QUESTION:

I have a problem with High-Frequency Electromagnetic Analysis when using ANSYS 8.1.
The following error messages occur and the analysis terminates:

*** FATAL *** CP= 11.850 TIME= 18:32:15
Internal memory manager error during solution (too many blocks per
tag). Please report this system error to your ASD or ANSYS, Inc.

This error occurred with the Windows XP 64-bit for Itanium2. It does not occurr in Windows XP 32-bit.


The specification of a computer is as follows:
model - HP zx6000
CPU - Itanium2,1.5GHz
memory - 16GB
OS - Windows XP 64bit Edition Version 2003


ANSWER:


This appears to be related to a bug which was corrected during 9.0 development that affected symbolic assembly and HF emag problems only (HF119 and HF120).
For the PCG memory manager, we estimate the size of each memory block based on the number of defined nodes times the number of DOF per node at the start of the SOLVE command. These two element types use only 1 DOF per node and it appears that internal nodes/DOF are created later during the solution. Thus, the memory block size estimation was too low and some HF emag models resulted in a PCG memory manager error.

Below are two suggested workarounds.

1. Define a shell63 element in the input deck. ANSYS then assumes 6 dofs per node in its memory calculations and it bypasses the problem.

2. Use the assopt,frontal command which uses frontal assembly for the matrices. This is really the recommended workaround.


QUESTION:

I have a problem with High-Frequency Electromagnetic Analysis when using ANSYS 8.1.
The following error messages occur and the analysis terminates:

*** FATAL *** CP= 11.850 TIME= 18:32:15
Internal memory manager error during solution (too many blocks per
tag). Please report this system error to your ASD or ANSYS, Inc.

This error occurred with the Windows XP 64-bit for Itanium2. It does not occurr in Windows XP 32-bit.


The specification of a computer is as follows:
model - HP zx6000
CPU - Itanium2,1.5GHz
memory - 16GB
OS - Windows XP 64bit Edition Version 2003


ANSWER:


This appears to be related to a bug which was corrected during 9.0 development that affected symbolic assembly and HF emag problems only (HF119 and HF120).
For the PCG memory manager, we estimate the size of each memory block based on the number of defined nodes times the number of DOF per node at the start of the SOLVE command. These two element types use only 1 DOF per node and it appears that internal nodes/DOF are created later during the solution. Thus, the memory block size estimation was too low and some HF emag models resulted in a PCG memory manager error.

Below are two suggested workarounds.

1. Define a shell63 element in the input deck. ANSYS then assumes 6 dofs per node in its memory calculations and it bypasses the problem.

2. Use the assopt,frontal command which uses frontal assembly for the matrices. This is really the recommended workaround.





Show Form
No comments yet. Be the first to add a comment!