A modular AC optimal power flow implementation for distribution grid planning

Abstract

We present a computational tool for solving semidefinite relaxations of multi-period AC optimal power flow (OPF) problems. Chordal conversion techniques are used to exploit problem sparsity. Three features set it apart from similar implementations: First, a new, concise real-valued model exploits the problem structure and avoids introducing redundant constraints. Second, a dynamic choice of constraint type improves computation time for grids with extensive radial subgraphs. Third, a modular software design enables the easy integration of additional models for photovoltaic inverters, optimal storage placement, etc. Benchmark results indicate that our computational improvements significantly enhance performance compared to a standard implementation. This holds in particular for large-scale networks and power grids with large radial subgraphs. Finally, a case study showcases the potential of our modular OPF software design.