An Intelligent Decision System for Sustainable Food Production: Mitigating Waste via CVaR Stochastic Optimisation

Abstract We present an intelligent decision framework for sustainable manufacturing in a high throughput perishable meat factory. We address this challenge of food waste in a facility with a two-week turnover exceeding €1 million. We develop a risk aware stochastic linear program with hierarchical objectives: (i) satisfy demand, (ii) balance daily workload to limit labour volatility, and (iii) minimise environmental and economic waste using a Conditional Value-at-Risk (CVaR) term calibrated from historical data. We analyse age-agnostic retrieval and age-aware retrieval warehouses and the impact on the production schedule. On retrospective industrial data, the proposed system reduces CVaR risk by $$\approx$$ 59% under ageaware retrieval and $$\approx$$ 18% under age-agnostic retrieval at unchanged throughput. We find that analysing sensitivity to the CVaR confidence level yields practical guidelines for sustainable production planning in the factory. We discuss the system’s deployment implications for digital transformation in factories handling short-lived SKUs.