Solving Local Stress Peaks with Ansys Mechanical Submodeling
Learn how to use Ansys Mechanical submodeling to resolve local stress concentrations without the computational cost of a globally refined mesh.
31 May 2026, 04:01 UTC

The Mesh Density Dilemma
In structural analysis, you often face a conflict: you need a coarse mesh for the bulk of a large assembly to keep solve times reasonable, but you need an incredibly fine mesh around a small fillet or hole to capture peak stresses. If you refine the mesh globally to satisfy a single high-stress region, the element count explodes, leading to excessive memory usage and long solve times.
The solution is submodeling. This is a global-to-local workflow where you solve a coarse "global" model first, then extract the displacements from a specific region of interest to drive a highly refined "submodel." Instead of remeshing the whole part, you only pay the computational cost for the area that actually needs the detail.
The Submodeling Workflow
Submodeling relies on the principle that if the displacements on the boundary of a local region are known, the internal stresses of that region can be calculated independently of the rest of the structure.
- Global Analysis: Create a simplified model with a mesh coarse enough to capture overall stiffness but not necessarily local stress gradients. Solve this model as usual.
- Defining the Submodel: Create a separate geometry (or a sliced version of the original) that encompasses the region of interest. This region must be large enough that the "cut boundaries" (the edges where you sliced the geometry) are located in areas of low stress gradients.
- Displacement Mapping: In Ansys Mechanical, you define the submodel and link it to the global solution. The software maps the nodal displacements from the coarse global mesh onto the refined nodes of the submodel's cut boundaries.
- Local Solve: The submodel is solved using these displacements as boundary conditions, allowing you to use a dense mesh in the critical area.
Example: Stress Analysis of a Bracket Fillet
Consider a heavy-duty steel mounting bracket with a small 2mm fillet where the vertical flange meets the base. To capture the stress concentration at that fillet, a mesh size of 0.5mm might be required. Applying that 0.5mm mesh to a 500mm bracket would create millions of unnecessary elements.
The Setup
| Parameter | Global Model | Submodel |
|---|---|---|
| Mesh Size | 10mm - 20mm | 0.2mm - 0.5mm (at fillet) |
| Geometry | Full Bracket | Small cube surrounding the fillet |
| Boundary Conditions | Fixed supports and remote loads | Cut boundary displacements from Global |
To implement this in Ansys Mechanical, you would typically:
- Solve the Global Model.
- Insert a Submodeling system in the Workbench project schematic and link the Global Solution to the Submodel Setup.
- In the Mechanical window for the submodel, select the Cut Boundary faces and set the Submodeling option to
Yes. - Assign the corresponding Global Model as the source for the displacement data.
Critical Trade-offs and Limitations
Submodeling is a powerful tool, but it is not a "magic button." Its accuracy depends entirely on the quality of the global model's displacements.
Boundary Selection: If your cut boundary is too close to the stress concentration, the coarse global mesh will provide inaccurate displacements, and the submodel will inherit those errors. Always ensure the cut boundary is in a region where the stress is relatively uniform.
Linearity Assumptions: While Ansys supports various analysis types, mapping displacements from a linear global model to a nonlinear submodel (or vice versa) requires caution. Ensure the physics—such as material nonlinearity or large deflections—are consistently handled or that the mapping is appropriate for the intended result.
Verifying the Result
To ensure the submodel is providing a valid result, check the displacement continuity at the cut boundary. If the displacements on the submodel boundary show erratic jumps or discontinuities compared to the global model, your global mesh may be too coarse to provide a stable input. A practical check is to slightly expand the submodel region; if the peak stress at the fillet changes significantly, your cut boundary was likely too close to the area of interest.
0 replies
A thoughtful contribution can make all the difference. Be the first to share one.