Time History Problem

jboshra · Post by **jboshra** » Sun Apr 26, 2009 10:20 pm

Hello
I performed a time history non linear analysis (using a given ground record) on a 2D frame representing a three-span bridge with two piers fixed at bottom and hinged to the deck at their top. I scaled the records incrementally and for each scale value I recorded the maximum absolute curvature of each pier (doing Incremental Dynamic Analysis). The analysis has been done twice: once WITH and another time WITHOUT the “minmax” material. The results are as follows:
*At scale factor of 2.0 , the max absolute curvature using minmax =0.00439, while the max absolute curvature WITHOUT using minmax=0.00715
*At scale factor of 2.24, the max absolute curvature using minmax =347 (looks like a numerically divergent solution occurred), while the max absolute curvature WITHOUT using minmax=0.00939.
*At scale factor of 2.45 , the max absolute curvature using minmax =0.0044, while
the max absolute curvature WITHOUT using minmax=0.0206

I am quite confident that the results using minmax at scale factors 2.0 and 2.45 makes much more sense than without using the minmax material. However, I hope you can help me find the reason for the divergent (nonsense) result occurring at the intermediate Scale Factor of 2.24. Kindly note that I get randomly these non-converged (or erratic) values of curvature at other high scale factors with the minmax material used; but whenever I have non-erratic values for the curvature at these high scale factors, the results still make much more sense than the case without the minmax material.

*NOTE: I am using Opensees program Version 1.7. Thanks

silvia · Post by **silvia** » Mon Apr 27, 2009 6:02 am

try it with the latest release of opensees.

jboshra · Post by **jboshra** » Mon Apr 27, 2009 12:13 pm

Thanks for your prompt response
I tried it with the latest release but i still have the same problem.

jboshra · Post by **jboshra** » Sun May 03, 2009 9:38 pm

Dear,
I tried to use the latest release of the opensees, also I tried to change the convergence tolerance of the code; but I still have a numerical divergence at the same scale factors. I hope you can help me to solve this problem; I can send my code. Thanks

silvia · Post by **silvia** » Mon May 04, 2009 6:32 am

look at your recorders

jboshra · Post by **jboshra** » Wed May 06, 2009 4:53 am

I checked the recorders; they are the same as the examples manual.

silvia · Post by **silvia** » Wed May 06, 2009 5:33 am

you must have convergence issues for that case.
do you have a convergence script, like in the examples?

jboshra · Post by **jboshra** » Mon May 11, 2009 3:10 am

Here is the convergence script when I run the program. The analysis is completed successfully but I still got nonsense results.

CTestNormDispIncr::test() - iteration: 1 current Norm: 0.000253082 (max: 0.01, Norm deltaR: 4.95783e-005)
CTestNormDispIncr::test() - iteration: 1 current Norm: 0.000253013 (max: 0.01, Norm deltaR: 4.95646e-005)
Transient analysis completed SUCCESSFULLY
eigen values at start of transient: 1.076724e-001 1.671086e+001

silvia · Post by **silvia** » Mon May 11, 2009 5:34 am

that's the opensees output.
do you have part of the script where you say:
set ok [analyze ...]
if {$ok !=0} { ...

that's what deals with convergence, see the examples manual, i think example 4 on....

jboshra · Post by **jboshra** » Wed May 13, 2009 4:20 am

yes i have this part of the code and it is as follows:

code:
while {$ok == 0 && $tCurrent < $tFinal} {

set ok [analyze 1 .01]

# if the analysis fails try initial tangent iteration
if {$ok != 0} {
puts "regular newton failed .. lets try an initail stiffness for this step"
test NormDispIncr 1.0e-8 100
algorithm ModifiedNewton -initial
set ok [analyze 1 .01]
if {$ok == 0} {puts "that worked .. back to regular newton"}
test NormDispIncr 1.0e-8 10
algorithm Newton
}

set tCurrent [getTime]
}

# Print a message to indicate if analysis succesfull or not
if {$ok == 0} {
puts "Transient analysis completed SUCCESSFULLY";
} else {
puts "Transient analysis completed FAILED";
}

I hope you can help me to solve this problem. Thanks a lot.

silvia · Post by **silvia** » Wed May 13, 2009 5:38 am

i don't know what to say, check the model.