Mesh pre-optimization
Description
The concept of mesh pre-optimization has been under development for several years. It aims at transforming all types of data involved in the finite element process into a size map. The calculated size map must then be taken into account by a mesh generator in order to obtain a pre-optimized mesh.
The following types of data are taken into account in this concept:
- Form features of the geometric model
- Discretization error of the geometric model (distance between exact and discretized boundaries)
- Boundary conditions and efforts applied to the model
- Materials and physical constants involved in the finite element problem
The last two types of data may imply the definition of size maps as tensors instead of scalars (anisotropic materials, CFD). In this case, the size map is described as a metric.
Example : half of a pump casing
The two figures above illustrate the mesh of the pump casing with a constant density map.
- The mesh is automatically refined around three types of features: holes, concave edges, concave fillets.
- The nodal spacing has been set to 0.6 in non-refined zones and varies linearly down to 0.12 around refinement features.
- The boundary mesh contains 4010 triangles.
- The mesh is automatically refined in order to respect a tolerance between the true shape and the triangulation.
- The nodal spacing has been set to 0.7 in non-refined zones and the tolerance has been set to 0.01 (maximum distance between the true shape and the triangulation).
- The boundary mesh contains 4480 triangles.