Integrated Production Scheduling and Delivery Routing: Complexity Results and Column Generation

Abstract

Abstract. In this paper, we study an integrated production scheduling and delivery routing problem. The manufacturer has to schedule a set of jobs on a single machine without preemption and to deliver them to multiple customers. A single vehicle with limited capacity is used for the delivery. For each job are associated: a processing time, a size and a specific customer location. The problem consists then to determine the production sequence, to constitute batches and to find the best delivery sequence for each batch. The objectives of the proposed problems are to find a coordinated production and a delivery schedule that minimizes the total completion time (makespan) or the sum of the delivery times of the products. We present complexity results for particular cases and a column generation scheme to solve a relaxed version of the problem, leading to a lower bound of high quality. Some computational results show the good performances of the method.

Two-stage stochastic/robust scheduling based on permutable operation groups

“…In previous approaches based on G-SEQ (Artigues, Billaut, & Esswein, 2005;Artigues, Jean-Charles, Cheref, Mebarki, & Yahouni, 2016;Billaut & Roubellat, 1996;Cheref, Artigues, & Billaut, 2016a, 2016bErschler & Roubellat, 1989), a G-SEQ is a valid solution iff any J-SEQ constructed by ordering the tasks within a group is valid in the sense of Definition 1. For the single machine problem, that means that a G-SEQ is valid iff for any pair (i, j) ∈ E, the group of job i is ordered strictly before the group of job j in the G-SEQ.…”

“…For partially valid group sequences, the worst case objective is approximated using the algorithm described in section 4.2.3. Optimizing without worst-case (WC=False) instead takes into account the second-stage heuristic and aims to find the G-SEQ solution that will have the best score over all training scenarios, when using the second-stage heuristic ( as in (Cheref, Artigues, & Billaut, 2016b)).…”

“…Artigues, Jean-Charles, Cheref, Mebarki, and Yahouni (2016); Cheref, Artigues, and Billaut (2016aBillaut ( , 2016b introduce for the single-machine problem a MILP approach to the computation of optimal group sequences w.r.t. the worst-case maximum lateness for a discrete set of scenarios given that the second-stage schedules the jobs according to the earliest release date first policy.…”

“…There were however several limitations. The MILP proposed for the G-SEQ model in (Artigues et al, 2016;Cheref et al, 2016aCheref et al, , 2016b) is only able to solve very small problems and a single tabu search heuristic was proposed for the G-SEQ approach. Only the robust setting in the 1-P problem was studied.…”

Efficient heuristics to minimize the total tardiness of chemotherapy drug production and delivery

“…For a good overview of the integration the production scheduling and vehicle routing problems, please see the review by Moons et al (2017). In all of these studies, the scheduling problem is often based on a single machine environment (Li and Ferrell 2011;Jamili et al 2016) or a parallel machine environment (Ullrich 2013;Belo-Filho et al 2015), and the distribution is done with either a single vehicle (Cheref et al 2016) or a fleet of vehicles (Li et al 2016).…”