Dynamic Convergence with a post very small elastic material

[jackmil]

hello dera frank

I have a model of three story braced frame with initial material steel02 (dynamic analysis) with a strain limitation . so when the fiber-stressStrain of bracing member reach the strain limit,the stressStrain relation change to a extremely small elastic material (e.g. K=0.006 MPA) as a postMaterial for modeling the fracture. I have checked the model for cyclic analysis and have got my interest result for experimental tests,but in dynamic analysis after the bracing fracture I face with convergence error after pga 2 and higher .

I am thinking of the problem about the test parameters for dynamic analysis:

would you please give me your magic solutions about this problem for converging?

constraints Transformation;
numberer RCM ;
system BandGeneral;
test NormDispIncr 1.0e-6 1000;
algorithm KrylovNewton;
integrator Newmark 0.5 0.25
analysis Transient;
#_________________________________________________
set deltaT 0.01;
set ok 0.0
set currentTime 0.0
while {$ok == 0 && $currentTime < $during} {
set ok [analyze 1 $deltaT]
# -------------------------------------------------------------------------------------------------------------------------
if {$ok != 0} {
puts "........................ Trying with Newton ............................."
algorithm Newton
set ok [analyze 1 $deltaT]
algorithm KrylovNewton
}
# ---------------------------------------------------------------------------------------------------------------------------
if {$ok != 0} {
puts "........................ Trying with initial ............................."
algorithm Newton -initial
set ok [analyze 1 $deltaT]
algorithm KrylovNewton;
}
# -----------------------------------------------------------------------------------------------------------------------------
if {$ok != 0} {
puts "........................ Trying NewtonLineSearch ............................."
algorithm NewtonLineSearch 0.8
set ok [analyze 1 $deltaT]
algorithm KrylovNewton;
}
# ------------------------------------------------------------------------------------------------------------------------------
if {$ok != 0} {
puts "........................ Trying initialCurrent ............................."
algorithm Newton -initialCurrent
set ok [analyze 1 $deltaT]
algorithm KrylovNewton;
}
# --------------------------------------------------------------------------------------------------------------------------------
if {$ok != 0} {
puts "..........................................................................................................
.................................................................................................................
........................... deltaT will be changed to =====> deltaT1 .............................
.................................................................................................................
.................................................................................................................."
for {set i 1 } {$i <= 50 } {incr i 1 } {
set ok [analyze 1 [expr $deltaT/50]];
if {$ok != 0} {set i 60}
}
};

Sincerely Yours
Mehdi

[fmk]

don't jave a magical bullet for this. commercial code just keep cutting the time step until it passes the convergence criteria (and it eventually will if time step is small enough).

if the braces do fracture, it might be best to remove them and replace them with some element with the minimal stiffness you are providing. this should be better than keeping the brace there and using initial stifness iterations.

[jackmil]

hello
and Again, as always, constructive suggestions, likes the final bullet in my head.

if I dont change my bracing model,what is your suggestion about using different tolerance, iteration or deltaT for a constant time step? I will remove "for" command.

[fmk]

did you see if increasing the number of iterations when you switch to initial iterations will work.

[jackmil]

dear frank
believe me I did it with 15000 iterations but not worked...

frank
I really appreciate your concern. I will solve it as soon as possible. I think the problem is one other thing I know.