AI Server & Data Center Liquid Cooling Measurement Solutions

As the computing density of AI, GPU, and high-performance computing (HPC) servers continues to increase, the heat load generated by each server rack has risen significantly. As a result, data center thermal management is gradually expanding from conventional air cooling to liquid cooling systems. Liquid cooling removes heat generated by high-power components such as CPUs and GPUs directly or indirectly through coolant circulation, transferring the heat through coolant distribution units (CDUs), heat exchangers, and facility cooling systems to complete the cooling cycle. It has become an important thermal management solution for high-density AI data centers.

A liquid cooling system typically includes facility cooling systems, heat exchange sections, coolant distribution units, supply and return mains, branch piping, and liquid-cooled server racks. Changes in supply and return temperatures, circulating flow, pump pressure, equipment differential pressure, or reservoir level may indicate changes in cooling capacity, circulation conditions, or equipment operating status. Without continuous real-time measurement data, maintenance personnel may only begin troubleshooting after cooling performance has already declined or an abnormal condition has occurred.

eyc-tech provides a comprehensive range of measurement products for AI servers and data center liquid cooling systems, covering temperature, flow, pressure, differential pressure, liquid level, and local display. Measurement points can be configured according to different system sections, pipe sizes, and installation conditions, helping establish a measurement architecture extending from the facility cooling system to the liquid-cooled server racks. Continuous data recording and comparison across different sections can provide valuable information for anomaly identification, equipment maintenance, and system adjustment.

Facility Cooling System Measurement

The facility cooling system is located at the upstream end of the liquid cooling infrastructure. Cooling capacity is typically provided by chillers, cooling towers, or other cooling equipment and transferred to the heat exchanger through the primary-side cooling water loop. Changes in supply and return water temperatures, circulating flow, or water storage levels may affect downstream heat exchange performance and server-side cooling conditions. Temperature, flow, and liquid level are therefore important measurement parameters at the facility level.

TP05 Temperature Transmitter can be installed on supply and return water piping to continuously measure cooling water temperature. For larger-diameter cooling water mains, the FUM06 Ultrasonic Flowmeter can be used for flow measurement. If the system includes cooling tower basins, make-up water tanks, or other water storage equipment, the L051 Submersible Pressure Transmitter can also be applied for continuous level monitoring. Continuous measurement of temperature, flow, and liquid level provides a more complete view of cooling capacity, cooling water circulation, and make-up water conditions, serving as a measurement basis for subsequent equipment inspection and system adjustment.

Heat Exchange Section Measurement

The heat exchanger is located between the facility cooling system and the data center liquid cooling loop, transferring heat absorbed from the servers to the facility cooling water. Heat exchange performance is influenced not only by inlet and outlet temperatures but also by internal flow resistance and coolant circulation conditions. Temperature and differential pressure are therefore important measurement parameters in this section.

By installing TP05 Temperature Transmitter at the inlet and outlet piping of the heat exchanger, inlet and outlet temperatures can be continuously monitored. A P063 Differential Pressure Transmitter can also be used to measure the liquid differential pressure across the heat exchanger. Comparing temperature and differential pressure trends helps monitor changes in heat exchange conditions and fluid resistance over time. Abnormal temperature differences or differential pressure trends can serve as references for heat exchanger inspection, cleaning, and preventive maintenance.

Coolant Distribution Unit Measurement

The coolant distribution unit (CDU) is a key piece of equipment connecting the facility cooling system with the server liquid cooling loop. A CDU typically integrates a heat exchanger, circulation pumps, filters, reservoir, and supply and return piping. Monitoring temperature, flow, pressure, differential pressure, and liquid level is therefore essential for establishing a complete measurement foundation for CDU operation.

On the CDU supply and return piping, TP05 Temperature Transmitter can be paired with the DPM03 Multifunction Signal Display Monitor for temperature measurement and local display. Depending on system requirements, DPM03 can also provide alarm, signal processing, and retransmission functions. For secondary-side or smaller-diameter coolant piping, the FUM03 Clamp-on Ultrasonic Flow Meter can be used to monitor circulation flow. A P048 Universal Pressure Transmitter can be installed at the pump outlet to measure supply pressure, while a P063 Differential Pressure Transmitter can be installed across the filter to monitor changes in flow resistance. The L051 Submersible Pressure Transmitter can also be used to monitor coolant level in the reservoir. By comparing multiple measurement parameters, operators can obtain a more complete view of CDU circulation, filtration, coolant supply, and reservoir conditions.

