Project No: 649972 Title: Data centre EURECA Project (EURECA) · 2018-09-15 · The data centre...

DocumentRef:EURECA-DEL-5.1-maki Issue:1.0 Date:2015-10-30 Page0of54

EURECA Project Document Use or disclosure is subject to the restrictions on the first page.

0

ProjectNo:649972

Title:DatacentreEURECAProject(EURECA)

Call:H2020-EE-2014-3-MarketUptake

DeliverableNo: D5.1

DeliverableTitle: Report on the evaluation method formeasuring the energy savings andenvironmental benefit of the project viainnovativepublicprocurement

LeadBeneficiary: makiConsultingGmbH

DisseminationLevel

PU Public PU



1

ChangeHistory

Issue Section ChangeDescription Author Date

0.1 All First draft issue, structure,conceptperchapter Marc-AndreeWolf 28/09/15

0.15 6 Insertedpartsofdraftchapter6 FrankVerhagen 18/10/15

0.2 5 Insertednotesonchapter5 KiranaWolf 18/10/15

0.25 annexInserteddraftannexonautonomousdevelopmentandsavingspotential

FrankVerhagen 18/10/15

0.3 4 Inserteddraftchapter4 Marc-AndreeWolf 18-20/10/15

0.35 5 Inserteddraftchapter5 JohnBooth 20/10/15

0.4 7 Inserteddraftchapter7 JohnBooth 22/10/15

0.45 5 Replacedtextwithrewrittenchapter5 KiranaWolf 23/10/15

0.5 3 Inserteddraftchapter3 Marc-AndreeWolf 24/10/15

0.55 6,7 Furtherdevelopedchapter6and7 JohnBooth 27/10/15

0.6

Execsum,

references

Insertedexecutivesummary,preparedReferenceschapter

KiranaWolf 27/10/15

0.9 6,all

Furtherdevelopedchapter6,finalisedallchapters,formatting,internalcross-references

Marc-AndreeWolf 27-28/10/15

1.0 All Finalread-throughandfine-tuning RabihBashroush 30/10/15



2

Contents

1 EXECUTIVESUMMARY..................................................................................................61.1 General..................................................................................................................61.2 Purpose,scopeandapproachofthisdocument.....................................................6

2 DEFINITIONOFTERMSANDACRONYMS..................................................................................83 EURECAUPTAKERATES...................................................................................................113.1 Purpose...............................................................................................................113.2 Approach.............................................................................................................113.3 UptakerateEURECA-runtrainings,workshops.....................................................113.4 UptakerateforusingEURECAdevelopmentsoutsidetheproject........................12

4 EURECA-TRIGGEREDENERGYANDENVIRONMENTSAVINGS.......................................................144.1 Purpose...............................................................................................................144.2 Introductionandstatusquo.................................................................................144.3 Approach.............................................................................................................194.4 Methodology.......................................................................................................22

4.4.1 Calculated and expected net savings of EURECA-supported (or avoided)procurementsandself-improvementsbyEURECApartnersinitiatedduringtheproject.. ............................................................................................................................224.4.2 ExpectednetsavingsduetoEURECA-runtrainings............................................244.4.3 ExpectednetsavingsusingEURECAdevelopmentsoutsidetheproject............25

5 EURECACOST/BENEFITANALYSIS.......................................................................................285.1 Purpose...............................................................................................................285.2 Introduction.........................................................................................................285.3 Approach.............................................................................................................295.4 Methodology.......................................................................................................30

6 INCREASEDSKILLSOFTHEPROCURERS...................................................................................337 INNOVATIVESOLUTIONSUPTAKE..........................................................................................378 REFERENCES/SOURCES....................................................................................................399 ANNEX..........................................................................................................................439.1 ‘Autonomous’developmentandimprovementpotentialforenergyefficiencyandenvironmentalperformance.........................................................................................43

9.1.1 Hardwareenergyefficiency................................................................................439.1.2 Traditionalcoolinghardware..............................................................................449.1.3 Innovativecoolingmethodsandhardware........................................................449.1.4 ICThardware(servers)........................................................................................45



3

9.1.5 Datacentreefficiencyingeneral........................................................................469.1.6 ThePUEasaproxyfordatacentreefficiency.....................................................479.1.7 Limitations of the PUE for comparisons and for energy and environmentalimpacts/savings...............................................................................................................489.1.8 SpeedofPUEimprovements..............................................................................499.1.9 Energyefficiencyandreuse................................................................................509.1.10 Useofpowerindatacentres...............................................................................519.1.11 TriasEnergetica..................................................................................................519.1.12 Conclusions.........................................................................................................52



4

Figures

Figure1 ProjectionofGlobalandUSdatacentreelectricityuse,bySRIAnalystofAllianceTrustInvestment,Parker(2013).............................................................................................15

Figure2 Best, expected, and worst case electricity consumption during use of datacentresworldwidebyAndrae&Edler(2015).........................................................................16

Figure3 UsephaseelectricitysavingseffectwhenchangingfromPUE1.8to1.2.Source:Future-Tech:http://www.future-tech.co.uk/data-centre-energy-efficiency/........................17

Figure4 Schematic viewof life cycle of a server, focussing on production andprimaryenergyuse,ontheleveloftheinventoryresults,forillustration..........................................20

Figure5 SavingsofprimaryenergyandenvironmentalimpactduetothevariousEURECAmeasures;illustrative.............................................................................................................22

Figure6 illustrates life cycle based calculation of net benefit of EURECA-supported (oravoided)procurementsinitiatedduringprojecttimeplan,bytheEURECApartners;withtheexampleofprimaryenergy.....................................................................................................23

Figure7 Illustration of environmental Life Cycle Costing, and its components TCO andexternalcosting.......................................................................................................................30

Figure8 Typical profile of energy consumers in a data centre, contribution to overallkWh/year(Source:42U,2015)................................................................................................43

Figure9 GoogleactivelypromotesitsprogressonreducingitsPUE(Google2015)........45

Figure10 Moore’s Law of increased computational performance with time.(http://phys.org/news/2011-09-koomeys-law-power-efficiency-parallels.html#jCp)............46

Figure11 PUEandDCiEasproxiesfordatacentreenergyefficiency(GreenGrid2012)..48

Figure12 PUE values’ change with time, based on the survey participants’ self-assessment (Uptime Institute, 2014). From Bron :http://www.datacenterknowledge.com/archives/2014/06/02/survey-industry-average-data-center-pue-stays-nearly-flat-four-years/................................................................................50

Figure13 Globaldatacentreenergydemandforecastof2013(DatacenterDynamics2013) ............................................................................................................................51

Figure14 TheTriasEnergeticaconcept;seetext(basedonKorbeeetal.1979)...............52



5

Tables

Table1 Thedefaultcostforemissions(EuropeanCommission,2015c)...............................32

Table2 Equipment efficiency improvement of high capacity water chillers(ANSI/ASHRAE/IESNA2004,2007,2010)COP=coefficientofperformance, IPLV=IntegratedPartLoadValue.......................................................................................................................44



6

1 EXECUTIVESUMMARY

1.1 GeneralThedatacentreindustryhasasustainedandhighgrowthrate:Astudypredictsthatthedatacentrerevenueacross19oftheEU28countrieswillincreasefrom3.2billionEuroperannum(2010) to 7.3 billion Euro (2015), i.e. at an annual growth rate of 25% (Telecom pricing,accessed2014),(studypredictsin-housedatacentre,accessed2015).However, the rise of the ICT sector and the data centre industry comes with increasingenvironmental pressures, despite of growing IT performance, the growth is outpacing thetechnologicalprogress(Wolf,2014).ThekeyobjectiveofEURECAistosupportenergy/resourceefficientandenvironmentallysoundprocurementactionswithin theEuropeanPublic Sector fordata centreand relatedproductsandservices.

1.2 Purpose,scopeandapproachofthisdocumentThe overall purpose of this deliverable is to describe approaches and methodologies forquantifyingsavingsofprimaryenergyandofvariousenvironmentalimpactsaswellascostsavings, triggered by innovative public procurement using EURECA‘s tool and otherdeliverables.ThemethodswillbeusedinD5.3toevaluatetheproject’scontributiontothecalltargets,aswellasEU2020targetsandfurtherbenefits.

TheDoW (Description ofWork) specifies the following activities to be performed for this“Report on the evaluation method for measuring the energy savings and environmentalbenefitoftheprojectviainnovativepublicprocurement”withinworkpackage5:

- To develop a methodology for quantifying captured energy and environmentalsavingtriggeredbyEURECAproject

- To develop a methodology that captures economic savings triggered by EURECAproject

Thishasbeentranslatedintothefollowingapproach:

- Taking intoaccount the finding fromD1.1,D5.1buildson the frameworkandtooldevelopedunderWP2, incombinationwithestimatesforthelikelyuptakerateofthe EURECA developments among procurers. It presents the methodology forquantifyingprimaryenergyandenvironmentalsavingstriggeredbyEURECAprojectbothduringandaftertheprojecttime.Thiswill,nexttothedatacentreusephase,takeintoaccounttheproductionandend-of-lifeoftherelevantdatacentrecapital



7

goods ina lifecycleperspective. Itwill covernext toprimaryenergy, theshareofrenewable energy and other environmental benefits including avoided climatechangeimpacts.

- This reportalsoprovides themethodology that captures cost savings triggeredbythe EURECA project in a TCO perspective, with complimentary external costinformation,providinganoverallLCCinformation.ThisequallybuildsontherelatedlifecyclecostingmethodandtoolsupportdevelopedunderWP2.

- The increasedskillsofpublicprocurerswillbe judgedalong insightsand feedbackfromthenetworkingandtrainingtopublicsectorstaffinthepilotcasesandthroughtraining,whichbuildonthefindingsinWP1andWP4duetoEURECA.AnapproachtojudgetheEURECAtoolsuccesswillbeprovidedaswell.

- Market uptake of innovative data centre solutionswill be evaluated equally frominsightsoffirstprocurementspecificationsandcasesduringtheproject.



8

2 DefinitionofTermsandAcronymsC3ITstandsfortheEURECApartnerCarbon3ITLtdCBAstandsforCostBenefitAnalysisCGstandsfortheEURECApartnerCertiosB.V.

