Phased Deep Retrofit Project: Real-Time Measurement … · 2017 International Energy Program Evaluation Conference, Baltimore, MD Phased Deep Retrofit Project: Real-Time Measurement

Phased Deep Retrofit Project: Real-Time Measurement of Energy

End-uses and Retrofit Opportunities

FSEC-PF-470-17

August 10, 2017

Submitted to

2017 International Energy Program Evaluation Conference Baltimore, MD

Authors

Karen Fenaughty, Danny Parker, and Eric Martin

This article or paper was published 2017 International Energy Program Evaluation Conference, Baltimore, MD. Material contained in conference proceedings may be reproduced if the appropriate citations are included and credit is given to the Conference and the authors.

International Energy Program Evaluation Conference (IEPEC)

Madison, Wisconsin 53705

phone: 608.231.2552 • web: https://www.iepec.org/

Disclaimer The Florida Solar Energy Center/University of Central Florida nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the Florida Solar Energy Center/University of Central Florida or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the Florida Solar Energy Center/University of Central Florida or any agency thereof.

2017 International Energy Program Evaluation Conference, Baltimore, MD

PhasedDeepRetrofitProject:Real-TimeMeasurementofEnergyEnd-usesandRetrofitOpportunities

KarenFenaughty,DannyParker,andEricMartin

FloridaSolarEnergyCenter/UniversityofCentralFlorida,Cocoa,FL

ABSTRACT

Thispaperdescribeshowutilitiescan“makereductionsreal”throughreal-timemeasurementofenergy end-uses and corresponding retrofit opportunities.A field evaluationof themethodologywasconducted from 2012 to 2016 with Florida Power and Light (FPL), an investor owned utility. AcollaborativeprogrambetweentheU.S.DepartmentofEnergy(DOE)BuildingAmericaandFPLledtoanambitious residentialenergy-efficiencyretrofit studyaimedto representFPL’scustomerbase.Fifty-sixexisting,allelectric,occupiedFloridahomeswereinstrumentedtocollectone-minutedataonmostallenergy end-uses in advance of energy-efficiency retrofits. Baseline measurements enabled thedevelopmentofend-useprofilesfortheutility’sserviceterritory.Thesamplethenservedasatestbedto evaluate and quantify energy and peak demand reductions from a variety of packaged retrofits(“shallow” and “deep”) and individual emerging technologies. Many of the measures producedimpressive energy-use savings for homeowners and reduced demand during utility-coincident peaksummer and winter hours. This paper presents details on the recruitment, monitoring equipment,statistical evaluation, and the innovative data platform used to collect and manage millions of datapoints. Using lessons learned from the Florida study, a similar project is in the planning stages forCalifornia. In addition to efficiency retrofits, the California study aims to evaluate advanced meterinfrastructure(AMI)datadisaggregationschemes,solarelectricoutputbytilt,orientationandlocation,influenceofelectricvehiclecharging,anddistributedelectricalstorage.

Introduction

The University of Central Florida’s Florida Solar Energy Center, with funding from the USDepartment of Energy’s Building America program, collaborated with Florida Power & Light (FPL) toconductapilotphasedresidentialenergy-efficiencyretrofitprogram.Thepartnershipwasformedgivencomplementaryinterestsofthepartners:BuildingAmericahasagoaloflarge,wholehouseenergyusereduction for existing homes, and seeks solutions to technical andmarket adoption barriers. FPL, inaddition to updating information on the magnitude of various energy end-uses across their serviceterritory, was interested in evaluating individual component technologies and retrofit packages thatmighthelpmeetfutureenergyuseandpeakreductiongoals.

