Hey
I am trying to make a windcross, with bars there are defined as trussbar - tension bar. The problem is that with some combinations Robot is making the bar a compression bar. It is not all the bars I have defined, which there are problems with.
Hello,
model would be welcome to check. Can you attach it after compressing to ZIP (additionally it may be saved without results to reduce the size).
Best regards,
Model received outside of forum. Main conclusions of troubleshooting are listed below to give some hints for others.
Observations during analysis of the model:
many instabilities of type 1 and 2 reported during analysis and then no convergence for most of load cases.
Type 1 and 2 instabilities cannot be ignored (type 3 sometimes cannot be avoided - see http://docs.autodesk.com/RSA/2012/ENU/filesROBOT/GUID-52C292E1-5C73-4FB1-B4E0-E5E8107777D-615.htm ) and model should be corrected to eliminate them.
No convergence of non-linear analysis means in typical situation not applying all load and not completing analysis with success. In case of tension-only members it means for instance no complete switch off of these members - their compression stiffness is gradually reduced to zero while incrementing the load up to 100% (Process parameter = 1) and when analysis is not compleded because of convergence problems some compression can be observed in tension-only members.
In this specific case:
1/ Type 1 and 2 instabilities resulted from too many bar releases close to pinned supports. Model corrected deleting some unnecesary releases and replacing some pinned supports by fixed supports (in case when all adjoining bars with pinned releases)
2/ After correcting the model to avoid instabilities there were still some convergence problems.
That is why the method of non-linear iterations for all load cases was changed from default BFGS to more powerful (but slower) full Newton-Raphson method (stiffness matrix update after each iteration). It is shown in the attached screen capture.
After above modifications of the model the analysis was convergent for all load cases and results were correct.
If this post answer your question please click "Accept as Solution". It will help everyone to find answer more quickly!
Best regards,