A new hybrid distance and similarity based scenario reduction approach for stochastic economic operation of microgrid

Abstract

This paper attempts to develop a multi-time period stochastic optimization model for economic operations of a typical microgrid by employing a scenario-based analysis approach to exploit various uncertainties associated with variable renewable energy (VRE) generations, electricity prices, and load demand. This stochastic model is aimed at generating the optimum schedules for various dispatchable generating resources such as micro-turbines, fuel cells, utility grid, energy storage devices as per the availability of the various VRE resources to meet the uncertain demand for a day-to-day microgrid operation. Further, a suitable scenario reduction approach named hybrid distance and similarity (HDS) approach is proposed to cater for two diverse objectives i.e., minimization of the Manhattan distance and maximization of the similarity index between an optimal scenario pair for generating a reduced scenario set by eliminating large redundant scenarios from its original large set. To verify the effectiveness of the proposed HDS, its performance is compared with three well developed distinct methods such as SBR (simultaneous backward reduction method), FFS (fast forward selection method), and SIMCOR (similarity-correlation method) on two different stochastic optimization problems including one real-life economic microgrid problem. All the competing scenario reduction methods are compared in terms of various performance indices i.e. OSDI (Output Sample Deviation Index), PSRI (Percentage Scenario Reduction Index), objective values, and computation time to verify their suitability and effectiveness on complex optimization problems. The proposed HDS method is found to be capable in achieving the lowest OSDI value of 5.68 at 98 % scenario reduction while compared to other competing methods i.e. 12.95 by SBR, 14.76 by FFS, and 16.32 by SIMCOR for the real-life microgrid problem. Moreover, the proposed HDS methods also outperforms the other three competing methods in terms of their objective function values after 98 % scenario reduction with a least computation time burden i.e. 87.6 %, 1.11 %, and 53 % less computing times are needed by HDS, FFS, and SIMCOR, respectively. These comprehensive simulation results reveal that the proposed HDS method is capable to generate high-quality scenarios, better approximation, superior stability, and with lower computation time burden as compared to the other three competing scenario reduction approaches.