Finding Robust Transition Paths for Industrial Ecosystems

More Info
expand_more

Abstract

In order to achieve the goal of sustainability, industries will have to make a transition to renewable energy and a more circular production. This requires substantial investments as well as constructive cooperation between the various companies in industrial clusters. The problem addressed in this thesis is to find ways of determining optimal investment paths for industrial clusters under specific constraints: the investments should contribute to sustainability (e.g., reduce emissions), provide a positive return for a cluster as a whole, and allow for a distribution of costs and benefits (e.g., through contracts) such that the companies that make the investment have strong incentives for cooperation.
The idea that underlies this thesis is to represent industrial clusters as networks of processes that are owned and operated by different companies, and interdependent via their input-output flows. Such representation should then facilitate analysis from an industrial ecosystem perspective, and identification of potential investment options that would improve the cluster’s sustainability. From a game theory perspective, these investment options can then be seen as potential moves, and the companies as players. Assuming a time horizon (e.g., somewhere between 10 and 40 years), any combination of investments in that period constitutes a strategy, while for each company the difference in cumulative cash flow for that company with/without these investments constitutes the players’ payoffs. Analysis of such a multi-company investment game will reveal rational strategies. We have elaborated and tested this idea by using the Linny-R modelling language and its associated MILP optimisation tool that is being developed at TU Delft first to represent and analyse a variety of simple process configurations with only one or two investment options. Conducting this first series of small-scale simulation experiments, and verifying their outcomes has demonstrated the feasibility of our approach. Subsequently, we have applied the same approach to a realistic, albeit simplified and stylised, industrial cluster that comprises three companies. After identifying a set of potential investment options, we have used the resulting Linny R model to conduct a second series of experiments to simulate and analyse solitary investment strategies per company, a cluster-wide cooperative strategy, as well as competitive strategies with various contractual arrangements. The results of this study show that we can indeed use Linny-R as a modelling language and simulation tool to represent and analyse investment decisions as multi-actor games, which then allow us to infer and evaluate cooperative as well as competitive investment strategies. This study has several limitations. We did not investigate the scalability of the method. We experimented with a relatively small and simplified cluster; upscaling to a cluster with 10 or 100 times more processes could become computationally infeasible. Another limitation is that we did not test in practice whether models in the Linny-R notation will indeed effectively support communication and negotiation between companies. Thirdly, the set of categories of investment options that we have identified is not exhaustive. Our recommendations for future research hence are to explore the computational limits using a state-of-the-art commercial solver, and to conduct real-world case studies, meanwhile extending and refining the categories of investment options that can be instrumental in furthering the transition towards more sustainable industrial clusters.