To Configure the NASTRAN Real Memory
You can configure the amount of memory resources NASTRAN is allowed to use. The default (16 MB) is usually sufficient for small to medium models. Larger models however will often need a higher value. You can increase the NASTRAN real memory setting with the following command:
(ferrum::ferrum-set-nastran-real-mem <value_in_MB>)
The following table shows examples for recommended real memory settings dependent on the mesh size for Linear Static analysis. When staying below these values the analysis will usually still run, but performance may decrease considerably. On the other hand increasing the real memory setting for medium to large models will increase performance.
Model size Recommended real memory
5000 nodes 16 MB (default value)
15000 nodes 32 MB
20000 nodes 36 MB
30000 nodes 48 MB
60000 nodes 128 MB
150000 nodes 336 MB
300000 nodes 560 MB
Note:
When using H-adaptivity the values above only refer to the first iteration with the initial mesh. Subsequent iterations typically require considerably more real memory, therefore this real memory setting also needs to be high enough for the final (refined) mesh.
These values are only a rough guideline and also depend on the complexity of the load case you want to analyze.