The OpenSees Community

Thanks.
This will save a lot of computational time.

Thanks for your reply.
Indeed, I use that element, and I would be gratefull if this modification-addition could be done.

Hi,

Regarding the DATABASE command, I noticed that it doesn't seem not to work with TripleFrictionPendulum elements.
Is there any alternative way to restore a model in the state just after static analysis when doing IDA and want to exclude the static analysis part from the IDA loop?
When using the RESET command within the IDA loop, not only the dynamic but also the static analysis response results are deleted (even though the static analysis part is outside the loop).

In all the examples at http://cyclic.ucsd.edu/opensees/ the mass proportional damping coef. in set to zero. They are about single quadrilateral elements or columns of quadrilateral elements.
On the other hand, in geotechnical examples at http://opensees.berkeley.edu/wiki/index.php/Examples, mass proportional damping is considered.
I am still sceptical whether it is right to exclude mass prop. damping or not and if the Rayleigh damping is appropriate for geotechnical problems since the source of the soil damping unlike the structural damping is radiation and material.

Thank you

I was using constraints Penalty 1.0e18 1.0e18.
I tried with Transformation method and it worked fine. The response in now rational.
Could you please give a guideline when one should use the penalty method and when the transformation method?

Hi

I have a problem with a simple model which is not structural but maybe someone could help me here, because in soil-modelling subforum nobody helps me.

I try to impose displacement time series at all the base nodes of a rectangular 2D soil layer that is simulated with ElasticIsotropic nDMaterial for the quad elements that are used.
All base nodes are fixed in 2 dimensions, as well as the nodes of the right boundary.
The displacement time series is sinusoidal with period 1sec and amplitude 1, is applied in x-direction (=1st DOF), at 0.05 increments and is located in file displ.txt
In order to impose the ground motion Ι use:


set GMdirection 1 ;# ground-motion direction
set DtGround 0.05 ;# time-step for input ground motion
set IDloadTag 2;
set IDgm 0
# numXnode= number of base nodes in x direction
# totalNodes= total number of nodes

timeSeries Path 1 -dt $DtGround -filePath displ.txt
pattern MultipleSupport $IDloadTag {
for {set k 1} {$k <= $numXnode} {incr k 1} {
set IDgm [expr $IDgm +1]
groundMotion $IDgm Plain -disp 1
imposedMotion $k $GMdirection $IDgm
}
}

I use the following to record the absolute displacements in some nodes at the base (middle) and on the surface (middle, left and right end).

recorder Node -file displX.out -time -node [expr int(0.5*$numXnode)] [expr ($totalNodes- int(0.5*$numXnode))] [expr ($totalNodes-$numXnode+1)] [expr $totalNodes] -dof 1 -dT 0.05 disp


The problem is that at the node in middle of the base the recorded displacement is half of the imposed (i.e. the time series is sinusoidal with amplitude 0.5 instead of 1 which is the imposed).
The results in the other 3 nodes are also half of those expected.

I don't think that is a matter of relative-absolute displacement, because the results still don't make sense.
What could be wrong? I'm looking forward for some help.

I've seen in many OpenSees geotechnical examples that mass matrix-proportional coefficient (a) of the Rayleigh damping expression is is taken equal to zero.
I was wondering if (a) is rational to be zero in geotechnical problems.
Typically, the stiffness proportional coeff. represents the loss of energy due to the deformation, what does (a) represents in the damping of soil that should be taken account for?

Indeed. It was a misunderstanding.
Thank you very much.

Hi

It's the first time I'm working on soil-modelling, so I'm encountering a few problems with the analysis.
The problem is about a elastic 2D soil layer 980x50m, meshed by 2x2m quad elements, fixed in both directions in its base and right boundary.
This layer is being excited by an horizontal sinusoidal base shaking sin(2πt) for 10sec.
It goes like this:

# Displacement time history
# The excitation is applied at all the 491 nodes of the base
timeSeries Path 1 -dt 0.05 -filePath displ.txt -factor 1;
pattern Plain 1 1 {
for {set k 1} {$k <= 491} {incr k 1} {
sp $k 1 1
}
}
puts "Load pattern created"
###########################
# create the Analysis
constraints Penalty 1.0e18 1.0e18 ;
test NormDispIncr 1.e-12 25 0
algorithm Newton
numberer RCM
system ProfileSPD
rayleigh 0.0 0.0 0.05 0.
integrator Newmark 0.5 [expr pow(0.5+0.5, 2)/4]
analysis Transient
puts "Analysis object created"


The problem is that the results are not as expected, so I would be grateful if someone could help me with this. I think that the application of the displacement time history (which is in the file displ.txt) might be the problem or something with the analysis parameters (e.g constraints Penalty?)
The displ.txt is:
0.00 0.00
0.05 0.31
0.10 0.59
0.15 0.81
0.20 0.95
0.25 1.00
0.30 0.95
0.35 0.81
0.40 0.59
0.45 0.31
0.50 0.00
etc

Thank you for your reply.
Your explanation is clear that Vb=lamda=time when Σ(Fi)=1. I had misunderstood the concept of all load patterns (even static ones) being associated with a time series.

However, the thing that still puzzles me is why lambda (=time) can be decreased in displacement-controlled pushover analysis, (i.e. after the maxVb). I mean in that case, time has no natural meaning.
Well, it doesn't represent something even for t before maxVb,i.e I think it is completely not related to time (that someone would expect to see in the first column of an .out file e.g. such as the one created by the recorder: recorder Node -file Vb.out -time -nodeRange 1 5 -dof 1 reaction) because it's only a multiplier for the load vector. Could we conclude something more from -time column in that case?

vesna wrote:
> Yes, it is a force.

Dear Vesna,

I wish you a happy new year!
Just to make a note on your reply, I'm wondering why the time (i.e. the 1st) column of the output file represents the base shear ONLY in the case that the sum of the load vector's component at each elevation is equal to 1 (i.e. when the load vector is [0.1 0.3 0.6] => 0.1+0.3+0.6=1)
Why is force related with time and why only in the above case the base shears equals to time?

Thank you

thank you Vesna

I guess you are right.
I was wondering if it is more appropriate to use corotational transformation in pushover analysis instead of PDelta, due to the large displacements.

What do you mean that the P-delta might not work well...?
Apart from the Steel01 and the transformation, is there somenthing else I should check, or leave it as it is?

Thanks Vesna

I used the expession:
[expr ($fu-$fy)/($eu-$ey)/$fy*$ey] where
fy 500000
fu 600000
ey 0.0025
eu 0.025
So, b= 0.022.
I think it's low.Isn't it?

Hi all.

I have a question regarding the pushover curve I'm getting as a result of the static pushover analysis of a 4-strorey plane RC frame.
Although PDelta geometric transformation is used for the column elements, the curve is increasing up (could be approximated as a bilinear) to the target displacement (1m) , without a 'backbone' curve shape (degradation after the maximum base shear), i.e. the max base shear occurs at the target displacement.
Forcebeamcolumn elements with fiber sections (with Concrete01+Steel01) have been used.
The lateral loads are based only on the first mode shape.

I would like to ask if this is normal, if it is due to the materials that have been used, or if there is something else I can't think of.
The pushover curve of an 8-storey frame with the same elements, materials and similar sections, has the 'backbone' shape I described above.
Looking forward for your help.

Evangelos