Technical guidePDF2019

Development of European Green Public Procurement Criteria for Data Centres - Preliminary report (JRC118550, EUR 29945 EN)

European Commission JRC · Felice Alfieri, Nicholas Dodd et al.

The JRC preliminary report that preceded the 2020 final technical report for the EU GPP data centre criteria. It covers scope and definitions, relevant standards and KPIs, an EU market analysis, a review of life cycle assessment (LCA) studies and life cycle cost structures, and identifies the environmental hotspots that later became criteria.

CC BY 4.0 — share with attribution

On this page

Overview

The report presents the first three tasks (product scope, market analysis, environmental and technical analysis) of a study carried out by the JRC with a consultant consortium led by Viegand Maagoe for the Commission's Directorate-General for the Environment; the fourth task is the draft criteria proposal. The aim is clear and ambitious GPP criteria based on a life-cycle approach and scientific evidence. The report has 13 chapters and includes stakeholder survey results. It reviews statistical classifications (Prodcom, NACE 63.11, CPV), legislation and standards (ecodesign, WEEE and RoHS, ETS, EN 50600, ISO/IEC 30134, ISO 14001 and 50001, ETSI GS OEU 001, ENERGY STAR, EU Code of Conduct, Blue Angel, EMAS, EURECA).

The scope proposal divides data centres into three types: enterprise, colocation and managed service provider (MSP). The proposed scope covers IT equipment (servers, storage, network), power conditioning (transformers, UPS) and white-space mechanical systems (CRAC/CRAH), and the system as a whole or a managed service; the building fabric is excluded. Market data puts about 96% of EU data centres in the enterprise segment, but only about 57% of white space (colocation 40%, MSP 3%); average white space is 60 m2 for enterprise, 1,152 m2 for colocation and 1,123 m2 for MSP. The enterprise count includes centres with at least 25 kW IT capacity and excludes server rooms.

The energy chapter puts EU data centre consumption at 55 TWh in 2010 and 74 TWh in 2015, rising on internal projections to 104 TWh in 2020 and 160 TWh in 2030, with annual growth slowing from 9% to 3%. Enterprise centres account for 62% of consumption in 2015 and 22% in 2030, while MSP centres rise from 26% to 62%. The IT share is assumed at 52% in 2010 and 67% in 2030, with a PUE of 1.5 used for 2030. The report says M&E improvement is more achievable and less disruptive to technology and resilience, while IT utilisation improvement (consolidation, virtualisation) is more complex but has high potential.

The LCA review covers ten studies; results are not fully comparable but hotspots are similar. One study attributes about 67% of global warming potential to the use stage and 33% to procured assets (production, transport, waste management), with the IT platform contributing about 67% of GWP and the facility (M&E and building) about 37%. The priority order is electricity mix, IT efficiency and utilisation, M&E consumption, equipment manufacture, and water discharged from cooling. Best practices such as virtualisation in enterprise data centres give about 15x reduction against the worst case and about 7x against the average. Because adiabatic cooling can raise water use, water is proposed for inclusion in the criteria.

The life cycle cost chapter gives separate structures for owners and customers: MSP customers pay no CAPEX but much higher OPEX, and colocation redundancy and resilience drive high CAPEX. According to BSRIA, building services installation CAPEX for data centres is about GBP 8,300/m2 (about EUR 9,400/m2), of which 62% is electrical installation and 17% space heating, cooling and air treatment. The report is an unfinished preliminary study: the number of data centre facilities could not be estimated directly at EU level, future projections remain uncertain and the criteria themselves were developed in later stages. It also notes that the perceived trade-off between reliability and sustainability can sustain poor legacy practice.

Key points

  1. About 96% of EU data centres are enterprise but they hold only about 57% of white space; colocation holds about 40% and MSP about 3%.
  2. Average white space is 60 m2 for enterprise, 1,152 m2 for colocation and 1,123 m2 for MSP; the enterprise count includes centres with at least 25 kW IT capacity.
  3. EU data centre energy consumption is 55 TWh in 2010, 74 TWh in 2015, and projected at 104 TWh in 2020, 134 TWh in 2025 and 160 TWh in 2030.
  4. Enterprise share of consumption falls from 62% in 2015 to 22% in 2030; MSP rises from 26% to 62% and colocation from 12% to 17%.
  5. IT energy share is assumed at 52% in 2010, 57% in 2015 and 67% in 2030, with a PUE of 1.5 for 2030.
  6. One reviewed LCA attributes about 67% of GWP to the use stage and 33% to procured assets; the IT platform accounts for about 67% and the facility 37%.
  7. LCA priority order: electricity mix, IT efficiency and utilisation, M&E consumption, equipment manufacture, cooling water discharge.
  8. Best practices in enterprise data centres, including virtualisation, can cut carbon footprint about 8-15 times.
  9. Building services installation CAPEX (BSRIA) is about GBP 8,300/m2, 62% electrical and 17% heating, cooling and air treatment.
  10. The temperature of cooled air supplied must be distinguished from the temperature of air entering IT equipment, which is 1 C to up to 20 C higher.

