Sensor Selection and Deployment Practices for Data Center Environmental Monitoring

Industry consensus recommends that rack inlet temperatures be maintained between 18–27°C, with relative humidity between 40–60%. However, in many operating data centers, temperature sensors are still placed at CRAC return air grilles or in the middle of server rooms, failing to accurately reflect actual rack inlet temperatures. This sensor placement can result in localized hot spots going undetected, causing IT equipment to operate under elevated temperatures for extended periods, which affects service life and compute performance.

Deploying temperature and humidity transmitters directly at rack inlet zones in cold aisles is now common practice. A recommended density is one sensor per 6–8 racks, or one every 2–3 meters along the cold aisle.

A long-standing challenge in data center airflow management is that most operations teams do not actually know how much air each rack or CRAC unit is delivering. They may know fan speed and duct static pressure, but airflow volume is often estimated rather than measured.

The consequence is straightforward: when supply airflow exceeds demand, fans waste energy; when supply airflow falls short, rack inlet temperatures rise, leading to IT equipment throttling or even shutdown.

The choice of airflow sensing technology is central to accurate measurement. Two main approaches are currently available:

Thermal airflow sensors measure velocity based on heat loss from a heated element as airflow passes over it. They offer good sensitivity at low velocities and are suitable for laminar flow or relatively uniform velocity profiles. However, they are affected by changes in media temperature and pressure, and installation must avoid turbulent or swirling flow zones.

Pitot tube / differential pressure-based airflow sensors derive velocity by measuring the differential between dynamic pressure and static pressure. With no moving parts, they offer good long-term stability in data center environments. They require relatively long straight pipe sections for proper installation and have certain directional requirements.

Each technology has its appropriate use cases—selection depends on installation constraints, accuracy requirements, and maintenance considerations.

Underfloor static pressure is a critical parameter for cooling delivery. Insufficient static pressure indicates inadequate airflow for rack cooling demands, while excessive static pressure suggests fan energy consumption is higher than necessary, leaving room for optimization. Micro-differential pressure transmitters are typically installed beneath raised floors and within cold aisles to continuously monitor the pressure differential between these zones.

In some projects, micro-differential pressure transmitter data has been integrated into DCIM systems to enable variable frequency drive (VFD) control of fans. Operating data shows that adjusting static pressure set points from fixed values to demand-based regulation can reduce fan energy consumption by 15–20% without compromising cold aisle temperature uniformity.

A flow loss in the cooling water system can trigger chiller protective shutdown within minutes. The function of a flow switch is to provide early alarm notification when flow drops abnormally, allowing operators time to intervene or automatically switching to standby pumps. The KWFS series flow switches have been adopted by multiple data center operators, installed at chiller outlet pipes, plate heat exchanger secondary loops, and cooling water pump discharge lines. Field feedback indicates that their actuation accuracy and response speed are comparable to major European brands, with a significant advantage in delivery lead time.

Differential pressure switches are used to monitor pressure differentials across the primary and secondary sides of plate heat exchangers, as well as chiller evaporator and condenser pressure differentials. Abnormal pressure differentials typically indicate fouling, clogging, or abnormal refrigerant charge levels. By setting appropriate alarm thresholds, operations teams can receive early warnings before issues escalate.

As pre-filters and medium filters in air handling units accumulate dust, airflow resistance increases, requiring fans to operate at higher speeds to maintain set airflow. If filters are not replaced in a timely manner, fans running continuously at elevated speeds result in significantly higher electricity costs.

Using differential pressure switches to monitor pressure differentials across filter media—with an alarm triggered when the differential exceeds a set threshold—is standard industry practice. Maintenance records from one data center show that regular filter replacement reduces fan current by 10–15%. For a single AHU with a 30kW fan, this translates to approximately 20,000–30,000 kWh of annual electricity savings per unit.

Signal Interfaces: 4-20mA analog remains the dominant input method, but RS485 Modbus is seeing rapid adoption, particularly in new data center projects where direct digital signal integration into DCIM systems is becoming more common.

Ingress Protection (IP) Rating: While data center environments are relatively clean, precision cooling zones may have condensate water present. A minimum IP54 rating is recommended for sensors in these areas. Flow switches near liquid piping should be specified with IP65 rating.

Installation and Maintenance: The ability to adjust set points on-site is frequently cited by operations staff as a valuable feature. Maintenance personnel prefer to complete set point adjustments without requiring specialized tools, reducing fault response times.

Reliability Requirements: Data centers demand extremely high reliability from sensors. False alarms can trigger unnecessary air conditioning cycling or ineffective maintenance patrols, reducing operational efficiency. In multiple project deployments, the KCL series differential pressure switches and KWFS series flow switches have demonstrated a field false alarm rate of less than 1% over one year of operation, in line with industry standard performance levels.

Based on operational data from multiple data centers, sensor deployment density, measurement point placement, output signal type, and long-term stability all directly influence cooling system efficiency and PUE metrics. As data centers transition from air-cooled to hybrid air/liquid cooling architectures, sensors will continue to evolve toward greater precision and intelligence. We will continue to monitor industry requirements and provide data center operators with reliable environmental monitoring sensor products.