Data-driven Optimization of Compressor Configuration and Dispatch in Industrial Compressed Air Systems

Compressed air systems (CASs) are widely used in industrial environments and account for a significant share of electricity consumption, often characterized by low overall efficiency. This work presents a data-driven methodology for the energy analysis and optimization of industrial compressed air systems. The proposed approach is based on real operational data collected from an industrial plant located in Veneto (Italy), which are used to perform a detailed energy performance analysis. This phase includes the identification of fixed-speed compressor operating states and the construction of representative demand scenarios, enabling the characterization of system inefficiencies and demand variability. Building on these insights, a Mixed-Integer Linear Programming (MILP) optimization framework is developed to address both compressor dispatch and system configuration. The model leverages data-driven performance representations of different compressor technologies and evaluates optimal operating strategies under varying demand conditions. The methodology is validated on multiple industrial case studies, including additional compressor rooms located in Thailand, demonstrating its applicability across different system configurations and demand profiles. The results highlight the effectiveness of the proposed approach, showing energy savings of up to 14% through optimized dispatch and up to 62% through configuration redesign. Overall, this work provides a practical and scalable framework for improving the energy efficiency of compressed air systems by combining data-driven analysis with optimization techniques grounded in real industrial operation.