This paper analyzes the problem of delivering perishable products from a depot to stores and introduces the option of performing a second delivery per day to allow a retailer to better deal with uncertainty. For each store we have to determine the initial delivery quantity, whether or not the store will receive a resupply, the timing and quantity of resupply and the sequence in which stores have to be visited.
The problem consists of into two intertwined sub-problems, a stochastic inventory optimization problem for given delivery routes and delivery times (second stage), and a deterministic profitable tour problem determining the stores receiving a second resupply, the sequence of the stores, and the resupply timing.
We provide lower and upper bounds on profits and propose a hybrid solution procedure. The stochastic inventory optimization problem is solved by stochastic dynamic programming, while the profitable tour problem is solved using variable neighborhood search. In order to provide an efficient solution method, the second stage profit is not always computed exactly but approximations and bounds are used whenever possible. The algorithm is tested on randomly generated test instances. Furthermore, a real-world scenario is investigated. The results show a considerable improvement compared to the "single-order" model.
Titel: A One-Warehouse, Multi-Newsvendor Distribution System with Resupply and Vehicle Routing
When: Monday October 10, 2016, 15:30 – 16:30
Room: C75 Leeuwenborch
By Professor Dr. Stefan Minner
Logistics and Supply Chain Management, Technical University München, Germany