D1.1meanstheDeliverableidentifiedasnumberD1.1withinWorkPackage1oftheEURECAprojectDeliverablemeansaformalcontractdeliverableitemundertheEURECAprojectDoWmeansDescriptionofWork.TheEURECAprojectsignedaprojectagreementidentifiedasprojectnumber649972foraprojectunderthecallH2020-EE-2014-3-MarketUptake.Thisdocument contains a tablewithwork plans, and it is this information towhich this tablerefersECmeanstheEuropeanCommissioneLCCstandsforenvironmental(orexternal)lifecyclecosting

Energy Usage Effectiveness = Annual Energy Consumption Data Centre [kWh] / AnnualEnergyConsumptionIT-Equipment[kWh]ETSImeansEuropeanTelecommunicationsStandardsInstituteEUCoCmeansEuropeanCodeofConductforenergyefficientdatacentresEUEmeansEnergyUsageEffectivenessEURECAmeanstheDatacentreEURECAProject

EnvironmentallySound stands for“A lowoverallenvironmental impactperprovidedDataCentreservice (computation/dataservices)basedonpresentdayavailablesolutions.”This‘environmental impact’ includes impacts such as climate change, acidification, particulatematter,etc.butalsoprimaryenergyconsumptionandwaterscarcity.GITAstandsfortheEURECApartnerGreenITAmsterdamGreenstandsfor:see‘EnvironmentallySound’GHGstandsforGreenHousGas(ses)

(Procurement) Scenario(s) provides an indication of the scenario the Public Sector bodyshouldinitiateatenderforthatmeetstheactualprocurementneed(relatedtodatacentreproducts or services). By providing an assessment to determine the actual needs, theEURECA framework and tool can help establish the right Procurement Scenario fortendering.ITTstandsforInvitationtoTenderIndustrystandsfordatacentreandrelatedICTindustry



9

eLCCstandsfor(environmental)LifeCycleCostingLCAstandsforLifeCycleAssessmentMAKIstandsfortheEURECApartnermakiConsultingGmbHPCPstandforPre-CommercialProcurementorPre-CompetitiveProcurementPPIstandsforPublicProcurementofInnovativesolutionsPracticestandsfortheuseofastandard,framework,guideline,specificationorKPI/metricPUmeansPublicdisseminationlevelPUE means Power Usage Effectiveness, (Total facility Power [kW] / IT-Equipment Power[kW])RFIstandsforRequestforInformationRFQstandsforRequestforQuotationRFPstandsforRequestforProposalTask 1.1 stands for the first task as described in the EURECA project’s DOW underWP1,consistingofaRegionalanalysisofgreendatacentreprocurement.

Task1.2standsforthesecondtaskasdescribedintheEURECAproject’sDOWunderWP1,consisting of a SWOT analysis of existing procurement of environmentally sound datacentresandofrelatedproductsandservices.

Task 1.3 stands for the third task as described in the EURECAproject’sDOWunderWP1,consisting of a GAP analysis between existing procurement and environmentally soundprocurement.

Task 2.1 stands for the first task as described in the EURECA project’s DOW underWP2,consisting of comprehensive framework that can guide procurers in the process ofproductionofRFI’sandRFP’s,andhelp themdifferentiatedatacentreand relatedserviceofferingsbytheirenergy-efficiencyandenvironmentalperformance.

Task2.2standsforthesecondtaskasdescribedintheEURECAproject’sDOWunderWP2,consisting of sound and robust software tool based on the requirements identified underTask2.1.

Task 5.1 stands for the first task as described in the EURECA project’s DOW underWP5,consistingof:

- methodologiestoquantifyenergysavings,otherenvironmental improvementsandeconomically saving, triggered by GPP using this project’s tools and otherdeliverables

- reasonableuptakerateoftheEURECAprojectandalsoinnovativesolution

- overviewevaluationapproachtoevaluatetheincreasedskillofprocurers.



10

TCITstandsfortheEURECApartnerTeleCityTCOstandsforTotalCostofOwnershipUELstandsfortheEURECApartnerUniversityofEastLondonWorkPackage1(orWP1)oftheEURECAprojectcovers‘GreenDCProcurementAnalysis’WorkPackage2(orWP2)oftheEURECAprojectcovers‘ProcurementframeworkandTool‘WorkPackage3(orWP3)oftheEURECAprojectcovers‘KnowledgeSharing’WorkPackage6(orWP6)oftheEURECAprojectcovers‘Dissemination’



11

3 EURECAuptakerates

3.1 PurposeIn order to estimate the actual benefit of the EURECA deliverables outside the EURECApartners,weneedanuptakerateestimate,thatwillbeusedinchapter4and5,aspartofthemethod to calculate the energy and environmental savings benefit thanks to EURECAdevelopments.ThiswillleveragethefactthatnoteverybodyamongthepublicsectorITparticipantsthatwetrainwillactuallyusetheEURECAframeworkandtoolanddirectoryresourcesforimprovingtheenergyandenvironmentalperformanceoftheirpublicbodies’datacentresolutions,notevery participating procurement officer will build on it in tenders. And also not everyprocurementresponsiblethatwe informaboutEURECAofferings inpresentationsorpressreleases or finds the EURECA website will use the information for DC improvement, ofcourse.

3.2 ApproachThe uptake of EURECA developments is evaluated from insights of first procurementspecificationsandcasesduringtheproject:Basedontheshareofprocurementpartnersthatparticipate in the project, first, the reasonable best and reasonable worse case generaluptake rate andeffectiveuseof theproject’s toolsbypublicorganisations areestimated.Theexpecteddifferencesinuptakeratefromtrainings-wherewewillhavealessintensiveandclosecollaborationthanwiththeEURECApartners-andtheevenlesscloserelationshipwithwebsitevisitorswillbeestimatedalongadditionalconsiderations,asfollows:

3.3 UptakerateEURECA-runtrainings,workshopsThe following assumptions were made that will be used to estimate the energy andenvironmentalsavingsviatheEURECA-runtrainings,workshopsmechanism:

- EURECAistotrainanddirectlyinvolveatleast500peopleduringthelifespanoftheproject. This will be achieved via addressing these people via special interestseminars (GPP, IT procurement or Tendering processes), webinars and courses.Teaching and informing these people can be done in several ways. The EURECAprojectwillmake these trainees awareof thedevelopments of EURECA,wakeupinitiatives,evenbeforetheEURECAtooliscompletedandthemarketdirectoryfullyavailable and filled. The dissemination of the EURECA framework will signpostanswers automatically in the audience's day to day work. The final number of



12

participants and the absolute number of represented data centres will be used(since several participants can come from the same data centre, in other casesparticipants that are involved in the procurement of more than one data centreproductorservice).

- Notallof thetraineesthatarewillingandabletoconvert theEURECAknowledgeinto energy savings in a datacentre, moreover, will do this in the time of thedurationoftheproject(refitorrenewfraction)

- Finally, not all initiatives to improve the datacentrewill be able to enjoy the fullpotentialofthesavings,onlyapartofit.

Combiningthesefactors -eachwithareasonablybestandworstcase-wewillobtaintheuptake rate α for EURECA-run trainings that will be used to estimate the energy andenvironmentalsavingsthatcanbeexpectedfromthismechanism/deliverable(seechapter4).

Itcanbealreadyanticipatedatthispointthatthesavingsviatrainingswillonlystartduringtheprojectduration,butmainly after its end,whenall EURECAdevelopments are fully inplaceandavailable to the trainees, and theyactually face the task to improveanexistingdatacentreorserverroomorgoforrelatedprocurements.

3.4 UptakerateforusingEURECAdevelopmentsoutsidetheproject

InaverysimilarwayasashareofthetraineesofEURECA-runtrainingswillmakeuseofthevariousEURECAdevelopments(framework,tool,directory...)toimprovetheirdatacentres/server rooms or affect procurement of more energy and environmentally efficientinnovativesolutions,alsoDCoperatorsandprocurersatotherpublicbodies, thatEURECAprojecthadnodirectcontactwith,willdoso.Purposeofthisseconduptakerateishencetoobtainareasonableandrobustestimateforthisusergroup.Atthesametime,wecanincreasetheamountofpeoplethatwilllearnaboutEURECA(nexttomakingthedeliverablesaccessibleandeasytouse,inthefirstplace).Insteadofrelyingoninternetsearchenginesandword-of-mouthalonetoreachthislargepotentialusergroups,weanticipateandalreadyengage inarangeofoutreachactivitiesto informanas largeaspossiblenumberoftargetgrouppublicsectorprocurersanddatacentreexperts,aboutourdevelopments. This happens via presentations at general conferences, via press releases,scientificpresentations,papersandposters,socialmediapostssuchasviaTwitterandattherelevant LinkedIn interest groups, aswell as using third-partywebsite calendar and newspostingoptions,newsblogs,andmailinglistsofmultipliersinvariousEUmemberstateswitharelevantstakeholderreader/memberbase.



13

Basedonestimatesorwhereweknowthenumberofrecipients,suchasformailinglistsandsomeothermechanisms,wecananticipate thenumberofEurope-widepublicsectordatacentreandprocurementexpertsthatwillbeinformedabouttheEURECAdevelopments.

At the same time, and similar as for the EURECA-run training participants, not everybodywho will be informed, will actually use the EURECA developments. Since the depth ofinteractionisclearlyless,comparedtoin-persontrainingsandespeciallythedirectsupporttotheEURECApilotpartners,wehavetoassumeaconsiderablelowershareofactualusersoftheEURECAdevelopmentshere.

Intheend,thefinaluptakerateβwilltakealltheseaspectsandfactorsintoaccountandbeexpressed by the share of all EU-wide data centre procurements and self-improvementactivitiesthatwillmakefullorpartialuseofEURECAdevelopments(seechapter4).



14

4 EURECA-triggeredenergyandenvironmentsavings

4.1 PurposeIn order to ensure a reliable evaluation of the project’s achievements and in closecoordinationwithtasks1.4and1.5,andbasedontheframeworkandtooltobedevelopedintasks2.1and2.2,arobustapproachandmethodologyisdevelopedhereforquantifyingenergy savings and other environmental improvements, triggered by innovative publicprocurementofenvironmentallypreferablesolutionsinsupportbytheEURECAframework,toolandotherdeliverables.Thischapterdescribesthisapproachandmethodology.ItwillbeusedinD5.3toreportontheachievementofthecalltargets intermsofsavedenergyperCommissionfinancialcontributiontotheprojectaswellastheproject’scontributionstotheEU2020targetsandincreasedresource-efficiencyinvariousenvironmentalareas.

The call target is to save at least 25 GWh annual overall primary energy savings (orreplacement of fossil by renewable energy production) per 1 Mio EUR Commissioncontributiontotheproject,i.e.about37.5GWhannuallyforEURECA.

TheEU2020-theEU'sgrowthstrategyforthecomingdecade–aimsattransformingtheEUinto a smart, sustainable and inclusive economy, under the Digital Agenda (EUROPE2020,accessed2015).Regardingresource-efficiencyitaimsat:

“Climatechangeandenergysustainabilitygreenhousegasemissions20%lowerthan1990,20%ofenergyfromrenewable,20%increase inenergyefficiency”(EuropeanCommission,2010)

InadditiontotheDigitalAgenda,anotherimportantflagshipistheResourceEfficiencyunderSustainable growth, aiming equally at a resource efficient, greener andmore competitiveeconomy. The scope of the Resource Efficiency initiative thereby goes beyond climatechange and energy. It covers a much broader protection of the environment, reducingvarious emissions, and preventing biodiversity loss. The Resource topic has beenstrengthened by the Commission lately and with a special focus on material resourceefficiency, in form of the new Initiative on the Circular Economy (European Commission,2015). D5.3. will report on the tangible andwider expected contributions of the EURECAprojecttothesepolicytargets.

