CFD analysis of a data center: thermal and airflow optimization

The increasing computational demands and heat loads in modern data centers create significant challenges for maintaining efficient thermal management and preventing hotspots. This thesis presents a comprehensive Computational Fluid Dynamics (CFD) study using ANSYS Fluent to analyze and optimize airflow, temperature distribution, and cooling strategies in a high-density data center environment. A validated reference model based on the CQ-University data center was replicated and then substantially improved by implementing a cold aisle containment strategy, which reduced maximum rack temperatures by 33.8 % and successfully eliminated hotspot formation. The research further investigates retrofitting for higher rack density, comparing several rack arrangements and identifying an alternating high/low-density configuration as the most effective for uniform cooling. Using real-world equipment sizing methods (Vertiv's Hi-rating and GRS 2.0 tools), the study then scales up to model a mega data center with a total IT load of 4.48 MW. Both normal and extreme operating conditions are simulated, with results confirming that fine-tuned cooling strategies and targeted airflow management can optimize Power Usage Effectiveness (PUE) from 1.06 to 1.021 and reduce annual energy costs by €374,760 (a 4.75% savings). The outcomes of this thesis demonstrate the critical value of CFD-guided analysis and literature-backed best practices in designing, validating, and operating thermally efficient data centers. By systematically addressing thermal management challenges, this work offers a robust framework for sustainable, cost-effective operation of high-density digital infrastructure.