Selecting and applying distribution optimization methods

In order to determine the design of a distribution system a planner must determine three aspects of design simultaneously: layout (the manner in which the distribution pathways are split and resplit on their way from a few sources, or substation buses, to many customers), line size (the selection of conductor size, number of phases, and line type for each segment), and radial switching pattern (determining the normally open points within the network that force an electrical flow into a radial pattern). These three aspects of design interact heavily with one another, and generally the true least-cost plan can be found only by considering all three simultaneously. In most real-world situations, the number of possible combinations of layout, line sizing, and switching is so large that the probability of a smart, experienced engineer finding the very best configuration unaided is rather remote. Optimization methods that analyze all three aspects simultaneously to minimize cost can provide a real benefit. But no optimization method encompasses all aspects of distribution design, and no method is completely free of approximations and limitations. They are merely tools that can provide great help in distribution design if properly applied. Their good performance depends on proper selection and fitting to the problem and engineering CAD environment. This article focuses on selection and application of optimization methods for distribution design. The methods are numerical optimisation (both nontransshipment and transshipment), genetic algorithms, and null point load flow.