A virtual model of data centre cooling

This study presents a calibrated, hybrid spreadsheet-based model developed to simulate the cooling energy consumption of a commercial-scale data centre in Dublin, Ireland. The model leverages five years of high-resolution, sub metered data, including chilled water flow rates, ambient conditions, and IT electrical load, offering a rare empirical foundation for system-level energy analysis. Existing modelling approaches often rely on simplified laboratory case studies, synthetic datasets, or isolated component analysis, limiting their relevance to operational facilities. This study introduces an innovative hybrid modelling framework that bridges the gap between black-box machine learning and rigid physics-only simulations. A key innovation is the use of a modular, Excel-based architecture that maintains high predictive fidelity (MAPE <2%) while remaining accessible to facility managers without specialised coding skills. By leveraging a rare 5-year empirical dataset, the model's novel diagnostic capability is demonstrated through the detection of sub-metering faults that traditional PUE-based metrics overlook. The analysis also revealed that IT equipment accounts for 86% of total electricity consumption, while cooling systems contribute 10–12%. Beyond prediction, the model supports benchmarking, performance tracking, and diagnostic evaluation under variable load and weather conditions. Its structure enables scenario testing for control optimisation, free cooling strategies, and setpoint adjustments. By grounding simulation in real-world measurements, this research provides actionable insights for energy and facilities managers, while contributing to the limited body of empirical work on DC cooling energy performance. Although developed for a single anonymised facility, the modular Excel-based structure is transferable and can be applied to other sites through systematic recalibration of equipment parameters and control logic. The model is particularly suited to Environmental, Social, and Governance (ESG) contexts, where robust, transparent energy and carbon metrics are increasingly required for disclosure and investment decisions.