After the simulation results have been produced, we have several possibilities to visualize cavities in the debris. Blender of course does not have a ready optimized routine for this, but a combination of several tools can be used to do exactly that. We made several tests with a calculable cluster of elements, a few hundred objects, which looked promising. Of course these techniques need to be evaluated in more realistic scenarios with ten thousands if not hundred thousends of debris elements.
1. The first approach uses a boolean operation method in which the rubble was merged into one mesh object. This object was then substracted from a solid cube. The boolean operator in Blender is not the most advanced and the program was subjected to crashes to get the result. Blender will not be able to deal with meshes with high face count.
2. A second very elegant method allows also interactive investigation of voids in the debris by using the dynamic paint feature in Blender: A movable section plane acts as canvas while the debris mesh functions as a brush. Those areas of the canvas, that are within the brush volume are tinted with color. the uncolored areas then represent the voids).
3. A third approach uses a sort of cell grid to detect void volumes. The void is filled with a bubble like mesh representation. The accuracy of the detection can be adjusted with thresholds. A section plane can be moved interactively through the rubble area. The cavity representation is updated with some couple of seconds delay. This technique is stable but is -depending on the resolution- quite slow.