Collaborative optimization of supply chains with restricted coalitions



Agents in a supply chain can often improve system-wide performance if they coordinate their decisions instead of acting in isolation. In tactical supply chain planning, each agent typically solves an optimization problem relating to either production, inventory, distribution or facility location. By collaborating, they pool their decisions and solve an integrated problem, now acting as a single decision maker. Significant improvements are often possible through integrated optimization, as emphasized by the large body of research on integrated optimization problems in logistics and supply chain management. Well-known examples of such problems are the vehicle routing extensions of inventory routing, production routing and location routing.

In practice, however, full integration across a supply chain is not always possible due to e.g. the complexity of managing large coalitions and information sharing and trust issues. In this talk we consider integrated supply chain optimization in multi-tier supply chains subject to restrictions regarding which coalitions may form; that is, we are looking for a partition of the agents into permissible coalitions given certain restrictions, together with a sequence in which the coalitions will act so as to minimize the total supply chain cost. We use mixed integer programming to solve the corresponding integrated optimization problems of the coalitions that form in order to calculate the total supply chain cost. In particular, we address the following questions: (a) How close to the performance of full integration can we get given the restrictions? (b) How should costs be allocated to the individual members of a coalition? In order to make comparisons across different types of supply chains, we provide results based on realistic data representing four different industries.