By Michael Young, Pr. Eng.

Airflow is extremely important inside a data centre – it is the medium that cools the central processing unit (CPU) chip in IT equipment.

MICHAEL YOUNG Michael Young is a trainer, coach and a pre-sales engineer in the HVAC industry. He graduated from the University of the Witwatersrand in the field of Mechanical Engineering (B.Sc Mech Eng) in 2008 and qualified as a Professional Engineer (Pr.Eng) in 2013. Michael is passionate about promoting knowledge and helping other young engineers grow within the industry through his training workshops and coaching sessions.

The challenge is that we can experience server failure not because there’s insufficient cooling but because there’s insufficient airflow that is being delivered to certain servers. When this occurs, some servers may not receive an adequate amount of cold air which results in a temperature rise of the CPU and the server will shut down. The opposite can also occur where excessive airflow results in higher energy consumption of the cooling unit.

Ensuring optimal and sufficient airflow in a data centre has emerged as a critical factor in maintaining the efficiency and reliability of IT equipment. Today, HVAC engineers are developing innovative strategies to tackle airflow challenges, thereby preventing potential server disruptions and optimising energy consumption from the cooling unit.

So how do we now accomplish this? Inside a computer room air conditioning (CRAC) unit are multiple different types of airflow control strategies and today, we are going to discuss two general methods that are used by some suppliers:

  • One prominent method involves utilising the pressure difference between the raised floor and the data centre environment to control the speed of the fans. This approach makes use of a pressure differential measurement device that monitors the air pressure difference   under the raised floor and outside the cooling unit. By comparing these measurements against a predefined pressure difference set point, the system will dynamically adjust the fan speed. If the sensor measures a high-pressure difference value, it indicates that over-pressurisation is occurring and there is excessive airflow. The onboard controller within the cooling unit immediately reduces the fan speed, thereby reducing the airflow into the raised floor and electrical consumption of the unit. Conversely, if pressure difference readings are below the set point, the unit recognises that there is insufficient air under the raised floor and a signal is sent to increase the fan speed. 
  • A second approach is the use of remote temperature sensors that are strategically placed throughout the data centre. These sensors continuously monitor temperature variations and relay this data to the cooling unit’s controller via an advanced communication protocol like Modbus. Once this data is received from the controller, it can adjust the speed of the fans according to either the highest temperature sensor reading or the average temperature of all the temperature readings. This one variable is compared against the setpoint of the unit and the fan speed is adjusted accordingly. So, if temperature readings are above the setpoint, the unit interprets this as insufficient airflow and increases the fan speed. Conversely, if temperature readings are below the setpoint, the unit interprets that there is excessive airflow and increases the fan speed.

So those are the two general ways on how we are now moving forward within data centre cooling with the focus on what we call airflow optimisation. To summarise, central to both methods is the concept of airflow optimisation, where the goal is not simply to maximise cooling capacity but to ensure that cooling resources are utilised in the most efficient manner possible. By fine-tuning airflow based on real-time data and feedback, data centre operators can achieve significant improvements in energy efficiency while concurrently enhancing the reliability of IT operations.

The implementation of these airflow control strategies underscores a broader industry trend towards greater energy efficiency and sustainability. As the demand for computing power continues to grow exponentially, so too is the need to reduce the negative environmental footprint of data centres. Having energy saving components such as variable speed fans within a cooling unit does not immediately result in a lower energy consumption. The way these components are controlled to meet the ever-changing data centre cooling requirement is the way true energy savings are accomplished. Therefore, clients and end-users alike are increasingly prioritising solutions that can adapt and respond faster to meet the demands of the data centre in the ultimate goal of accomplishing sustainability.