Data-Driven Optimisation of Refinery Energy Systems: Exploring Flexibility and Integrated Decarbonisation Options with PyPSA

Industrial refineries operate as complex multi-energy systems where electricity, process fuels, steam, and multiple energy carriers interact across tightly coupled processes. This work presents a data-driven optimisation framework implemented in PyPSA to analyse operational dispatch and evaluate alternative decarbonisation technologies in a refinery energy system. The model integrates detailed historical operational data and hourly efficiency profiles to replicate assets’ behaviour and explore different technology configurations over a reference year. The objective function includes annualised capital expenditures and operating costs, including maintenance and energy carrier costs, enabling a consistent comparison of candidate technologies within the operational optimisation. The analysis provides insights into the potential contributions of existing assets’ flexibility and sector coupling to the identification of cost-efficient refinery decarbonisation strategies. Key performance indicators include total system cost, technology utilisation factors, and the resulting configuration of the energy system. The proposed full-year, hourly resolved LP approach offers a data-driven tool for exploring operational improvements and evaluating technology options under realistic operating conditions in industrial energy systems. The identified cost-optimal solutions for the multi-energy sub-system object of the study achieve annual emissions reductions of up to 7.2% of around 600 ktonCO2eq/y while maintaining annual system costs comparable to those of the reference case. Across all scenarios, flexible scheduling of the existing generation assets consistently emerges as a key driver of operational savings, highlighting the potential of industrial multi-energy systems to provide market-responsive demand-side flexibility through the adaptation of energy dispatch in response to electricity market signals.