Joined: 10 Jan 2012
Location: Mumbai, India
|Posted: Wed Oct 07, 2015 1:28 am
I am performing nonlinear dynamic analysis for 10 materials and 50 earthquake time histories, totaling 500 analyses. I am using BEAM4 and COMBIN40 elements. My model is very small (No of equations - 272) and dynamic analysis steps in the range of 1200 to 3000 time steps. each analysis runs for 6 to 15 minutes.
system available with me offers huge resource - 128GB RAM and 16 core CPU (2.6GHz each). My problem is I am not able to utilize it to fullest.
My program flow (algorithm) is like this -
!solve model for ii material and jj earthquake
!update material property to "ii"
!solve model for jj earthquake time history
!extract desired result, store in a matrix and reset the model.
I am using HPC setup with SMP and all 16 processors.
But I understand that this will not result into much time saving as long as my APDL code gives
commands to solve sequentially.
Is there any way in which I can ask ANSYS to run "n" number of DO loops in parallel. results of each of my 500 analyses are independent of each other. So if there is any way in which I can run, say 5 analyses, at the same time, I will save huge amount of time.
Thanks for your time and advise.
Dept of Civil Engineering
Indian Institute of Technology - Bombay (IIT-B)