Data driven multi-objective economic-environmental robust optimization for refinery planning with multiple modes under uncertainty

Abstract

In the context of the global energy transition, the refining industry is facing a major challenge between economic profitability and environmental protection. Planning refineries using traditional methods ignores the complex interactions between market dynamics and environmental factors, which causes unsatisfactory decisions. This study proposes a novel model-driven multi-objective robust optimization framework designed to tackle both economic and environmental challenges faced by the refining industry during its transformation process. The framework aims to maximize refinery profitability while reducing price uncertainty and achieving low carbon emissions and university energy use. Firstly, we introduce a nonlinear mechanism model of key production units to construct a multi-objective planning model that aims to simultaneously maximize profits and minimize carbon emissions and energy consumption. Secondly, to capture market dynamics uncertainty, we construct an uncertainty set based on the principal component analysis-robust kernel density estimation technique. Thirdly, using robust optimization theory, we transform this uncertainty set into a solvable binary problem. Finally, a case study is presented to demonstrate the effectiveness of this optimization framework in real-world applications, where we can keep CO2 emissions constant and still make the refinery profitable up to 95% in the worst-case uncertainty optimization. Additionally, uncertainty optimization’s profitability increases as conservatism decreases, illustrating the framework’s flexibility in responding to changing markets. Using this framework for decision support can assist the refining industry in protecting the environment while safeguarding economic profitability.