Fast Decoupled Probabilistic Power Flow for Islanded Microgrids

This paper presents a fast decoupled power flow algorithm to estimate voltages and frequency in microgrids with islanded operation. The algorithm is derived from dividing the original Newton’s linearized system into subsystems: Pθω and QV. Furthermore, the proposed approach uses a stationary Newton Method (constant Jacobian) to reduce the computational burden. The proposed method is compared with the conventional Newton-Raphson Method (NRM) and with a Fixed-Point method based on the Zbus matrix. This comparison was performed through a probabilistic power flow study in a meshed microgrid with 310 nodes. The probabilistic power flow used Monte Carlo Simulation to model uncertainties associated with load forecast errors and component outages. The tests results showed that the proposed approach has the same accuracy as Newton and Fixed-Point methods, but with significant computational time savings over NRM, e.g., the proposed approach is about 19 times faster than NRM.