Skip to main content

The Hidden Science of Data Center Cooling

The Hidden Science of Data Center Cooling
Data Centers

How CFD Reveals Hidden Airflow Risks, Tests Critical Scenarios, and Validates Design Improvements

Software: Siemens Simcenter STAR-CCM+

At a glance

Challenge
Over the years, data center clients have asked Predictive Engineering for help with three common issues. These include hidden airflow risks in new facilities, maintaining cooling when a CRAC unit fails, & choosing the best equipment layout for extreme ambient conditions.
Approach
Predictive Engineering built CFD models of each layout using Simcenter Star-CCM+. From there, we analyzed the airflow to identify risks through racks & underfloor regions, simulated CRAC unit failures, & tested external chiller placements to limit intake of generator exhaust.
Result
Simulation results for airflow and temperatures revealed hot spots and reverse floor-tile flow, confirmed high-powered racks stayed cooled during CRAC failures, & uncovered thermal challenges specific to proposed chiller layouts. This gave our clients design insights prior to construction.

The rapid expansion of artificial intelligence and high-performance computing has accelerated the construction of data centers worldwide, making reliable thermal performance a critical design challenge. Even well-designed facilities can experience airflow inefficiencies and localized thermal issues that are difficult to identify without simulation. Computational Fluid Dynamics (CFD) addresses this by evaluating design performance under a range of operating conditions.

Drawing on the expertise of Predictive Engineering's CFD engineers, this case study consolidates engineering insights from multiple data center projects to highlight common airflow challenges, cooling strategies, and best practices for improving thermal performance and system reliability.

Full CFD case study (PDF)