A first optimization approach to parameterize demand-driven MRP in the presence of multiple products and finite capacity

Demand-Driven Material Requirements Planning (DDMRP) is a promising production planning and control system that first appeared in 2011. The main novelty of DDMRP is that it protects critical references with buffer stocks and generates replenishment orders based on real-time demand and stock. Many scholars have studied its performance relative to more established production planning systems and demonstrated its competitiveness. More recently, studies dealing with the parameterization of Demand-Driven MRP have emerged. These papers present algorithms to fix DDMRP parameters with the objective of maximizing the On-Time Delivery (OTD) that is the percentage of customer orders delivered on-time or minimizing the average on-hand inventory. These studies all consider either a single reference or multiple references without managing conflicts between references competing for a bottleneck resource. This paper presents a first study to parameterize DDMRP in the presence of multiple products and finite capacity. Capacity limitation is modeled as a limitation of WIP (work-in-progress). A multi-objective genetic algorithm, which we initially suggested for a single reference, is extended in this paper to multiple references and finite capacity. The optimization algorithm is tested and analyzed on 21 data instances with 10 references.