Removal of long connections

[olwijn]

A rule could be introduced to remove long connections based on a local distance criterion considering the largest distance among the nearest N nodes. Consider e.g. the red node. If N=3, the nearest nodes are those indicated in green. The connection to any node that is further away from the red node than the distance to the lower-right green node would be removed. This would remove the very long connection to the blue node. Note that this is just a geometrical argument for removing connections. Instead of distances to nearest nodes, it could be based on distances to nearest wells, and perhaps also consider the types of those wells.

[olwijn]

On further thought, some refinement may be necessary. For example, if all nearest nodes are to the right, any connections to the left would be removed, which may be undesirable. This suggests that some directional criterion would need to be included as well.

[edubarrosTNO]

On further thought, some refinement may be necessary. For example, if all nearest nodes are to the right, any connections to the left would be removed, which may be undesirable. This suggests that some directional criterion would need to be included as well.

But would this situation ever occur when using the Mitchel best candidate's node placement approach in a convex hull?

Your suggestion to consider directional criterial goes a bit into the path of the suggestion that I had made about ensuring additional nodes surrounding the well nodes before running the Mitchel best candidate's algorithm