Modeling and Analysis of Actor Behavior in Supply Chains using Bilevel Optimization

In the contemporary world economy, characterized by global interconnection, supply chains serve as the foundational element of production and distribution networks that span international borders. The dynamic interplay among diverse actors, comprising producers, suppliers, wholesalers, and retailers, shapes the efficiency and resilience of these networks. While the integration of these entities through vertical integration can enhance supply chain performance, challenges arise from conflicting interests and competitive behaviors among stakeholders. This dynamic often gives rise to inefficiencies, escalated costs, and disruptions, particularly in highly competitive industries such as the energy sector. 

Thus, a central problem in this field is the question of how actors at different hierarchical levels of the supply chain can make optimal decisions when their interests do not coincide. In particular, the design of price contracts, production quantities, investment decisions, or schedules is often characterized by a relationship of mutual dependency. This issue can be expressed by so-called bilevel optimization models, which make it possible to model and analyze decision-making processes on several hierarchical levels simultaneously. In these models, one actor (Leader) makes a strategic decision, which is then responded to by one or several subsequent actors (Followers), who solve their own optimization problems.

Thus, the objective of this dissertation is to make innovative contributions to the modeling and solution of practice-relevant bilevel optimization problems in supply chains. The focus is on the following central aspects:

• Modeling of multistage, competitive supply chains under realistic assumptions, for example, in the energy sector, where actors compete for limited resources and market shares.
• The impact of price agreements and incentive mechanisms on the design and operation of supply networks, exploring how diverse contract models influence the decision-making processes of stakeholders and the overall efficiency of these networks.
• Solutions for bilevel optimization problems by integrating (meta-)heuristics, with the objective of reducing the computational complexity and enabling practical solutions for large problem instances.