Problem in convergence of nonlinear time history analysis

[rejina]

Hello everyone,

I have been modeling a confined rocking wall which can rotate at its both ends, and also has a post-tensioned tendon at its center. I could successfully run static monotonic and cyclic pushover analysis, however, I am stuck with convergence issue while running non-linear time history analysis. I have been using forcebeamcolumn element to model columns while dispbeamcolumn element to model beams. There are many rocking walls in my models so I am using fiber models to represent a rocking wall. There are lots of rigid elements too. Following are the dynamic analysis parameters that I have been using till now, although I have tried with different integrators too but its not working out. Please suggest me the way out so that I can reach for the solution.

constraints Penalty 1.e16 1.e16;
numberer RCM
system UmfPack -lvalueFact 60
test EnergyIncr 1.e-3 2000 0;
algorithm NewtonLineSearch;
integrator Newmark 0.5 0.25
analysis Transient

PS. It runs for some time duration only.

Thank you in advance.

Rejina

[fmk]

try alternative solution strategies inside a for or while loop that handles the analysis part of the script.
see following for simple example that tries some other solution algorithm

http://opensees.berkeley.edu/wiki/index ... ple3.3.tcl

[rejina]

Dear fmk,

I have tried that already with different loops inside as you said too but, it fails ultimately. I have also tried to reduce the step size within the loop for solving convergence as well but it is not working out. However static pushover analysis runs up to 4% drift although with some convergence issue. I am confused whether it signifies issues in my model or not. Please give me some suggestions, to check out the problem.