Computationally Efficient Formulations for Fault Isolation and Service Restoration in Distribution Systems

This paper proposes two computationally efficient mixed integer optimization models, in quadratic-constrained (QC) and linear forms, for optimal real-time fault isolation and service restoration (FISR) in distribution systems. The proposed models integrate an efficient integer programming formulation to model the automated switching operation, which use only one set of binary variables, and therefore are suitable to use in real-time switching application because of their relatively lower computational burden. The proposed models are implemented on the 32-bus and 123-bus test distribution systems under multiple fault scenarios in order to compare their accuracy and speed of computations. The simulations results show that the linear model can provide FISR solutions in almost real-time, with an acceptable accuracy compared to the QC model.