Background

The mesh stripping procedure operates in two separate modes:

• to prepare the mesh for shape optimization and
• to reduce the model size.

The two modes operate in a different way and require a different set of parameters, as discussed in the following.

█ Stripping the mesh for shape optimization

When the mesh is stripped for subsequent shape optimization, the finite elements have to be cut precisely along the material-void boundary. High cutting precision, however, results in significant extent of distorted elements along the cutting surface. This negative effect can be mitigated only by lowering the cutting precision.

For this reason the user can provide some tradeoff parameter to decide either to:

• request lower precision cutting (and get less distorted elements), or
• request higher precision cutting (and get more distorted elements).

The figure below illustrates a stripped surface obtained by requesting the lowest and highest cutting precisions.

  

Figure. Mesh after: low precision (left) and high precision (right) cutting.

█ Stripping the mesh for model reduction

Stripping the mesh for model reduction makes sense only if the intent is to reduce the model before continuing with topology optimization. This is especially useful if a mesh refinement is also planned before continuing.

This operation may be quite useful, but it may also result in undesired and harmful consequences. Therefore, the following guidelines should be followed to minimize the occurrence of possible negative effects.

• Remove only unrecoverable elements. When removing void elements, make sure to remove only those elements that have obviously no chance to become material elements again during subsequent optimization cycles. In other words, the surface stripped in this operation should never become an actual material surface again.
• Expand the material region by a safe bandwidth of at least 2 to 3 elements. This will not actually expand the material region, but will keep a layer of void elements along of the material-void boundary. This layer of elements is absolutely necessary to preserve the stability and smooth material-void boundaries in subsequent topology optimization cycles.

The figures below illustrate a safe mesh stripping operation.

Figure. The initial model; all finite elements are visible.

Figure. Topology optimized model; void elements are still present but not visible.

Figure. Stripped model obtained by using an expansion bandwidth of 2 elements; material and void elements are visible.

In the above shown example one can be quite sure that the removed elements will be never needed again. And the expansion bandwidth of 2 elements also seems to be quite sufficient to assure a stable continuation of topology optimization cycles.