InitStrain Material

[gerardoreilly3]

Hi,

I am trying to put a tension force into an element and then load into tension. I am doing this using a corotTrussSection element with a material InitStrainMaterial that applies an initial strain to a previously defined material, which for this case I am just using a Linear Elastic Material. The model is shown below, and for no initial strain (PT equal to zero) it works fine, with the Load Displacement results shown exactly what would be expected. However, when any form of force ( PT > 0) is applied, the results go crazy, with forces and displacements in completely unrealistic values.

My question is am i going about applying a post-tensioning force in the right way here? or is this an analysis problem? Your help would be greatly appreciated.

Many thanks for your time,
Gerard


wipe;
model BasicBuilder -ndm 3 -ndf 6

# Define Variables
set d 0.010
set L 1.0
set A [expr $d*$d]
set PT 0.0
set E 200e6

# Define Nodes
node 1 0.0 0.0 0.0
node 2 $L 0.0 0.0

# Define Boundary Conditions
fix 1 1 1 1 1 1 1
fix 2 0 1 1 1 1 1

# Define Materials
uniaxialMaterial Elastic 1 $E
uniaxialMaterial InitStrainMaterial 2 1 [expr $PT/($A*$E)]

# Define Sections
section Fiber 1 {
fiber 0 0 $A 2
}

# Define Element
element corotTrussSection 1 1 2 1

# Define Load
pattern Plain 1 Linear -factor 1 {
load 2 1.0 0.0 0.0 0 0 0
}

# Define Recorder
recorder Node -file LoadDispPT.txt -time -node 2 -dof 1 disp

# Define Analysis
constraints Plain
numberer RCM
system BandGeneral
test EnergyIncr 1.0e-30 100 0
algorithm Newton
analysis Static

integrator DisplacementControl 2 1 +0.00001
analyze 500

wipe;

[vesna]

Your code looks fine to me but I can not run it as I do not have LoadDispPT.txt. Since your element is elastic it is very easy to manually calculate the response of the truss element and to compare it with OpenSees results. I would do this for both cases: PT=0, and PT>0.

[gerardoreilly3]

Thanks for your reply. The file LoadDispPT.txt is actually the recorder file that the results are outputted to, so this can be obtained by just copying and pasting the code I've copied in above, and running in OpenSees.

When I use a PT = 0, the response is exactly as expected - a linear force-displacement plot with slope E up to the target displacement. The actual problem is when any form of initial strain is inserted into the element (PT > 0).

The plot goes to about 7.0e+4 m displacement, with a force of -2.5e+25 and is anything but linear. I was expecting a plot similar to PT=0 but with the force starting at the initial value of PT at zero displacement, but these actual results are completely bizarre. If you try running this code, you will see what i mean.

I also tried the InitStress material and using this does not effect the results at all.

Thanks for your help.
Gerard.

[vesna]

The problems you were seeing are because of your analysis objects. I simplified your code a little bit but it models the same thing. Please look at the code below, it should solve your problem:

wipe;
model BasicBuilder -ndm 2 -ndf 3

# Define Variables
set d 0.010
set L 1.0
set A [expr $d*$d]
set PT 10.
set E 200.0e6
puts "initStrain [expr $PT/($A*$E)]"

# Define Nodes
node 1 0.0 0.0
node 2 $L 0.0

# Define Boundary Conditions
fix 1 1 1 1
fix 2 0 1 1

# Define Materials
uniaxialMaterial Elastic 1 $E
uniaxialMaterial InitStrainMaterial 2 1 [expr $PT/($A*$E)]

# Define Element
element corotTruss 1 1 2 $A 2

# Define Recorder
recorder Node -file LoadDispPT.txt -time -node 2 -dof 1 disp

# Define Load
pattern Plain 1 Linear {
load 2 1.0 0.0 0.0
}

# Define Analysis
constraints Plain
integrator LoadControl 20.0
numberer Plain
system BandGeneral
test NormDispIncr 1.0e-7 100
algorithm Newton
analysis Static

analyze 500

wipe;

[gerardoreilly3]

Thanks for that.

Does this mean it is not possible to use a displacement control integrator when using a initstrain element?

[vesna]

You can also use displacement control but it is trickier to make it work properly.

[gerardoreilly3]

What are the parts of the analysis that need to be changed to make it work properly?

This forms one element of a system i will be testing using a displacement history so i am trying to get the single element with an initial strain before i build it into the final model.

Thanks for your help.

[mnazari]

Hello,

I am modelling an unbonded post-tensioned tendon in a concrete wall with a truss element using Steel02 material, with an initial stress of 69.2 ksi:
set initStress 69.2
"uniaxialMaterial Steel02 $PT 245.0 [expr (245.0/0.0086)] [expr (((270.0-245.0)/(0.04-0.0086))/(245.0/0.0086))] 18 0.925 0.15 0 1 0 1 $initStress"

During base excitation tendons yielded (stress>245 ksi) and initial stress (69.2 ksi) dropped to almost 15 ksi. After yielding of tendons and for the rest of the motion, stress in tendons change from this 15 ksi! I am not sure if this is the best way to model PT tendons while yielding takes place and if so, how can I control the PT loss using Steel02 material.

Thanks a lot!

[internationalyoga]

Hi folks i read your post that is very informative.

[gor]

Have you tried defining a new initStrain material from the initial Steel02 material?

Otherwise, if you have a problem with an initial loss of prestress before you start any analysis (you apply 100kN but the recorders say that the initial load is 95kN) then its because the elastic deformation of the surrounding concrete wall in your case is allowing a strain relaxation and in turn a reduction in initial stress, which means the initial load is reduced. To get around this, you will need to work out the relative stiffnesses of the components such that the initial force you apply will be slightly higher (eg 110kN) but taking the relaxation into account, results in an initial force of 100kN in your recorders. If i didnt answer your question sorry, but it wasnt very clear.