A Deep Dive into an Air Cooled Data Centre's Mechanical Cooling System Energy Consumption

As data centre electricity usage grows rapidly in response to global digital demand, there is increasing pressure to understand and optimise cooling system energy performance. This study presents an open-source, hybrid simulation model developed in Microsoft Excel to replicate and analyse the cooling energy profile of a large-scale commercial data centre in Dublin, Ireland. The model integrates five years of sub metered data across over 90 electrical and thermal measurement points and incorporates both regression-based load estimation and first-principles physics to calculate mechanical cooling energy use. Designed for flexibility and transparency, the model enables site-specific evaluation of air cooled chillers, adiabatic dry coolers, CRAC units, AHUs, and pumping systems under real-world control logic and weather conditions. The framework also supports scenario testing, control optimisation, and diagnostic fault detection. Validation against metered data shows strong alignment, with a modelled annual PUE of 1.27 and sub-system accuracies exceeding 98% in the simulated cooling load category The results provide valuable insights into free cooling opportunities, control inefficiencies, and underreported energy use, supporting both operational improvements and broader ESG reporting goals.