Increasing output time step increment

[Armin]

I am using a script to run a parameter study on a model. I was using a transient analysis with constant time steps in windows base system.

1-because of small time increment in transient analysis the output files are very large. is there any way to increase recorders time step increment to reduce the size of output files?
For example in analysis with dt = 0.001s outputs record only in each 0.01s.

2-Can opensees.exe use dual core CPU features in MS WINDOWS XP systems?

[fmk]

1. include -dT value? in the args for the recorder .. after the -file fileName? would be good.

2. OpenSees is not thread safe, there is a version for parameter studies that will run
on parallel machines that have MPI installed; i know of know such version for sual-core machines running XP.

[quhui]

Hi, every friends:

I meet the same questions as Armin. According on the suggestions of fmk,
I modified my command as follows:

recorder Node $fName.out 0.01 disp -load -node 10 -dof 1


but opensees gave me an error.

Would you like give me an example?

Thanks a lot!

[quhui]

Hello, fmk

I modified the command again as follows:

recorder Node $fName.out disp 0.01 -load -node 10 -dof 1

Opensees run this time, but the result file was empty!

Why?

Would you like give me a hand?

Many thanks!

[silvia]

I think you should follow Frank's suggestion to use " -dT $value" argument after the filename:


recorder Node -file Rnode.out -dT 0.01 -time -node 1 3 5 -dof 1 2 reaction

this should work, please let me know if it doesn't.

[quhui]

Dear silvia:

Thank you very much!

I had modified the command as you told me.

It worked, but I found the results were recorded at every increment, not every 0.01 increment.

[silvia]

mmm, that's why this is an undocumented feature, as it doesn't work with all recorders.
let's see what frank says, otherwise, i'd reduce the data in post-processing.

[quhui]

My result file is 580M. It is so large that no software can open it.

[Armin]

it is a sample recorder command that I used in my scripts and works well.

recorder Node -file rott.out -dT 0.02 -time -node 3 -dof 6 disp

In this line the z axes rotation of node 3 with 0.02s time increment will be recorded in file rott.out

best regards

[quhui]

Many thanks for everyone!

I am sorry that my script still doesn't work. Is there a mistake in my script?

It is part of my script below, perhaps the problem is in procRC.tcl. But I cann't check it out.

It is very long

#######################################
pattern Plain 2 Linear {

load 10 0 -285100 0 -const

}


system ProfileSPD

constraints Plain

integrator LoadControl 0 1 0 0

test NormDispIncr 1e-8 150

algorithm Newton

numberer RCM

analysis Static

analyze 1

loadConst -time 0.0


pattern Plain 1 Linear {

#load nd? Fx? Fy? Mz?

load 10 1 0 0

}


recorder Node -file pu.out -dT 0.01 -time -node 10 -dof 1 disp

set peakpts [list 0.01 5 5 10 10 15 15 20 20 25 25 30 30 40 40]

set increment 250

set nodeTag 10

set dofTag 1

procRC $increment $nodeTag $dofTag $peakpts

# print the results at node and at all elements

print node

#print element

#########################################
########################################################################################################

#

# procRC.tcl

## procedure for setting up a reversed cycle loading scheme. The input are mainly the

## peak points for the loading.

## The procedure primarily uses Displacement control for loading, if it fails uses ArcLength control

## created : QU HUI (quhui@fzu.edu.cn) dated: AUGUST. 2006

########################################################################################################



proc procRC { incre nodeTag dofTag peakpts } {


set displayTag 0;

set numTimes 1500;


set x [lindex $peakpts 0];

set dU [expr $x/$incre];

#set dU0 [expr $dU/1000];

set dU0 [expr $dU/10000];

integrator DisplacementControl $nodeTag $dofTag 0.0 1 $dU $dU

analysis Static

analyze $incre


integrator DisplacementControl $nodeTag $dofTag 0.0 1 [expr -$dU] [expr -$dU]

analyze [expr 2*$incre]


integrator DisplacementControl $nodeTag $dofTag 0.0 1 $dU $dU

analyze $incre


## end the first peak pt start for others


for {set j 1} {$j < [llength $peakpts]} {incr j 1} {


set y [lindex $peakpts $j]

set dSt [expr $y/$dU]

set dS [expr int($dSt)]


test NormDispIncr 1e-8 $numTimes $displayTag

algorithm Newton


############# start loading cycle ##################

set t 0;


while {$t != $dS} {

integrator DisplacementControl $nodeTag $dofTag 0.0 1 $dU $dU

set ok [analyze 1]

incr t 1;


if {$ok != 0} {

# if {$t == $dS} {break};

puts "Displacement control failed ..... trying Arc-Length control"

set currentDisp [nodeDisp $nodeTag $dofTag]

puts "Current Displacement is $currentDisp"

