Skip to Content
SIL3X
  • Work with us
    • Our skills
    • Our tools
    • Support services
    • Trainings
  • Our activities
    • Nuclear
    • Datacentres
    • Industry
  • News
    • Articles
    • Blogs
  • About us
    • The team
    • Contact us
    • Join us
    • Terms and Conditions
  • English (UK) Français
  • Sign in
SIL3X
      • Work with us
        • Our skills
        • Our tools
        • Support services
        • Trainings
      • Our activities
        • Nuclear
        • Datacentres
        • Industry
      • News
        • Articles
        • Blogs
      • About us
        • The team
        • Contact us
        • Join us
        • Terms and Conditions
    • English (UK) Français
    • Sign in

    PUE-WUE Calculations

    Evaluate and optimise the energy and water performance of data centres, from design to operation.

    Contact us

    Assessment of the electrical and water consumption of data centres

    PUE and WUE

    The indicators of data centres

    ​

    Power Usage Efficiency (PUE)

    The PUE represents the energy efficiency of a data centre, it represents the ratio between the energy consumed by the entire data centre (IT, cooling, security, lighting, etc ...) and the energy consumed to operate the servers. 

    PUE = Total electrical energy / IT electrical energy

    Note: The PUE can be seen as an integral quantity (comparing energy amounts) or instantaneous (comparing powers).

    The closer the PUE is to 1, the more efficient the data centre is from an energy perspective. The average in France is 1.55 and 1.36 in Europe [1]. It is currently considered that a data centre with a PUE of less than 1.3 is a high-performing site. However, it should be noted that the PUE is not an economic indicator that takes into account electricity market prices or a carbon impact indicator that considers the source of electricity (nuclear, coal, solar, ...). It is simply an indicator of energy performance.

    Some jurisdictions are beginning to emerge to compel data centres to operate more optimally. In China, the "Three-Year Action Plan on New Data Centers" sets as the goal a PUE < 1.3 for new data centres . Three different ratings of MEPS (Minimum Energy Performance Standard) are assigned to data centres ranging from a PUE < 1.2 (Grade 1) to a PUE < 1.5, (Grade 3). In Germany, new data centres must, since 1st July 2026, achieve a PUE < 1.2 and reuse 10% of their energy. Existing data centres must reduce their PUE to below 1.5 before 1st July 2027 and below 1.3 before 1st July 2030 [2].

    One of the major challenges for data centre operators and designers is therefore to reduce the PUE as much as possible.

    Cooling

    The vast majority of the difference between the energy consumed by the data centre and the energy actually dedicated to the operation of the servers comes from the cooling of these servers. The main lever for reducing the PUE of data centres is therefore the consumption of cooling equipment. This includes the consumption of pumps circulating cold water, fans, or even the compressors of the cooling units. 

    There are two main ways to produce the cold needed for data centres:

    • Air-cooled chillers
    • Water-cooled chillers 

    In the case of a water-cooled chiller, it is quite rare to pump water directly from a cold source (sea, river) and the most common architectures are those combining a water-cooled chiller with a rooftop air exchanger. These air exchangers can be of different types: 

    • Dry cooler
    • Cooling tower
    • Hybrid Dry Adiabatic Cooler (HDAC)

    In order to minimise energy consumption associated with cooling production, some data centres integrate into their chilled water production modes of free chilling cooling (cf free chilling article)

    Water Usage Efficiency (WUE)

    Just as the PUE indicates the energy performance of a data centre, the WUE is an indicator of the water consumption of a data centre.Water consumption in a data centre can come from several sources (from the most significant to the least significant):

    • Water evaporated by evaporative cooling technologies
    • Water consumed during the production of osmosis water
    • Water discharged after multiple evaporation cycles (too high salt concentration)
    • Air humidification by Air Treatment Units (ATU)
    • Sanitary uses on site

    The counted water is only the water evaporated or returned to the sewers. Water taken from a river or the sea and then discharged warmed by a few degrees does not fall within the scope of the WUE.

    The WUE is therefore calculated as follows:

    WUE [l/kWh] = Volume of water consumed / IT Energy

    Note: In the same way as for the PUE, an instantaneous WUE can be calculated by taking the ratio of the flow rate of water consumed to the IT power.

