Negative Stiffness Eigenvalues

[alwaeli]

Dear members,

Using ETABS 16.2 version, when setting the stiffness modifiers for my slabs (shell-thin) to 1.0, the run goes smooth without warring and I will get reasonable mode periods. However, when I modify the m11, m22, and m12 stiffness to 0.5 for my slabs and run the model again, I got some negative stiffness eigenvalues in my P-Delta analysis and no modal analysis is made! When I tried 0.9 instead of 0.5, no warnings! Tried 0.8, warnings came back!

By the way, there is no warnings and or errors  when perform "check model". Everything is ok. Attached is my ETABS file for your reference. Please advice if you have faced similar problem?

ETABS Model.rar

[UmarMakhzumi]

Here is the answer from CSI Wiki. I have also highlighted the portion of response that I think is most applicable to your case.

Thanks.

Quote

to start of metadata

 

Why do I get negative stiffness during P-Delta analysis?

Extended Question: When I get negative stiffness(i.e Diagonal<0), and P-Delta analysis does not converge, does this indicate instability and buckling?

Answer: Convergence problems may correlate with P-Delta behavior. Inadequate member capacity and insufficient structural support may lead to instability and buckling. Excessively large stiffness values may also disrupt convergence by causing numerical problems which adversely affect computation.

When a model fails under P-Delta, increase the size and/or quantity of columns and bracing to improve lateral capacity, and thereby avoid buckling. Walls that have been modeled uising membrane type area element (never recommended) or shells with unrealistic thickness and or low stiffness modifiers can buckle and cause convergence issues.  As an alternative, you may also refine the factors and/or loads applied during analysis.

It may be best to troubleshoot the model without P-Delta until you are confident that behavior is appropriate. If problems persist with geometric non-linearity, it may be helpful to run, after the P-Delta load case, a modal analysis which uses Stiffness At The End Of This Stage. Modal analysis may then reveal the source and location of instability.

Buckling can also occur in horizontal floor elements  such as beams and floors, if the floors take in plane loads (i.e have no diaphragm or semi rigid diaphragm) and have insufficient capacity, due to structural properties or small stiffness  modifiers. Using  rigid diaphragms and checking if the analysis converges can help rule out this as the cause of the problem.

Additional suggestions which may resolve negative-stiffness errors include:

• Remove auto line constraints. Select all deck/shell objects in the model, then select Assign > Shell Elements > Auto Line Constraint and uncheck the Apply to Full Structure option. This will remove auto line constraints and possibly resolve instabilities.

 

[alwaeli]

Dear Umar,

Your response is appreciated. I will try the suggestions in your quote and update you with the results.

Meantime, I need to ask you about the best way to mesh floor slabs in multi-story building ETABS model? Will it program default AutoMesh or manual meshing? Keep in mind that on each floor, I have different values of superimposed dead/live loads on different locations in the slab. I really need to know your perspective on that as most of the problems I normally face with the results are caused by the slab meshing.

 

Regards,

Eng. Wael

[alwaeli]

Additional suggestions which may resolve negative-stiffness errors include:

• Remove auto line constraints. Select all deck/shell objects in the model, then select Assign > Shell Elements > Auto Line Constraint and uncheck the Apply to Full Structure option. This will remove auto line constraints and possibly resolve instabilities.

 

Dear Umar,

The above suggestion works! thank you.

However, I would love to understand the engineering behind it. How line constraint is affecting the stability of the floors in P-Delta analysis when the slab stiffness modifier is reduced from 1.0 to 0.5 for instance? Do you have some sort of technical explanation to that?

 

Best regards,

Wael

[UmarMakhzumi]

13 hours ago, alwaeli said:

Dear Umar,

Your response is appreciated. I will try the suggestions in your quote and update you with the results.

Meantime, I need to ask you about the best way to mesh floor slabs in multi-story building ETABS model? Will it program default AutoMesh or manual meshing? Keep in mind that on each floor, I have different values of superimposed dead/live loads on different locations in the slab. I really need to know your perspective on that as most of the problems I normally face with the results are caused by the slab meshing.

 

Regards,

Eng. Wael

Engr. Wael,

The best approach that I have seen is to apply slab loads in ETABS directly on Beams.

For that you will need to manually resolve slabs forces to beams which doesn’t take much time. The upside of this approach is that your model is very clean and without flimsy meshing errors. That is the recommended approach based on my experience. 

You can always later draw slabs in SAFE and design them there. Now regarding why did auto-line constraint fixed your error; I don’t exactly know the answer to that. CSI might be able to answer this question. You can ask them if you have a licensed software and it would be great if you can share the response here as well

Thanks.