Bug Report

[fran]

If you push 350cm length cantilever beam IPE500 (h=50cm b=20cm tf=1.6cm tweb=1.02 cm without neck curvatures) to 40cm top displacement, the moment value of the base section does not increase any more and you obtain the Mp value of the section. OPENSEES gives 77871 kNcm with fiber sections, displacement based elements and with fy=36kN/cm^2 and zero strain hardening. However if you apply a top moment 60000kNcm as a dead load before the pushover analysis, the moment value reaches to 79460 kNcm at the end of the push over analysis. The section plastic moment changes with the applied dead load ! It sould have not been affected by dead load. DRAIN2DX is not affected for example.

(Meanwhile the true value of the sectiom plastic moment is 75863 kNcm without neck curvatures)

[vesna]

Would you post your file so I can check it?

[fran]

Thank you very much vesna.
The dead load moment 60000kNcm may be high but it is affected for lower values also.
My code is the following:

wipe;
model BasicBuilder -ndm 2 -ndf 3;
source Wsection.tcl;

# node coordinates
node 14 0.00000 350.00000
node 13 0.00000 325.00000
node 12 0.00000 297.91667
node 11 0.00000 270.83333
node 10 0.00000 243.75000
node 9 0.00000 216.66667
node 8 0.00000 189.58333
node 7 0.00000 162.50000
node 6 0.00000 135.41667
node 5 0.00000 108.33333
node 4 0.00000 81.25000
node 3 0.00000 54.16667
node 2 0.00000 27.08333
node 1 0.00000 0.00000


# node restraints
fix 1 1 1 1

# materials
set fy 36.00000;
set Es 21000.00000;
uniaxialMaterial Steel01 1 $fy $Es 0.0

# section properties
Wsection 1 1 50.0 20.0 1.6 1.02 16 2 16 4; # IPE500
set IPE500 1;

# transformation
geomTransf Linear 1;

# Columns (from up to bottom)
element dispBeamColumn 13 14 13 4 $IPE500 1
element dispBeamColumn 12 13 12 4 $IPE500 1
element dispBeamColumn 11 12 11 4 $IPE500 1
element dispBeamColumn 10 11 10 4 $IPE500 1
element dispBeamColumn 9 10 9 4 $IPE500 1
element dispBeamColumn 8 9 8 4 $IPE500 1
element dispBeamColumn 7 8 7 4 $IPE500 1
element dispBeamColumn 6 7 6 4 $IPE500 1
element dispBeamColumn 5 6 5 4 $IPE500 1
element dispBeamColumn 4 5 4 4 $IPE500 1
element dispBeamColumn 3 4 3 4 $IPE500 1
element dispBeamColumn 2 3 2 4 $IPE500 1
element dispBeamColumn 1 2 1 4 $IPE500 1

# Dead load
pattern Plain 1 Linear {
load 14 0.0 0.0 0.0
#load 14 0.0 0.0 60000
}

# Dead load analysis
constraints Transformation ;
numberer RCM;
system BandGeneral ;
test NormDispIncr 1.0e-8 6 ;
algorithm Newton;
integrator LoadControl 0.1;
analysis Static;
analyze 10;
loadConst -time 0.0;

puts "Dead load analysis complete"


# pushover recorders
recorder Node -file topnodedisp.out -time -node 14 -dof 1 2 3 disp;
recorder Element -file basecolglobalForce.out -time -ele 1 globalForce;

# Pushover load pattern
pattern Plain 200 Linear {
load 14 1.0 0.0 0.0;
}

# pushover displacement parameters
set IDctrlNode 14;
set IDctrlDOF 1;
set Dmax 40; # maximum displacement of pushover: 11% roof drift
set Dincr 0.1; # displacement increment

# pushover analysis commands
constraints Plain;
numberer RCM;
system BandGeneral;
test NormUnbalance 1.0e-6 400;
algorithm Newton;
integrator DisplacementControl $IDctrlNode $IDctrlDOF $Dincr;
analysis Static;
set Nsteps [expr int($Dmax/$Dincr)];
set ok [analyze $Nsteps];
puts "Pushover complete";
wipe all;

[vesna]

There are two things that you can do:
1) change solver to BandSPD
or
2) change post-yielding stiffness to 1.0e-15

Look at the forces of the section that corresponds to fixed end of element 1 to compare them with the Mp of the section. Element end force is going to be different then Mp due to Gauss integration along the element length.

[fran]

Using BandSPD solver and 1.0e-15 strain hardening did not change the result, 77871. Using 2, 4, 8 and 10 Gaus points for integration resulted 77871, 77871, 77869 and 77869 plastic moments, respectively which has a little effect on the plastic moment value.

However, more importantly, above values change with applied dead load before the pushover analysis. For example think about two case: (1) firstly dead load is aplied and then pushover is performed. (2) no dead load applied and pushover is performed. The plastic moment should have not been affected by the dead load. This is because the section deos not know how the moment on it is produced, by a dead load or from pushover or from wind forces or from fluid pressure, etc.

[vesna]

In my last message I asked you to: "Look at the forces of the section that correspond to fixed end of element 1 to compare them with the Mp of the section. Element end force is going to be different then Mp due to Gauss integration along the element length". What are the forces that you are looking at?

[fran]

You are right. The section force gives the exact value of 75863 and is not affected by dead loads. Thank you very much. I thought that one of the Gauss point is at the element end and the recorder gives its moment as the element end moment.

By the way, where can we use element end forces?