Carbon-Aware Orchestration in Cloud Kitchens: A Multi-objective Field Experiment Integrating Menu, Inventory and Dispatch (MINDs) for Emissions, Waste and Throughput

Abstract

The rapid expansion of cloud kitchens has intensified concerns about the environmental footprint of app-mediated food delivery, yet operational sustainability mechanisms in this sector remain underexamined. This study investigates whether integrating environmental information into operational decision processes can improve sustainability performance without reducing service efficiency. We develop and empirically evaluate a carbon-aware orchestration framework MINDs (Menu–Inventory–Dispatch Scheduling) that incorporates real-time carbon and waste signals into production timing, preparation quantities and delivery routing decisions. The framework was tested using a 12-week stepped-wedge field experiment across 18 multi-cuisine cloud kitchens in Delhi and Bangalore, supported by digital-twin simulation. Data sources included IoT-based energy telemetry, point-of-sale records, waste logs and courier routing data. Two-way fixed-effects models estimate within-kitchen changes following the sequential adoption of carbon-aware slotting, predictive inventory alignment and emission-weighted dispatch routing. Results show reductions in environmental intensity after implementation: per-order CO₂e emissions decreased by approximately 21%–28% and edible waste by 15%–19%, while order fulfilment time and customer satisfaction remained statistically unchanged. Digital-twin benchmarking indicates convergence towards modelled efficiency frontiers, suggesting improvements arose from improved coordination rather than reduced operational activity. The findings indicate that environmental metrics can function as operational control variables, enabling measurable decarbonization without major infrastructural changes.