Solver Performance
Solver Performance
(OP)
I have what might be an odd question.
I, personally, don't work a lot with STAAD but have done so in the past. I have, however, collegues that do and when the modells get larger, say 20000 gridpoints, especially with plate elements the solver gets very slooow.
Since I today work with FEM-software we have compared and what in STAAD takes hours in FEM-software typically takes minutes.
Does anybody have similar experiences or are we doing something wrong?
I have heard that the new solvers in for example ver 2004 is faster. If so, how much faster? (We haven't recieved it yet).
The result today is that STAAD is useless for some applications so I'm curious about the future.
Regards
Thomas
Ps The "slow" part is the triangulization.
I, personally, don't work a lot with STAAD but have done so in the past. I have, however, collegues that do and when the modells get larger, say 20000 gridpoints, especially with plate elements the solver gets very slooow.
Since I today work with FEM-software we have compared and what in STAAD takes hours in FEM-software typically takes minutes.
Does anybody have similar experiences or are we doing something wrong?
I have heard that the new solvers in for example ver 2004 is faster. If so, how much faster? (We haven't recieved it yet).
The result today is that STAAD is useless for some applications so I'm curious about the future.
Regards
Thomas
Ps The "slow" part is the triangulization.





RE: Solver Performance
RE: Solver Performance
No there are no master/slave in the models. Actually they are a pretty straightforward mix of beams (members) and plates (elements).There are lot of internal releases if that means anything.
I don't know if I can send you the models due to proprietary rights, but I will definetly ask. I see that you work for rei and that makes your input very interesting.
I will get back to you asap.
Regards
Thomas
RE: Solver Performance
Just out of curiousity, whay kind if triangulization does STAAD use? Is it full-matrix or some more efficient method. I've seen that there is a band-width-reduction.
One problem might be that the stucture we are working with will give a larger bandwidth than the "average" spaceframe.
Thank you RAYC1 for trying.