Learning Forum

[Yingnan Wang]

Hi there,

I am using Ansys 2023 to simulate the elastic modulus of composites. There were issues/errors in the "result part." 

I attached images which showed the error type.  "MAPDL internal data has been corrupted."

Thanks!

[Ashish Khemka]

Hi,

SIG$SEGV is short for signal segment violation. Technically, it is described as an incorrect use of pointer or a buffer overflow (memory problems). It is a bug, whenever it happens, but it is difficult to relate the cause to any one particular issue. Based on previous call records, it has happened in the past with distributed solver and rezoning, but is not limited to these applications. If you are using a distributed solver then please uncheck that option and then run the analysis

Regards,

Ashish Khemka

[Yingnan Wang]

Hi Ashish,

 

Thanks for the reply. I unchecked the distributed solver and re-run my model. And the same error happened. 

Do you have any other suggestions?

Thanks! 

[Yingnan Wang]

 I found another similar problem, which had been closed from a new reply. I attached the possible answer here.

The issue was that I tried the "-blas intel" on the MAPDL command line, but it showed this is not a recognized Begin command.  Can I get any suggestions based on this solution idea? 

Thanks! 

[Ashish Khemka]

Hi,

Thanks for pointing this out. I think this post explains the steps: Ansys student 2022 R1

Regards,

Ashish Khemka

[mrife]

Hi Yingnan

I'd like to add to what Ashish said regarding the SIG$SEGV error - the cause most likely is a programming issue but it could be something else.  I once saw the cause be a bad network card!  What we also need to focus on is the traceback that is listed - that is the trace of the functions/libraries.  So MAPDL was using a function in con_int3 when the error happened.  That along with the error being a SIG$SEGV is what we need to look into...and I was not able to find an prior example of this happening with an actual MAPDL release.  Well...it was reported once but when the user ran the model again it did not happen.  And never happened again.  

Are you a commercial customer?  If so please open an technical support call in the standard manner for you.  If you can share the model with technical support and development they may be able to track down the cause.  Development has the skills and tools to track what function MAPDL is calling and the memory stack to figure out what is going on.  Well, the first step would be to see if the issue is repeatable using a different computer to solve.

Mike

p.s. using the Intel BLAS library is not a generic solution.  

[Yingnan Wang]

Hello Ashish and Mike,

Thanks a lot for your reply. Actually, I tried to use the Static structure function instead of the Material Designer to do modeling for the sample, it didn't give any error. 

But the issue is that I have three phases for my composites, i.e., matrix, fiber and particles, when mesh the geometry, the nodes among  the three phase are not matched. 

FYI, I tried use the spaceclaim share topology function for the three phases geometry, but it showed "Bodies failed to share topology. Transferring without shared topology." when I opened the "model. 

By the way, I am a student in the university, can we continue the talk here or shall I open a new post? 

 

Thanks for your help! 

Regards!

Yingnan