I have a nonlinear calculation of a complex model with beams.
I have calculated both linear and nonlinear geometrical (TH3) and mechanical (NSTR).
My problem is with reactions, when i perform a mechanical non linear calculation (NSTR) with a custom calculatoric stress strain curve, total model reaction is reduced compared with linear and TH3 calculations.
It seams that NSTR is reducing somehow the loads acting on the structure. May be the material density is affected?
When the sum of reactions and loadings is shown in the output you can see the difference in total z force between sum of reactions and sum of loadings.
I have checked this happens also in the example file plc_on_csm_cracked_pier.dat
In this example both load cases 105 and 106 have same loads but different reactions. LC 105 is calculated with th3 while 106 is calculated th3+nstr.
Total reaction difference in this case is about 2% but in my case it reaches up to 6% depending of load case.
Is there any problem inside sofistik with NSTR?
Today we have made some findings. Difference in reactions sum only happens when factd=1.35 is applied to the selfweight in a NSTR calculation.
In addition the difference of reacions sum is exactly the 35% of the weight of the groups where the NSTR is applied.
In the example file plc_on_csm_cracked_pier.dat NSTR calculation is applied to the group 9 (column). The self weight of this column is 1553kN and 0.35*1553=543kN. This figure matches exactly the difference between loadings sum and reactions sum: 15304-14761=543kN. We found the same in our model.
Therefore it seams there is a bug in NSTR calculations where the factd factor is not working properly.
Hello,
we can reproduce this problem and will try to solve it.
In the meantime you can use the workaround with:
search for “pier with gam 0 Bellmann 23.02.2024”
in the attached file where we used GAM 0 for section 9 and applied the dead load of this pier as SOFILOAD LC 9.
Then the equilibrium is correct.
Hi Pablo,
this data set is an official example from SOFiSTiK with a hint for a workaround. And in my opinion it’s not a problem to share it in this forum.