Scalable Power System Line Upgrade Planning With Policy Constraints: A "Branch and Benders" Approach
The integration of more renewable energy sources into the power system is presenting system operators with various challenges. At the distribution system level, voltage magnitudes that violate operating limits near large photovoltaic installations have been observed. While these issues can be partially mitigated with more advanced control, hardware upgrades are required at some point. This work presents a scalable, optimization-based approach for deciding which lines in a network to upgrade. Compared to existing approaches, it explicitly takes the operating policy of the system into account and provides both reasonable solutions in short computation times as well as globally optimal solutions when run to completion. Compared to earlier work on the same topic, an extended computational approach is taken that can simultaneously optimize for many load scenarios across arbitrary configurations of machines and CPU cores per machine in a scalable manner by using the Benders decomposition. In addition to the theory, numerical experiments are presented along with a discussion of the scaling properties of the Benders-based approach, giving potential users a better basis to decide whether their problem is big enough for the approach to make sense.