Dynamic Optimization of Nuclear Expansion in Ontario’s Electricity System with Endogenous Learning Effects

This paper evaluates the magnitude of future nuclear expansion in the economically optimized 2049 Ontario power mix. A dynamic cost-minimization power mix model is used in conjunction with mixed integer linear programming (MILP) elements to generate several optimized scenarios for power mix expansion over a 25-year period. The linear model uses hourly timesteps to optimize supply management with high time-resolution and multistage decision-making to ensure that endogenous short-term decisions account for future carbon reduction goals and achieve overall cost-minimization. MILP integration provides in-depth analysis of potential future cost reductions that arise from experiential learning in production and deployment processes for new technologies. Multiple scenarios were generated to study optimal system expansion and final power mix composition in 2049 under differing assumptions for learning rates and large nuclear reactor (LN) construction time. The assumed learning rates closely follow technology-specific projections for expected learning effects and, where available, historically demonstrated learning rates. Simulation results show that large reactor focused nuclear expansion is generally encouraged across all candidate scenarios, including the baseline case (no learning assumed). When learning is applied in scenarios, results showed a shifting of system priorities away from current expectations of short-term natural gas (NG) expansion, instead encouraging the construction of additional nuclear technologies and storage systems. Simulated hourly operations, considering seasonal demand changes, showed these alternatives to be capable of providing similar system flexibility to existing gas. When assuming extended construction periods for LN compared to SMRs (10 years versus 5 years) and applying moderate learning rates, scenarios suggest that SMR deployment could be an economically competitive pathway for supply expansion. Under such conditions, early deployment of SMRs is favored to address immediate supply growth needs over other technologies (e.g., NG) and lessens large reactor expansion in the scenario. Through multistage decision-making, results conclude that sufficient learning and short, consistent deployment schedules render SMRs as a cost-effective long-term investment due to impending carbon restrictions, providing incremental supply growth during prolonged construction of larger nuclear projects. Finally, significant grid-scale battery energy storage system (BESS) expansion was observed in high-learning scenarios to provide ramping for system flexibility. BESS inclusion greatly reduced system dependence on continued NG usage for response capabilities but did not significantly impact base-load decision making. All costs in this analysis refer to 2024 CAD unless otherwise specified.