OR-Based Menu Planning Cuts Daily Food Costs by 18% Across Correctional Facilities
The Client:
A public sector food service operator responsible for generating monthly menus for correctional facilities — a process governed by strict nutritional guidelines, budget constraints, and procurement realities.
The Engagement:
SSI implemented a Google OR-Tools-based optimization model that automates monthly menu generation satisfying nutritional requirements and procurement constraints while minimizing daily food cost.
The Challenge:
Dieticians were building monthly menus manually; a time-intensive process that required balancing nutritional standards, ingredient availability, and cost targets simultaneously. Manual iteration meant each menu took significant effort to finalize.
There was no systematic cost optimization in the process. Dieticians made reasonable choices, but without a model exploring the full solution space, significant savings were being left on the table.
The process also didn’t scale well. Adding a new facility or changing nutritional guidelines required rebuilding menus from scratch, with no reusable framework.
The Solution:
SSI modeled the menu planning problem as a constrained optimization task: given a set of ingredients, nutritional requirements per meal, and cost parameters, find the monthly menu that minimizes total daily cost while satisfying all constraints.
The model was built using Google OR-Tools, which efficiently explores the full solution space and returns an optimal menu plan something no manual process could achieve.
The system is parameterized: nutritional guidelines, ingredient costs, and facility-specific requirements can be updated independently, enabling rapid adaptation when procurement conditions or regulations change.
The Impact:
Average daily food costs dropped by more than 18%; a significant saving at institutional scale, where the cost delta compounds across hundreds of meals per day.
Dieticians redirected the time saved from menu construction to nutritional quality review and program development, higher-value work that the optimization model cannot replace.