4.2 IntroductionandstatusquoThe data centre industry has a high growth rate: A study predicts that the data centrerevenueacross19oftheEU28countrieswillincreasefrom3.2billionEuroperannum(2010)to7.3billionEuro (2015), i.e.atanannualgrowth rateof25% (Telecompricing,accessed



15

2014), (study predicts in-house data centre, accessed 2015). However, the rise of the ICTsectorandthedatacentreindustrycomeswithincreasingenvironmentalpressures:

Data centres areoneof the largest and fastest growing consumersof electricity. In 2013,U.S. data centres consumed an estimated 91 TWh of electricity -- enough electricity topowerallthehouseholdsinNewYorkCitytwiceover--andareon-tracktoreach140TWhby 2020 (Delforge P., 2014). This, despite of the fast technological progress of highercomputationalperformanceandeffortstowardsenergy-efficiency.Theevenfasterincreasein demand for data services is outpacing the technological and energy-efficiency progress(Wolf,2014).On global level, the electricity consumption by data centres is estimated to be about 2%(Koomey, 2011) (i.e. more than all air transport together) and to increase to 70 TWh inEuropeby2020(UBA,2015).In2013theamountofdatacentrepowerknownintheworldis38.84 GW (growth 2012-2013: +7,2%), in Europe 13.47 GW (growth 2012-2013: +6.0%),according to the Data Centre Dynamics’ Industry Census 2013. The world consumed 340TWhinitsdatacentresin2013.Europeconsumed118TWh.Figure1illustratestheoutlookofthisdevelopment(SRIAnalystofAllianceTrustInvestment,Parker,2013):

Figure1 Projection of Global and US data centre electricity use, by SRI Analyst of Alliance TrustInvestment,Parker(2013).

Another research (Andrae & Edler, 2015) comes to similar results, furthermoredifferentiatingelectricityusageintofourprincipalcategories:



16

1. consumer devices, including personal computers, mobile phones, TVs and homeentertainmentsystems;

2. networkinfrastructure;

3. datacentrecomputationandstorage;andlastly

4. productionoftheabovecategoriesTheforecastisbasedonexpectedannualgrowthofglobaldatacentreInternetProtocol(IP)traffic between 2010 and 2030, electricity used per traffic unit, and improvements ofelectricity efficiencies to be achieved year by year; the capital goods production data is arougherestimate.Figure2showsthebestandworstcaseestimatesfordatacentreelectricityuse:

Figure2 Best, expected, andworst case electricity consumption during use of data centresworldwide byAndrae&Edler(2015).

--------------------------

Insert:whichsavingspotentialhastheusephaseelectricityconsumption?

The average public datacentre has an ICT load of 500 kW (not taking into account localadministrationetc.),whileweaimatimprovingtheaccuracyofthatfigure,amongothersviafeedbackfromthetrainees.



17

Active energy efficiency management of Data Centres can help to reduce up to 30% ofenergy consumption by implementing permanent changes through measurement,monitoring, and control of energy usage. More efforts shall be done to improve coolingmanagementat smallerdata centregranularity levelsand tominimizedata centreenergyconsumption(Feindt,2015).ThesavingsofusephaseprimaryenergyviaanimprovedPUEare within reach for an average datacentre when best practises are followed andimplemented. Particularly the substantial savings achieved in German data centres onfederallevelthathavebeenanalysedandquantifiedindetailbytheGermanUBA(EPA)willfeedinquantitativeinformation.Anexample(Figure3):

Figure3 Use phase electricity savings effect when changing from PUE 1.8 to 1.2. Source: Future-Tech:http://www.future-tech.co.uk/data-centre-energy-efficiency/

---------------------------------

Additionalsavings inbothprimaryenergyandcapitalgoodswill stemfromeffortssuchasvirtualisation and consolidation, that will have a similar level of savings on top. A moreaccurateestimatewillbederivedfromanalysingthebusinesscasesofEURECApilotpartnersandexternalbusinesscases.Electricity consumption however is only the tip of the iceberg: the primary energyconsumption is typically several times as high and a range of environmental impacts areassociatedwith energy carrier extraction, conversion until the electricity is delivered to adatacentre.



18

In view of the climate change impacts, projections (Climate Group and the Global e-SustainabilityInitiative(GeSI),2008)seeICTgrowthskyrocketingintheforeseeablefuture.Thatreportcontendsthatevenwithcloud-computingcompaniesfocusingmoreintentlyonenergy efficiency, overall ICT-based greenhouse gas emission will increase by over 70percentfrom2007to2020,whilethemostrecentGeSireportSmart2030somehowrevisesthisfinding(GeSI,2015).Specifically,theSMART2020numbersshowworldwideICTmetrictonnecarbondioxideequivalentin2020tobe1,430MtCO2e,withdatacentresaccountingfor18percentofthatamount(257MtCO2e),telecoms25percent(358MtCO2e),andPCsandperipherals57percent(815MtCO2e)(WebbM.,2010).

However, the trend of using renewable power is strong (Goiri and others, 2015;Oro andothers,2015)andlikelymanydatacentrescanberunGHGefficient,eveniftheydonotfindwaystoreducetheabsoluteelectricityusage(Andrae&Edler,2015).

Nexttotheusephaseelectricityconsumption,whatisthefocusoftheabovedata,anotherand increasingly relevant contribution comeswith the capital goods production: the datacentre building, theM&E plant as well as the servers and other IT hardware productioncontributesahighshareaswell:twomoredetailedstudieshavecalculatedthiscontributionofthecapitalgoodstotheprimaryenergyorexergyconsumptiontobeabout30%to50%(MezaJ.etal,2010,HonéeC.etal,2012).Itishencenotsufficienttoexclusivelytakeintoaccountthedatacentreusephaseenergyconsumption.Moreover,capitalgoodsgenerallyhave a higher relative contribution than electricity consumption for many otherenvironmental impacts, ranging from acidifying and particulate emissions to depletion ofmetal ores. The production of capital goods – and the savings with the help of EURECAadvice, such as virtualisation, consolidation,workload shifting and others that particularlyreduce the amount of hardware needed for the same service – is hence to be fullyconsidered to capture the situation in a non-distorted way. For data centres that useexclusively clean renewable energy (e.g. wind power plus storage, geothermal) for theiroperations, the energy that was used to produce all the hardware will still typicallypredominantlycomefromfossilenergy.Next,waterconsumptionbydatacentreshascomeintofocus, incontextoftheotherwisebeneficialshifttowater-basedchillers.Whilelanduseisnotanimportantenvironmentalimpactfordatacentres,itisnotedthatintheEU,thetotaldatacentre‘white’spacedeployedamountsto10.5millionsquaremetersin2012.Fourmarketscommandoveronemillionsquaremeterseach,rankedinorder:TheUnitedKingdom,Germany,FranceandItaly.SpainandtheNetherlandsbothhavebetween600,000and950,000squaremetres(DCDIntelligent,2012).Thedatacentresinthepublicsectorthereforecanplayakeycontributiontohelpmeetingthe EU2020 targets of increased energy-efficiency, increased share of green energy, andreducedgreenhousegasemissions,aswellasthewiderEUenvironmentaltargets–iftheir



19

environmentalfootprintcanbestrongerreducedthanexpected.ThisiswhereEURECAwillcontribute.

4.3 ApproachAstotheDoW,aquantitativelifecyclebasedapproachwillbeusedinEURECAtocalculatethe overall project “impact” (i.e. saving of primary energy and environmental impacts) bytheEURECAproject’sdeliverables.Bytakingintoaccountthewholelifecycleofdatacentres,itwillbeassuredthatallrelevantdatacentrerelatedenergyandenvironmentalimpactsareconsidered-fromrawmaterialextraction,componentproduction,usephaseuntiltheendoflife.Thisisnecessaryalsotoavoidashiftingofburdensfromusephase,whichoftenistheonlyscopeofanalysisofICTproducts,tootherproductstages.

The lifecyclewideenvironmental impactsavingsdue toEURECAof thedatacentrecapitalgoods production, use and end-of-life are calculated using the EURECA framework (D2.1)and tool (D2.2). Applying the EURECA framework and tool to evaluate the impact of theEURECA project is also in line with the latest Impact Assessment Directive of the EU(European Commission, 2015b), which refers explicitly to using the developments of theEuropeanPlatformon LCA,whicharenamely the ILCDHandbook, theELCDdatabaseandthePEFguidewhichformthebasisofthequantitativelifecycleapproachoftheframeworkD2.1.AsD2.1isduebyendofNovember,i.e.onemonthafterD5.1,abriefsummaryisgivenherebelow:

---------------------------------

Insert:summaryofthequantitativelifecyclepartoftheEURECAframework,ofD2.1:

Formula1showsthehigh levelcalculationoftheannualenvironmentalburdenassociatedwithadatacentreperyear.Figure4showstheconceptschematically.

Formula1 E=(Er+Eeol)/n+Eu–Ec/n

With

E=environmentalandenergyburdenofdatacentre (e.g.primaryenergy, climatechangeetc.)peryear

Er=environmentalburdenandenergy inrawmaterialextractionandproductionphaseofcapitalgoods(building,servers,UPS…)

Eeol=environmentalandenergyburdenintheend-of-lifephaseofthecapitalgoods



20

Eu = annual environmental and energy burden during data centre use phase (electricity,coolantloss,waterconsumption,…)

Ec=environmentalandenergycredit fromrecoveryof secondarymaterialsandenergyatend-of-lifetreatment

n=lifetimeofcapitalgoods[a]

Figure4 Schematic viewof life cycle of a server, focussing on production and primary energy use, on thelevel of the inventory results, for illustration. After further aggregation, all energy and environmentalimpacts for thewhole lifecyclewillbeexpressed in15numbers for thewholedatacentre (i.e. for the15impact and resource categories, one of which is the primary energy). These can optionally be furtheraggregated to 1 number representing the annual overall environmental and resource impact of the datacentre,forclearerdecisionsupport.



21

The quantitative life cycle part of the framework builds on the relevant EuropeanCommission life cycle method developments and data, with the Product EnvironmentalFootprint guide and the relatedPEFCRs forUPS,HDD,metal sheets, copper andbatteries(PEFCRUPS2015,PEFCRHDD2015,PEFCRMetalsheets2015,OEFSRCopper2015,PEFCRBatteries2015),thepreparatorystudiesforEnterpriseservers(BiobyDeloitte&FraunhoferIZM,2015),andthestudyonenterpriseserversbytheCommission’sJRC(TalenPeiroL.&ArdenteF.,2015).Sector-specificrecommendationsbyETSI(ETSI,2011),thegreengridonlifecycleassessmentofdatacentres(Greengrid,2012),aswellastheGHGProtocol(GHGProtocol,2013)areusedasguidingprinciples.Atthesametime,andgiventhecomplexityofICT products and data centres and limitations of the availability of comparable life cycleguides for data centres, as well as to ease the effort for data centre goods and servicesvendors,somesimplificationsaremade.Moreover,wherespecificinformationordataisnotavailable,suitabledefaultvaluesareused. Inresult, theframeworkallowstoquantify inarobust manner the environmental and energy efficiency of the data centre, substantiallymorecompleteandreliable than thebestpractice inprocurement thatusesweakproxiesforusephaseenergyonly.