Using it in design

  • Seed the embodied-versus-operational carbon module with the roughly one third procured assets and two thirds use-stage split as a default sanity check for LCA outputs.
  • Use the EU white space split and average white space per segment (60, 1,152, 1,123 m2) as market defaults in templates.
  • Define supply cooled air temperature and IT inlet air temperature as separate design variables in cooling design; the inlet is always higher (1 C to 20 C).
  • Plan air management (hot/cold aisle separation and similar) as the first step; the report names it as normally the most cost-effective starting point for energy efficiency.
  • When evaluating adiabatic cooling options, add water consumption and discharge to the evaluation output, because energy-efficient designs can raise water use.
  • Keep this document as background for the GPP criteria checker (see eu-gpp-criteria-data-centres-swd-2020-55).

Monthly newsletter · free

Keep your reading list current

Once a month: new library sources, changes to Turkish regulations and upcoming data centre events.

SoSo CAD

Compare the cost of design decisions

Evaluate equipment costs, energy expenses and total cost of ownership (TCO) using your inputs and assumptions in SoSo CAD.

Request a demo
CAPEX and TCO estimates in the SoSo CAD cost workspace

Who should read it

Engineers, public procurement specialists and policy analysts reviewing the market, energy and LCA evidence behind the EU GPP data centre criteria.

Document contents

  1. Background
  2. Review of existing scope, product types, product classification and metrics
  3. Product classification in statistical sources (Prodcom, NACE, CPV)
  4. Scope, categorisation and metrics in existing legislation, standards, voluntary agreements and procurement criteria
  5. Definition of product and service scope
  6. Data centre types
  7. Proposed definition and scope
  8. Procurement practices
  9. Key actors in data centres value chain
  10. Life Cycle Assessment (LCA) of Data Centres
  11. System-oriented analysis
  12. Identification of priority improvement options for products and services
  13. Data Centre energy consumption in the EU market
  14. Data Centres market volumes
  15. Technical state-of-play of data centres
  16. Key performance indicators (KPIs) and overview of metrics
  17. Life Cycle Costs structure of data centres
  18. Main outcomes from stakeholders' survey in relation to proposed criteria

Related standards

  • EN 50600-4
  • ISO/IEC 30134
  • ETSI ES 205 200
  • ETSI GS OEU 001

Read the document

jrc-eu-gpp-data-centres-preliminary-2019.pdf

PDF · 3.3 MB · English

CC BY 4.0 — share with attribution

The document is provided unaltered because its licence permits redistribution.

Publication details

Authors
Felice Alfieri; Nicholas Dodd; Miguel Gama-Caldas; Oliver Wolf; Larisa Maya-Drysdale; Baijia Huang; Jan Viegand; Sophia Flucker; Robert Tozer; Beth Whitehead
Publisher
European Commission JRC
Year
2019
Document number
EUR 29945 EN / JRC118550
Document language
English
Source type
Government / public
Access
Open licence — downloadable
Licence
CC BY 4.0 — share with attribution
Licence evidence
op.europa.eu
Original licence note

Copyright page (PDF p.2): Commission reuse policy, Decision 2011/833/EU; reuse authorised under CC BY 4.0 except where otherwise noted.

Citation

Felice Alfieri; Nicholas Dodd; Miguel Gama-Caldas; Oliver Wolf; Larisa Maya-Drysdale; Baijia Huang; Jan Viegand; Sophia Flucker; Robert Tozer; Beth Whitehead (2019). Development of European Green Public Procurement Criteria for Data Centres - Preliminary report (JRC118550, EUR 29945 EN). European Commission JRC. EUR 29945 EN / JRC118550. https://doi.org/10.2760/327087

All content and rights belong to the respective authors and publishers. SoSo CAD provides only the citation, its own summary and a link to the source, and is not responsible for the sources' content.