Abstract
As the load and some photovoltaic (PV) resources of microgrids (MGs) bring uncertainty to the system, they potentially affect the solution of conventional optimal MG planning. The scenario-based models do not guarantee the provision of continuous power in the possible worst-case scenario. This paper therefore develops a robust framework to address the uncertainty caused by load and PV generation. The operational constraints are also considered where the MG is connected to the grid and participates in the day-ahead market. Considering an energy storage system (ESS) and its sizing in the MG planning problem makes the determination of the worst-case scenario more challenging. A bi-level optimization framework is used to address this issue. One level tries to find the worst case through those variables directly being affected by uncertainties. The other level attempts to optimize the solutions to the MG planning problem. The Column-and-Constraint Generation algorithm is utilized to solve the proposed two-stage problem. The non-linearity involved in the model is transformed into linear equivalents to reduce complexity. It has been applied to a real case in Western Australia, and the results are discussed for both grid-connected and islanded modes by incorporating the effect of a PV tracker system in the planning problem.
