Network Reduction Assisted Acceleration of Risk-Aware Islanding for Distribution System Resiliency

This paper presents a computationally efficient method to solve optimization problem for power network resiliency applications, which can have many binary variables. One of such applications is microgrid formation in the distribution grid, which enables to optimally form islands to enable operation of the distribution grid without relying on risk prone assets. To make the solution faster, we propose the use of network reduction algorithm to remove the non-critical information from the network without masking the presence of essential components, such as switches and critical loads that are needed to make the island formation decisions. The proposed method reduces both the number of nodes and lines, while accounting for the load criticality of the aggregated loads placed at the nodes of the reduced network. Subsequently, this reduced network can be used to solve the islanding optimization problem. We validate the proposed method on the IEEE 123-bus and 8500-bus systems.