SMRΦ: Cyber-Physical System Framework for Market-Responsive SMR Dispatch Optimisation via MILP-MPC

Abstract

Global decarbonisation necessitates flexible, low-carbon dispatchable generation to complement intermittent renewables. Small modular reactors (SMRs) offer this capability. However, their typical inflexible baseload operation fails to exploit dynamic market opportunities. This paper proposes and validates a Cyber-Physical System framework that predictively optimises SMR dispatch to minimise operational costs and carbon liabilities while maximising carbon-credit arbitrage revenue. With the levelised cost of electricity £40-60/MWh and the power purchase agreement (PPA) €60-80/MWh, a cyber-physical system (CPS)-model predictive control framework exploiting carbon-price signals offers compelling economic value. The methodology integrates autoregressive integrated moving average (ARIMA), random forest and long short-term memory models for hourly forecasting with a mixed-integer linear programming (MILP) optimisation model, dynamically aligning SMR dispatch with electricity demand and carbon-price signals, using historical data from 2022-2025. The MILP model incorporates real-world operational constraints, including ramp-rate limits and minimum up or down times derived from xenon poisoning dynamics. Simulation of the full-horizon MILP (Model II) achieves a 180.32% improvement over the static baseline, rolling-horizon model predictive control (MPC) (Model III) is 193.74% and the concatenated MPC (Model IV), which augments the prediction horizon with machine learning forecast over a hybrid window of measured and projected data. These figures reflect the aggregate performance over the full simulation horizon. Results validate the framework’s potential to enhance the economic and environmental value of SMRs in modern energy markets.