• State Suggested Answer
  • Replies 3 replies
  • Answers 2 answers
  • Subscribers 111 subscribers
  • Views 2540 views
  • Users 0 members are here

RAM Elements, Version 13.4 - Nonlinear Analysis with Temperature Load

Eriks Jekabsons
Offline

I am currently working on a concrete frame model using RAM Elements.  I have been able to get the non-linear analysis to run on the model; however, when I add a temperature load, the non-linear analysis crashes.  Can anyone provide me with any pointers on this?

  • 0

    It's working OK for me with 13.04.01.208.  Is your model non-linear due to the presence of tension-only members, compression-only springs or just from using P-Delta?  Are the thermal loads applied to linear members, shells or both?  Better yet, send over your current file for review.



  • 0 Offline in reply to Seth Guthrie
    Thank you for the response. The non-linear analysis is only for P-delta. The temperature load is applied to both linear members as well as shells. In earlier runs, we were able to get the analysis to finish, but it took at least a day to complete. The input file is over 20 mb, so I will need to upload it on the secure file upload page. Let me know if you see any places where we can better optimize the run time.
  • 0 in reply to Eriks Jekabsons
    I got your file, but it does not seem to include the custom sections and materials from your file so I can't rerun it. What I can tell is that after meshing the model has about 500 nodes and 5000 shells. It has 35 load cases, but more importantly 293 load combinations. When you switch to P-Delta analysis it's necessary to analyze each load combinations (and load case) and iterate on each until deflections converge. So if it takes 3 iterations to get convergence in a P-Delta analysis and you run all load combinations it would take at least (293+35)/35)*2.5 = 23 times as long as first order analysis. It's also possible that it's running out of memory of course.

    To test it a little more rigorously I would first run dead load and thermal load only in first order analysis to see how long it takes. Then try running a combination of dead and thermal (only) in P-Delta analysis.

    If the analysis time is too long to make these tests practical try increasing the mesh size first. Also, make the P-Delta convergence tolerance something larger, e.g. 0.001.

    If it really does turn out to be a problem with thermal loads and P-Delta, and not simply a matter of time/memory, let me know and I'll dig in further.



Communities
Support and Services
Training and Learning
Social Media