Implementation of nonlinear variable-cost network optimization models for technology assessment in the petrochemicals industry

Abstract

Innovation in petrochemical technology is ongoing, whether to exploit opportunities, such as increased natural gas liquid supplies, or to face new challenges, including the need to decarbonize and to adapt to a future circular economy. Thus, there is a continuing need for petrochemical technology assessment. We consider methods for implementing optimization-based, superstructure models that assess new technology in the context of the entire hydrocarbon ecosystem, thereby capturing not just first-order, local impacts, but also higher-order, industry-wide impacts. These models are nonlinear, variable-cost, network representations of the industry, incorporating a dominant-producer price leadership approach. They are formulated either as a mixed-integer nonlinear programming problem or as a special-purpose successive linear programming problem. The performance of the different approaches is evaluated based on quality of results and computational efficiency. The comparisons initially use small prototype networks that feature structural properties of the industrial network, followed by application to full-scale, industry-wide problems.