ForthePhasedDeepRetrofitproject(PDR),atotalof56allelectrichomeswererecruited,withtwo years of pre-enrollment monthly utility data obtained for each research site. The homes werespread across the utility partner’s territory in east Central Florida, southeast Florida, and southwestFlorida. Thestudysiteswereauditedwithadetailedprotocol, includingblowerdoorandduct leakagetesting.Theywerethen instrumentedtocaptureenergyconsumptiononupto18end-uses.Detailed,monitored end-use datawere collected pre- and post-retrofit, alongwithmonthly utility billing data.Increasinglyforutilityprogramevaluations,algorithm-basedAMIdisaggregationmethodsareusedforformalizedmeasurementandverificationpurposes.Whilerigorousandwidelyaccepted,suchmethodshavetheirdrawbacksincludingtheneedforverylargesamplesizesforstatisticalsignificance,andlargeend-useestimationerrors.For instance, ina645-homestudy,Cetin,Siemann,andSloop(2016)found


predictionofdisaggregatedHVACloadserredbyalmost18%.Incontrast,directdetailedmeasurementofenergyend-usesprovidesformuchgreateraccuracyandthereforesmallersamplesizescanbeusedwith greater confidence. Moreover, direct measurement allows for potential improvement to AMIdisaggregationestimationschemesbyprovidingtruepowerforrefinedestimationalgorithms.

SampleRecruitment,Characteristics,andRepresentativeness

The 56 all-electric, single-family homes, located in Central and South Florida, comprised anopportunitysample.Thestudyoriginallyrecruited60sites,butfourhomeswerelostthroughattrition.FPL assisted with recruitment via press releases, and participants were largely self-selected withinprogramlimits.Thismediaoutreachgeneratedanoverwhelmingresponseasparticipantswereeagertoreceive freehome improvementmeasures.Sampleselectionavoided thenewesthomesand targetedmoderately-sizeddwellingstomakeresultsmoreappropriatetoretrofitprogramsaimingtoreacholder,less efficient houses. Living area ranged from 1,000 to 2,650 ft2; vintage ranged from 1942 to 2006;ceilinginsulationR-valueaveraged22hr/ft2-oF/Btu;andairtightnessaveraged8.5airchangesperhourat50Pa.AsrepresentativeforFlorida,atypicalstudyhomehadsingle-glazedwindows,slab-on-gradefoundation,masonrywalls, asphalt shingle roof, electric resistancewater heating and a 10-year pluscentralairconditioningsystem.Onethirdofthehomeshadpools.PDRprojectintent:

• Bestatisticallymeaningful,representingall-electrichomesgeographicallyinFlorida• Includeonlyhomesthatweretobeoccupiedyear-round(notseasonal)• Betypicalofexistingsingle-familyhousingwithconstructionfrom2006orearlier• Includearepresentativesaturation(33%)ofswimmingpools.

Withtheseselectionguidelines, itwashopedthatthemeasuredelectricityusewouldbetypicalofallelectricnon-seasonal,single-familyhomesintheFPLserviceterritory.

MeasurementandEquipment

Detailedauditdatawereobtainedfromallhomes:housesizeandgeometry, insulation levels,materials,finishes,andequipment.Eachhomereceivedandenvelopeairtightnesstestconductedwithablowerdoorandaductleakagetestusingductblasterequipment.Photographswerealsotakenofthehome exteriors, appliances, equipment, thermostats, and associated labels. Showerhead flow ratesweremeasuredusingaflow-catchapparatusshowingarangeofauditedflowsof0.9–4.4gpm.

House power and the various end-uses were monitored by a 24-channel data logger(PowerhouseDynamicsSiteSage)using20-and50-ampcurrenttransducers.Whilemostend-usesweredirectlymeasuredatthecircuit,“lightingandother”wereobtainedbysubtractingalltheend-usesfromthemeasuredhousetotalpower.Thesedataloggershaveastatedaccuracyof±1%between10%and130%oftheirratedoutput.Aportablepowerlogger(WattsUp?)wasusedtoobtainenergyusedataonsomeremoteend-usesthatwerenotonisolatedcircuitbreakers,i.e.washingmachines,themainhomeentertainment center, game systems, and home office and computer workstations. TheWattsUp? isaccurate to 1.5% of stated full load. Portable loggers (Point Six andOnset HOBO)were used to taketemperatureandhumiditydata.Thesehaveastatedaccuracyof±0.95°Ffortemperatureand±3.5%RHforrelativehumiditiesupto85%.DataareretrieveddailyovertheInternetviabroadbandconnectionona1-hour timestep.Forgreater resolution,1-minutedataare retrievable forallend-uses.Ambienttemperature and relative humidity (RH) data were obtained from nearby National Weather Servicestations,typicallylessthan20milesawayfromthestudysitewithastatedaccuracyof±1°F.


