2150. Dynamic resource flow optimization in industrial symbiosis networks
Invited abstract in session WA-42: Industrial Symbiosis, stream Circular & Sustainable Supply Chains.
Wednesday, 8:30-10:00Room: Newlyn GR.02
Authors (first author is the speaker)
| 1. | Jiayun Wang
|
| Industrial Engineering and Business Information Systems, University of Twente | |
| 2. | Haohui Zhang
|
| University of Twente | |
| 3. | Alessio Trivella
|
| Industrial Engineering and Business Information Systems, University of Twente | |
| 4. | Daniela Guericke
|
| Department of Industrial Engineering and Business Information Systems, University of Twente | |
| 5. | Devrim Murat Yazan
|
| Department of Industrial Engineering and Business Information Systems, University of Twente |
Abstract
Industrial symbiosis networks (ISNs) are becoming increasingly important in sustainable resource management. They are clusters of industries interacting in a way that one’s waste, e.g., materials and energy, can be used as input of another, potentially after treatment. While most of the extant literature focuses on the design of such networks, this work studies their day-to-day optimal operations under uncertainty in the price, demand, and availability of resources and wastes. Adopting a centralized ISN management perspective, we formulate a Markov decision process (MDP) where a set of interconnected (manufacturing and treatment) plants dynamically decide on production planning, inventory levels, and resource exchanges. As this MDP is intractable to solve optimally due to the exponential growth of state and discrete (combinatorial) action spaces with the network size, we compute heuristic policies based on forecast-based and scenario-based reoptimization, and on adaptations of deep reinforcement learning (DRL) methods. Numerical experiments show that DRL is more effective at capturing uncertainty and can thus be valuable for optimizing ISN operations, while reoptimization policies have an edge in large ISNs where DRL methods are difficult to train. Our results also highlight the value of collaboration and provide insights into production and procurement strategies in relation to the evolution of uncertainties.
Keywords
- Supply Chain Management
- Stochastic Optimization
Status: accepted
Back to the list of papers