Siemens Optimized HT Data Center Using AI Technology

Siemens has successfully completed the White Space Cooling Optimization (WSCO) project at Hrvatski Telekom's data center. Thanks to an intelligent sensor layout and AI-based cooling control, HT’s data center in Zagreb now operates more efficiently and sustainably.

“Going forward, Hrvatski Telekom will use an AI solution with machine-learning capabilities that dynamically manages the cooling process in the data center. Energy savings in the six-figure euro range are expected because of the WSCO solution, along with the avoidance of potential failures caused by excessive temperatures,” said Martin Lang, Head of Smart Infrastructure Buildings business unit at Siemens Austria.

“Thanks to the WSCO solution and dynamic AI-based cooling, we have achieved significantly more efficient cooling of IT and network equipment in our largest data center. The implementation has led to improved temperature control in system rooms, the elimination of so-called hot spots, reduced operating hours of cooling units, and lower maintenance costs,” added Ivan Visković, Director of Core Network and Services Sector at Hrvatski Telekom.

White Space Cooling Optimization (WSCO) consists of both technical and process components. Thermistors measure temperature changes on servers and network racks based on electrical resistance. The technical component includes wireless sensors and control modules, an AI engine, and a user interface. From a process perspective, an optimization loop is used to ensure continuous improvement.

The process begins with an analysis based on data collected by sensors monitoring current operating conditions. WSCO then uses artificial intelligence to automatically regulate fan airflow and identify issues. Cooling processes are dynamically adjusted based on the acquired data. Continuous system operation and machine learning enable the elimination of up to 99% of so-called hot spots in a data center. In addition to ongoing optimization, the solution also improves the quality of preventive maintenance by identifying faulty components.