Periodically,Floridautilitiesarerequiredtoperformahomeenergysurvey(HES)ofthehousingcharacteristics in their service territory for submission to the Public Service Commission. The 2010survey provides a convenient method to compare the characteristics of the homes in the PDRopportunitysampletothelargerstatisticallydrawnsurveyevaluatedbyFPL(FPL2010).Acomparisonofthe HES Public Service Commission’s survey and PDR data revealed that the samples are quitecomparable relative to both electricity use and demographics (Sutherland et al. 2016). Indeed, theaverage totalmeasuredannual electricityuse in the sampleand in theFPL surveywerewithin5%ofeachother(16,963vs.17,843kWh).

DataManagementPlatformPre-RetrofitData

Adedicatedwebsite(www.infomonitors.com/pdr/)wasdevelopedtohostthelargequantityofmonitored energy data from the project. Hourly data are available for each site and for each energyend-use.Theplatformprovidesdetailedandsummarydataintableorgraphicalform,withflexibilityinchosendatesandsitesforevaluation.Whilethewebsitehasunrestrictedaccess,tominimizeinfluenceon behavior, study participantswere notmade aware of its existence nor their unique identificationnumberuntiltheendofthestudy.Figure1depictsanend-usebarforeachsiteforthefirstfullyearofmonitoring(2013).Thedifferentcolorsineachbarrepresenttheamountofenergyusedbyvaryingend-usesanddemonstratesthateachhomehasauniqueenergyuseprofile.Spacecooling(brightblue) isthedominantend-use(asexpectedinahot-humidclimate),howevercoolingisnotthehighestuseforeachsite.Othertypical largeend-usesare lightingandotherplug loads(orange)andpoolpump(lightgreen).Theinteriortemperature(orangedots)andrelativehumidity(tealdots)areindicatedatthetopofthefigure.

Figure1.PDRhomeend-useduring2013,bysite.

Figure2representsthesesamedatainaggregate.Andwhilespacecooling(brightblue)makesup37%ofthewholehousedailyconsumptionforthesampleaverage,itisnotablethattherearemanyotherend-usesthatrepresentlargecontributionstoaggregateuse.Theselargeend-usesincludewaterheating(11%,brightred),lighting(8%,orange),fansandplugloads(8%,darkred),poolpumps(7%withitscontributiondiminishedoverthefullsample,lightblue),andrefrigeration(7%,darkandlightgreen).


Figure 2. PDR home end-use during 2013, in aggregate. Average total whole houseenergyuse=44.1kWh/day;16,080kWh/year.

Finally,byaveragingtheenergyuseforallhomesbyhourofday,wecanseethemakeupofthedailydemandprofileasshowninFigure3.Beyondspacecooling,lightingandother(orange)andwaterheating(red)aresignificantcontributorstopeakload.

Figure3.Dailydemandprofilebyend-useforthePDRhomesduring2013.

The plots in Figure 4 below provide examples of how the data can highlight time-related

tendencies foran individualend-use–waterheatingenergy in this case.Theplatformalsoallows forend-useexaminationofindividualsites,groupsofspecificsites(e.g.allthedeepretrofithomes),ortheentirePDRsample.Inthegraphicpresentation,theaveragewaterheatingenergyuseforallPDRsitesplottedhourly,daily,andbytimeofday.


