BEELINE: BilevEl dEcomposition aLgorithm for synthesis of Industrial eNergy systEms

Energy transition to a more sustainable basis is the most significant and complex challenge facing industry. On most industrial sites, the largest single energy user is the utility system that produces the heat and power necessary for the site. The heavy reliance of current utility systems on fossil fuels and the requirement of strategic measures to ensure a sustainable future has prompted researchers to explore different energy sources, technologies and pathways for evolving existing systems to a sustainable basis for future utility systems. Whether based on renewable or non-renewable energy sources, it is essential to minimize energy demand and mitigate emissions. The present work focuses on developing cost-effective solutions for the synthesis of process utility systems, considering site-wide energy integration. This will provide a platform in future work for the efficient introduction of renewable energy sources in the transition to sustainable systems. Utility system performance is generally determined by system configuration and operational load. Steam mains selection, in terms of pressures and superheating have a critical role in the utility system performance and site energy integration. Therefore, the synthesis of energy-efficient utility systems involves optimizing the configuration of utility components and number and operating conditions of steam mains simultaneously. Due to nonlinearities and non-convexities from underlying physics and binary decisions involved, the resulting Mixed-Integer Non-Linear Problem (MINLP) presents challenges for advanced state-of-art solvers to solve real-world problems. In past work, a number of important practical issues have been oversimplified in order to make the solution tractable. However, the oversimplifications also lead to misleading results. In this research, a mathematical formulation for simultaneous optimization of comprehensive utility system configurations and operating conditions is for the first time combined with more realistic steam system configurations and operating conditions to represent the utility systems. Its framework is constructed via a bilevel decomposition algorithm based on piecewise MILP relaxation, McCormick relaxation, and linearization of steam properties. In addition, the solution pool feature of the CPLEX solver is incorporated to enhance the performance and convergence of the algorithm. This work presents the fundamental problem formulation that has not been sufficiently addressed previously. Indeed, this methodology sets out the basis for synthesizing energy-efficient utility systems for the future and allows for many energy conversion technologies and sources to be added to the framework that have previously not been possible to include.