The following environmental and energy indicators will be calculated: primary energy,renewable primary energy, climate change, acidification, eutrophication, ozone depletion,summer smog,particulatematter, ionising radiation,human toxicity, andecotoxicity, landuse,waterscarcity,allinlifecycleperspective.---------------------------------

ThecalculationofEURECA-relatedsavingsofprimaryenergyandenvironmental impacts isdividedintothreecomponents.Thethreecomponentsare:

- Net savings due to EURECA framework and tool supported (or avoided)procurementsandself-improvementsbytheEURECApilotpartners.Someofthesesavingswillbeinitiatedduringtheprojectduration,othersonlyafterwards.

- NetsavingsduetoEURECA-runtrainings,usingthevariousEURECAdevelopments.

- Net savings of EURECA framework and tool supported (or avoided) procurementsandself-improvementsinitiatedoutsidetheEURECAproject,until2020.

Foreachofthese,thedirect(i.e.DCoperation)andindirect(capitalgoodsproductionandend-of-lifetreatment/benefication)willbereportedseparately.Notethatonlysavingsaretakenintoaccountthatoccurbeforetheendof2020.

Figure5illustrateshowthecomponentsoftheoverallexpectednetsavingsduetoproject’sdeliverablesareanticipated:



22

Figure5 Savings of primary energy and environmental impact due to the various EURECA measures;illustrative. Each bar represents one data centre, with all impacts of production, use and end-of-life areshownaveragedovertheexpecteddurationofthedatacentreuse(e.g.5years).Notethatthedataisnotstacked,i.e.thetotalisthesumofthesavingsduetothethreemeasures.ThehighestabsolutesavingsareexpectedfromuseoftheEURECAdevelopmentsbypublicbodiesoutsidethepilotpartnersandtheEURECA-run trainings, includingdue to itsmaintenanceafter theendof theprojectwithupdating technologyandotherdata,expandingtheonlinedirectory,etc.

4.4 MethodologyAs introducedabove, theoverall savingsdue to theEURECAdeliverablesconsistsof threeparts:

4.4.1 CalculatedandexpectednetsavingsofEURECA-supported(oravoided) procurements and self-improvements by EURECApartnersinitiatedduringtheproject

Theabsoluteannualenergy-savingsandenvironmentalbenefitsarecalculatedcomparedtoa“business-as-usual”orautonomousprocurementscenario(seeFormula2).ThisapproachestimatesexpectedsavingsandbenefitsasillustratedinFigure6:

Formula2 Bp=Eb-Ee



23

With

Bp = Annual savings of data centre energy and environmental impact from EURECA, viaEURECApartnersEb=Annualenvironmentalburdenofdatacentrein“businessasusual”scenarioEe=AnnualenvironmentalburdenofdatacentreusingEURECAframeworkand/ortool

Figure6 illustrates life cycle based calculation of net benefit of EURECA-supported (or avoided)procurements initiated during project time plan, by the EURECA partners; with the example of primaryenergy

For calculation the environmental burdens of the “business as usual” and the EURECA-supported procurement, specific data provided by the procurers is preferable. Therequested input data that will be used by the framework to calculate the energy andenvironmentalperformanceare:(selectionofkeydata,detailsdescribedinD2.1)

- Typeandserver/storageperformance/capacity

- Usage/loadofserversandstorage

- Location

- Electricityconsumption

- Waterconsumption

- Specificationandperformanceofcoolingsystem

- Specificationandperformanceofserverandstorageequipment

- SpecificationandperformanceofUPS

- Buildingspecificationonhighlevel



24

Ifthespecificdataisnotpossible,e.g.producttypeaverageorcountryspecificinformationwill be used as default values (details described in D2.1). For selected products it will bepossibletoprovidespecificenvironmentalperformancedatabytheprovider(optional).

Thiscombinationofdetailedvalueswithfall-backoptionsensuresthatmissingdatawillnotimpacttheabilitytoobtainfigures.

4.4.2 ExpectednetsavingsduetoEURECA-runtrainingsDue to participation in the EURECA trainings, procurers will have a better knowledge onenvironmental(includingenergy)efficiency.Theprocurersthereforewillbeabletoprocuremoreenvironmentallysoundandenergy-efficientdatacentresolutions.However, based on various studies, there are ongoing improvements in the market alsowithoutEURECA,ofcourse(i.e.autonomousdevelopment);detailsseeannex9.1.Thustheexpected savings due to EURECA-run trainings are to be calculated from the differencesbetweeninterpolation/extrapolationofenergysavinginthepilotswiththepotentialsavingwithoutEURECA(seeFormula3):

Formula3 Bt=[Bte-Btw]*α

WithBt =BenefitfromEURECAduetoEURECA-runtrainings

Bte =Estimatedsavingsofdatacentreenergyandenvironmental impactusingEURECAdevelopments,peryearBtw = Estimated savings of data centre energy and environmental impact withoutEURECA,peryear

α =uptakerateamongEURECA-runtrainingparticipants(≤100%);seechapter4.4.2

The estimated savings are interpolated and extrapolated savings from pilots that areengaged in theproject (i.e. theEURECApilotpartners frompublic sector), calculatedwiththeEURECAframework/tool.Thisconsidersatleastthesizeofthedatacentre(ifavailable)andcountry-specificdifferenceswhenextrapolating.Anotherimportantelementisthecurrentstatusofthedatacentresthatarerepresentedbythe participants in the training. It is therefore beneficial to get the necessary informationfromprocurerswhoparticipate inthetrainingeitherby intervieworsurvey.Thefollowingquestions are examples to indicate the ex post evaluation after the training that will bedetailedunderWP4:

- Are the trained people able to influence the energy efficiency of “their” publicorganisation’sdatacentre(s)?



25

- Whataretheindividual‘takeaways’fromthetraining,whataretheyactuallygoingtodowiththenewlyacquiredknowledge?

- Whereisthecurrentdatacentrelocated?

- Whatisthecomputationandstorageloadofdatacentre?

- What is the size (kW) of their data centre(s) andwhat is/are the PUEs (orwhichdefaultwecanassumeshouldthatinfonotbeavailable)?

- Is there already a planning for refit or renewal of the data centre(products/services)?

Theabovewillbeusedtofine-tunetheinitialestimateoffeasiblesavings(seechapter3.3ontheuptakerate).

4.4.3 ExpectednetsavingsusingEURECAdevelopmentsoutsidetheproject

The estimated annual energy-savings and environmental benefits are again calculatedcompared to an autonomous development scenario. The key difference to savings byEURECApilotpartnerswherewehavedirectaccesstoinformation,isthatherethenumberand size of procurement cases and self-improvement activities is to be estimated via themarket uptake rate β (see chapter 3.4): the estimated environmental and energy benefitcannot be related to specific data centres / procurers (as in the EURECA partners andEURECA-run trainings). Theaverage/typical situationwill beused,drawingon theaverageachievementsthatwereidentifiedfromthemoreaccuratedataobtainedfromtheEURECApilotpartners.

Also these wider EURECA users will use any of the following developments: EURECAframeworkand/ortoolsupported(oravoided)procurements,non-EURECA-runtrainingsandworkshops,andviauseoftheonlinedirectoryincl.webinars.

The estimated savings insights without EURECA come again from specific business casesinsideandoutsideEURECA(examplesarebelow),transformingtheinformationfromthosecases(e.g.PUEchange,procurementofnewservers,annualelectricitypurchase,estimatedcoolant loss, virtualization, switch to free cooling etc.) into input data for theframework/tool (details inD2.1). This is done, inorder thewhole rangeof environmentalandenergy savingscanbecalculated in theanalogouswayandcompletenessasdone forthebenefitsviatheEURECApilotpartners.SeeFormula4:



26

Formula4 Bo=[Boe-Bow]*β

WithBo =AnnualnetsavingsandbenefitsfromEURECA-developmentsoutsidetheprojectBoe = Estimated saving of data centre impact using EURECAdevelopments outside theproject,peryearBow =EstimatedsavingwithoutEURECA,peryearβ=uptakerateamongALLEUpublicdatacentres(≤100%);seechapter3.4Thebenefitsdue to these threeoutreachmechanisms constitute the total savingsbenefitduetoEURECA,whilethedifferentrobustnessoftheseestimateswillbeconsidered.

SpecificbusinesscasesoutsideEURECA,examples/insightsonautonomousdevelopment:

Asmentionedabove,independentfromtheproject,thereareongoingeffortsandsavingsinpublic data centres due to interests in (energy and hardware) cost savings andenvironmentalawareness.Thisandotherbusinesscasesand insightswillbeusedtogetabetterpictureattheachievablesavingsinindividualpublicsectordatacentres,whichforman important factor when estimating the benefit of EURECA outside the project ‘s pilotpartners.For example, business case building permits in the Amsterdam region (GemeenteAmsterdam,2015)require:

- restrictingthenewdatacentrestoanEUEofamaximumof1.2

- prescribingadatacentreroom(inlet)temperaturebetween24-27Celsius.

The environmentalmanagement act in theNetherlands (DutchGovernment - "Overheid",2014), enforced by the “omgevingsdiensten” has been very effective in the Amsterdamregion.Atthecoreofthisactisthelegalobligationforeverycompanyconsumingover200MWh/yeartotakeallenergyefficiencymeasuresthathaveapaybackperiodshorterthan5years (legislation: Milieuwet, Article 2.15, 2014). Although this approach might appeardifferentthandemanddrivenefficiencyimprovement,thecommonelementisthatallplansandactionsareformulated,initiatedandactuatedbytheconsumer.SuchstronglegislativepushwillhowevermeanthatEURECAhastodeliverontop–orhelptoidentifysuchsavingpotentialswheretheywerepreviouslyundetected.AnotherexampleisthemunicipalityofAmsterdamthatintendstoprocuresustainableICT,resulting in voluntary actions taken by the data centre industry. The 40 major DC’s inAmsterdamarea,use11%ofthecity’spowerusedbyallofits22,000companies(Gemeente



27

Amsterdam, 2015), a total of 460GWh in 2013, butwith plans to lower this usageby 68GWh,aimingatsavingsof15%inthenearfuture(GemeenteAmsterdam,2014).