Figure4.PDRsamplewaterheatingenergyplottedhourlyaverage(upperleft),dailyaverage(upperright),andaveragedailyloadshape(lower).

The hourly average data plot (upper left) illustrates that water heating energy use is aboutdoubleinwinterthaninsummer;thedailyaverageplot(upperright)showswaterheatingenergyusespikesdramaticallyduringtheThanksgivingandChristmasholidays;thedailyaverageloadshape(lower)displays thebi-modaldistributionofwaterheatingenergy thought theday,with the firstandhighestpeakdemandat7:00amandthenasecond,smaller,butbroaderpeakfromabout5:00to8:00pm.

RetrofitScope

The shallow retrofits were installed on the whole 56-home sample by project staff. ShallowmeasuresincludedtheinstallationofcompactfluorescentandLEDlampstoreducelightingenergyuse.Toreducedomestichotwaterenergy,waterheatertankwrapswereappliedandlow-flowshowerheadswereinstallediftheflowoftheexistingheadexceeded2.2gallonsperminute.Refrigeratorcoilswerecleanedifdirty.Also,poolpumptimerswereresettoreduceofoperatinghourswhentheyexceeded5hoursperday.“Smartplugs”wereprovidedforhomeentertainmentcenterswhenmeasuredstandbypower loads exceeded 10 watts continuous demand. The installations were installed at thehomeowners’discretion(for instancesmartplugswereoftennot installedevenwhereapplicable,duetohomeownerpushback).Auditsconductedatthetimeoftheshallowretrofitsprovidedmoredetailedinputdatatothetailoreddesignofeachdeepretrofit.

The deep retrofits, installed by local contractors,were applied to a sub-sample of 10 homes.Deep retrofit efficiency measures included replacement of air-source heat pumps, duct repair, andsubstitution of conventional thermostats with learning thermostats. Heat pump water heaters wereinstalled to reducewaterheatingenergyuse. Pool pumpswere changed to variable-speedunits, andceiling insulation was augmented where deficient. Old and inefficient refrigerators and dishwasherswerereplacedwithmoreefficientunitswhenindicated.

PhaseIIofthePDRprojectincludedevaluationofsingle-measureadvancedtechnologiesappliedto homes that could be studied in isolation and used to refine a retrofit package and identify


technologies less well-proven. Phase II involved the installation of eight energy-efficiency retrofitmeasures: Supplemental mini-split heat pump (MSHP), complete central system replacement with amini-splitormulti-splitheatpump,ductedandspace-coupledheatpumpwaterheater(HPWH),exteriorinsulationfinishsystem(EIFS)forwalls,high-efficiencywindowretrofit,learningthermostat,heatpumpclothesdryer,andvariable-speedpoolpump.

Methodology

Severalevaluationmethodswereusedtoassesstheenergyimpactsofinterventionsdescribed.Evaluations includedmeasured impactsonwholehouseenergysavingsaswellas individualend-uses.Most often, aweather-normalized, space-conditioning disaggregated utility data analysis compared 1year pre-retrofit to 1 year post-retrofit for both the shallow and deep retrofit packages to evaluatewhole-houseandspaceconditioningsavings.EnergydemandimpactswereassessedontheFPLsystempeakwinterandsummerhoursforallmeasures.

