Modern data centers must process enormous amounts of data. Cloud applications, artificial intelligence, high-performance computing, and data-intensive business processes require powerful processors, graphics processing units, and storage systems. However, as computing power increases, so does the amount of heat generated inside the servers. This heat must be reliably dissipated so that the technical components can operate within their specified temperature ranges.

This is precisely where thermal management comes into play. Thermal management encompasses all technical measures used to absorb, transport, regulate, and dissipate heat within a data center. These include traditional air cooling as well as modern liquid cooling, heat pumps, heat exchangers, pumps, coolant circuits, and the associated sensor technology. Temperature and pressure sensors from i2s can provide important measurement data for these systems, thereby supporting reliable monitoring and demand-based control of the cooling process.

Thermal management describes the entire path of heat, from its generation to its dissipation into the environment or its reuse.

Heat is generated primarily in electronic components such as:

• CPUs
• GPUs and AI accelerators
• RAM
• Power supplies
• Storage systems
• Network components
• Power supply and power electronics

This heat must first be absorbed by the individual components. It is then transported away from the server via air or liquid. From there, it travels through server racks, cooling circuits, and heat exchangers before finally reaching the data center’s central cooling system.

Electronic components operate reliably only within certain temperature ranges. If a processor gets too hot, it may automatically reduce its performance. This process is often referred to as thermal throttling. If the temperature continues to rise, protective mechanisms may shut down individual components or entire systems. Persistently high temperatures can also accelerate the aging of electronic components.

Effective thermal management therefore serves several purposes:

• reliable operation of IT systems
• prevention of overheating
• even temperature distribution
• demand-based cooling
• reduction of unnecessary cooling capacity
• early detection of unusual operating conditions
• protection of servers and infrastructure
• maximum energy efficiency

As power density increases in modern data centers, this task becomes more challenging. Artificial intelligence servers, in particular, contain high-performance GPUs and accelerators that can generate large amounts of heat in a small space. As a result, liquid cooling systems are increasingly being used alongside traditional air cooling.

The basic principle is simple: Heat is absorbed at one location and then transported to another.

A typical heat path might look like this, for example:

• A processor generates heat.
• A heat sink or cold plate absorbs the heat.
• Air or coolant transports it out of the server.
• A heat exchanger takes over the heat.
• Another cooling circuit transports it out of the data center.
• Heat pumps can be used to bring the heat to a different temperature level and make it usable for other applications.
• The heat is dissipated into the outside air or reused.

In practice, multiple cooling circuits can be interconnected. Especially in large data centers, this results in a complex interplay of pumps, valves, heat exchangers, cooling units, and sensors. To ensure this system operates reliably, key operating parameters must be continuously monitored.

In a liquid cooling system, the coolant must be reliably transported through pipes, heat exchangers, and heat sinks. Pumps generate the necessary pressure for this. At the same time, pipes, filters, valves, and cold plates create flow resistance.

Pressure sensors can help monitor these operating conditions.

For example, they can be used to:

• Monitoring system pressure,
• Checking pump operation
• Detecting unusual pressure fluctuations
• Monitoring individual cooling loops
• Checking heat exchangers
• Detecting increasing pressure losses
• Monitoring filters

Since pipes, connections, heat exchangers, and seals are designed for specific pressure ranges, pressure monitoring is an essential part of safe system operation.

A sensor in a cooling system must provide reliable readings over a long period of time. It may be in constant contact with coolants and exposed to fluctuating temperatures and pressures. Therefore, several factors are important when selecting a sensor.

Media Compatibility
All parts of the sensor that come into contact with the cooling medium must be compatible with that medium. Relevant examples include water, deionized water, water-glycol mixtures, and refrigerants used in heat pump and refrigeration systems. Not only the fluid itself, but also its concentration, temperature, and pressure influence the choice of materials.

Fast Response Time
Temperatures within a cooling system can change dynamically. As server load increases, heat generation can also rise rapidly. Fast temperature measurement enables the control system to detect changes in a timely manner.

Accuracy
The required accuracy depends on the specific application. A different level of accuracy may be sufficient for simple monitoring than for precise control. It is important that the sensor delivers stable readings across the entire intended operating range.

Long-Term Stability
Data centers operate around the clock. Sensors should therefore provide reliable and reproducible readings even over long periods of operation. High long-term stability also makes it easier to detect long-term changes.

Compact Design
Space is often limited inside server racks, CDUs, and coolant distributors. Compact sensors are easier to integrate into existing systems.

Electrical Interfaces
The measured values must be transmitted to a control system. Depending on the application, different signals and interfaces may be required. Electrical connections and connectors can also be critical for integration.

A modern cooling system should, as much as possible, provide only as much capacity as is actually needed. If pumps, fans, and refrigeration systems operate continuously at maximum capacity, energy may be wasted. On the other hand, reducing cooling capacity too much can lead to elevated temperatures. Sensors provide the foundation for controlled operation.

Based on the measured values, control systems can, for example:

• Increase or decrease pump speeds
• Open or close valves
• Activate cooling loops
• Regulate heat exchangers
• Adjust fan speeds
• Trigger warning messages

The sensor itself does not control these components. However, it provides the information required for intelligent control.

Cooling is one of the major sources of energy consumption in a data center. Therefore, a key goal of thermal management is to provide the necessary cooling capacity as efficiently as possible. Pressure and temperature measurements can help provide a better understanding of the actual operating conditions. Instead of operating with large safety margins, a system can be controlled based on real-time measurements. Sensors do not directly save energy, but they do enable more precise control of the systems.

Heat pumps can be used in heat management to transfer heat specifically between different temperature levels. To do this, they utilize a refrigerant cycle in which pressure and temperature are closely related. For efficient and reliable operation, these operating parameters must therefore be continuously monitored.

Pressure and temperature sensors provide important measurement data for controlling the refrigerant circuit and can be used, for example, to monitor compressors, evaporators, and condensers. With its CCT pressure/temperature sensors, i2s offers combined sensor solutions for heat pump applications. Our pressure/temperature sensors are also available in a stainless steel version, which ensures reliable long-term operation even when used with corrosive media.