Toward Online Optimal Power Flow of a Networked DC Microgrid System

Abstract

Microgrids with distributed generation and storage assets often form an underdetermined system of power flow equations for balancing loads and generation. This feature is compounded for networked microgrid topologies. Advanced control strategies offer a solution to this power flow problem and often require a feedforward reference. This provides an opportunity to compute energy optimal references but with the limitation of solution computation time. Therefore, four optimal reference command generators were developed focusing on solution time and scalability to both asset quantity and number of microgrids networked. Strategies explored were pure numerical, closed form, and numerical hybrids, and a Lagrange multiplier method. A tradeoff existed between smaller solution time of the Lagrange multiplier method and guaranteeing a feasible solution intrinsic to the hybrid approaches. Timing trials were performed where generation assets per microgrid ranged from 1 to 130000 and networked microgrid quantity ranged from 1 to 250. These trials quantified the bounds where subsecond updates rates could be achieved for two types of topologies.