Fast reconfiguration of distribution systems considering loss minimization

This paper addresses the problem of finding the state of switching devices (open or closed) in primary distribution networks so that the total loss is minimum. Radiality and capacity constraints are taken into account. This optimization problem is a mixed-integer nonlinear optimization problem, in which the integer variables represent the state of the switches, and the continuous variables represent the current flowing through the branches. The standard Newton method (with second derivatives) is used to compute branch currents at each stage within the integer search, which, in turn, is implemented as a simple best-first search. Although a best-first search cannot normally guarantee the optimality of the solution, the high quality of the suboptimal solutions found, together with the high processing speed, make this approach very attractive for real-size distribution systems. Results from the application of the proposed methodology to a 1128-branch, 129-switch, real-world distribution system are presented and discussed.