Energy impact evaluations for retrofit measures not targeting space conditioning wereperformedbycomparingpre-andpost-monitoringresults.Theseincludedlighting,waterheating,andapplianceenergyuse reductions. In the caseof lightingwhichwasnot an isolatedmeasurement, thecategoryof‘lightingandotherplugloads’isthedifferencebetweenwholehouseenergyuseandalltheremainingmeasured circuits combined. Using data on installed lighting wattage collected during theshallowretrofits,wedeterminedthatthepre-retrofitlightingconsumptionwasroughly51%ofthe"plugloadsandother"end-use.Measuredenergyusefortheshallowretrofitswereevaluated intwoways.The first method was 30-day pre- versus 30-days post-retrofit, which occurred between spring andsummer2013.Becauseweather influencesmanyend-use loads, savingswerenormalized forweatherdifferences. This was necessary to avoid underestimating savings for measures in which energy usenaturally increases with higher seasonal temperatures (e.g., refrigerators) or overestimating savingsfromothers thatnaturallydrop (waterheating).Thesecondenergyuseevaluationmethodcomparedthemonth of October (amoderateweathermonth) before and after the retrofit. In using the samecalendar month for each period, this investigation essentially excluded space-conditioning changes.Therearebenefitsofoneevaluationovertheother; theshort-livednatureof thefirstevaluation(30-daypre/post)minimizes influencesoutsideof the intervention, suchasbehavioral changes,while thesecondevaluation(Octobers)considerssavingspersistencethatthefirstevaluationmethodmaymiss.

StatisticalEvaluationofSpaceCoolingandHeatingMeasures

Measures impacting space conditioning required more sophisticated treatment. Linearregressionanalysisagainstoutdoortemperaturewasusedtoprojectsavingsforthedeepandadvancedtechnology measures that influence space-cooling and space-heating energy use. This included theinstallation of space conditioning equipment and air sealing, space-coupled heat pumpwater heater,wall exterior insulation finish system (EIFS), advanced window replacement, and the learningthermostat.Thesamegeneralmodel–usingmeasuredcoolingandheatingelectricalpowerandthenmodelingagainstoutdoorweatherconditions–wassuccessfullyappliedforeachoftheseevaluations.

Fromanevaluation standpoint,we found thatweatherhad the strongest statistical power toaccountfordifferencesinaveragedailyspaceheatingandcoolingenergyuse.Dailyaveragesweremuchsuperior to hourly data since many building elements such as slab, concrete walls and high-densityfurnishings respond slowly to thedaily temperatureand solar irradianceharmonics that arenaturallyassociatedwiththedailycycle.TimelapsedtemperaturesforregressionswerealsofoundtobeinferiortotheuseofasimpledailyaverageasprescribedintheASHRAE“toolkit”guidetoestimatingresidentialenergysavings(Haberl,Culp,andClaridge,2005).Averagingthehourlytemperaturesintodailyaverages


wasabetterstatisticalpredictorofspace-conditioningenergythanestimatingheatingdegreedaysandcoolingdegreedays at a 65°Fbase for the sameperiods as anticipatedby theASHRAE “toolkit.” Thecoefficients of determination tended to be much superior, mainly because heating degree days andcooling degree day periods with zero or negative numbers that were truncated by the degree-dayprocedure influencedailyspace-conditioningneeds.Forexample,predawnperiodswithtemperaturesbelow65°Freducetherequiredcooling,whereasthedegreedaycalculationsassumethatthesehourshave a cooling degree day value of zero; as a result, daily average temperatures were used for theanalysis. Space-conditioning energy was then plotted against average outdoor temperature, and thedailyaveragebalancetemperature forheatingandcoolingwasdetermined. Insomehomeswithverytighttemperaturecontrol,thesewereoftenthesame.Thetypicaldailybalancepointwasapproximately65°F,althoughthisvariedfromonesitetothenext.

Toestimatepre-andpost-retrofitannualspaceconditioningenergyuse,regressionswereusedtonormalizedailyaverage temperaturesagainstmonitoreddailyHVACenergyuse; thenweassumedthe same outside temperatures were applied to the resulting site-specific, pre- and post-retrofitregressionresults.Theperiodafterthemeasureinstallationwasthencomparedtothepre-installationperiod.Thisallowedforanevaluationofhowenergyusechangedaftertheretrofit.

