Data centers already use around 2% of the world’s energy production, and their energy demand is set to increase eightfold — to some 3,200 terawatt hours (TWh) by 2030. On average, 40% of this energy is used by cooling systems to ensure the facility remains operational 24/7. That is why there has never been a greater focus on improving data center power usage effectiveness (PUE) through the increased energy efficiency of the fans, pumps, and compressors that form the heart of data center cooling systems.
Variable frequency drives (VFDs) control speeds of electric motors used in cooling applications so that they use the precise amount of energy required to produce the airflow at any given time. But, not only do they offer increased energy efficient, they also offer reliability, availability, operability, security, serviceability, scalability, and sustainability.
Data center cooling systems are sized to handle peak loads under the most adverse conditions, such as in the heat of summer or in the event of a component failure. But, in most cases, they are not operating at their design loads. For much of their life span, they operate in a lightly loaded state. The challenge is to ensure the cooling system can adjust to match the data center’s varying load profile to maintain high system efficiency even at partial loads.
There are numerous options when it comes to motors and speed control solutions for fans, pumps, and compressors. When considering their efficiency, it is important to consider the total effect of both electrical and mechanical performance. A high-efficiency motor operating in combination with a VFD can offer superior overall efficiency.
Reliability is critical for data centers, and their cooling systems play a vital role in ensuring the facility is always up and running. Harmonics in the power network is one of the key factors affecting reliability.
While VFDs make a very significant contribution to energy efficiency, together with other non-linear loads (like UPS and battery storage systems) they cause current and voltage wave form distortion, increasing troublesome harmonics in the power supply. Harmonic currents generate power losses that add heat to the power chain. This can increase the risk of downtime by causing equipment malfunction and failure.
Harmonic currents need to be minimized, and from a system perspective, there is, of course, the opportunity to simply over-dimension the cables, fuses, and switchgear. But if there is no mitigation at all, as is the case with some EC fans, the need for over-dimensioning can be up to 30%.
In addition to costs incurred due to reduced equipment lifetime, process issues, and constant maintenance of malfunctioning devices, operating costs might be further increased by penalty charges from electrical utilities for harmonics content in the network.
The answer is to use ultra-low harmonic (ULH) drives designed specifically for HVACR applications as they can maintain low harmonic content, even at partial loads right at the source. With a proper system analysis, adding ULH to the chillers, large pump and fan sets can eliminate the need for external harmonic mitigation solutions for the whole system.
There is also a need to consider the type of load on the network, since it affects power factor (PF) — a measure of how effectively equipment uses the electricity. Data servers are capacitive loads. Cooling equipment has large inductive loads. Both load types are sources of reactive power that impact the power factor and, again, cause energy losses as well as unstable operation for power system equipment, including UPSs and generators.
For effective electricity usage and power system reliability, data center PF should be close or equal to unity (1).
The final element in power quality is radio frequency interference (RFI). Data centers use a high amount of variable-speed equipment to save energy. At the same time, all variable-speed equipment generates electromagnetic noise, both radiated to the near field and cable borne, which is transmitted throughout the data center and to the neighboring buildings. This radiation might also affect IT equipment and data center security and performance. Any variable-speed product must comply with the EMC standard for power drive systems EN61800-3.
Negligence in working to the correct EMC standards can cause severe issues. Not only will there be a need to install external filters at additional cost later on, managing high-frequency noise after the installation is completed can be an almost impossible task. This is because high frequencies might be contained within the cables also also radiate. If, as an example, mains and motor cables are installed in the same cable tray, the RFI filters are essentially without function, and the only solution is to move one of the two cables to a separate tray at substantial cost.
Operability and Serviceability
There are several approaches to motor speed control for fans in data centers. One approach is to use an EC fan unit — a speed control solution packaged with a motor and a fan wheel. If these are installed in a fan array, it offers redundancy if one of the EC motors fails. But the units are usually controlled via fieldbus, and the drawback is that there is no “hand” mode, so speed control is lost if connection with the control system fails. It is also not possible to use a bypass.
The alternative is to use multiple AC motors, each controlled by its own VFD. This offers individual fan control in a fan array application or control via fieldbus. BACnet connectivity can provide an easy-to-use interface. “Hand” and “auto” modes allow fan control even if the control system fails. There is also redundancy so that, if one or several units fail, the rest will continue working with increased speed.
It is also possible to configure systems for fan array control via a single drive. The redundant drive is kept in standby mode to maintain the required airflow should the main drive fail. Bypass usage to run the fans direct-on-line is also possible if no redundant drive is employed.
With fan arrays in particular, it is important to consider what happens should the motor or drive fail. For an EC fan or fan with integrated motor, it is often necessary to replace the complete unit. This is costly, can involve a long lead time, and impacts sustainability since the unit gets scrapped. Should an Integrated motor drive fail, the components can usually be replaced individually. However, the drive and motor spare parts are frequently locked to a package supplier, resulting in long lead times and higher costs to replace them.
The optimum choice for better serviceability is to specify standalone motors, drives, and fans, as in most cases every component can be replaced the same day. This also has a positive impact on sustainability.
Cooling System Availability
The functionality built into VFDs designed specifically for HVACR can make a significant contribution to availability across the entire cooling and ventilation system. For example, soft motor start eliminates the mechanical stress on the piping system as well as on applications like pumps, compressors, or fans. Additional functionalities include indication of incipient bearing failure through torque monitoring; pipe pressure monitoring to raise an alarm should a pipe start leaking or become blocked; fireman’s override mode to help the fire suppression system react properly and protect server equipment as well as workers should a fire occur.
The one sure thing is that the server density of data centers will continue to increase. That means that heat loads will also increase. This is where VFDs developed specifically for HVACR applications offer an important advantage, as they are designed with flexibility and scalability built in.