Fast Batched Solution for Real-Time Optimal Power Flow With Penetration of Renewable Energy

Renewable energy systems have become an integral part of modern power grid operation, where the forecasting error is inevitable even though advanced prediction techniques are utilized. To improve the solution efficiency and accuracy of real-time optimal power flow (RTOPF), a three-stage framework for parallel processing is employed in this paper. In Stage 1, uncertainties from renewable generators and demand loads are characterized with scenarios. Large numbers of RTOPFs corresponding to each scenario are formulated and addressed in Stage 2, where the linear systems are regulated into the same sparsity pattern and then tackled in a batched style with the graphics processing unit (GPU). Results from Stage 2 are utilized in Stage 3 to perform a hot-start RTOPF, where the forecasting error can be minimized. Case studies are implemented on the IEEE 14-bus, 57-bus, 118-bus, and 300-bus systems with 1024 scenarios. The superiority of the batched GPU solution has been validated by comparisons with regular GPU, parallel CPU, and sequential CPU implementations. Discussions on the batch size and hot-start strategy are also presented.