Optimal Transmission Switching: Improving Solver Performance Using Heuristics

The optimal transmission switching problem (OTSP) is an NP-hard problem of changing the topology of a power grid to obtain an improved dispatch by controlling the operational status of the transmission lines. Exact solution techniques based on mixed-integer programming (MIP) like branch-and-bound (B&B) guarantee identifying global optimal solutions if they exist but are potentially intractable in realistic power grids. Heuristic methods, on the other hand, can provide tractable solution approaches but potentially cut off optimal solutions. Heuristics are implemented along with B&B in modern MIP solvers to take advantage of both approaches. This paper proposes solving the full OTSP formulation alongside parallel heuristics that generate good candidate solutions to speed up conventional B&B algorithms. The innovative aspect of this work is a new asynchronous parallel algorithmic architecture and the exploitation of domain-specific knowledge in parallel heuristics. Heuristics generate solutions asynchronously to be injected into the full OTSP solution procedure during run time. Our method is tested on 14 instances of the pglib-opf library: The largest problem consists of 13659 buses and 20467 branches. Our results show good performance for large problem instances, with consistent improvements over off-the-shelf solver performance.