Stochastic dual dynamic programming for transportation planning under demand uncertainty

This paper addresses the problem of transportation procurement process in a stochastic context and proposes a resolution approach aiming to obtain an optimal policy over a finite planning horizon. Based on a conceptual framework which integrates both strategic and operational decision making levels, this paper addresses the problem from an operational point of view while integrating the transportation decisions already taken at the strategic level. Strategic decisions provide a set of core carriers selected using a combinatorial auction mechanism in which carriers compete by submitting package bids on shipper s' requests. Operational decisions involve a set of core and spot carriers competing to procure transportation services and ship loads from a set of warehouses to a set of distribution centres over a finite planning horizon. The problem is modelled as a Stochastic Linear Multistage Program and the Stochastic Dual Dynamic Programming SDDP is adapted to solve it. To illustrate the practical usefulness and the behaviour of the obtained results, experimentations and sensitivity analyses are carried out.