LRB Isolator convergence problem

[hbma]

Hi,
I have modeled a 3D bridge model isolated by LRB using "KikuchiAiken" material. My model converges for ap=.0339 m2(area of lead plug). But as soon as I reduce it by a half, it does not converge even for deadload analysis; indeed by using "puts" command before and after "Analyze 1" in deadload analysis, I found out that it even cannot do the deadload analysis. But when I divide the "ap" by 1.5, the model does converge for both deadload and seismic analysis and this factor (1.5) is the border of convergence. I mean by dividing to 1.6, the model answers for deadload but does not answer for the seismic load and is hindered at 6.23 sec. To explain more, for the factor of 1.7, both analysis does not converge and so on.
In all above cases of divergence, there is not an error message in the dos window of analysis, bu the model does not proceed and is hindered, while the CPU is working. For example, when dividing "ap" by factor 2, the analysis is hindered in deadload analysis, interestingly without any error message on screen, but simultaneously the CPU is involved and the analysis is being done but without any results. I donno what is the problem.
regards

[fmk]

with that element when it undergoes one of it'sbig changes in stiffness you need really small time increments to get the model to work.

[hbma]

I have an analysis engine including loops in which dt of analysis is decreased and even is divided into1000 in some loops. But the problem is that it does not enter these loops, indeed it does not fail during the part of code for analysis before the loops and in a specified time, it is hindered without entering the loops, it just lasts so long. If I wanna decrease my time step so much just before the loops, it lasts so long to do the whole analysis since the discretization is so small and is applied in all time step including ones in which such a small time step is not needed.

[fmk]

nothing i can do to help other than suggest you try something else. BTW I have students at stanford who use time steps of 1e-7 for analysis of better behaving elements than this.

[hbma]

For using "variableTransient" analyze command, I have written the following code. But it seems not to work since the analysis is hindered while it doesn't decrease the time step. What is the problem with my code?
"set dtAnalysis 0.01
set dtmin 0.001
set dtmax 0.01
set dtRecord 0.02
set TAnalysis [expr 40+5.0]
set accelx "Series -dt $dtRecord -filePath EQ_recorded_x91.acn -factor 9.81"
pattern UniformExcitation 3 1 -accel $accelx
set Tol 1.e-5
constraints Penalty 1.e18 1.e18
numberer RCM
system ProfileSPD
test NormDispIncr $Tol 5
algorithm KrylovNewton
integrator Newmark 0.5 [expr 1./4.]
analysis VariableTransient
set ok [analyze [expr int($TAnalysis/$dtAnalysis)] $dtmin $dtmax 100]