Computing technology continues to advance rapidly. Increasing processing power for artificial intelligence necessitates larger server rooms that require robust cooling solutions. Depending on thermal load and humidity, it is possible to establish different temperature zones within a single room, allowing for uniform temperature control despite varying thermal demands. Improper climate control in IT environments can significantly shorten the service life of electronic components. Even minor deviations can lead to overheating of the technical infrastructure and subsequent substantial data loss. Today, approximately 800 watts per square metre must be compensated for by air conditioning, with internet data centres requiring around 1,500 watts or more. The trajectory of this development is evident in new IBM supercomputers designed for climate change or genetic research, which generate 1.5 megawatts across approximately 250 square metres—equivalent to 6,000 watts per square metre. These resulting thermal loads place enormous demands on climate control technology.
To remain prepared for future challenges in IT air conditioning, a holistic approach—and thus precise knowledge of flow and temperature conditions in server rooms and server cabinets—is essential. Consequently, FlowMotion analysed the airflow in a server room using Computational Fluid Dynamics (CFD). These simulations incorporated all flow-relevant details, such as the positioning of server cabinets and racks, air inlets and outlets, raised floors, and the operational settings of the air conditioning systems. The influence of free convection and thermal radiation on room airflow and temperature distribution was also accounted for.
The insights gained from these analyses enable the development of new concepts. Through the flow-optimised positioning of all server cabinets, racks, and air inlets and outlets, improved cooling of computer equipment can be achieved without increasing investment costs, thereby ensuring significantly higher operational reliability even as thermal loads increase.
