Combining the high interpretability and accuracy of high-fidelity models (HFMs) with the lower computational cost of machine learning (ML) opens avenues for improving baseline processes as well as modeling emerging systems that are not entirely understood. Moreover, the assessment and optimization of emerging electrified chemical, energy conversion, and storage systems are of great importance for highlighting the potential cost-effectiveness and sustainability of these systems. Led by these motivations, this work presents two complementary frameworks for: (i) modeling dynamic systems using hybrid physics–machine learning models; and (ii) assessing economic and environmental aspects of emerging electrified systems. The first framework is a generalizable approach for obtaining dynamic discrepancy reduced-order models (DD-ROMs) that balance the differences between HFMs and ROMs using Gaussian Processes (GPs). The framework offers a comprehensive step-by-step process, including: (i) where to place the discrepancy terms; (ii) how to compute the discrepancy values; and (iii) how to train the discrepancy profiles using GPs. Specific results obtained by employing the framework have demonstrated that DD-ROMs can mimic the dynamic trajectories of an HFM with high accuracy while achieving significant computational gains. Meanwhile, the second framework consists of a systematic techno-economic analysis (TEA) approach to assess the cost and environmental aspects of emerging electrified technologies and storage systems. The developed framework extends conventional cost approaches to assess the capital and operating costs of these systems on a large scale. The proposed approach provides information on profitability and environmental metrics to evaluate the feasibility and trade-offs of enabling novel process designs, taking into account the inherent modularity of these technologies. Results obtained using the TEA framework for selected case studies of emerging systems suggest that such electrified technologies can be economically and environmentally feasible under specified conditions. Together, the developed frameworks enable efficient and interpretable modeling of dynamic systems while supporting the systematic assessment and optimization of emerging electrified technologies. Therefore, contributing to the improvement of baseline processes and the advancement of decarbonization efforts in the industrial sector.
Research Interests:
My capabilities lie at the intersection of Process Systems Engineering & Optimization, Fluid Mechanics, and Machine Learning, with research interests focusing on developing new systematic frameworks for modeling, TEA/LCA assessments, and optimization. These new frameworks aim to enhance the operational, economic, and environmental performance of industry-relevant baseline processes, as well as accelerate the deployment of next-generation chemical, energy conversion, and storage systems.