Supply and Return Main Measurement

After leaving the CDU, coolant is distributed through the main supply to different areas and server racks. After absorbing heat, the coolant returns to the CDU through the return main. Because these mains carry the primary coolant flow throughout the liquid cooling system, monitoring main flow is an important method for understanding overall circulation conditions.

For larger-diameter supply and return mains, the FUM06 Ultrasonic Flowmeter can be used for flow measurement, with the resulting data compared against existing temperature and pressure measurements. When main flow decreases, operators can further inspect the pump, filter, valves, or piping for changes in resistance. If main flow remains normal while cooling conditions in a specific area become abnormal, measurement data from downstream branches or server racks can then be compared to help narrow the scope of inspection.

Branch Flow Monitoring

Large data centers typically distribute coolant from the main piping into multiple areas or individual server racks. Normal total flow in the main piping does not necessarily mean that every branch is receiving sufficient coolant. When flow decreases in a single branch, the resulting change in total main flow may be relatively small. Branch-level flow measurement therefore provides greater visibility into coolant distribution between different areas.

The FUM03 Clamp-on Ultrasonic Flow Meter is suitable for smaller-diameter liquid cooling branch piping. Its external clamp-on measurement method reduces the need for pipe cutting, system shutdown, and coolant leakage during installation. In existing data centers, it can be applied when expanding systems, adding new measurement points, balancing flow, or performing maintenance inspections. Comparing flow data among individual branches helps verify coolant distribution across different areas and server racks.

Liquid-Cooled Server Rack Measurement

After coolant enters a server rack, it is distributed through the supply manifold to cold plates installed on CPUs, GPUs, and other high-heat components. After absorbing heat, the coolant flows through the return manifold and back to the main piping. Because this section is closest to the actual server heat sources, individual racks may still experience insufficient flow due to valve position, piping resistance, manifold conditions, or changes in cold plate flow paths, even when CDU and main flow remain normal. Flow velocity and temperature are therefore important measurement parameters at the rack level.

The FTC06 Temperature and Flow Transmitter can be installed on supply and return manifolds or other critical liquid cooling branches and paired with the DPM04 Flow Totalizer. The FTC06 simultaneously measures coolant flow velocity and temperature, while the DPM04 provides local display of instantaneous and totalized flow, together with alarm, signal processing, and retransmission functions. Where additional supply and return temperature monitoring is required, TP05 Temperature Transmitter can also be installed for temperature differential comparison. Cross-referencing flow velocity, flow rate, and temperature data provides a useful measurement basis for inspecting piping, valves, distribution manifolds, and cold plate circulation conditions.

From Facility Cooling to Server Racks: eyc-tech Measurement Solutions for AI Data Center Liquid Cooling

AI data center liquid cooling systems involve multiple sections, including facility cooling systems, heat exchangers, coolant distribution units, liquid cooling mains, branch piping, and server racks. eyc-tech provides measurement products covering temperature, flow, pressure, differential pressure, liquid level, and local display. Products can be configured according to pipe diameter, equipment location, and actual application requirements to establish a measurement architecture extending from the facility side all the way to the server rack.

In addition to local inspection and on-site value confirmation, measurement data can, depending on product specifications, be integrated into existing BMS, SCADA, PLC, DCIM, or data acquisition systems through analog signals, relay outputs, or RS-485 communication. Where local alarm, signal processing, or retransmission functions are required, corresponding display controllers can also be integrated according to the application.

By establishing continuous and traceable measurement points across different sections of the system, temperature, flow, pressure, differential pressure, and liquid level data can be further applied to anomaly comparison, trend monitoring, equipment inspection, and preventive maintenance, providing AI data center liquid cooling systems with a comprehensive and continuously traceable measurement foundation.