Robustness of reconfigurable complex systems by a multi-agent simulation: Application on power distribution systems

Abstract

A strategy to improve robustness and reduce performance degradation due to damaging events is by system reconfiguration, taking advantage of backup connections that ensure the interaction between undamaged elements, and consequently, reduce performance losses. A prominent example of a system for which such strategy works are power distribution systems, where the system reconfiguration can reduce the number of unserved loads after failures. In this paper, we assess how reconfiguration affects the robustness of complex systems. The analysis is based on a multi-agent simulation over a power distribution system composed of distribution networks with backup connections between them. We model the power distribution network as a System-of-Systems (SoS). The robustness of the SoS is assessed at varying values of tolerance, each posing different limits to the maximum capacity of each distribution network. Results show that the reconfiguration enhances system robustness, and such improvement is directly related to the tolerance parameter. Furthermore, the simulation cost is higher in situations of low tolerance, reflecting the difficulty of finding a reconfiguration solution with low capacity limits.