The problem during dynamic history analysis

ionwang · Post by **ionwang** » Thu Apr 09, 2009 2:04 pm

Hello everybody,

I try a simple example to understand the dynamic history analysis, the odd result comes out. In the test, I decompose the whole series to several series and use "for-loop" to connect them again. The testing model is "Ex1a.Canti2D".

Does anybody know the reason of inducing the difference?

Here is the code.

wipe; # clear opensees model
model basic -ndm 2 -ndf 3; # 2 dimensions, 3 dof per node
file mkdir data; # create data directory

# define GEOMETRY -------------------------------------------------------------
# nodal coordinates:
node 1 0 0; # node#, X Y
node 2 0 432

# Single point constraints -- Boundary Conditions
fix 1 1 1 1; # node DX DY RZ

# nodal masses:
mass 2 5.18 1.e-9 0.; # node#, Mx My Mz, Mass=Weight/g.

# Define ELEMENTS -------------------------------------------------------------
# define geometric transformation: performs a linear geometric transformation of beam stiffness and resisting force from the basic system to the global-coordinate system
geomTransf Linear 1; # associate a tag to transformation

# connectivity: (make A very large, 10e6 times its actual value)
element elasticBeamColumn 1 1 2 3600000000 4227 1080000 1; # element elasticBeamColumn $eleTag $iNode $jNode $A $E $Iz $transfTag

# Define RECORDERS -------------------------------------------------------------
recorder Node -file Data/DFree2.out -time -node 2 -dof 1 2 3 disp; # displacements of free nodes
recorder Node -file Data/VFree2.out -time -node 2 -dof 1 2 3 vel;
recorder Node -file Data/DBase2.out -time -node 1 -dof 1 2 3 disp; # displacements of support nodes
recorder Node -file Data/RBase2.out -time -node 1 -dof 1 2 3 reaction; # support reaction
recorder Drift -file Data/Drift2.out -time -iNode 1 -jNode 2 -dof 1 -perpDirn 2 ; # lateral drift
recorder Element -file Data/FCol2.out -time -ele 1 globalForce; # element forces -- column
recorder Element -file Data/DCol2.out -time -ele 1 deformation; # element deformations -- column

# define GRAVITY -------------------------------------------------------------
pattern Plain 1 Linear {
load 2 0. -2000. 0.; # node#, FX FY MZ -- superstructure-weight
}
constraints Plain; # how it handles boundary conditions
numberer Plain; # renumber dof's to minimize band-width (optimization), if you want to
system BandGeneral; # how to store and solve the system of equations in the analysis
test NormDispIncr 1.0e-8 6 ; # determine if convergence has been achieved at the end of an iteration step
algorithm Newton; # use Newton's solution algorithm: updates tangent stiffness at every iteration
integrator LoadControl 0.1; # determine the next time step for an analysis, # apply gravity in 10 steps
analysis Static # define type of analysis static or transient
analyze 10; # perform gravity analysis
loadConst -time 0.0; # hold gravity constant and restart time

rayleigh 0. 0. 0. [expr 2*0.02/pow([eigen 1],0.5)]; # set damping based on first eigen mode

# DYNAMIC history analysis -------------------------------------------------------------
# create load list
set loadlist "0.1 1.0 3.0 20.0 80.0 100.0 70.0 30.0 -40.0 -70.0 -20.0 10.0 0.0 ";

# create the analysis
wipeAnalysis; # clear previously-define analysis parameters
constraints Plain; # how it handles boundary conditions
numberer Plain; # renumber dof's to minimize band-width (optimization), if you want to
system BandGeneral; # how to store and solve the system of equations in the analysis
test NormDispIncr 1.0e-8 10; # determine if convergence has been achieved at the end of an iteration step
algorithm Newton; # use Newton's solution algorithm: updates tangent stiffness at every iteration
integrator Newmark 0.5 0.25 ; # determine the next time step for an analysis
analysis Transient; # define type of analysis: time-dependent

set DtAnalysis [expr 0.02]; # time-step Dt
set Nstep [llength $loadlist];

pattern Plain 100 "Series -time $DtAnalysis -values $loadlist -factor 100" {
load 2 1.0 0. 0.;
}
analyze $Nstep $DtAnalysis;

puts "Done!"

