Research Projects

As India seeks to achieve carbon neutrality by 2070, it is critical that the growing energy demands are satisfied using low-carbon technologies. Renewable energy and carbon capture and storage are the two important technologies that can mitigate CO₂ emissions. The two technologies have been primarily developed independently. However, their hybridisation can offer complementary benefits and lower the costs of greenhouse gas abatement. Accordingly, in this proposal, we develop a novel carbon-neutral process OFSET, Oxy Fuel combustion with Solar power and thermochemical Energy sTorage, to produce electricity. We optimize the process’ economic and environmental performance considering variability in solar irradiance and electricity prices by combining unsupervised machine learning (ML), mathematical optimization (MO), and supervised ML models. Specifically, the OFSET process employs redox-based thermochemical energy storage (TCES) materials for energy storage and carbon capture and concentrated solar power (CSP) and fuels (biomass, natural gas, coal) combustor to generate electricity. Our preliminary results indicate that compared to CSP-TCES, the levelized cost of electricity (LCOE) of the OFSET process is 25% lower for an 85% capacity factor. Previous studies on the economic analysis of CSP-based systems use LCOE as the performance metric and do not consider the seasonal and daily variability in solar irradiance. While LCOE is an essential metric for understanding the major cost drivers, it does not capture the economic opportunities arising from time-varying electricity prices. Finally, it is unclear how the OFSET process can contribute to achieving India's decarbonization targets in the presence of other electricity production (wind turbine, photovoltaics) and storage (battery, thermal storage) technologies. The project aims to develop two open-source computational tools. The first tool will be based on a stochastic programming model that considers seasonal and hourly solar irradiance and electricity price variability. The model results will yield the optimal design (e.g., solar field area, receiver size, reactor volume, heat exchange area), operating conditions (mass and energy flows, unit temperatures, fuel consumption), and schedule (charging and discharging times, power production rate) that maximizes the profit of the OFSET process. We propose to reduce the model complexity by selecting representative days from the entire year by employing unsupervised ML techniques (e.g., k-means clustering). We propose a novel algorithm that combines supervised ML and MO with GPU computing to enable solving large problem instances. The second tool would focus on evaluating the decarbonization potential of the OFSET process by developing a mixed-integer linear programming-based supply chain network model and providing insights into the installation capacity of the OFSET process needed at various locations to satisfy the time-varying demand.