According to another report, the energy savings potential is an annual 80 GWh in theNetherlands, according to extrapolations (Gemeente Amsterdam Dienst Milieu enBouwtoezicht, 2012). 60 GWh of which has been found attainable through more energyefficientcooling,anadditional20GWhbyusingbetterICThardware.

InGermany, on federal level the aimwas set to reduceoverall energy consumption from2008until2013by40%[Köhn2015].Thiswasindeedachieved(-36%),evennottakingintoaccount the increased amount of digital services that were provided. These savings hadvirtualisationandconsolidationas themain contributors.Beyond2014, it is foreseen thatthe overall energy consumption of federal data centres shall not increase, despite ofincreasedamountofdatamanagedandstored[Köhn2015].Theseresultsdemonstratetheeffectthatcanbeobtainedbydemandingbetterdatacentreperformance,eitherbygovernmentsorbycustomers.Somebusinesscasesofactualsavingsinpublicorganizationdatacentresarebelowtoseehowsavingsarereaped.

- Business case “City of Amsterdam”: to change the data centre from an oldinefficient one (PUE 2.5) into a new 500 KWh data centre (PUE 1.3), savesAmsterdam annually 5.2 GWh (see Dutch presentation movie:https://youtu.be/9H_DOprUVJU).

- BusinesscaseHHD:30%savings,thankstotheoutsourcingofitssmalldatacentreofaround30KWofICTload,thedatacentresavesannuallyaround0.2GWh.

- Business case Dutch Central Government: 107 GWh annually by reducing thenumberofitsdatacentresfrom64to4(Kerssens,2015).

- Business case of Amsterdam, Omgevingsdienst Noordzeekanaalgebied (ODNZKG):through the application of the environmentalmanagement act, 68 GWh is savedannually(Harryvan,2014).

TheseandothercaseswillbeusedtocalculatetheenvironmentalbenefitwithoutEURECAactiveinvolvement.



28

5 EURECAcost/benefitanalysis

5.1 PurposeInordertoensureareliableevaluationoftheproject’seconomicachievementsandinclosecoordinationwithandbasedontheframeworkdevelopedintask2.1,arobustapproachandmethodology isdevelopedhereforquantifyingcostsavings, triggeredby innovativepublicprocurementofenvironmentallypreferablesolutionsinsupportbytheEURECAframework,toolandotherdeliverables.Thischapterdescribesthisapproachandmethodology.ItwillbeusedinD5.3.

5.2 IntroductionBeforewe ascertain the cost/benefit analysis approach and calculationmethod,wemustfirstadvisethatthereareanumberoffactorsthatwillaffecttheresult:

- “CloudFirst”policies invariousEUcountrieswilldiluteandrender largely invisiblethe energy and environmental efficiency of the individual data centres thatcontribute to the cloud and that are run by various private organisations. Theseorganisationsmaynothave thewill toundertakedetailed studieson theirenergyuseorenvironmentalimpactunlessitwillbebusinessrelevant.Thisisvalid,exceptfor government cloud solutions with government-run data centres only. On theotherhand,inotherEUmemberstates,thepoliticalaimtokeepphysicalcontrolofthe data centre location - i.e. “no cloud” - are relevant developments, such as inGermany,thatbringtheindividualdatacentresintofocus.

- Rapid pace of technology development particularly for the IT componentsmeansthatdatacouldbeoutofdateinarelativelyshortperiodoftime.

- Notreplacingservicesorequipmentbutaddingtothemduetoalackofknowledge,i.e. where additions are made to data centres or server rooms withoutdecommissioning existing services, results in an increased energy andenvironmentalburden.

- ParticularlyforinnovativetechnologiesthetrueTotalCostofOwnership(TCO)willnotbeknownthatprecisely,asexperienceislackingbydefinition.

- Externalcoststosocietyduetoemissionsoverthelifecycleofthedatacentreareless “hard” figures, compared to TCO figures, what needs to be understood andconsideredincost-baseddecisionmaking.



29

TheprojecthasatargetofassistingtheEUPublicsectorwithatleasttenpublishedtenders(D5.5,dueMonth30).

Themostreliableandhencerelevant informationsourceforthecostsavings(orcase-wiseincrease)duetousingtheEURECAdevelopmentsarethetendersthatarepublishedorevenimplemented by the EURECA pilot partners. At this stage, we are still in the process ofidentifyingsuitablepublicsectororganisationsthatareseekingtoissuetendersfornewdatacentres.Wehave11possiblecandidatesatthetimeofpublicationofthisreport(Oct2015)withdiscussionsgoingonwithothers.

5.3 ApproachAstotheDoW,aquantitativelifecyclebasedapproachwillbeusedinEURECAtocalculatetheoverallproject “cost impact” (i.e. cost savingsorhighercosts inTCOperspective,plusreduced externality cost due to reduced environmental impacts over the life cycle of thedatacentreorotherprocuredgoodsandservices).ReportingbothTCOandexternalcosts,weprovideancompleteenvironmentallifecyclecosting(eLCC)informationoftheEURECAproject’sdeliverables’impact(seeFigure7).

Bytakingintoaccountthewholelifecycleofdatacentres,itwillbeassuredthatallrelevantcosts of data centre are considered – investment cost (hardware and other capital costs,data centre operation electricity costs, Software Licensing Costs, and Personnel Costs(Facilities/IT)), annual operation cost (during use-stage), end-of-lifemanagement cost andalsoexternalitycoststosociety.ThelifecyclecostbenefitduetoEURECAofthedatacentrecapital goods production, use and end-of-life are calculated using the EURECA framework(D2.1)andtool(D2.2).Thedetailed TCOapproach for data centres is described and formulated in theupcomingD2.1thatformstheformalbasisofthisD5.1.ThecomplementaryeLCCdetailsaredescribedheremorebelow.



30

Figure7 IllustrationofenvironmentalLifeCycleCosting,anditscomponentsTCOandexternalcosting.

5.4 MethodologyAnalogously as for theenvironmental benefit in chapter 4, the cost saving canbedividedinto three outreach/involvement levels of public procurers and data centre experts (seeFormula5):

Formula5 NetSaving=NetFra+NetTra+NetOth

Where



31

NetFra = Net savings due to EURECA framework and tool supported (or avoided)procurementsandself-improvementsbytheEURECApilotpartners.NetTra=NetsavingsduetoEURECA-runtrainings,usingthevariousEURECAdevelopmentsNetOth=Net savingsofEURECA frameworkand tool supported (oravoided)procurementsandself-improvementsinitiatedoutsidetheEURECAproject,until2020Someofthesesavingswillbeinitiatedduringtheprojectduration,otheronlyafterwardsThe absolute savings are calculated compared to a “business-as-usual” or autonomousprocurementscenario(Detailinchapter5)(seeFormula6).

Formula6 NetXxx=eLCCeureca-eLCCbau

Where

Netxxx = either NetFra (Net savings due to EURECA framework and tool supported (oravoided)procurementsbytheEURECApilotpartners)orNetTra(NetsavingsduetoEURECA-run trainings, using the various EURECA developments) or NetOth (Net savings of EURECAframeworkandtoolsupported (oravoided)procurementsandself-improvements initiatedoutsidetheEURECAproject,until2020).Expressedin[Euro]

eLCCeureca = environmental life cycle costing when implementation according to EURECAframeworkand/ortool[Euro]eLCCbau=environmentallifecyclecostingfor“businessasusual”case[Euro]

OntopofreportingtheTCOandexternalcostfiguresseparately,theoverallenvironmentallife cycle costing can be calculated combining the TCO (total cost of ownership) plus theexternality cost; see Formula 7. To be in line with Clean Vehicle Directive, only fourelementary flows in climate changeare foreseen tobe considered i.e. CO2,NOx,NMVOCandParticulatematter.Table1showsthedefaultcostfactors.Thelifecyclewideemissiondata are the result of the EURECA tool calculations along the modelling and calculationmethodasdetailedinD2.1.

Formula7 eLCC=TCO+Ext

WhereeLCC =environmentalLifeCycleCost[Euro]TCO =Investmentcost+Operationalcost+End-of-lifemanagementcost[Euro]Ext =externalitycost[Euro]TheTCOcostswilltakeintoaccountcapitalcosts/interestratesetc.



32

Theexternalitycostiscalculatedasfollows:

Formula8 Ext=f*Emission

f =costfactorforindividualemission;seeTable1[Euro/kg]Emission =emissionflow[kg]

Table1 Thedefaultcostforemissions(EuropeanCommission,2015c).

Emission Externalcosttosociety

CO2 0.03-0.04EUR/kg

NOx 4.4EUR/kg

NMHC 1EUR/kg

Particulatematter 87EUR/kg

It is to be highlighted that the Clean Vehicle Directive does not take into account otheremissions and resource depletions than from the vehicles off gas, hence will be clearlyincomplete for data centres and miss out e.g. coolant loss, water scarcity and otheremissions and resources. We nevertheless will provide the necessarily incompleteexternalitycostresults,toimprovetheavailableinformationformakingdecisions.WhenpresentingthecostresultsinD5.3,itisforeseentoseparatelypresenttheresultsofthe TCO (for eachof the three contributing figures) andeLCC, given thedifferent level ofrobustnessandnatureofthecostdata.



33

6 IncreasedskillsoftheprocurersFromthefeedbackatourfirstpublicevents,interviewsandsurveys,itisclearthattheskilllevelsofpublicsectorprocurementandtechnicalstaffislimitedwhenappliedtotheuseofenergy efficiency best practices, global best practices for operations and procurement ofinnovativedatacentrerelatedsolutions.Theprojectteamfeelsthatpublicsectoremployeesareoftenandlargelyexcludedfromthedatacentreindustryperse,withrespecttoattendanceatindustryeventsandpublicationsand therefore are not aware of the latest innovations in cooling, the implementation ofenergyefficientbestpracticeandemerginginternationalstandardssuchasISO30132/33/34forPUEandERFandtheEUCodeofConduct forDataCentres (EnergyEfficiency)andtheGreenGrid’sMaturityModel.Thisalsoaffectsprocurementof innovative,energy-efficientand more environmentally sound data centre solutions in public sector organisations.Criteria on energy and environment are not included or limited (see also D1.1). Notabledifferences exist at least on Federal level in Germany, thanks also to the efforts ondevelopingthefirstTypeIEcolabel(BlueAngel)fordatacentresafewyearsago.TheuseoftheEURECAtoolsetandtheassociatedtrainingprogrammewillhoweveraddressthismorecommonissueoflimitedknowledgeandusageofmoreadvancedwaystocapturetheenvironmentalandenergyperformanceofdatacentreproductsandservices.

Whilstthecontentofthetrainingcurriculumhasyettobeformallystructuredanddefined,as this is contained within WP4, our approach will be to develop two sets of trainingprogrammes,directedattwodistincttargetaudiences:Thefirstprogrammewillbedirectedatprocurementstaffandthesecondprogrammeattechnicalstaff.

