Problem in modeling CBF bracing assembly

[AbRiahi]

Hi everyone
I have a problem with modeling CBF bracing assembly. I know how to model a single strut or bracing members in order to capture its buckling behavior, i.e. dividing it into at least two segments and using beam column elements and assigning an appropriate imperfection value ..., and the results are acceptable. but when I use some of such elements in a structure or a subassemby the results are not good. I mean that it seems that all braces even those with tensile loads show buckling force-deformation response. I run some simple files, like the following one, in opensees and I noticed that even the tensile brace has buckling force-deformation .
Does anybody know how to tackle this problem?

## Units kips-inches
wipe;
model BasicBuilder -ndm 2 -ndf 3
file mkdir data;
set L_subassembly 80.0; # Assembly Height
set H_subassembly 52.75; # Assembly Length
set x_brace [expr 0.5*$L_subassembly];
set y_brace $H_subassembly;
set L_brace [expr pow(($x_brace*$x_brace+$y_brace*$y_brace),0.5)];
set tanx [expr $y_brace/$x_brace];
set cosx [expr 1.0/pow((1.0+$tanx*$tanx),0.5)];
set sinx [expr pow((1.0-$cosx*$cosx),0.5)];
puts "tanx=$tanx"
puts "cosx=$cosx"
puts "sinx=$sinx"
puts "L_brace=$L_brace"
set N_Ele 2; # No. of sub-elements in a strut, an even No.
set Delta [expr 0.001*$L_brace]; # Imperfection value
set numIntgrPts 5;

# Materials
set Fy 59.8; # Yield strength
set E 25000.0; # Modulus of elasticity
set b 0.005; # strain hardening
uniaxialMaterial Steel02 1 $Fy $E $b 20 0.925 0.15

### Pipe Section (4.5” diameter, 0.337” wall thickness)
# Pipe 4x0.357
set D 4.5; # Pipe Diameter
set t 0.337; # Pipe Thickness
set startAng 0.0;
set endAng 360.0;
set numSubdivCirc 60;
set numSubdivRad 10;
set extRad [expr $D/2];
set intRad [expr $D/2-$t];
set yCenter 0.0;
set zCenter 0.0;
section fiberSec 1 {
patch circ 1 $numSubdivCirc $numSubdivRad $yCenter $zCenter $intRad $extRad $startAng $endAng
}

### Transformation
geomTransf Corotational 1
### Define Model
# Nodes
node 1 0.0 0.0 ;
node 3 [expr $x_brace/2.0-$Delta*$sinx] [expr $y_brace/2.0+$Delta*$cosx] ;
node 5 $x_brace $y_brace;
node 6 $x_brace $y_brace;
node 7 [expr 3*$x_brace/2.0+$Delta*$sinx] [expr $y_brace/2.0+$Delta*$cosx] ;
node 8 $L_subassembly 0.0;

## Defin elements
element nonlinearBeamColumn 1 1 3 $numIntgrPts 1 1 ;
element nonlinearBeamColumn 2 3 5 $numIntgrPts 1 1 ;
element nonlinearBeamColumn 3 6 7 $numIntgrPts 1 1 ;
element nonlinearBeamColumn 4 7 8 $numIntgrPts 1 1 ;

#Boundary Conditions
fix 1 1 1 0
fix 8 1 1 0
equalDOF 5 6 1 2

## Apply the nodal Load pattern
pattern Plain 1 Linear {load 5 1.0 0.0 0.0}
### Recorder
recorder Node -file Data/DFree.out -time -node 5 -dof 1 disp;
recorder Node -file Data/RBase_Left.out -time -node 1 -dof 1 reaction;
recorder Node -file Data/RBase_Right.out -time -node 8 -dof 1 reaction;
recorder Element -file Data/ele1local.out -time -ele 1 localForce
recorder Element -file Data/ele2local.out -time -ele 2 localForce
recorder Element -file Data/ele3local.out -time -ele 3 localForce
recorder Element -file Data/ele4local.out -time -ele 4 localForce
# Monotonic for testing
test EnergyIncr 1.0e-8 300 0
algorithm KrylovNewton
system UmfPack
numberer RCM
constraints Plain
set dU [expr 0.028*$H_subassembly/50.0]
integrator DisplacementControl 5 1 $dU 10 $dU $dU
analysis Static
analyze 50

[vesna]

Are your braces fixed or pinned at their ends? I prefer to have them fixed. I also have rigid elements at their ends to simulate gusset plates. I do not define any geometric imperfection as small numerical imperfection will produce the same effect.

[AbRiahi]

Dear vesna
Thanks a lot for your reply
I should mention that I agree that it is better to define fix boundary condition and rigid beams for gusset plate. About your comment on modeling the imperfection I am not sure to get exactly what you mean. Do you mean there is any other way rather than what I used in this code (changing the coordinates of the middle nodes of braces to have an initial camber based on Uriz et al. proposition)? How? By the way, my problem was about why in a assembly or structure tensile braces showed buckling response? Can you please offer any proposition or modification to tackle this problem in the aformentioned code or send me an example that solve this problem in the case of a subassembly?

[vesna]

Currently, I do not have small model to post as an example. I will add it to my to do list and hopefully post it soon.

If the brace ends are fixed you do not need to define geometric imperfections.

[AbRiahi]

Dear vesna
Thanks a lot for your reply. I will wait for you to send that model, but I want to know did you notice any problem in my code resulting to such suspect results?

[vesna]

Here is the simple example I created for you:
http://opensees.berkeley.edu/wiki/index.php/SCBF_Model

Try to run it and see if it is going to work for you.

[AbRiahi]

Dear vesna
Thank you very much for your kindness and reply
I am going to use it

[sirvan]

Dear vesna
i m going to model pinned braces.but when the model is run,there is a difference between period of structure which is modeled by ETABS and OpenSEES.
Please guide me.

wipe
model basic -ndm 2 -ndf 3

node 01 0 0
node 02 6 0
node 11 0 4.5
node 12 6 4.5
node 9110 0. 0.
node 91122 6. 4.5

#node 9110 0. 0.
#node 91122 6. 4.5

#### mass
mass 11 1200 1e-9 1e-9
mass 12 1200 1e-9 1e-9
### Beam
geomTransf Linear 1
element elasticBeamColumn 3 11 12 4.187e-3 2.1e10 4.912e-5 1
### Column
geomTransf PDelta 2
element elasticBeamColumn 1 01 11 0.024 2.1e10 3.976e-4 2
element elasticBeamColumn 2 02 12 0.024 2.1e10 3.976e-4 2
### Bracing
uniaxialMaterial Elastic 1 2.1e10
#element corotTruss 4 9110 91122 0.01 1
## nBr10x10 FiberSec (HSS-Section) ###

section Fiber 1 {
patch quad 1 15 15 -0.05 -0.05 0.05 -0.05 0.05 0.05 -0.05 0.05

}
element nonlinearBeamColumn 4 9110 91122 7 1 2

### Regidity
fix 01 1 1 1
fix 02 1 1 1
fix 9110 1 1 0

equalDOF 91122 12 1 2
equalDOF 01 9110 1 2
equalDOF 11 12 1

puts "2"
set Pi [expr 4*atan(1)]
puts "
PROGRAM CALCULATED PARAMETERS:---------
w^2=[eigen 1] (rad/sec)^2

[vesna]

For the first iteration model all of your elements as elastic and assign the same mass to your nodes in both OpenSees and ETABS. You should get the same period.