Quadratic programming-based grid optimization algorithms for real-time applications

Present applications in smart grid operation and planning require fast algorithms to optimize the reactive power of distributed generators. We present three methods to speed up a quadratic programming reactive power optimizer in which sensitivity matrices are computed by ""brute force"" load flows: selective updating of sensitivity matrices, linear approximation of the power flows and extracting sensitivity matrices from the power flow solver. Including the ""brute force"" reference case, we thus compare four reactive power optimization algorithms on their running time and ability to meet the optimization set point. Promising gains in running time can be reached, though the reactive power dispatch differs strongly between some of the methods.