# algorithm Linear

test NormDispIncr 1e-6 $numTimes $displayTag

#algorithm ModifiedNewton

# integrator DisplacementControl $nodeTag $dofTag 0.0 1 $dU0 $dU0

# integrator DisplacementControl $nodeTag $dofTag 0.0 10 $dU0 $dU0

integrator ArcLength [expr $dU0] 1.0

# set ok [analyze 1]

analyze 1

}

# puts "that worked ..... back to regular Newton "

test NormDispIncr 1e-8 $numTimes $displayTag

# algorithm Newton

}


################## end of loading cycle, start unloading cycle ########


set t 0;


while {$t != [expr 2*$dS]} {

integrator DisplacementControl $nodeTag $dofTag 0.0 1 [expr -$dU] [expr -$dU]

set ok [analyze 1]

incr t 1;


if {$ok != 0} {

# if {$t == [expr 2*$dS]} {break};

puts "Displacement control failed ..... trying Arc-Length control"

set currentDisp [nodeDisp $nodeTag $dofTag]

puts "Current Displacement is $currentDisp"

# algorithm Linear

test NormDispIncr 1e-6 $numTimes $displayTag

#algorithm ModifiedNewton

# integrator DisplacementControl $nodeTag $dofTag 0.0 1 [expr -$dU0] [expr -$dU0]

# integrator DisplacementControl $nodeTag $dofTag 0.0 10 [expr -$dU0] [expr -$dU0]

integrator ArcLength [expr $dU0] 1.0

# set ok [analyze 1]

analyze 1

}

# puts "that worked .... back to regular Newton "

test NormDispIncr 1e-8 $numTimes $displayTag

# algorithm Newton

}


############# end of unloading cycle, start reloading cycle ###########


set t 0;


while {$t != $dS} {

integrator DisplacementControl $nodeTag $dofTag 0.0 1 $dU $dU

set ok [analyze 1]

incr t 1;


if {$ok != 0} {

# if {$t == $dS} {break};

puts "Displacement control failed ..... trying Arc-Length control"

set currentDisp [nodeDisp $nodeTag $dofTag]

puts "Current Displacement is $currentDisp"

# algorithm Linear

test NormDispIncr 1e-6 $numTimes $displayTag

#algorithm ModifiedNewton

# integrator DisplacementControl $nodeTag $dofTag 0.0 1 $dU0 $dU0

# integrator DisplacementControl $nodeTag $dofTag 0.0 10 $dU0 $dU0

integrator ArcLength [expr $dU0] 1.0

# set ok [analyze 1]

analyze 1

}

# puts "that worked .... back to regular Newton "

test NormDispIncr 1e-8 $numTimes $displayTag

# algorithm Newton

}


######## reloading cycle completed #############################


if {$ok == 0} {

puts "analysis succesful at $y mm displacement";

} else {

puts "analysis could not proceed fine beyond $y mm displacement";

}


}

}

[silvia]

are you on the latest version?

[quhui]

I used the lasted version: opensees exe.1.7.2

[silvia]

so,
i let your analysis run for a while and then i plotted the time increments in the output and after a while you do get the time increment you wanted.
somehow, I think your time step gets messed up with your proc.
Based on my experience, I don’t think you should specify the number of steps, as these lead to disp increments that are too small for small peaks and too large for large ones, and that’s why you get soo much data.
I think you should specify the displacement increment in your analysis and calculate the number of steps based on that. Your data won’t be sooo busy.

[quhui]

I have tried again and again, but I fail.

In my proc, I use Displacement control, not transient analysis. If is it wrong to use :

recorder Node -file $fName.out -dt 0.01 -time -load -node 10 -dof 1 disp?

Maybe it is :

recorder Node -file $fName.out -du 0.01 -time -load -node 10 -dof 1 disp

or something else?

[silvia]

even though it is static, the load is applied over time.
the only difference between static and transient, i believe, is the mass and damping.

still, you have -dt, when maybe it should be dT....