Operationalizing Sustainability through Data-Driven Batch Sequencing : A Case Study on Minimizing Waste in Dairy Production

Sammanfattning

The process industry faces increasing sustainability requirements that highlight a persistent gap between strategic sustainability goals and operational reality. This study investigates how a hybrid modeling approach, combining constraint programming and discrete event simulation, can evaluate batch sequencing decisions to minimize waste in dairy production, while ensuring operational feasibility. Through a case study at a large dairy production facility in Sweden, qualitative interviews were combined with production data and technical drawings to build and validate an optimization model using CP-SAT and a simulation model in FACTS Analyzer. The results demonstrate that the optimized production plan decreased mixed-phase waste by an average of 52% and reduced late order volumes by 99% during a representative week. However, validating the schedules in a stochastic environment significantly increased missed deliveries, narrowing the performance gap. While the optimized plan saw more late orders in the stochastic simulation than under deterministic conditions, the original schedule also performed worse in the simulation than it did in the historical production. This shows that standard statistical distributions fail to capture how operators dynamically adapt and prioritize tasks during disruptions. Ultimately, deterministic optimization alone cannot handle factory volatility. Instead, digital tools should serve as decision support within a human-in-the-loop framework.