- This topic has 6 replies, 3 voices, and was last updated 2 months, 2 weeks ago by
.
-
Posts
-
Hello,
I’ve been testing the program, and I’m running into strange behaviour when running a p-delta analysis. I’m trying to replicate AISC’s benchmark Case 1 – Fig. C-C2.2. I tested two scenarios to better understand how the software behaves.
The first, I subdivided the member into four elements. With that one, I can replicate the results from AISC.
The second, I kept a single element, because I know some softwares have the built-in capability within their beam element formulation to capture p-delta effects. In this case, I’m getting significant end moments and the moment at mid-span doesn’t match the expected results. Worst case, I would expect the moment to match the first load case (no axial load applied) as a single element wouldn’t be able to capture the p-delta effects.
I attached the files and screenshots for reference.
Dear Alain,
the beam formulation for built-in solver does not account for automatic mesher or p-delta effects correction for a single beam element.
As a result, for p-delta and second-order effects, you have to consider more than one element for a model. Fortunately, all the model having internal nodes (e.g. structures) does not suffer from this.
Considering the single-beam model, the moment at middle-span for LC1 is the same for all loadcases, but translated of the initial/ending moment (e.g. 34.186 – 7.629 = 26.55). The unrealistic initial and ending moment is due to the fact that the iterative solver cannot compensate the residual in internal nodes.Thank you very much for the detailed explanation. Is there any work around to avoiding this effect? I’m just trying to see if there are potentially any unforeseen impacts with this behaviour with larger structural systems.
It’s not every structure/structural member that will be sensitive to p-delta effects, so I guess how can one rely on the analysis results, or should the default with this software be to always provide four elements whenever a member will have significant axial and bending effects?
The important thing to do is to have a model with at least an internal node, otherwise the solver tries, as said, to compensate the residuals (especially for rotations) on end nodes.
I may not be familiar with the terminology, but when you say an internal node, is that breaking up the member in two, or is there a way to mesh the line, similar to how in SAP2000 you have the line object that you can automatically mesh it without breaking up the member into multiple line elements.
As told, the program does not have internal automatic meshing. Hence, I simply mean breaking the beam in 2 pieces. Please note that this is required only for single beam model, general structures always have internal (not bounded) nodes.
hi, let i share experiences of p-delta analysis in NextFEM: large test i took has shown all results in good agreement with another FE software i.e SAP2000/Etabs, Mastan, STAAD, RISA. Only some minor discrepancy is in shear force distribution curve, OOFEM solver can consider shear deformation (Timoshenko) but OpenSees can not.
- You must be logged in to reply to this topic.