> Are the 3 cases enough? Since there is possibility that the 3 spheres > share one area (volume in 3D) and a cubic corner will also intersect > with them, then the situation will be very complicated, it is not just > a sphere cap, I attached a figure (intersection.gif) about this, a > possible solution might through Alpha-shape theory as described below: > > http://cnx.org/content/m11616/latest/ > > However, whether this theory is able to handle the cubic intersection > is unknown, although possible. So currently I still think the sampling > method might be the "best" solution:-| > I know it is simplified. But if the overlap size is (much) smaller than radii of spheres (which is a condition for DEM convergence, I think), you cannot get 3 spheres sharing 1 point. And if you create the boxes such that they encompass the red volume while stretching further away a bit, you will never get sphere being punched by the box edge as you pictured.
I bet this method would be quite fast (at least compared to sampling), while being comparably o even more precise. But that's just feeling; try out both and you will see. Vaclav _______________________________________________ Yade-users mailing list [email protected] https://lists.berlios.de/mailman/listinfo/yade-users
