A design concept for data-driven brewing: sensor-based system architecture and ML applications for sustainability in micro-breweries

Breweries, as part of the food manufacturing sector, face increasing demands to reconcile traditional craftsmanship with modern expectations for sustainability, efficiency, and consistent product quality. Recent advances in digitalization and machine learning (ML) offer new opportunities to monitor, predict, and optimize complex production processes such as brewing. For micro-breweries, however, the practical implementation of such technologies remains particularly challenging, as they often lack both the technical infrastructure and a clear framework for how process data can be systematically collected and used to enable predictive maintenance, quality forecasting, and real-time optimization. Here we show an integrated concept based on a sensor- and ML-driven system architecture that supports data-driven decision-making across critical stages of the brewing process, contributing to resource efficiency and process stability. The proposed concept defines specific use cases and maps them to relevant sensor signals and ML models, including Support Vector Machines, neural networks, and reinforcement learning. A modular system architecture is proposed, linking IO-Link sensors, a process control system, and a time-series database to enable real-time feedback and model-based process control. The study highlights the need for high-resolution sensor data to compensate for raw material variability and to stabilize biological processes, particularly during fermentation. These findings provide a practical blueprint for micro-breweries seeking to implement intelligent process monitoring and control. The proposed framework lays the foundation for future prototyping and empirical validation and supports the broader digital transformation of sustainable food production.