Data Center Energy Efficiency Measures: How Sensor Precision Improves PUE and Cuts Costs
Data centers currently consume around 485 TWh of electricity worldwide each year. According to an IEA report from 2025, this figure could reach 950 TWh by 2030. Artificial intelligence is one of the main drivers of this increase.
Companies looking for data center energy efficiency measures should not focus only on IT hardware. Cooling offers the greatest short-term potential. Depending on the location, utilisation and cooling concept, it can account for up to 25% of total facility energy consumption.
Precise and reliable measurement data enables targeted control of energy efficiency in data center cooling without compromising availability.
In this article, you will learn:
- What PUE and WUE mean and why they are important metrics.
- Which measurement points in the data center have the greatest impact on energy consumption.
- What precise sensor technology can achieve in practice.
What are the key metrics for data center energy efficiency measures?
What is PUE, and why is it the central energy efficiency KPI?PUE stands for Power Usage Effectiveness. The formula is straightforward: PUE = Total facility energy consumption / IT equipment energy consumption Consider a data center with a 10 MW IT load. If its PUE decreases from 1.5 to 1.2, the required total power decreases from 15 MW to 12 MW. At an electricity price of €0.12 per kWh and with year-round operation, this represents a calculated annual saving of approximately €3.15 million. What is WUE, and when is it relevant?WUE stands for Water Usage Effectiveness and describes a data center's site-specific water consumption in relation to its IT energy consumption. It is relevant when evaporative cooling or free cooling via cooling towers is used. Cooling capacity and the available free-cooling potential are largely determined by the wet-bulb temperature of the outside air. Precise and long-term stable measurement of outdoor temperature and humidity enables optimum control of cooling towers, free cooling and chillers. This supports data center energy efficiency measures and helps reduce energy consumption. |
What drives energy consumption in data centers?
1. Server inlet temperature
The temperature at the server rack inlet is one of the most important control points for data center energy efficiency measures.
According to the US Department of Energy and ENERGY STAR, each controlled increase of 0.5 °C in server inlet temperature saves approximately 4% to 5% of cooling energy.
The problem: For reasons of availability, the setpoint is often kept too conservative. This is understandable. A sensor with ±0.5 °C measurement accuracy requires a safety margin.
Highly accurate and long-term stable sensors allow the operating point to be moved closer to the permitted temperature limits without compromising operational reliability.
This requires suitable sensor placement and a careful assessment of the complete measurement system.
2. Airflow and containment
In data centers with raised floors, precise differential pressure measurement supports demand-based cooling air distribution.
The correct setpoint depends on the specific construction, air outlets, and required airflow volume. Deviations can lead to poor air distribution or bypass air.
This can cause cold air to bypass the servers without cooling them.
3. Humidification and dehumidification
Humidity control in a data center requires a balanced operating range. Air that is too dry can increase the risk of electrostatic discharge. Excessive humidity can increase the risk of condensation and corrosion on circuit boards and contacts.
Relative humidity changes when air temperature changes. Dew point, by contrast, describes the actual moisture content of the air. It can therefore provide a more stable control variable in many applications.
For air-cooled IT equipment in classes A1 to A4, ASHRAE recommends a dew point between -9 °C and 15 °C and a maximum relative humidity of 70% or 50%, depending on the level of corrosive contamination.
4. Outdoor climate
Outdoor temperature, relative humidity and wet-bulb temperature determine whether and for how long free cooling can be used. When outdoor conditions are favourable, mechanical cooling can be partially or fully switched off. Without continuous measurement, operators may miss usable free-cooling windows or switch too early, increasing the risk of condensation and corrosion.
An outdoor climate station can therefore make a substantial contribution to data center energy efficiency measures.
5. Cooling concept
Air cooling reaches its limits at rack power densities of around 30 to 50 kW. Liquid cooling and immersion cooling are increasingly taking over in these power classes.
