Spatial and temporal planning of renewable-based supply chain network considering multiple uncertainty

Abstract

Recently, the importance of a transition from a central power system based on fossil fuels to a distributed power system based on renewable energy has emerged. Accordingly, for the efficient design and operation of the renewable energy supply chain network, decisions on the optimal expansion timing and scale, operation strategy, and spatially specified design must be made. However, some factors make it difficult to determine the timing, scale, and operation strategy of expansion, such as long-term uncertainties such as increased demand for renewable energy and falling installation prices, high short-term volatility, and seasonality of renewable energy. In addition, the economic feasibility of renewable energy is greatly influenced by geographical factors, so a specific design of the supply chain is critically needed. Accordingly, this dissertation proposes a methodology for determining long-term expansion investment plans and energy supply chain management strategies in response to multi-time-scale uncertainties by converging reinforcement learning and mathematical modeling. As a case study, a long-term investment and operation plan for a distributed power system consisting of green hydrogen, batteries, and wind power generation is presented in Gapado, Jeju Island. In this dissertation, a methodology is porposed to determine the optimal design of an offshore wind power farm by integrating the geographic information system with mathematical modeling. As a case study, the design of turbines, power lines, offshore substations, and onshore substations is determined for four locations on Jeju Island. We further examine any changes to the design and economic viability obtained from stochastic optimization which consider wind uncertainty.