Our thoughts at this point are toprovide, for both technical andnon-technical routes, anoverview of energy efficiency and sustainability in the data centre per se, and this willinclude informationontheEUCodeofConduct forDataCentre (EnergyEfficiency) (or theimpending EN 50600 TR 99-1), the Green Grid’s Data Centre Maturity Model and otherrelevantdocumentsidentifiedinD1.1.Buildingon lifecyclecosting,which iswidelyknownandrequiredtobe including inpublicprocurement,moreover theenvironmental sideof life cycle thinkingwillbe introduced toprocurers, with information on available policy developments such as the Energy-relatedProductsDirectiveactivitiesonEnterpriseServersandindustry-wideactivitiestodeveloplifecyclebasedlabelsfore.g.UPSandharddisks.

Thefirstsessionwillalsoincludereallifeexamplesofpublicandprivatesectordatacentresthat have used innovative solutions and after some time the projects that have beenassistedwiththeEURECAtools.



34

The secondand subsequent sessionsare foreseen toprovideanoverviewof the toolsets,including detailed scripts and flowcharts of the information required in each module toachieveanoptimumresult.

The results will provide an organisation with the necessary information to build aprocurementtenderandassociateddocumentation.

A key message will be that collaboration between the various departments includingprocurement,legal,servicedelivery,estates/facilitiesandITfunctionswillberequiredandisessential tosuccessfullydesign,buildandoperateanenergyefficientandsustainabledatacentreusinginnovativesolutions.

The information provided by the training sessions will allow both procurement, non-technicalandtechnicalpersonnelwithinthepublicsectortohavethenecessaryskillsetstoimplementenergyefficientandsustainabledatacentresolutionsand/orservicesortoseekthesamefromthecolocationandcloudproviderswithintheEU.

Weaimtoprovidethenecessaryknowledgetoprocurementpersonneltoquestiontechnicalstaff regarding theprovisional designof the projectwith respect to the use of innovativesolutions, alternatives such as cloudor colocation in a recognised energy efficient facility,energyefficientequipmentetc.

Technicalstaffwillbemadeawareofthehighlevelviewofclimatechange,energyscarcity,thevariousglobalstandardsandguidelinesforuseindatacentres,areviewofsomeofthemost common innovative solutions available for use in the data centre environment andfinally introducingtheconceptof thecirculareconomywheredatacentreoutputssuchaswaste heat can be harnessed to provide heating for office environments or for otherindustrialuses.Training for the industry is covered elsewhere in the work packages under the EURECAprojectandwillfollowasimilarlinetothoseprovidedforthepublicsector.

This will allow the industry to be “EURECA” aware, and to provide upon request or incompetitivetenderingsituationstheinformationthatwewillimparttothepublicsector.

Theprojectaimstotrain500publicsectorpersonneltogainthenecessaryknowledgetoaskdata centre designers, colocation and cloud suppliers about their energy efficient andsustainableproductofferings.

The people thatwill have been trainedby EURECAwill be approachedwith the followingfrequency,inordertoidentifythesuccessofthetrainingandhowthetrainingandtheotherdeliverablesofEURECAhaveeffectivelycontributedtoincreasedskillsoftheprocurersandofpublicsectorITstaffthatareinvolvedindatacentreprocurementoroperation.

Thetrainingsessionswillstartwithinthelifespanoftheproject.Duringthewebregistrationofpeople tobe trained,peoplewillbe informedandconsent to the fact that theywillbeapproachedaccordingtothefrequencymentionedhereunder.



35

GettingdataonsuccessEURECA needs to get information on how successful the training is. During the EURECAtrainingandwebinars,thiswillextensivelyhavebeenexplained.Therewillbethefollowingwaysofgatheringdataand informationon thesuccessof theEURECAtrainingsandotherdeliverables:

1. Askingthetraineesdirectly.Timewillbespenttomotivatethetraineestoreactandanswerthesequestionsthatwillbeaskedthem,overtime.Thequestionswillbeexplainedandthetrainerswillexplainhowthetraineesshouldmeasureandgatherdatainordertocomeupwithanswers.

EURECAwill ask the following typeof questions, both before the training orwebinar andafterwards(potentiallywithafollow-up6monthslater).Somequestionsaremoredirectedattechnical(IT)staff:

- AreyouawareoftheEUCoC?

- DoyouknowandmeasurethePUEofyourdatacentre?

- Whichenvironmentalaspectsofthedatacentreoperationmattersrelevantly,nexttoelectricityconsumption?

- Whichenvironmentalandenergyaspectsarerelevantforservers,forUPS?

- Whichcoolingtechnologyisthemostenergy-efficient,isthemostenvironmentallyfriendlyone?

Otherquestionsaredirectedatprocurementofficers:

- What does the Procurement Directive say about using energy and environmentalcriteriainpublicprocurement?

- DidyoualreadyincludeenergyandenvironmentalcriteriainICTprocurement,inDCprocurement(andifsowhich)?

Thelogicofthequestionsandhowtointerprettheanswerswillbeworkedoutindetailandinclosecoordinationwithdevelopingthetrainingcurriculum,toderivecomparablelevelsoflearningsviaapointsystem.Thisinformationcanbeusedinfuturetrainingsessionstoo.Theanswerswillalsobeusedtoimprovethetool.

2. AskprocurersinEuropeEURECAwill ask procurers in Europe feedback on their knowledge of the innovative datacentretechnologies,ofenvironmentalandenergyaspectsofdatacentresingeneral,andofspecificelements.



36

Thiswill bedoneduringnetwork activities, seminars,with thehelpofquestionnaires andinterviews.Whereconsortiummembersspeak(networking,events,presentations),thereisalwaysapossibilitytosendandgetinformation.ThedatabaseofprocurersthatEURECAisbuilding, through the several events throughout of Europe will be having data on whoEURECAmayapproach(consent)askingquestions(mail,questionnaires,etc.).Overacertainamountoftime it isexpectedthattheamountofpeoplethatwillbeaskedquestions,willgrowduringthelifespanoftheproject.



37

7 InnovativesolutionsuptakeTheuptakeof innovativesolutions largelydependson the“riskappetite”of theprocuringbody,somepublicsectorbodieswillperceivethattheprovisionof“informationanddigitalsolutions” needs to be as simple and secure as possible and thus unlikely to adopt anyinnovative solutions, such as the adoption of the EUCoC andDCMMmethodologies, newcoolingsystems,loadsharingandrenewableenergysolutionsiftheyactalonetoprocureorspecifydatacentreservices.

Anaddedissueisthedrivetowardstheuseofcloudservices(SaaS,IaaS,PaaSandXaaS)bypublicsectororganisationswherethe“datacentre”islargelyhiddenfromview.Add the fact that most public sector organisations will be tied into Managed ServicesProviders (MSP) contracts that prevent the adoption of unusual solutions, indeed thedraftingoftheprincipalcontractmaynotevenincludeanyreferencetoenergysavingortheuseofinnovation.Some of these problems will be addressed by increasing skills of procurers addressed inChapter7,howevertheEURECAprojectneedstoengagewiththesuppliersofMSPsolutionstoensurethattheyareawareoftheEURECAtoolandthedirectoryofinnovativesolutions.The EURECA project team has engaged directly with end users and will commencediscussionswithMSP’stoidentifysuitablecandidatesfortheprojectsaswellasthepublicsectorbodiesthemselves,andasofthisdate(October2015)has2confirmedmeetingsthatwillcover11possibleprojects.Our approachwith regard to the uptake of innovative solutions is twofold: The first is toidentifysuitablepublicsectororganisationsthatareatthefirststageofpublicprocurementand who are seeking expert assistance; these were identified through the use of publicprocurementportals.The second is to engage withMSP’s with public sector clients, these include consortiumpartnersandotherorganisationsassociatedwiththeDCA,itisintendedthat6EURECApilotpartnersofthisgroupwillbeidentifiedbyDec2015viathisroute.Duetothenatureofaprocurementexercise,itisessentialtocaptureorganisationsatapretenderstage.TheuseoftheEUprocurementguidelines2014touseexternalconsultantsfordesignassists theEURECAproject in this requirement,however,notall EUcountrieshaveshiftedtothenewdirectiveyetandthismayhamperactivitiesincertaincountries.Ourmeasurementoptions(atpresent)willbelimitedtothoseorganisationsthattheprojectcollaborates with. This will be reported by a case study that provides the details on theinnovativesolutionsadoptedinaparticularfacility.Inthefuture,ourfollowuptaskstriggeredbytheuseofthetoolwillprovideadditionaldataforanalysis.



38

Fromourexperience,partsofthecolocationmarketareadaptingdesignstoincludesomeofthelessinnovativesolutions,suchasfreecooling,useofboreholesforcoolingandrecharge,some limited renewable energy either on or off site, but has yet to engage with otherorganisations to adopt load sharing, renewable energy rebalancing, and waste heatsolutions.Finally, special attentionwill have to be laid on the questionwhat actually and rightfullyconstitutesan“innovative”technologyorothermeasures,asvendorstendtomarketmuchof theirproducts as “innovative”.Wewill differentiate thisperprocurement scenarioandalsoreflectthisdifferentiationinthemarketdirectory.Ourinitialthoughtsaretomeasureonacasebycasebasis,ourrationaleisthatevenasmallchange to a design, for instance raising the inlet temperature to the higher levels ofASHRAE’sallowablerangewillyieldsignificantimprovement.TheadoptionoftheEUCodeofConductforDataCentres’155bestpracticesataverybasiclevel(thebestpracticesthatarenot capex intensive)will result in an even greater saving andmaynot be as a result of aprocurementexercise,butduetotheotherdeliverablesofEURECA.



