Numerical analysis of the individual and combined effects of the aisle containment layouts on the cooling performance of the edge data center

Abstract

This study examines the cooling performance of edge data centers deployed in urban environments, where physical constraints and low-latency demands limit the feasibility of large-scale hyperscale facilities. By applying high-fidelity CFD simulations to a real-scale system comprising 70 server racks and an underfloor air distribution configuration, the study quantitatively compares three types of aisle containment strategies: cold aisle containment, hot aisle containment, and their combined configuration. The results reveal that cold aisle containment markedly reduces bypass airflow, lowering the mean inlet temperature by 5.0 °C and eliminating overheated server racks across the entire server room. The combined cold and hot aisle containment further mitigates recirculation and modestly enhances temperature uniformity, but the extra reduction in mean inlet temperature is only about 0.2 °C. By contrast, using hot aisle containment without a ducted return yields only a 0.5 °C improvement and even worsens thermal imbalances of the server room, resulting in 18.6 % of server racks exceeding 27 °C due to geometric limitations. These findings highlight that the aisle containment strategies should not be applied uniformly across facilities, but rather tailored based on spatial and structural characteristics of the edge data center. This study underscores the necessity of high-fidelity CFD modeling for accurate airflow diagnosis and offers practical design guidelines for developing reliable and thermally efficient edge data centers in dense urban contexts.