Acting bending moment greater than ultimate (plastic) bending moment

[salvatorebenfratello]

I am new to SeismoStruct and I am solving a simple steel frame (two bays with different span, two floors, elastic pefectly plastic model with 235 MPa strength)
with one beam loaded by a distributed permanent element load and an acceleration time history applied to all the supports (all fixed).
My interest is to verify the plastic behaviour under different loading conditions and acceleration time histories.
To this aim, since this is only a test, I have amplified the time history by 25 to plasticize one beam.
However, when I plot R2(B)/M2(B) graph the maximum bending moment is equal to 305 kNm
and it exceeds the value of ultimate (plastic) bending moment (the section is an HEA280 with a corresponding plastic moment of 261 kNm).
If I plot the stress-strain graph in the middle upper point of the section B the curve is correct (it shows an elastic-perfectly plastic behaviour, as expected, with the imposed limit values).
Someone can help me to understand and/or to correct the probable error?

[huffte]

Just to be certain, it seems the ratio of moments is about 1.16. This is probably close to the ratio of Z/S. Before we begin investigating other possibilities, could you verify that you hand calculated moment of 260 is indeed Mp and not My? After you verify that, other possibilities can be explored.

[salvatorebenfratello]

Thanks for your reply.
The HEA280 section shows a resistance modulus Wel equal to 1013 cm^3 and a plastic one equal to 1112 cm^3. Taking into account a S235 steel (elastic-perfectly plastic behaviour) it follows that the elastic limit bending moment is equal to 238 kNm and ultimate (plastic) one is equal to 261 kNm.
The beam under examination is 6 metr long, possesses HEA280 section and it is loaded by a permanent downwards element load equal to 41.50 kN/m.
The frame is loaded by a simulated accelerogram at its supports.
I amplify the accelerogram by 25, run the software but it does not complete the analysis since it stops prematurely.
In the postprocessor section I plot the bending moment at the extreme section of the beam obtaining values of 256 kNm (A section) and 304 kNm (B section) being the second greater than the ultimate bending moment of the section.
If I plot the stress/strain curve at top and bottom points of both extreme sections (A and B) of the beam I obtain a corrrect curve (the stress does not exceed the maximum allowable one).
Why the bending moment exceeds the ultimate (plastic one) taking into account an elastic-perfectly plastic material behaviour?
Thanks in advance for your support

[huffte]

I am running SeismoStruct 2021, Release 2, Build 20. When I create a Material Class and specify the built-in S235, the software uses a default yield stress of 270.25 MPa, not 235 MPa. Presumably this is the mean expected yield stress as opposed to the lower bound yield stress. I can certainly change the value used to the lower bound if I wish. I wonder if this is what is happening with your model?

[salvatorebenfratello]

I am running SeismoStruct v2023 Release-1 Build-120
In the "Edit Material Properties" I have the following parameters:
Modulus of elasticity (kPa) 2,1000E+008
Yield strength (kPa) 235000,00
Strain hardening parameter (-) 0,00
Fracture/buckling strain (-) 0,20
Specific Weigth (kN/m3) 76,70

The stress/strain graph sketched in the tab reports the above parameters

[huffte]

Which element type are you using and are you subdividing the beam and column elements? I am going to try and reproduce the issue.

[salvatorebenfratello]

Thanks.
I adopt only one element type: infrmFB for all the beams and pillars.
If it is possible I can share the file to simplify the check.

[seismosoft]

Salvatorebenfratello hi,
We carried out a check with a cantilever with the infrmFB element class and the results come out reasonable, i.e. fy=fu=247-248kN. We also checked the dimensions of the HEA280 section and they are correct.
Note the following:
1) Our nonlinear implementation of the HEA280 section is through the adaption of the sits section in our library. It does not contain the curves in the connection of the web with the flanges, but this should not give an error of more than 1%
2) You cannot carry out correct section analysis with axial loads variable and different than zero
3) The dynamic response and the hysteretic rules of the selected material model further complicate things
4) In your checks try to use a large number of monitoring points, i.e. more than 250-300
5) SeismoStruct employs the mean material strengths. Do not compare the characteristic strengths of the section with the mean values.

Seismosoft Support