Iterative Solution Failed Due to Insufficient Memory (Nastran 5.1)

[brainLG]

Hello everybody!!

I started a CAE Analysis of a big assembly, and I have this elements:


Summary of Bulk Data cards written

+----------+----------+
| NAME | NUMBER |
+----------+----------+
| CHBDYE | 16729 |
| CONV | 16729 |
| CTETRA | 162885 |
| CTRIA6 | 2222 |
| GRID | 347294 |
| MAT4 | 4 |
| MPC | 72592 |
| NLPARM | 1 |
| PARAM | 4 |
| PCONV | 2 |
| PSHELL | 1 |
| PSOLID | 2 |
| SPC | 2 |
| SPCADD | 1 |
| SPOINT | 2 |
| TEMPD | 1 |
+----------+----------+


I need some help with this problem:


*** USER WARNING MESSAGE 3000 (SITDEL)
THE USE OF CTRIA6 OR CQUAD8 ELEMENTS IN THE ELEMENT BASED ITERATIVE
SOLVER REQUIRES THE USE OF THE PARAMETERS AUTOSPC AND ELITASPC.
WITHOUT THESE PARAMETERS, A SINGULAR MATRIX MAY OCCUR.
*** SYSTEM FATAL MESSAGE 3000 (SITDELC)
ITERATIVE SOLUTION FAILED DUE TO INSUFFICIENT MEMORY
* * * END OF JOB * * *


Also, in an other file from the solution:

Status:6
Version:6.0
PCGLSS0201
PCGLSS0201:Inadequate memory. Increase pre-allocated virtual memory for the solver
Source File
m:\qasimtst_nas5_pb_i386\nastran1\indep\casi\pcglss\preallocmemmanager.c
Line:156
******PCGLSS Error******

PCGLSS0201:Inadequate memory. Increase pre-allocated virtual memory for the solver
13:43:42 1676967 records read from JID file "D:\CAE_Analysis\Cesar_Reyna\Test\reinforce_sim1-deformation.dat"
13:43:42 NSEXIT: EXIT(0)
13:43:42 Analysis complete 0


&&&&& My current Workstation:

Dell Precision T3400
Windows® XP Professional
32B
Intel Core 2 Duo process E7400
2.80GHz 3MB L2 Cash
2GB DDR@ 667MHz SDRAM
Nvidia, Quadro NVS290
256MB
160GB 7200rpm SATA

The Virtual memory that I can set is: 8 GB.

Do you know how to increase this memory?
Somebody does know how can I complete the test?

David E.
Design Eng.
LG ELectronics.

 

[Onda]

I was suggesting you to use the Element based iterative solver (CASI SOLVER). You can activate with SMETHOD=ELEMENT.
But with this solver, very good for big models made of solid elements you cannot solve heat transfer problems.
I suppose you need to buy some more memory and maybe switch to 64bit machine.

Onda

 

[BlasMolero]

Dear David,
Forgot at all the iterative solver and use the regular Direct Sparse Solver. The iterative solver is targeted to solve problems with huge 3D solid models meshed with huge CTETRA/CHEXA solid elements, when you have in the model Shell or beam elements better use the direct sparse solver, is more robust than the iterative one. And your model is relative small (I see 350000 nodes), then you can solve the model perfectly in any 32-bits workstation.

Best regards,
Blas.