NP sub-optimization issue

[Naigege]

NP is not choosing the optimal MV line track. One possible reason is NP is not back checking after calculation. Below are the examples.

[chrisnatali]

This is possible due to the dynamic nature of mvMax, recalculated upon each new connection

A path between vertices may be bypassed because it's incident vertices do not meet mvMax criteria at the point they are tested within Kruskals....yet later on, a more costly path between those same vertices may pass mvMax due to the recalculation mentioned above (i.e. a vertex with large demand was subsequently connected to the subcomponent which boosted the mvMax of that subcomponent allowing for this more costly path).

As you suggest, addressing this might involve some sort of backtracking or "lookbehind" to determine if a shorter path now meets mvMax criteria.

[Naigege]

Scenarios found. On server: http://networkplanner.modilabs.org/
Black line: Buf1-Cluster-75m-HH-1000kWh SN=100 MVLines (scenarios 3929) 01/13/2015
Red line: Buf1-Cluster-75m-HH-500kWh SN=100 MVLines (scenarios 3928) 01/13/2015
Identical scenarios are running to double check if the same issue still shows up. Network model is modKruskal