Next, the pre- and post-retrofit regression results from theweather-normalization evaluationdescribedabovewereappliedtoregionalTMY3weatherdata.Thisallowsthesavingsestimatestobeextended to the various climate zones (Miami,West Palm Beach, FortMyers, andDaytona) that FPLtypically uses for forecasting purposes. For more details on the evaluation methodology includingparameters,resultsandstatisticalinference,seeParkeretal.2016andSutherlandetal.2016.

ShallowRetrofitResultsSummary

Predicted whole-house savings for the shallow retrofit package were similar regardless ofanalysismethod.Adjustingforweather-relatedchangesoverthe30daysbeforeandaftertheshallowretrofits,overallsavingsinhomesaveraged4.2kWh/dayor10.3%ofpre-retrofitmonthlyconsumption.Comparingpre-retrofitOctober topost-retrofitOctober forasubsetof thedataset, savingsaveraged3.6 kWh/day or 7.9% of pre-retrofitmonthly consumption. The utility bill data analysis indicated themoreseverepost-retrofitweathereroded25%oftheactualwhole-houseenergysavings.

Average annual post-retrofit energy bill reduction was 1,030 kWh (2.8 kWh/day; 7%); theweather-normalized post-retrofit energy savings was 1,356 kWh (3.7 kWh/day; 9%). The normalizedsavingswerefoundhighlysignificantata95%confidenceinterval.Whole-houseenergydemandduringtheFPLsystempeakhourwasreduced0.67kWinsummerat5PMand0.25kWinwinterat7AM,asshowninTable1.Theaveragecostincludinglaborfortheretrofitswas$374persite.

Although space conditioning energyuse reductionwasnot specifically targetedby anyof theshallowretrofitmeasures,significantinteractionsbetweentheshallowmeasuresandspaceheatingandcoolingwereobserved.Inparticular,annualcoolingenergyappearedtobestronglyaffected,likelyfromreducedinternalgainsfromthelightingretrofit,butalsosystematicchangestothermostatpreference.

Fromaparticipantperspective,thecost-effectivenessoftheshallowretrofitoutcomelooksverypromising forbroadapplication.Withanestimatedannualsavingsof1,310–1,530kWh/yearataper-site average cost of $374, a simple payback is reached in about 2 years, all measures included. Thecorresponding rateof returnon investment forparticipants isexceedinglypositive (higher than42%).Onepossibleprogrammaticissuewiththeshallowretrofit,confirmedbytheutilitybilldataanalysis,isthatitsmodestsavingslevelsmaybehiddenfromconsumersbybothseasonalweatherchangesaswellasweathervariationsbetweenpreandpostinterventionyears.


ShallowRetrofit:EvaluationofIndividualMeasures

Evaluationoftheindividualmeasuresindicatedthelightingretrofitmeasureasmosteffective.The initialanalysispresentedanaveragedaily savingsof1.2kWh/day; the2012versus2013Octoberanalysis reported even greater savings, 2.4 kWh/day (453 and 874 kWh/year, respectively). Simplepayback for the lighting retrofit averaged 4.9 and 2.7 years, depending on evaluationmethod. Tankinsulationwraps/ showerhead change-outs cut averagewater heating energy by 0.4 kWh/day or 7%.Refrigeratorcoilcleaningwasnotstatisticallyeffectiveanduptakeonsmartpowerstripswaspoor.

We discovered the shallow retrofit caused significant indirect changes to space-heating andspace-coolingenergyuse.Strongevidence indicated that the lighting retrofit’s reduction inheatgainswas responsible. An end-use disaggregation, completed in concert with the shallow retrofits in 41homes,showedannualspacecoolingdecreasedby1,353kWh(16%)coincidentwiththelightingretrofitwhennormalizedtopre-retrofitweather.Meanwhile,theevaluationpredictedthatpost-retrofitannualspaceheatingwouldnearlydouble,withanincreaseof629kWh.Thepredictedannualbaseloadsavingsof 632 kWh is about half the space-cooling energy savings. Detailed space-heating and space-coolingevaluation of hourly monitored thermostat setting data on nine study homes confirmed the aboveinteractionofthelightingretrofitonspace-conditioningenergyuse.

