Revenue optimization for less-than-truckload carriers in the Physical Internet: dynamic pricing and request selection

Abstract

This paper investigates a less-than-truckload (LTL) request pricing and selection problem taking forecasting and uncertainty of transport requests at the selected destination into consideration. An optimization model coupling Dynamic Programming and Integer Programming is developed to optimize carrier revenue based on historical data of transport flows. The proposed model is studied in the context of the Physical Internet (PI). PI can be considered as a global interconnected logistics system that connects logistics networks via open logistics hubs. In each hub, LTL requests of different volumes and destinations arrive continually and are immediately allocated or reallocated to carriers. Carriers can bid for these requests through participating auctions. Carriers are confronted with numerous heterogeneous requests and must select one or several requests to bid for while at the same time deciding on a bidding price to maximize profit. Moreover, the carrier needs to forecast the number of requests at the destination hub to improve total profit, for example by improving the backhaul fill-rate. In this research, the number of requests is formulated as a distribution function due to uncertainty. Then, the optimization model is used for a multi-leg dynamic pricing and request selection decision. An experimental study based on real data is conducted to demonstrate the feasibility of the model and the impact of transport forecasting uncertainty on carrier revenue.