Recommended Modeling Procedure
Solving geotechnical problems using the finite element method is a relatively complex task. But yet, most users attempt to analyze the entire complex structure right from the beginning - to find the cause of possible loss of convergence may then become rather difficult. We therefore recommend the following approach:
1) Define the whole topology of the structure
2) Assume elastic response of soils and contact elements (use)
3) Generate coarse mesh
4) Define all
5) Performof all calculation stages (it is sufficient to launch the analysis of the last stage of construction - analyses of all previous stages are carried out automatically).
6) Asses the course of analysis
If the analysis fails, the computational model is not correctly defined - e.g. beams have too many internal hinges resulting into a kinematically undetermined structure, props are not properly hooked to the structure, etc. The program contains a number of built-in checking procedures to warn the user for possible drawbacks in the model definition. Some of the errors, however, cannot be disclosed prior to running the program.
If all stages were successfully analyzed, we recommend the user to check the resulting displacements and this way also the objectivity of the used soil parameters and structure stiffness. Note that using nonlinear models always results into larger displacements in comparison to the pure elastic response - should the elastic displacements be already excessively large, we must first adjust the computational model before adopting any of the available plasticity models.
If the analysis succeeded and the displacements are reasonable, we may proceed as follows:
7) Replace linear models with suitable(Mohr-Coulomb, Drucker-Prager)
8) Perform analysis and evaluate the results according to step 6
9) Add nonlinear contact elements
10) Perform analysis and evaluate the results according to step 6
11) Refine and adjust the finite element mesh and perform the final analysis.
Although this approach may seem rather cumbersome and complicated, it may save a considerable amount of time when searching for the cause of failure (loss of convergence) of the analysis of complex problems.