    The average WUE in France is 0.25 and 0.58 in Europe. [1] but one must be cautious with averages as they include data centres not using evaporative means. The data centre operator EQUINIX clearly mentions this difference by indicating that on average its data centres have a WUE of 0.95 l/kWh but that the average of data centres using evaporative cooling is 1.55 l/kWh [3]. There is not yet any regulation imposing a limit on water consumption for data centres, but in light of the growing tensions on the issue of water stress, it cannot be ruled out that limitations may begin to appear. 

    Arbitration between WUE and PUE

    In view of what has been mentioned previously, it is clear that PUE and WUE are not independent indicators and that they are to be treated together. A data centre with a very low PUE due to having an evaporative cooling system may potentially have a greater environmental impact than a data centre consuming more electricity but no water. These arbitration questions between electrical and water consumption vary also throughout the seasons, with tensions on the water resource appearing more in summer in temperate regions. A data centre that is able to best adapt its water and electricity consumption to the available resources will be the one with the least environmental impact.

    The SIL3X method

    Model, simulate and quantify

    Digital twins

    SIL3X develops digital twins in Modelica modelling the behaviour of cold production installations. These digital twins model all the important equipment impacting the electrical and water consumption of the installation: 

    • Pumps
    • Fans
    • Heat exchangers (dry and adiabatic)
    • Valves (thermostatic, two-way valves, etc...)
    • Chillers
    • External conditions (temperature, relative humidity)

    The control system associated with this equipment is also reproduced and integrated into the models. 

    Operational

    By integrating climatic data representative of the site and the operating profiles of the equipment, we simulate the behaviour of the system over a full year. The electrical and water consumption are then calculated taking into account the external conditions, the loads of the data centre and the regulation strategies, in order to estimate the PUE and the WUE over the year (instantaneous, monthly or annually). These digital twins therefore initially allow us to model the existing installation and to analyse in depth the operation of the installation. 

    Moreover, with the aim of predicting the performance of an installation, it is possible to subject these models to future climatic scenarios in order to predict the future performance of the data centre.

    The graphs below present the results of a annual simulation of a site using a HDAC and having an architecture that allows it to operate in free chilling part of the year.  

    The first graph shows the instantaneous PUE and thus the PUE peak which reflects the peak consumption of the site (at constant IT power). It can be observed that the PUE varies throughout the year depending on temperatures, reaching maximums during the summer period due to higher fan consumption of the HDAC in particular.

    The second graph overlays the results obtained for a monthly PUE and WUE and shows the joint increase in electrical and water consumption of cooling production during the summer period. The average PUE only increases in very small proportions (1.11 to 1.15) mainly thanks to the water consumed, which is illustrated by a WUE rising from nearly 0 in winter to a peak of around 2.3 l/kWh in summer. This graph well illustrates the duality of the two indicators mentioned above. 

     


    Optimisation

    The models can be used in the optimisation phase of an existing installation by allowing to test modifications (replacement of equipment, control modifications, change of architecture, etc...) in order to quantify the impact over a year of operation.

    Design

    Finally, it is also possible to develop models in the design phase in order to address concrete issues regarding the choice of equipment, the sizing of equipment or regulation strategies.

    Any needs?

    Are these topics of PUE and WUE optimisation a significant challenge for you? Do not hesitate to contact us to discuss your issues and see together how we can address them.

    References

    [1] "Assessment of the energy performance and sustainability of data centres in EU", European Commission, July 2025

    [2] "Policy development on energy efficiency of data centres", Flona Brocdehurst, Ballarat Consulting, February 2024

    [3] https://blog.equinix.com/blog/2026/03/19/a-guide-to-responsible-water-use-in-data-centers/

    How can we help?

    Contact us anytime

    Call us

    +33 6.22.51.73.26

    Send us a message ​

    contact@sil3x.fr

    Suivez-nous


    • Home
    • •
    • About us
    • •
    • Services
    • •
    • Terms of Service
    • •
    • Privacy Policy
    Copyright © SIL3X
    English (UK) Français
    Powered by Odoo - Create a free website

    We use cookies to provide you with a better user experience on this site.Cookie policy

    What are essentialsI agree