*********************************************************
*********************************************************
wipe; # clear opensees model
model basic -ndm 2 -ndf 3; # 2 dimensions, 3 dof per node
file mkdir data; # create data directory

# define GEOMETRY -------------------------------------------------------------
# nodal coordinates:
node 1 0 0; # node#, X Y
node 2 0 432

# Single point constraints -- Boundary Conditions
fix 1 1 1 1; # node DX DY RZ

# nodal masses:
mass 2 5.18 1.e-9 0.; # node#, Mx My Mz, Mass=Weight/g.

# Define ELEMENTS -------------------------------------------------------------
# define geometric transformation: performs a linear geometric transformation of beam stiffness and resisting force from the basic system to the global-coordinate system
geomTransf Linear 1; # associate a tag to transformation

# connectivity: (make A very large, 10e6 times its actual value)
element elasticBeamColumn 1 1 2 3600000000 4227 1080000 1; # element elasticBeamColumn $eleTag $iNode $jNode $A $E $Iz $transfTag

# Define RECORDERS -------------------------------------------------------------
recorder Node -file Data/DFree1.out -time -node 2 -dof 1 2 3 disp; # displacements of free nodes
recorder Node -file Data/VFree1.out -time -node 2 -dof 1 2 3 vel;
recorder Node -file Data/DBase1.out -time -node 1 -dof 1 2 3 disp; # displacements of support nodes
recorder Node -file Data/RBase1.out -time -node 1 -dof 1 2 3 reaction; # support reaction
recorder Drift -file Data/Drift1.out -time -iNode 1 -jNode 2 -dof 1 -perpDirn 2 ; # lateral drift
recorder Element -file Data/FCol1.out -time -ele 1 globalForce; # element forces -- column
recorder Element -file Data/DCol1.out -time -ele 1 deformation; # element deformations -- column

# define GRAVITY -------------------------------------------------------------
pattern Plain 1 Linear {
load 2 0. -2000. 0.; # node#, FX FY MZ -- superstructure-weight
}
constraints Plain; # how it handles boundary conditions
numberer Plain; # renumber dof's to minimize band-width (optimization), if you want to
system BandGeneral; # how to store and solve the system of equations in the analysis
test NormDispIncr 1.0e-8 6 ; # determine if convergence has been achieved at the end of an iteration step
algorithm Newton; # use Newton's solution algorithm: updates tangent stiffness at every iteration
integrator LoadControl 0.1; # determine the next time step for an analysis, # apply gravity in 10 steps
analysis Static # define type of analysis static or transient
analyze 10; # perform gravity analysis
loadConst -time 0.0; # hold gravity constant and restart time

rayleigh 0. 0. 0. [expr 2*0.02/pow([eigen 1],0.5)]; # set damping based on first eigen mode

# DYNAMIC history analysis -------------------------------------------------------------
# create load list
set loadlist "0.1 1.0 3.0 20.0 80.0 100.0 70.0 30.0 -40.0 -70.0 -20.0 10.0 0.0 ";

# create the analysis
wipeAnalysis; # clear previously-define analysis parameters
constraints Plain; # how it handles boundary conditions
numberer Plain; # renumber dof's to minimize band-width (optimization), if you want to
system BandGeneral; # how to store and solve the system of equations in the analysis
test NormDispIncr 1.0e-8 10; # determine if convergence has been achieved at the end of an iteration step
algorithm Newton; # use Newton's solution algorithm: updates tangent stiffness at every iteration
integrator Newmark 0.5 0.25 ; # determine the next time step for an analysis
analysis Transient; # define type of analysis: time-dependent

set DtAnalysis [expr 0.02]; # time-step Dt
set Nstep [llength $loadlist];

