Demand clustering in freight logistics networks

• This paper investigates demand clustering in freight logistics networks.
• Price and volume data are used in a Latin Hypercube sampling process.
• A network of bilateral utility is estimated based on optimized scenarios.
• Community detection is used to find cluster in this network.
• This tractable model overcomes limitations of other freight clustering methods.

Demand clustering in freight logistics networks is an important strategic decision for carriers. It is used to incorporate new business to their networks, detecting potential economies, optimizing their operation, and developing revenue management strategies. A specific example of demand clustering is truckload combinatorial auctions where carriers bundle lanes of demand and price them taking advantage of economies of scope. This research presents a novel approach to cluster lanes of demand. Community detection is used to cluster the emergent network finding profitable collections of demand. Numerical results show the advantages of this method.