Mesh and Convergence Issues in 3D Dynamic Analysis with MIDAS GTS NX – High Strain and Topography Effects

Hi everyone,

I'm currently conducting research comparing the seismic performance of a filtered tailings dam using 1D (DeepSoil), 2D (PLAXIS), and 3D (MIDAS GTS NX) dynamic analyses. While Ive had success with the 1D and 2D models, Ive encountered serious meshing and convergence problems in the 3D model using MIDAS GTS NX, particularly during the dynamic stage.

This is my first time developing a 3D dynamic model in MIDAS, so Id appreciate any guidance or insights from experienced users. Below I summarize my current setup and the issues Ive encountered:

Model Context

• Objective: Compare 1D, 2D, and 3D dynamic responses of a filtered tailings dam.

• Software: MIDAS GTS NX for the 3D model.

• Dynamic Input: Spectrum-matched, filtered, and baseline-corrected seismic record.

• Material Properties: Same as the validated 2D model (HSS model with rayleight damping and calibrated properties).

• Boundary Conditions: Free-field conditions on lateral boundaries and a compliant base.

Main Issues Encountered

1. Excessive Deformations in Localized Zones

   • In specific areas of the model, especially near steep slopes and contact zones, the simulation experiences very high deviatoric strains (>10%) within the first second of shaking, which causes the analysis to stop.

   • These zones likely coincide with abrupt changes in geometry or surface irregularities derived from the imported topography.

2. Mesh Size and Quality

   • We are currently using a mesh nearly 10 times larger than recommended by the Lysmer & Kuhlemeyer (1969) criterion for dynamic wave propagation.

   • I initially avoided refining the mesh to focus on achieving convergence, but now I suspect that mesh resolution may be a major factor contributing to the numerical instability.

   • Additionally, I'm wondering if the issue could stem not from the mesh itself, but from topographic irregularities, and whether these zones should be locally stiffened or smoothed.

3. Mesh Generation in MIDAS GTS NX

   • Are there tools within MIDAS to assess mesh quality or perform localized refinement?

   • Would it be advisable to smooth the topography prior to import, or should I handle it within the software?

   • Is it common practice to apply material stiffening or artificial constraints in problematic zones to ensure convergence?

4. Boundary Conditions and Material Initialization

   • Boundary conditions seem correctly defined, but I'm open to recommendations regarding damping layers, absorbing boundaries, or other stabilization techniques.

   • Initial stress conditions are inherited from the 2D model, but Im unsure whether there is a preferred method in MIDAS for initializing stress states or Ko conditions in complex 3D geometries.

Request for Advice

I would greatly appreciate your insights on the following:

• Best practices for mesh generation and refinement in 3D dynamic models.

• How to handle zones with steep topography or contacts that induce high strains.

• Whether it's valid or common to apply stiffened zones or use local mesh controls to enhance stability.

• How to troubleshoot convergence failures due to topography-mesh interactions.

Any suggestions, references, or shared experiences would be very helpful.

Thank you in advance for your support!
 

Best regards,
Edwin Sanchez

Asked 07/05/2025 20:00, updated: 19/07/2025 16:45

• Edwin Alonso Sanchez Butron
  @ edwin-alonso-sanchez-butron
• Report user
• Send a message

Edwin Alonso Sanchez Butron