Production Planning and Scheduling
Most of this work has been done in collaboration with Stéphane Dauzère-Péres.
Integrated Planning and Scheduling. The emphasis is on the coherency between the "Planning" (or lotsizing) and "Scheduling" decision levels in the standard hierarchical decomposition. The context is a production by lots for small-medium series, and make-to-stock or make-to-order environments (for the latter the customer demand is supposed to be known on a sufficiently long horizon). In such a production context, we claim that most of the planning decision levels use lotsizing models with much too "naive" aggregate capacity constraints, which results in unfeasible production plans. We pretend that in a context of production by lots, the planning model cannot ignore the effects of sequencing decisions.
Therefore the traditional procedures are inappropriate and must be revisited. We have proposed the first integrated model of planning and scheduling which ensures the feasibility of the production plan, i.e., the existence of at least one feasible schedule to achieve that plan.
Scheduling. In parallel, we have also considered efficient heuristics for the scheduling problem in a complex production environment, i.e. with resource allocation, nonlinear routeing (assembly), etc ... An eficient Tabu search procedure has been developed and tested on a sample of problems. It is part of the scheduling module in the solving procedure of the integrated planning and scheduling model mentioned above.
Finally, we have also proposed an efficient "lot-streaming" procedure for the standard scheduling problem when one is allowed to split lots into independent sub-lots. Significant savings on the makespan can be achieved with very few sublots.