39

8 References/Sources42U(2015,April20).retrievedfromhttp://www.42u.com/

Andrae, S. G. & Edler, T. (2015, April 22). On Global Electricity Usage of CommunicationTechnology: Trends to 2030. Retrieved fromhttp://www.google.de/imgres?imgurl=http%3A%2F%2Fwww.mdpi.com%2Fchallenges%2Fchallenges-06-00117%2Farticle_deploy%2Fhtml%2Fimages%2Fchallenges-06-00117-g004-1024.png&imgrefurl=http%3A%2F%2Fwww.mdpi.com%2F2078-1547%2F6%2F1%2F117%2Fhtm&h=666&w=1024&tbnid=gbHFTT5NQBc7pM%3A&docid=0GC_WldezGD1BM&ei=Cy8vVtwmh7GwAeDAtegE&tbm=isch&iact=rc&uact=3&dur=1109&page=4&start=135&ndsp=48&ved=0CMoBEK0DMEI4ZGoVChMInNjGvJfiyAIVhxgsCh1gYA1N

BiobyDeloitte&Fraunhofer IZM(2015).Preparatorystudyfor implementingmeasuresofthe EcodesignDirective 2009/125/EC.DG ENTR Lot 9 - Enterprise servers and dataequipment.Task5:Environment&Economics.Draftfinalreport,June2015

ClimateGroupandtheGlobale-SustainabilityInitiative(GeSI).(2008).SMART2020:Enablingthe low carbon economy in the information age. Retrieved fromhttp://www.smart2020.org/_assets/files/02_smart2020Report.pdf

Cruz M., Pérez M., Almarza F., Cortés A., Pascual J., (2012). 2nd Annual Report onExploitationActionsandImpactAssessmentD7.5

DCD Intelligent (2012). 2012 Census Report: WESTERN EUROPE DATA CENTER MARKETTRENDS2012-2013AMarketResearch&AnalysisReport

Delforge P. (2014). America’sData Centers AreWastingHugeAmounts of Energy- CriticalactionNeededtosaveBillionsofDollarsandKilowattsIB14-08-AAugust2014

Dutch Government - "Overheid". (2014, February 26). Activiteitenbesluit milieubeheer,artikel 2.15. Retrieved fromhttp://wetten.overheid.nl/BWBR0022762/Hoofdstuk2/Afdeling26/Artikel215/geldigheidsdatum_26-02-2014

ETSI (2011).Technical specification:EnvironmentalEngineering (EE); LifeCycleAssessment(LCA)of ICTequipment,networksand services;Generalmethodologyandcommonrequirements103199V1.1.1

EuropeanCommission(2010).COM(2010)2020finalBrussels,(3.3.2010)EU2020Astrategyforsmart,sustainableandinclusivegrowth

EuropeanCommission(2015a).CircularEconomyStrategyroadmap



40

EuropeanCommission(20215b).COM(2015)215final-Betterregulationguideline

European Commission (2015c). Clean vehicle directive.http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32009L0033&from=EN

EUROPE2020 (2015, June 15). Retrieved from http://ec.europa.eu/europe2020/europe-2020-in-a-nutshell/priorities/sustainable-growth/index_en.htm

EuropeanCommission,Directorate-General JRC,JointResearchCentre. (2015).EUCodeOfConduct In Data Centres On Energy Efficiency. Retrieved fromhttp://iet.jrc.ec.europa.eu/energyefficiency/ict-codes-conduct/data-centres-energy-efficiency

Feindt, S. (2015). EU CODE OF CONDUCT IN DATA CENTRES ON ENERGY EFFICIENCY.Retrieved fromhttp://www.digitaleurope.org/Ourwork/MinimisingourEnvironmentalImpact/EnergyEfficiency/EUCodeofConductinDatacentresonEnergyEfficiency.aspx

GeSI(2015).Smarter2030Retrievedfromhttp://smarter2030.gesi.org/the-opportunity/

Gemeente Amsterdam. (2014). Amsterdam datacentres Save Energy!. Retrieved fromfile:///Users/Frank/Downloads/20140502_dro_datacenters_uk.pdf

Gemeente Amsterdam Dienst Milieu en Bouwtoezicht. (2012). Energiebesparing bijdatacentres. Retrieved fromfile:///Users/Frank/Downloads/eindrapport_energiebesparing_bij_datacenters.pdf

GemeenteAmsterdam.(2015,March13).Bouwbrief2014-129Energieprestatie-eisnieuwedatacentres. Retrieved from https://www.amsterdam.nl/wonen-leefomgeving/bouwen-verbouwen/bouwbrieven-0/bouwbrieven/bouwbrief-2014-129/

GHGProtocol (2013).GHGProtocolProduct LifeCycleAccountingandReportingStandardICTSectorGuidance

Greengrid(2012).DataCentreLifeCycleAssessmentGuidelines

Goiri, Í.;Haque,M.E.;Le,K.;Beauchea,R.;Nguyen,T.D.;Guitart, J.;Bianchini,R.Matchingrenewableenergysupplyanddemandingreendatacentres.AdHocNetw.2015,25,520–534.

Google (2015). Google Data Centers. Efficiency: How we do it. Website.https://www.google.com/about/datacenters/efficiency/internal/ . Accessed 14 Oct2015.



41

Harryvan,D. (2014,May21).ZERVERS.Retrieved fromhttps://www.amsterdam.nl/wonen-

leefomgeving/duurzaam-amsterdam/publicaties-duurzaam/rapport-zervers/

Honée C., Hedin D., St-Laurent J., FrölingM. (2012). Environmental Performance of DataCentres-ACaseStudyoftheSwedishNationalInsuranceAdministration

Kerssens, D. (2015,May 29). Smart sharing van ICT infrastructuur - kan het duurzamer?.Retrievedfromhttp://www.surfsites.nl/duurzaamheid/symposium/presentaties)

Köhn,M.(2015).EURECAinterview,August2015

KoomeyJ.(2011).GrowthinDatacenterelectricityuse2005to2010.Oakland,CA:AnalyticsPress.August1.http://www.analyticspress.com/datacenters.html

Koomey, J. G. (n.d.). Implications of Historical Trends in the Electrical Efficiency ofComputing.IEEEAnnalsoftheHistoryofComputing,03(33),46-54.

Korbee H., Smolders B., Stofberg F. (1979). Milieu voorop bij uitwerking van een globaalbestemmingsplan,THDelft,afd.Bouwkunde,in:BOUW,no.22,27oktober1979

Meza J., Shih R., Shah A., Ranganathan P., Chang J., Bash C.. (2010) Lifecycle-based DataCenterDesign.HPL-2010-117

Parker, H. (2013, February 19). Projection of datacentre Electricity Use. Retrieved fromhttp://www.alliancetrustinvestments.com/sri-hub/posts/Energy-efficient-data-centres

PEFCRUPS(2015a).ProductEnvironmentalCategoryRulesforUninterruptiblePowerSupply(UPS). First Draft PEFCR, European Commission Product Environmental Footprint(PEF)pilotphase.

PEFCR HDD (2015b). Product Environmental Footprint Categories Rules for IT Equipment(storage).FirstDraftPEFCR,EuropeanCommissionProductEnvironmentalFootprint(PEF)pilotphase.

PEFCRMetal sheets (2015c). Product Environmental Footprint Categories Rules formetalsheets for various applications. First Draft PEFCR, European Commission ProductEnvironmentalFootprint(PEF)pilotphase.

PEFCRBatteries(2015d).ProductEnvironmentalFootprintCategoriesRulesforHighspecificenergy rechargeable batteries formobile applications. First Draft PEFCR, EuropeanCommissionProductEnvironmentalFootprint(PEF)pilotphase.

Study_Predicts_Inhouse_Data_Centre (2015, June 15). Retrieved fromhttp://www.cio.com/article/501527/Study_Predicts_Inhouse_Data_Centre_Capacity_Decline



42

TalensPeiroL.,ArdenteF. (2015).EnvironmentalFootprintandMaterialEfficiencySupportforproductpolicy-AnalysisofmaterialefficiencyrequirementsofenterpriseserversISBN978-92-79-51893-5

Telecoms Pricing (2015, June 15). Telecoms Pricing- European Data Centre revenue todouble over the coming 5 years: Retrieved fromhttp://www.telecomspricing.com/news_detail.cfm?item=2592

Webb,M. (2010).SMART2020:Enabling the lowcarboneconomy in the informationage.Report,TheClimateGrouponbehalfoftheGlobaleSustainabilityInitiative(GeSI)

Oró,E.;Depoorter,V.;Garcia,A.;Salom,J.(2015).Energyefficiencyandrenewableenergyintegrationindatacentres.Strategiesandmodellingreview.Renew.Sustain.EnergyRev.2015,42,429–445.

UBA(2015).LeitfadenzurumweltfreundlichenöffentlichenBeschaffungvonProduktenundDienstleistungenfürRechenzentrenundServerräume

Wolf,M.-A.(2014).KPIsforGreenDataCentres.Backgroundreportforstakeholdermeeting1stApril2014,Brussels.



43

9 Annex

9.1 ‘Autonomous’developmentandimprovementpotentialforenergyefficiencyandenvironmentalperformance

9.1.1 HardwareenergyefficiencyTheimprovementofenergyefficiencyofICTanddatacentreinfrastructurehardwarecausesanautonomousincreaseinoverallICTenergyefficiency.Thinkofcoolingmachines,servers,switches, storage, UPS, generators. The main influencers here are the different ways ofcooling efficiency (traditional and innovative) and the ICT hardware energy consumption(mainlytheservers,alsoUPS);seeFigure8.

Figure8 Typicalprofileofenergyconsumersinadatacentre,contributiontooverallkWh/year(Source:42U,2015)

Saving IT energy saves a lotmore in primary energy: 30% of the datacentre energy is ITenergy.Of the100%energy that is being transported tobeused in thedatacentres, only33%isleftforuseinthedatacentre;only9,9%iseffectivelybeingusedintheITequipment



44

It is therefore very important to have energy efficient IT hardware. Servers consumingenergywithoutprocessingdata(idle),isveryinefficient.ProjectZERVERSpointedtoanother>66%savingpotential

9.1.2 TraditionalcoolinghardwareEquipment efficiency improvement of high capacitywater chillers is nearly flat, as can bederivedfromtheANSI/ASHRAE/IESNAStandard90.1-2004/2007/2010,specificallyshownintable6.8.1C(seebelow),theminimumefficiencyrequirementforelectricallydrivenpositivedisplacementchillershaschangedverylittleoveraperiodof6years.

Table2 Equipmentefficiencyimprovementofhighcapacitywaterchillers(ANSI/ASHRAE/IESNA2004,2007,2010)COP=coefficientofperformance,IPLV=IntegratedPartLoadValue

Yearofpublication minimumCOP minimumIPLV

2004 5,50 6,15

2007 5,50 6,15

2010 5,67 -

Thischillertype ismostcommonlyattheheartofthecooling infrastructureof largerdatacentres,whywecanderivethatcoolingefficiencyimprovementscannotbeexpectedfrompurechillerefficiencybutmustbeobtainedfrombetterpractices,includingbutnotlimitedto variable speed fans and pumps, higher ambient temperatures and broader humidityranges.

9.1.3 InnovativecoolingmethodsandhardwareThereisalotofdiscussionwithintheindustryoverthebestandmostefficientwaytocopewiththeproducedheat.Newmethodsofdealingwiththeproducedheatmaycausehugeenergysavings.GoingfromtraditionalenergyconsumingDXkindofcoolingsystemstonewsystemslikeadiabaticor‘free’coolingvariantsbringsdownthePUEdramatically.Manydatacentres have improved their cooling and gained on energy efficiency. Figure 9 shows theexampleofGoogle:



45

Figure9 GoogleactivelypromotesitsprogressonreducingitsPUE(Google2015).

