On the Robust Application of Loop Optimization Heuristics in Distribution Operations Planning

Loop optimization heuristics are widely used techniques in distribution operations planning. These heuristics are acknowledged as approximate optimization techniques. However, the approximation error and its dependence on the system have not been quantified with generality. Thus, being able to classify distribution systems by the extent of the expected approximation error is of interest for many. In this paper we project a set of heuristic errors on a defined system parameter, which we designate by network complexity. The projection results exhibit a two-stage evolution that suggests the existence of a complexity boundary for the robust application of loop optimization heuristics.