Modern A1 racks with power densities above 100 kW are also increasingly using direct-to-chip liquid cooling. However, air cooling often remains in place for less heat-intensive components. Depending on the system, the following parameters may be monitored:
- Supply and return temperature
- Flow rate
- Pressure
- Conductivity
- Potential leaks
- Water content in dielectric cooling fluids
For dielectric cooling fluids, monitoring the water content in the liquid may also be relevant.
Efficiency and availability: Which measurement variables are critical?
The following table shows which parameters need to be monitored in data centers to reduce energy costs. It also highlights the risks that can arise from inaccurate measurement data. E+E Elektronik provides precise sensors for all the measurement points listed above.
| Measurement variable | Critical location | Why the measurement matters | Suitable E+E sensors |
|---|---|---|---|
| Temperature and relative humidity | Server rack inlet, measured at the top, center, and bottom | A measurement deviation of \pm0.5 °C can require an additional safety margin. Operators may accept overcooling to prevent overheating. | HTS401, EE072 |
| Dew point | Cold aisle and chilled water loop | GPU racks above 30 kW can create strong local cooling. If a surface falls below the dew point, moisture can condense directly on hardware. | EE046 |
| Differential pressure | Raised floor and cold or warm aisle | If pressure deviates from the setpoint, cooling air can bypass the servers. | EE610, EE600 |
| Air velocity | CRAC outlet and rack inlet | Excessive fan speed consumes a disproportionate amount of energy. Targeted airflow control can reduce consumption. | EE650 |
| Outdoor climate | Cooling tower and free-cooling economizer | Wet-bulb temperature determines whether free cooling or evaporative cooling can operate effectively. | EE260, EE08, EE451 |
How should sensors be positioned in a data center?
Two standards provide clear guidance on sensor placement.
- DIN EN 50600 and ISO/IEC 22237 require continuous monitoring of environmental conditions in data centers. For meaningful temperature monitoring, sensors should be positioned as close as possible to the air inlets of the IT equipment.
Multiple measurement points along the rack height can be useful when racks are high or IT loads vary significantly.
- ASHRAE recommends temperature and humidity sensors at the lower, middle, and upper sections of the rack inlet.
Consistent sensor placement helps operators detect hot spots before they cause failures while also enabling targeted data center energy efficiency measures.
What can precise sensor technology achieve? A calculation example
Initial situation: 10 MW IT load, PUE 1.5, electricity price €0.12/kWh
- The server inlet temperature is increased by 0.4 °C. This is possible when the sensor provides ±0.1 °C rather than ±0.5 °C measurement accuracy. The calculation is based on the conservative estimate of 5% cooling energy savings for each 1 °C increase in server inlet temperature.
| Annual energy saving | Annual cost saving | New PUE | CO_2-reduction |
|---|---|---|---|
| 675 MWh | Approximately €81,000 | 1,492 | 169 t CO_2 |
Source: The calculation is based on the LBNL/DOE guide and ENERGY STAR data.
The result: Over five years, a 10 MW IT load and a 0.4 °C increase in server inlet temperature can deliver potential energy cost savings of up to €400,000.
How to implement data center energy efficiency measures
Data centers consume more energy than ever before. The pressure to reduce costs while preventing outages is increasing. Yet energy efficiency and availability are not mutually exclusive. They share the same technical foundation: reliable measurement data at the right points and with sufficient accuracy.
If operators can increase the server inlet setpoint by 0.4 °C because they trust the sensor data, a 10 MW site can save around €80,000 per year. Measuring differential pressure in the cold aisle helps prevent bypass air. Monitoring the outdoor climate enables the use of free-cooling windows that would otherwise remain unused.
Data center energy costs can be reduced safely and systematically, but only with measurement data that supports targeted operational decisions. PUE and WUE then become more than abstract metrics. They become the direct result of deliberate decisions made in daily operation and form the basis for effective data center energy efficiency measures.
E+E Elektronik supplies highly accurate and long-term stable sensors for all critical measurement points in the data center, from the rack inlet to the cooling fluid circuit, with fully documented measurement traceability and seamless BMS integration.