9.1.4 ICThardware(servers)Muchattentionhasbeengiventotheenergyconsumptionofservers.ICThardwarevendorsarecompetingonserver’senergyefficiency.Thecostsofenergyofaserverduringits(short)lifespan,ishigherthanthepricetobuyaserver.Energyconsumptionisthereforecrucialtomakepurchasedecisionsandtotheoperationcostsofadatacentre.Koomey(n.d.)showedthatcomputationcapacitydoublesevery18months.Thismeans,every18monthsserversbecometwiceasefficient.WiththesameIT loadonecoulddowithhalfofthenumberofserver.Youngserversinthedatacentressaveenergyduring,theyhaveabetterenergy-useperformance: “Using historical data, the research team created a graph comparing theamount of computing power of the average computer (from supercomputers to laptops)with the amount of electricity it needs and found that over time, energy efficiencyimprovementsfromthe1950’stillnow,havemovedinvirtuallockstepwithincreasesintheamountofprocessingpower1:energyefficiency,theyfoundeffectivelydoubledevery1.57year,or18months”(seeFigure10).

1Actually,Mooreusedthenumberofprocessorsonachip.Thismaynotbetheexactproxyfortheamountofprocessingpower.



46

Figure10 Moore’s Law of increased computational performance with time.(http://phys.org/news/2011-09-koomeys-law-power-efficiency-parallels.html#jCp).

9.1.5 DatacentreefficiencyingeneralData centres are improving the energy efficiency, also without EURECA. The energyefficiencyindatacentreshasarelationwiththework(proxy:trafficin)andtheenergyusedofdatacentres.TheEuropeancodeofconductforEnergyefficiencyinDataCentres(EUCoC)registersthePowerUsageEffectiveness(PUE)ofparticipants,withthePUEasproxyfordatacentreenergyefficiency(assumingnodifferencesincomputationalperformanceandload).The110participantswith their 235data centres (October2015), register theTotal FaciltyenergyandITenergyasseparateitems,andthecalculationforPUEisdeterminedfromthisinformation. The participants use altogether a total of over 2 TWh annually. The average



47

Power Usage Effectiveness (PUE) is on average around 1.7. The improvement potential,judgedfromcurrentstate-of-the-artdatacentresclaimsPUE’sbelow1.2.

There is a noticeable trendwithin the data centre industry to increase energy efficiency.Therewillbedatacentresthatareveryconcernedwiththeenvironment,whilesomeofthedriverswillbemoremundane:

- costreductionofoperationtoimprovemargins

- pressureonprices

- increasedcompetition

- consolidationofthenumberofdatacentres

- theinvestedsumsrequireprofits

- themanagement/financialneedtofillupdatacentresspaceasap

- Public bodies require PUE <=1.2, (Gemeente Amsterdam, 2015) for newconsolidated(from70to4)datacentrearchitecture,similarinGermany

- Repeatedthreatsofincreasedenergyprices.

9.1.6 ThePUEasaproxyfordatacentreefficiencyHowtocalculateaPUE isalreadyadiscussion forsometimes.Certain typesofmeasuringare inspiredby thewish to showa lowandpreferablePUE, forcommercialpurposes; seeFigure11.



48

Figure11 PUEandDCiEasproxiesfordatacentreenergyefficiency(GreenGrid2012).

Ownersofdatacentres,wanttoshowtheircustomersthattheirdatacentreisefficient.

9.1.7 Limitations of the PUE for comparisons and for energy andenvironmentalimpacts/savings

Havingsaidallthis,afewpeculiaritiesaretobehighlightedaboutthePUE:



49

- ThemoreenergyefficienttheIThardwarebecomes(servers,storageandswitches),thehigherthePUEgets

- AlsovirtualisationcancounteractthePUEorisnotcapturedbyit.Virtualisationwasthe by far main source for reducing Germany’s federal level ICT electricityconsumptionby40%from2010to2014(Köhn,2015).

- AsthePUEisunrelatedtotheperformanceoftheservers,twootherwiseidenticaldatacentres,onewithbrandnewservers,onewith20yearoldservermodelsandhenceordersofmagnitudelowercalculationperformance,willhavethesamePUE.

- ProductionofthecapitalgoodsisnotcapturedbythePUE,asare…

- theenergysourcessuedtoproducetheelectricity.ThePUEishencenotagoodproxyforenvironmentalandenergyperformanceofdifferentcentres, isvaluableonlyfor improvingcertainaspectsofexistingdatacentres,respectivelyneedstobepartofamorecomprehensivepackageofevaluationinformationandindicatorsto actually capture the energetic and environmental performance of data centres. Apartfrom this, the PUE alone does also not deliver information on the absolute amount ofelectricityspentandhenceprimaryenergyusedorsaved.

9.1.8 SpeedofPUEimprovementsHowever,thedatacentreenergyefficiencyimprovementisslow:DigitalRealtyTrust's2014research with interviews in the mid sector of datacentres learns that “only 27% of theinterviewee’s thought their firmmeasures PUE, while 42% said the plan to do so in thefuture.Of thosethat thoughtPUEwasbeingmeasured,73%couldnotcite thevalue.ThemedianPUEforthosethatdidknowwas2.0.Bottomlineisthatonly20%ofthese"seniorleveldecisionmakerswithresponsibilityfordatacenters"knewthemostbasicindicatoroftheirfacilities'energyefficiency”.

TheUptimeInstitute’sannualsurveyshowsthesamepicture:StartingwithPUE1.89in2011asanindustryresult,averagePUEamongthecompaniessurveyedwentdownto1.8in2012andfurtherdownto1.67thefollowingyearof2013.In2014,however,itwasslightlyupto1.7:



50

Figure12 PUE values’ change with time, based on the survey participants’ self-assessment (Uptime Institute, 2014). From Bron :http://www.datacenterknowledge.com/archives/2014/06/02/survey-industry-average-data-center-pue-stays-nearly-flat-four-years/

AnearlierDigitalRealtyTrust's2013 researchshowedmanydatacentresoperatingabove2.0PUE.

"12%oftherespondentsisunfamiliarwiththetermPUE.19%doesn’tknowtheirPUE.TheaveragereportedPUEis2.53.6%reportaPUEof3ormore;28%reportaPUEbelow2.0."

ThefeedbackreceivedfromtheEuropeanCodeofConductfordatacentresshowsthatthe±100participantsreportaPUEaround1.8-1.7.

9.1.9 EnergyefficiencyandreuseThe energy used in data centres is often restricted only to the energy the data centreconsumes,andtakesfromthegrid.Thewaste, i.e.theheatthatadatacentreproduces iscooledaway,withfreecooling(i.e.naturaloutsidecoldairorwaterflow)asitsmostenergyefficient variant. The reuse of energy is often overlooked but in certain data centres thenewest developments. Examples are a proprietary data centre (2010) of the City ofAmsterdam and the KPN Eindhoven datacentre (2014) that offers all kind of colocationservices. Both innovative datacentres offer the produced heat into an efficient heatingsystemforoffices intheproximity.These initiativesandexamplesarehardtofindbutthewaytogo.Thewastedheat(evenwith‘free’cooling)hasaresidualenergyvaluethathelps



51

increase the energy efficiency,when reused properly, i.e.when itmeets a real need andthenreplacesalternativewaystoproducetheheat(e.g.naturalgasbasedheating).

At the same time, one has to be aware that the loss of primary energywill still be large(giventhelossesofbetween40and70%whenconvertingenergycarrierstoelectricity)andthatrarelyayear-rounddemandforthewasteheatexists.Reuse(ormoreprecisely:furtheruse)ofthewasteheatishenceonlyanadditionaloption,whereitmakessense.Prioritywillalwaysbetoreduceenergyconsumptionfirst.

9.1.10 UseofpowerindatacentresThegeneralstatementisthattheenergydemandisgrowingfast.Researchindicatesthatthegrowthofthepowerdemandindatacentresismoreorlessstable.ThistrendisconfirmedbytheDatacentreDynamics2013report,thathasbeenforecastingaverysmalldecreaseintotaldatacentreenergyusefor2016,afteryearsofconstantgrowth,whatwouldreverselyconfirm the fact that average equipment utilization is either stable or slightly decreasing.Thefuturehastoshowtherealdevelopment.

Figure13 Globaldatacentreenergydemandforecastof2013(DatacenterDynamics2013)

9.1.11 TriasEnergeticaThemostenergyefficientwaytoreducetheuseofenergyorpowerneededistoreducethewasteofenergy,particularlytheamountofnon-essentialwork.Thisisnotwithinthescopeoftheproject.Thegrowingamountofremotedataprocessingandstorageisagivenfactfor



52

EURECA.Still,itwillbeanadvicetopublicsectororganisationtochecke.g.fordatastorageneed and – in same cases –whether video and other data intensive applications are themostsuitablewaytomeettheactualservicebehind.

Figure14 TheTriasEnergeticaconcept;seetext(basedonKorbeeetal.1979)

Theotherwayofsavingenergyistousedifferentwaysofcarryingoutthework.Usemoreenergyefficientcooling,software,hardware,etc.Itisalleasytofindwhattodo,whatkindofbestpractisestoimplementandreducetheunnecessaryinefficientuseofenergyinthedatacentre.

9.1.12 ConclusionsonautonomousdevelopmentAssumingnochangeinserverutilisationpercentage,aresultingdecreaseintotalenergyusemaybeexpected.

Combining the effects of a near flat PUE, a near flat ICT equipmentutilisation and an ICTcapacity (demand) usage growth of 60% CAGR that nearly matches the ICT equipmentefficiencyimprovement(70%CAGR)leadsustothephenomenalenergysavingspotentialofdemanddrivenbestpracticesindatacentres.



53

ItiswidelyacceptedthatstrictadherencetothebestpracticesdetailedintheEUCoCwouldleadtodatacentreswithanaveragePUEof1.2.ThepotentialenergysavingsresultingfromfollowingallrecommendationsoftheEUCoCadoptionwouldhencebe30%oftotalusage.

Thesesavingsreflectexclusivelythehowever importantusephaseefficiency improvementindatacentrefacilities.TheincreaseinICTequipmentutilisationfrom10%to40%averageutilisation,atthesametime,wouldlowerthetotalenergyrequiredforrunningthisICTloadwithafactorof3.Sincethecomponents interrelateinthedatacentresystem,thesavingscannotsimplybesummedup.Anestimationofthetotaleffectintheorderof70%decreasein total use phase energy requirement is however reasonable. Increasedutilizationof thehardwarewouldontopreducethenumberofserversandstorageequipment(andtheothercapital goods of the data centre that relevantly contribute in life cycle perspective) toprovidethesameservice,furtherandsubstantiallycontributingtheoverallsavingspotential.

In conclusion, there are large potentials beyond the PUE that are to be identified by theEURECAconsortiumoverthecomingmonths,beconnectedtoprocurementscenarios,andfinallytenderspecificationsthatmeetthespecificdemandofthepublicbody.

Project No: 649972 Title: Data centre EURECA Project (EURECA) · 2018-09-15 · The data centre...

Documents

Transcript of Project No: 649972 Title: Data centre EURECA Project (EURECA) · 2018-09-15 · The data centre...