A cross-disciplinary many-objective optimization framework for sustainable recipe design: A meatball case study with dynamic ingredient substitution

Food products must increasingly satisfy multiple, often conflicting requirements related to cost, sensory quality, nutritional value, and sustainability. Conventional recipe optimization studies in food engineering typically treat these aspects in isolation or restrict attention to at most two or three objectives, which limits the ability to explore realistic trade-offs. In this work the authors present a cross-disciplinary optimization framework that treats recipe design as a many-objective problem and integrates three domains: process, operation, and nutrition. The framework is demonstrated on the development of a protein-rich meatball recipe. Four ingredients are combined under compositional and process constraints, and six objectives are optimized simultaneously: production cost, production time, overall taste, acidity, energy density, and protein density. Sensory attributes are modeled by surrogate functions trained on data from a designed experiment and sensory evaluation, allowing rapid scoring of candidate recipes during optimization. System and nutritional attributes are linked to a Digital Food Identity Card (DFIC), which provides ingredient-specific information such as cost, cooking time, and nutritional composition. This enables dynamic ingredient substitution and adaptation of recipes to changing prices and availability. A custom multi-objective evolutionary algorithm is used to generate a six-dimensional Pareto set, which is then visualized with a novel representation based on a ring of congruent quadrilaterals (Quad2Quad). This advanced Pareto front visualisation enhances the interpretability of the trade-offs between objectives visually interpretable and supports the identification of balanced solutions that satisfy requirements across all three domains (i.e., process, operation and nutrition). The case study illustrates how the proposed framework can help food developers design recipes that are nutritionally dense, sensorially acceptable, economically viable, and robust to fluctuations in the supply chain. • Proposes a cross-domain many-objective optimization framework for food design. • Formulates meatball recipe design as a six-objective optimization problem. • Uses surrogate models built from experimental sensory evaluation data. • Integrates a Digital Food Identity Card for dynamic ingredient substitution. • Introduces a visualization method for identifying many-objective trade-offs.