Title : Models and methods for decentralized decision in logistics networks
Keywords : Combinatorial Auctions, Integer Programming, Physical Internet
Co-supervisors : E. Ballot (Ecole des Mines de Paris), L. Brotcorne (INRIA)
Do not hesitate to contact Luce Brotcorne : luce.brotcorne@inria.fr
Description:
This Ph.D. research project consists in designing innovative logistics services based on the concept of physical internet while taking into account environmental and economic objectives. More precisely, it aims to develop models and decision making algorithms to design simultaneously several sustainable logistics and transportation chains between origins and destinations given a logistics and transportation network. This mesh network will result from information published by logisticians and carriers from their networks, warehouses, line haul services, less-than-truckload services.
The problem of designing multimodal sustainable chains will be addressed in the case where the system is shared between actors and decentralized. The shipper will specify the general framework (potential multimodal platforms to be used) and the origins and destinations of flows. Then, transportation companies will put out their offers, which will be considered according to a combinatorial auction mechanism.
This will result in a shared logistics activity that optimizes the use of resources.
The objective of the first part of the thesis will be to define the mathematical models for the problems addressed. The task will be complex as it is a decentralized approach. Models are required at the level of each partner (shipper, carrier or logistics service providers bidding and auction agents).
Foreseen models are based on: Offer of logistics services; Price determination model for buyers of logistics services; Combinatorial auctions at the level of the auction agent for the “winner determination problem”; Game sharing solution, Shapley value for example, to allocate gains resulting from auction between the vendor of the logistics service, the buyer and the possible intermediate (cross-dock operations…).
The second part of the thesis will be devoted to the development of innovative optimization algorithms. The objective is to solve the two mains and more complex models: the winner determination problem (allocation of a logistics service to a set of shipments) and the sharing mechanism resulting of new allocation of services. These two problems are NP-hard problems in general. Ad-hoc efficient methods based on mathematical programming will have to be developed.
