Nonlinear analysis. Incorrect results for intermediate load steps.

[Andrey_Golenkin]

Hi Guys,

When dealing with nonlinear analysis of a model which contains unidiractional releases, I found out that results for intermediate load steps (if you swithed option to save them in analysis parameters) look very strange, actually the look like there is no unidirectional relesase at all. Anyway results for the last step look correct. What could be the reason for this problem?

Test model attached.

[Pawel.Pulak]

In case of nonlinear properties, which can result in complete deactivation of some stiffness (like unidirectional releases, uplift supports,  tension-only or compression-only members) the stiffness coresponding to them is reduced gradually during each increment of load. It is completely reduced to zero during the last load increment (when applying 100% of load).

That is why you see non-zero forces or moments in unidirectional releases when you check the load increment steps different than the last one.

---------------------------------------------
If this post answers your question please click the "Accept as Solution" button. It will help everyone to find answer more quickly!

Regards,

[Andrey_Golenkin]

Hi Pawel,

Thanks for the explanation. But still I wonder whether it is a feature of Robot solver, or a common practice for such types of nonlinear model objects? I calculated the same model in several other programs and their intermediate step results were correct. Sometimes it required to track changes in structure behaviour and having these results would be preferrable.

[Pawel.Pulak · Accepted solution]

Hi Andrey,

this approach is the feature of implementation in Robot.

In the past we have used the approach of complete deactivation of such elements but it was observed that for more complex structures and load conditions it could result in the oscillations of iterations and problems to obtain the convergence of the process.

That it why it was changed to gradual deactivation during load increments.

This approach does not give the possibility to directly track changes in structure behaviour but it gives bigger chance of convergent solution - that is why it was selected.

If you really need to track changes in structure behaviour for specific load case or combination the workaround for such structure is to do not use "Save results after each increment"  but instead of it to define additional combinations with factors 0.2; 0.4; 0.6; 0.8 etc. and to analyse and compare results for them.

Shortcoming of such approach is the impossibility to use Results>Advanced>Diagrams to graphically see inside Robot the changes when incrementing the load.

---------------------------------------------
If this post answers your question please click the "Accept as Solution" button. It will help everyone to find answer more quickly!

Regards,