Data Center Sensor Solution: Select the appropriate type from four categories in one go
Data center sensors: wind speed sensor, pressure difference sensor, temperature and humidity sensor, data center Temperature Sensor; PUE optimization; cold/hot aisles
What sensors are used in data centers? Temperature and humidity, wind speed, pressure, and temperature—each category requires its own sensor. Choosing just one wrong sensor will prevent the PUE from improving.
Room temperature and humidity serve as the fundamental data basis for PUE optimization. Temperature and humidity sensors are installed at the cold aisle inlet to continuously monitor whether the supply air temperature remains within the 18–24°C range. Three key parameters should be considered during system selection:
Accuracy: Temperature ±0.2°C, humidity ±2% RH. A sensor with a measurement accuracy of ±0.5°C may experience a deviation of 1°C under a cold channel environment at 23°C, leading to either excessive or insufficient cooling by the air conditioner and resulting in a 15%–20% waste of energy consumption.
Sampling rate: Prioritize sampling every second. Due to transient data center loads, sensors sampling longer than 5 seconds cannot keep up with the air conditioning synchronization rhythm; local hotspots may have already spread before an alarm is triggered.
Transmission method: Network interface (Modbus TCP) outperforms analog transmission. Ethernet latency is <100 ms with a packet loss rate <0.01%, significantly better than WiFi/LoRa, and supports POE power supply without additional wiring.
Installation locations: One unit per cabinet (primarily in CPU cooling zones); one unit every 5–8 meters along hot and cold aisles; UPS rooms and air conditioning rooms require separate installations.
The cold channel delivers air while the hot channel exhausts it; airflow velocity determines cooling efficiency. The recommended velocity for the cold channel ranges from 0.5 to 3 m/s—velocities below 0.5 m/s cause airflow stagnation, while velocities above 3 m/s lead to significant noise and energy consumption increases. Two key factors determine selection:
Type: Data center preferred – Hot-type (thermal convection). Compact size, fast response (<1 second), sensitive to light wind speeds (0.1–5 m/s) within cabinets. Mechanical type (wind cups/blades) offers low accuracy and wear-prone components, unsuitable for dense deployments. Ultrasonic type suitable for large spaces but costly.
Accuracy: ±2% FS meets the requirements for airflow monitoring in cold and hot aisles; ±0.5% FS represents research-grade precision, but offers poor cost-effectiveness in data center applications.
Measurement locations: Air supply velocity at the cold aisle entrance; exhaust velocity at the hot aisle outlet; and air supply velocity at the perforated floor bricks.
The cold channel must maintain a positive pressure of 5–10 Pa to prevent hot air backflow. Pressure differential sensors monitor the pressure differences between the cold and hot channels, between the front and rear of the cabinet, and between the upper and lower surfaces of the floor. Two key factors should be considered during selection:
Range:0–50 Pa or 0–100 Pa for low-range differential Pressure Sensors. Data center pressure differentials typically fall within the 5–20 Pa range, making a universal 0–10 kPa differential Pressure Sensor impractical due to insufficient accuracy and resolution.
Resolution: 0.1 Pa resolution is preferred. A ±1 Pa deviation in a 5 Pa positive pressure differential indicates a 20% control error and significantly increases the risk of thermal backflow.
Installation locations: between the cold aisle and hot aisle (to monitor positive pressure in the cold aisle); front and rear of the cabinet (to monitor airflow penetration); beneath the floor and within the data center space (to monitor supply air pressure).
The temperature difference between inlet and outlet water in a liquid-cooled data center (CDU) directly determines cooling efficiency. A precision deviation of 0.5°C in this temperature difference can cause the PUE to shift from 1.3 to 1.4. When selecting equipment, consider two key parameters:
Accuracy:±0.3°C for probes from the same batch. Inlet and outlet water Temperature Sensors must be of the same batch and identical accuracy; otherwise, the temperature difference calculation will introduce a systematic error of ±0.6°C, far exceeding the design allowable deviation.
Encapsulation: Glass encapsulation for NTCs is preferred, with response time <3 seconds. Metal encapsulation for NTCs takes over 5 seconds to respond, and CDU flow regulation cannot keep up with temperature changes.
Installation locations: CDU inlet, CDU outlet, cold plate inlet, and cold plate outlet – achieving full coverage across all four temperature differential points.
Only the coordinated operation of four types of sensors can reduce the PUE from 1.5 to below 1.3. Key parameters include: environmental acceptance for temperature and humidity sensors, airflow efficiency for wind speed sensors, thermal isolation for pressure differential sensors, and liquid cooling accuracy for Temperature Sensors—the absence of any one type creates a monitoring blind spot. Selection comparison table:
Data center sensor solutions: Clarify all four selection categories in one go – only then can PUE optimization be supported by solid data foundations.