time step changes

[jefferyr]

During a transient (EQ) analysis, have you used scripts where the time step is changed on the fly? So that a coarser time step is used, and when convergence issues arise, then the time step is made smaller and smaller until convergence or a limit is reached? Then the time step is returned to the larger value to continue the analysis?

If you have scripts that illustrate this technique, could you post them?

thanks,

Jeff

[silvia]

here is my latest:

[code] # ----------------------------------------------first analyze command------------------------
set ok [analyze $Nsteps $DtAnalysis]
# ----------------------------------------------if convergence failure-------------------------
if {$ok != 0} {; # performance is slower inside this loop
puts "Trying Reducing This Time Step by 2 .."
set ok [analyze 1 [expr $DtAnalysis/2]]
if {$ok != 0} {
puts "Trying Reducing This Time Step by 10 .."
set ok [analyze 1 [expr $DtAnalysis/10]]
}
if {$ok != 0} {
puts "Trying Reducing This Time Step by 100 .."
set ok [analyze 1 [expr $DtAnalysis/100]]
}
if {$ok != 0} {
puts "Trying Newton with Initial Tangent .."
test $testType $Tolerance $maxNumIterConverge $printFlagConverge
algorithm Newton -initial
set ok [analyze 1 $DtAnalysis]
test $testType $Tolerance $maxNumIter $printFlag
algorithm $algorithmType
}
if {$ok != 0} {
puts "Trying Broyden .."
algorithm Broyden 8
set ok [analyze 1 $DtAnalysis]
algorithm $algorithmType
}
if {$ok != 0} {
puts "Trying NewtonWithLineSearch .."
algorithm NewtonLineSearch $NewtonLineSearchRatio
set ok [analyze 1 $DtAnalysis]
algorithm $algorithmType
}
if {$ok == 0} {puts "Converged at this step, continuing...."}
}
# -----------------------------------------------------------------------------------------------------[/code]

[bkosbab]

I tried this code, but for some reason, the analysis stops after "Converged at this step, continuing...." What forces the analysis to continue where it left off, that I'm missing?

[silvia]

does it really stop, or does it continue?

[bkosbab]

it thinks it's done, and stops. I went back to the convergence helper code from the examples manual, and added the decreasing time step, and it continues running. The only difference I see is the extra while loop in the examples manual.