Multi-Stage Stochastic Programming Under Endogenous Uncertainty of Integrated Sustainable Chemical Process Design and Expansion Planning

Abstract

Effective decision-making is essential for minimizing the environmental footprint while strengthening the competitiveness of the chemical industry. This paper proposes a multistage stochastic programming (MSSP) framework to take into account the design and expansion planning simultaneously for a sustainable multinetwork that integrates the water treatment, renewable energy supply, and carbon capture, utilization and storage (CCUS) with a main chemical processing sector under endogenous uncertainty related to conversion rates arising from the possible processing steps. Both economic and environmental concerns are involved in the optimization model for the entire system. The resulting mixed-integer linear programming (MILP) model is then solved using the Lagrangean decomposition algorithm. The proposed framework is further implemented in xylitol process design problems. Four cases are established based on different combinations of uncertain parameter distributions and time horizons, with model complexity increasing sequentially. The effectiveness of the proposed framework is further validated under new randomized sampling scenarios. The results indicate that integrating multinetworks can significantly reduce carbon emissions, thereby mitigating environmental impacts while satisfying production demands. Specifically, carbon dioxide (CO₂) can be fully captured and optimally utilized, wastewater can be completely treated, and economic benefits can be effectively maximized. In comparison to the deterministic model, the MSSP counterpart offers a sustainable, robust, and reliable solution for integrated design and expansion planning over the specified time horizon. Additionally, it achieves the value of stochastic solution (VSS) of 2%, potentially saving millions of dollars in long-term capacity planning, thereby underscoring the advantages of incorporating endogenous uncertainty into the problem formulation.