Turning cheese production planning into a daily optimization process

Challenge

Grupo Lactogal is the largest dairy group in the Iberian Peninsula, processing more than 1.2 billion liters of milk annually across 14 industrial sites and six product families. However, its production plans depended on manual work, spreadsheets and the experience of a small number of planners. Each plant planned largely in isolation, while production and transfer decisions were made separately. This limited visibility across the network and made it harder to align plans with actual line capacity, milk availability, product shelf life and operational constraints.

The planning process focused on short-term execution. Teams had limited ability to anticipate demand peaks, assess the impact of different decisions or consistently account for cleaning, setup and downtime. Developing a feasible plan required repeated coordination between planning and production.

Solution

LTPlabs developed a planning platform powered by a mixed integer linear programming model, with Gurobi as the solver. It brings together daily data on demand, orders, production capacity, downtime, milk intake forecasts, quality requirements and coverage targets.

The model generates production, transfer and purchase suggestions across products, lines and planning periods. It plans these decisions together across plants, taking into account interdependent flows, milk availability, capacity, production constraints, curing and quarantine times, and product shelf life. It also supports production ahead of demand peaks and includes a drying plan to help ensure forecast milk is used.

The model’s priorities reflect Grupo Lactogal’s operating objectives: first reduce stockouts, then limit deviations from target coverage, reduce production connection changes and manage inventory costs. Planners review and validate the recommendations, and production orders can then be created in SAP.

Results

In the pilot with cheese production across three plants, service level increased by 10 percentage points.

The first version of the planning platform was running within three weeks. Planning knowledge that had previously resided with individual planners was captured in the system, giving the team a shared basis for reviewing plans and scenarios. The solution is already running out across company’s 14 plants and six product families.