Without going into too much unnecessary detail here is what I am trying to accomplish:
I have a six-story moment frame experimental specimen that I have modeled in OpenSEES that has local nonlinearities at each of the beam-column interfaces.
I can adjust the degree of non-linearity such that the joint at the beam-column interface exhibits nonlinear hysteretic behavior with a specific moment capacity (Mcap).
This Mcap is essentially based on how tight I tighten the connection using a torque wrench.
Based on data I have collected I can relate the torque used to tighten the joint to a specific Mcap.
There is, however, some slight variability between the anticipated Mcap (based on the tightening torque) and the measured Mcap behavior of the joint.
I therefore have 12 probability distribution functions that describe the relationship between the tightening torque and the moment capacity of the joint.
I have 2 loading configurations, 2 resistance modification (R) factors, and 5 suites of 39 ground motions.
I therefore have (2x2x5x39 = 780) 780 distinct and pre-defined input scenarios. I would like to perform 500 simulations of each of the pre-defined input scenarios based on the probability distributions obtained for each of the 12 beam-column interfaces so that I can develop a confidence interval for the response of the experimental model
Is there any way I can set up something similar to “for” loops in my OpenSEES (.tcl) files so I can send one script to our research computing cluster that can handle the (500x780 = 390000) 390,000 analyses?
Any help or advice would be greatly appreciated.
Parallel Processing for Monte Carlo Simulation
Moderator: selimgunay