Preffered analysis options

[ospinag]

Hi everyone,

I have a model with +4100 nodes, containing elastic beam elements and shell MITC4 elements. I'm trying to make a time history analysis but it's taking very long, which seems weird to me. I'm using the following analysis options:

integrator Newmark 0.5 0.25;
test EnergyIncr 1.0e-12 10 0
algorithm Newton
numberer RCM
constraints Transformation
system ProfileSPD
analysis Transient;
analyze 159 0.0175

can somebody make a suggestion of what options to change to make it faster? Currently is has taken almost 24hrs running in the Neeshub site, and the analysis has not finished.

Thanks!

[vesna]

To me it looks like the analysis is hanging somewhere (it got stuck).

For large models use Sparse solvers (e.g. UmfPack). Your tolerance may be too low. I would not go lower than 1e-8.

make use of "puts" command to check the progress of your analysis. Do your anlaysis in a loop and after each step of analysis output the time step.

[ospinag]

.

[ospinag]

vesna wrote:
> To me it looks like the analysis is hanging somewhere (it got stuck).
>
> For large models use Sparse solvers (e.g. UmfPack). Your tolerance may be
> too low. I would not go lower than 1e-8.
>
> make use of "puts" command to check the progress of your
> analysis. Do your anlaysis in a loop and after each step of analysis output
> the time step.

Thanks!