Potential savings from reducingpool pumpinghours appeared significant, but in practicewasdifficult to achieve. In the 19 project homes with pools, nine were already operating less than 5hours/dayandwerenotaltered.Eachofthe10homesforwhichhourswerereducedsavedanaverageof4.6kWh/day.However,thereductioninenergyusewasshort-lived.Thesavingsobservedduringtheimmediatepost-retrofitanalysiswasmarkedlydiminishedintheevaluationlookingseveralmonthsafterthe intervention. Many pump timers were likely moved back to pre-retrofit settings given poolmaintenance pushback on hours of operation. Subsequent research showed that variable-speed poolpumpsofferabetteroptiontoreducepoolpumpenergyusewithverygoodcustomeracceptance. Giventheinterestinhouseholdstandbyloads(clocks,GFIs,computersandfans),wealsoevaluatedminutedatatoexaminethelowestelectricitydemandfortheresidualloadsovertheentireyearof2013.In53homeswithsuitabledata,wefoundthattheaverageminimumresidualdemand-notincludingthemeasuredend-uses-was86Watts(range25-203Watts).ThetimeoftheminimumdemandtypicallycameduringearlymorninghoursinFebruaryorMarch-aperiodoflittlespaceconditioninginFlorida.ResultsforshallowretrofitmeasuresarediscussedindetailinParkeretal.2016.

DeepRetrofitResultsSummary

Forthedeepretrofits,ananalysiscomparedoneyearpre-retrofittooneyearpost-retrofit forthe 10 deep intervention sites to evaluate energy savings. The results show that, accounting forweather, average post-retrofit annual cooling energy use was reduced by 46% (4,336 kWh savings),spaceheatingby33%(854kWh),andbase-loadby17%(1,878kWh).Whole-housesavingswere38%(7,067 kWh). The savings range for individual homes was 22%–52%. The average overall utility billsavingswereslightly lower.Utilitycoincidentpeakdemandreductionaveraged39%forpeaksummerhour(excludingtheshallowretrofitdemandreduction),and60%forpeakwinterhour. PeaksummerdemandreductiononutilityreportedpeakdaysforthedeepsitesisdisplayedinFigure5andinTable1.

Using the incremental package costs at an average of $7,074, simple payback for theimprovementswas8.3years fora12%simpleafter-taxrateofreturn. If theretrofitswerecompletedoutright as in this study andwith an average full cost of $14,323, the economics are less attractive.However,ausefulmodel forautility“deepretrofitprogram”wouldtargethomeownerswhoneedtoreplacetheirairconditioningandheatingsystems—atwhichpointalltheotherimprovementswouldbeperformedoutright.Thisscenarioachievesa10.5-yearpayback.


Figure5.Deepretrofitpeaksummerhourdemandreductionwas1.96kW,39%overpre-retrofit(left).Winterreductionwas2.71kW,60%overpre-retrofit(right).DeepRetrofit:EvaluationofIndividualMeasures

Theindividualdeepretrofitcomponentswerealsoevaluatedseparately(Parkeretal.2016).Thepre- topost-retrofitevaluationof the10HVACretrofitsshowedthat theheatpumpreplacementandduct repair saved an average of 40% of pre-retrofit HVAC consumption, but that lower interiortemperaturesweregenerallychosen(byanaverageof~1°F),evenwiththelearningthermostat.Despitethis“takeback,”coolingsavingswereabout15.4kWh/day(37%).Anothernoteworthyfinding:theeightheat pump water heaters replacing electric resistance units showed consistently large energy usereductionswithsavingsof69%(5.3kWh/day).

