Undefined Source of Non-Convergence OpenSees

[ZouatineMohamed]

Dear All, I wish you a happy new year.

I have a nonlinear simple model in OpenSees , where the nonlinearity is model by zerolength element at the 4 support point of the model. ( I attached a picture to visualize the model) .

To ensure that the model is running correctly I did run Gravity – Modal -Pushover analysis , and all these three different cases are running correctly . So I want to increase the intensity of the GM , I started with very low intensity 0.2g up to 0.689 g is running fine ( note that 0.689 Is the start of the nonlinearity in the model) . start from here I start getting convergence issues.

So for the solution, I've tried various solutions like changing convergence tests, algorithms, and material parameters, but the issue persists.
I've attached images showing zerolength element deformations in cases with and without convergence problems.

Is this problem with OpenSees itself or do I have some issue in the modelling?

Your help is much appreciated !







The Error :
NormDispIncr KrylovNewton 0
WARNING: CTestNormUnbalance::test() - failed to converge
after: 25 iterations current Norm: 909944 (max: 1e-06, Norm deltaX: 22.8693)
NewtonRaphson::solveCurrentStep() -the ConvergenceTest object failed in test()
DirectIntegrationAnalysis::analyze() - the Algorithm failed at time 15.56
OpenSees > analyze failed, returned: -3 error flag

[FreakoValiko]

Dear ZouatineMohamed,

Thank you for providing the details of your OpenSees model and the convergence issue you are facing. The error message you shared indicates a convergence problem during the dynamic analysis. Below are some suggestions to help you address the non-convergence issue:

Increase Convergence Tolerance:
Try increasing the convergence tolerance (Tol) in your analysis. Sometimes, a more relaxed tolerance can help the analysis converge, especially in cases with non-linear behavior.

Code: Select all

Tol = 1e-9  # or a higher value

Check Time Integration Parameters:
Adjust the Newmark time integration parameters. You can experiment with different values for the gamma and beta parameters.

Code: Select all

ops.integrator('Newmark', 0.6, 0.3)  # Adjust gamma and beta values

Check Damping Parameters:
Ensure that the damping parameters are appropriate for your model. You may experiment with different damping ratios to see their effect on convergence.

Code: Select all

xDamp = 0.03  # Adjust damping ratio

Check Initial Stiffness Proportional Damping:
If your model experiences abrupt changes, you might consider including initial stiffness proportional damping.

Code: Select all

betaKinit = 0.02  # or another suitable value

Check Load Pattern:
Verify the load pattern and the applied acceleration time series. Ensure that they are consistent with your expectations.

Check Eigenvalue Analysis:
Perform an eigenvalue analysis separately to ensure the stability of the structure and the correctness of the applied damping.

Review Material Parameters:
Double-check the material properties and their behavior under dynamic loading. Ensure that the material models are suitable for the range of deformations encountered.

Use Smaller Time Steps:
Try reducing the time step size (DtAnalysis). A smaller time step can sometimes help in capturing the dynamic behavior more accurately.

Code: Select all

DtAnalysis = 0.01  # or a smaller value

Check for Instabilities:
Investigate if there are any instabilities in your model that might be causing the convergence issues. This includes checking for negative or zero stiffness, element distortion, or other numerical instability sources.

After making adjustments, gradually increase the intensity of the ground motion and observe if the convergence issues persist. If the problem still exists, you may need to perform a more detailed analysis of your model's dynamic behavior and review the applied loads and boundary conditions.

I hope these suggestions help in resolving the convergence issues in your OpenSees model. If you have further questions or encounter any difficulties, feel free to provide additional details for more targeted assistance.

Best regards,

[ZouatineMohamed]

Dear FreakoValiko,

Thank you for your reply and your recommendations , unfortunatly i have been through all your recommendation but the problem remains. i am not able to figure out where the issue could be since all the necessary cases to run the model are working Grav-Modal-Pushover and even for THA analysis for linear analysis it is running as well as up to a certain level of nonlinearity .