Proceedings of the PowerUp Conference

To ensure the security of power systems, operators solve security-constrained optimal power flow (SCOPF) problems to determine setpoints that satisfy operational constraints under credible contingencies. As the size of the contingency increases, the problem becomes computationally expensive, especially for very large systems with thousands of buses and lines. Solving SCOPF problems with large systems using the typical B-θ model can become intractable. Instead, a more tractable approach is to use a sensitivity-based model based on Power Transfer Distri- bution Factors (PTDFs). Sensitivity-based models solve SCOPF problems more efficiently when a subset of the constraints is active at optimal solutions. Although solving SCOPF with the PTDF has many advantages over the B-θ model, the PTDF matrix is very dense in real-world systems, resulting in slow computational performance. This paper presents a method for sparsifying the PTDF matrix without compromising solution quality. The proposed method is exact and yields faster computa- tion speed. We demonstrate the proposed sparsification method on large-scale test systems representing the eastern and western interconnections of North America, with up to 78,000 buses and 10,000 contingencies. The results show that the proposed method solves the SCOPF problem 3 to 7 times faster than the common PTDF model.