for {set i 1} {$i <= $Nstep} {incr i 1} {
set timelist "[format {%5.2f} [getTime]] [format {%5.2f} [expr [getTime]+$DtAnalysis]]"
puts $timelist
if {$i == 1} {
set loadinput "0 [lindex $loadlist 0]";
} else {
set loadinput "[lindex $loadlist [expr $i-2]] [lindex $loadlist [expr $i-1]]";
}
puts $loadinput
pattern Plain [expr $i+1000] "Series -time $timelist -values $loadinput -factor 100" {
load 2 1.0 0. 0.;
}
analyze 1 $DtAnalysis;
}
puts "Done!"

silvia · Post by **silvia** » Thu Apr 09, 2009 7:28 pm

i don't understand the question.

ionwang · Post by **ionwang** » Fri Apr 10, 2009 7:13 am

Sorry about that.
The question is:
*(1)We can use the Series command to execute a dynamic history analysis under a series loading. Such as:
set DtAnalysis [expr 0.02]; # time-step Dt
set Nstep [llength $loadlist];
pattern Plain 100 "Series -time $DtAnalysis -values $loadlist -factor 100" {
load 2 1.0 0. 0.;
}
analyze $Nstep $DtAnalysis;

*(2) On the other hand, i want to use "for-loop" to complete this process.
So i need decompose the whole series into a series of short segments(for example: the origin series is "0. 0.1 0.3 0.4 0.8 0.9",after decomposing, it is "0. 0.1", "0.1 0.3", "0.3 0.4", "0.4 0.8", "0.8 0.9" ). the code is as following.

for {set i 1} {$i <= $Nstep} {incr i 1} {
set timelist "[format {%5.2f} [getTime]] [format {%5.2f} [expr [getTime]+$DtAnalysis]]" ##get each time segement corresponding to the loading data segment.
puts $timelist
if {$i == 1} {
set loadinput "0 [lindex $loadlist 0]";
} else {
set loadinput "[lindex $loadlist [expr $i-2]] [lindex $loadlist [expr $i-1]]";
} #get each loading data segment
puts $loadinput
pattern Plain [expr $i+1000] "Series -time $timelist -values $loadinput -factor 100" {
load 2 1.0 0. 0.;
}
analyze 1 $DtAnalysis;
}

***I thought the results would be same, but it did not. The results were totally different.
I wonder the reason of this difference.

silvia · Post by **silvia** » Fri Apr 10, 2009 10:58 am

try it on a simple model.

ionwang · Post by **ionwang** » Fri Apr 10, 2009 1:25 pm

Silvia,
I have tried it using "Ex1a.Canti2D". The response of free node 2 is different due to these two ways. I'm confused on this. I don't know the reason. Could you help me to figure out this problem?

Thank you very much!

silvia · Post by **silvia** » Fri Apr 10, 2009 2:09 pm

please e-mail me the script.

ionwang · Post by **ionwang** » Fri Apr 10, 2009 5:53 pm

Silvia,
I have sent the script to your "silvia at peer dot berkeley dat edu"

Thank you!

silvia · Post by **silvia** » Fri Apr 10, 2009 7:55 pm

nope, old address. use opensees-support at berkeley dot edu.

however, i'm in the midst of preparing for heading out to L'aquila next week, so i won't be able to look at it until i return on the 24th.
please try to put some puts here and there to test your algorithm.

ionwang · Post by **ionwang** » Sat Apr 11, 2009 6:09 am

That is OK.

In first floor I have posted the integrated code. This example is based on "Ex1a.Canti2D". The transient analysis parameters is the same as "Ex1a.Canti2D.EQ". I just change the "pattern" and "analyze" parts.

The point is how the transient analysis invokes the Series load. The interpolation should be used to handle the series. So I keep to use the path time series command rather than to use the rectanglular time series.
I hope the same results, not the approximated results. However, it doesn't work. Under the "for-loop" load, the response is all zero.