- Machine Learning
In the manufacturing industry, short-term production planning and scheduling requires multiple trade-offs to account for service targets, capacity utilization, setup, on-time delivery, costs and profit. If many SKUs flow in the same production line, the challenge is how to plan and schedule in such a way that an optimal trade-off between customer service, operational performance, and cost of goods sold can be achieved while maximizing gross profit. This research project provides a novel mixed integer linear model formulation that optimizes lot sizes in a CG factory such that manufacturing capacities and efficiencies, production, inventory, holding and setup costs are considered simultaneously while maximizing the expected profit. The model solves a multi-echelon production and inventory network and quantifies the advantages by comparing different baselines. The model application evaluated against the simulated Sponsor Company reference baseline proves to be on average 4% more profitable every week, in a quarter of a year period, in the most conservative scenarios. The scenario analysis provides interesting managerial insights into what to expect when improvement efforts focus on minimum production lots, decoupling buffers or less-than-full deliveries and how they increase even further the overall profitability.