2025 AIChE Annual Meeting

(680e) Multi-Period Optimization of Multi-Agent Circular Economy Networks: Application to Single-Use Plastics and Textiles

Authors

Ana I. Torres, Facultad De Ingeniería Udelar
The transition to a circular economy (CE) in which the value of end-of-life products is retained through reuse and recycling requires multiple agents in supply chains (SCs) with competing objectives to take certain initiatives over a multi-year time span. Furthermore, the decisions taken by different agents are often complex and interconnected, and the combined effects of different combinations of initiatives are not well understood. Although dynamic models of circular economies have been developed to explore the effects of different scenarios involving different initiatives, there is a lack of methods for systematic optimization of the transition to a circular economy over time. Such methods would help stakeholders prioritize efforts to maximize circularity while minimizing cost, environmental impact, burden-shifting, and trade-offs between different agents’ objectives.

In our previous work, we presented a generic system-dynamics based framework for modeling closed-loop SC networks. This framework is built up from the actor level and includes upstream raw material manufacturers, product manufacturers, consumers, a material recovery facility (MRF), recyclers, and the Earth. It considers the effects of consumer reuse while estimating the required capacity expansion and material quality lost due to recycling. Environmental impacts are quantified by combining LCA with the planetary boundaries framework by considering flows of raw material and pollution between the SC and the Earth. The framework was used to study a simplified version of the SC for single-use polyethylene terephthalate (PET) plastic packaging in the U.S. that included one actor of each type and a single product.

Here, we extend that work to use a multi-period state-task network (STN) formulation to define a superstructure representation for a generic circular SC network that now includes multiple manufacturers, products, and recyclers, considers material streams with non-uniform compositions, and distinguishes between durable and nondurable products with different lifetimes. We use this model to optimize an extended version of the PET supply chain that includes glycolysis-based chemical recycling and polyester textiles, as well as PET bottles and clamshells. We consider different objectives, including circularity, environmental impact, and agents’ profits and use multi-objective optimization to find trade-off solutions that balance different objectives.

Preliminary results show that glycolysis-based chemical recycling maximizes circularity and recycling profit while mechanical recycling minimizes environmental impact while maximizing manufacturer profit. However, a large amount of mechanical recycling combined with a smaller amount of chemical recycling results in an optimal trade-off between sustainability and circularity, achieving almost the same circularity level but significantly reduced emissions relative to the maximum-circularity solution. These results reveal that although some degree of chemical recycling is needed achieve closed-loop recycling of waste plastic that cannot be mechanically recycled back into the same product due to quality loss or excess thermoform content, processing the majority of post-consumer waste by mechanical recycling results in a solution that harmonizes circularity the greatest.