 3.0 are the same, as is the accuracy setting. I have resolved/fixed the first failed area and rerun the analysis only to get another message with another geometry issue. I do have the concern that I may only be seeing the tip of the iceburg, as it were, in regards to geometry failures. Has anyone else run
into this problem? Is there something that has changed in WF 3.0 that makes it more sensitive in the calculation portion of an analysis but not in the meshing portion? Is there any type of work around short of going through and resolving the geometry issues step by step? Since the failures don't show up until the analysis, finding where the bad gepmetry is with a "dbid number" is cumbersome. Thanks in advance for your help. David W. Ambrose, P.E. Principal Engineer Compaction Wacker Corporation N92 W15000 Anthony Avenue Menomonee Falls, WI 53051-1504 USA