ShallowPlus:EvaluatingSpecificAdvancedTechnologies

The “shallow plus” evaluation segment of the project examined individual promisingtechnologies that were evaluated singly so impacts could be isolated. Results of all “shallow plus”measuresanalyzedinthisphaseoftheprojectcanbefoundinSutherlandetal.2016.Highlights:

Mini-splitHeatPumps

Verysubstantialsavingswerefoundfromapplicationofductlessmini-splitheatpumps(MSHP).Thesesystemshavenoductsystemandoftenhavehighenergyefficiencylevels.One-tonhigh-efficiency25.5 seasonal energy efficiency ratio (SEER), 12 heating seasonal performance factor ductlessMSHPswere installed in the main living area of 10 Central Florida homes. These supplemental units wereinstalledwiththegoalofreducingspace-heatingandspace-coolingenergybyminimizingtheruntimeofthe less-efficient existing central system. Results suggest cooling energy use savings of 33% (2,007kWh/yearor7.0kWh/day)andheatingenergyusesavingsof59%(390kWh/yearor6.8kWh/day),foratotalannualsavingsof34%.Theaveragepercentheatingenergyreductionswereconsiderablygreaterthancooling for thesixhomeswithelectric resistancecentralheating.While thecost-benefitanalysissuggestsapaybackof14yearsandanannualrateofreturnof7%,improvedeconomicsareexpectedasthe MSHP market continues to mature with lower costs. A large added non-energy benefit to theconsumerisaredundantheatingandcoolingsystem—highlydesirablegiventhefailurerateofcentral


systems,which tend to be replaced every 12 years and serviced evenmore often. Electrical demandreductionsduringpeaksystemhourswereverygood:0.50kW(16%)forsummerand2.06kW(56%)forwinter,asshowninTable1.FurtherresearchonthesupplementalMSHPinacooling-dominatedclimateiswarranted.Inadditiontoachievinglargeenergysavings,thesupplementalMSHPshowedapotentialtoimproveinteriortemperatureandrelativehumidityconditions.Fullchange-outsfromcentralsystemstomulti-splitsystemsinlargersamplesaredesirable.Athoroughdescriptionoftheproject’smini-splitheatpumpevaluationisgiveninSutherland,Parker,andMartin(2016).

LearningThermostats

“Learning” or connected thermostats regulate the home temperature by self-programmingdepending on heuristic evaluation of user control habits as well as sensed homeowner occupancy.Evaluations of 22 Nest thermostats showed an average space cooling energy savings of 9.6% (498kWh/year or 2.1 kWh/day)—butwith a veryhighdegreeof variation. Themedian savingswere6.3%(219kWh/yearor1.0kWh/day).Sixofthe22sitesexperiencednegativesavings,whichwaslargelyanartifactofpre-retrofitthermostathabits.Averageheatingseasonsavingswere9.5%(39kWh/yearor1.1kWh/day), although themedianwas higher at 18.5% (35 kWh/year or 1.9 kWh/day). Simple paybackbasedonmediansavingsfortheNest isestimatedtobeapproximately4yearswithanannualrateofreturnof24%.Electricaldemandreductionsduringpeaksystemhourswere0.18kW(7%)forsummerand0.25kW(14%)forwinter.Onasite-by-sitebasis,wefoundthatpre-installationthermostatbehaviorand consumers’willingness to use availableNest featuresmade an appreciable difference in realizedsavings. In particular, defeating the occupancy-sensing “away” function appeared to adversely affectsavings.Withitslowcostandquickpayback,thelearningthermostatisagoodadditiontotheshallowretrofitpackage.Evaluationofthismeasure-describedbyParker,Sutherland,andChasar(2016)-foundthat learningthermostatsresultedinincreasestotime-weightedinteriortemperaturesforcoolinganddecreasesforheating.ThesechangeswereassociatedwithobservedHVACenergyusereductions.

HeatPumpClothesDryers

Electric clothes dryers represent 5% (790 kWh) of annual energy use in Florida homes—thesecond largestapplianceenergy consumption,behind refrigeration. Ineightproject test sites,electricresistance clothes dryerswere replacedwith a new fully-condensing unventedWhirlpool Heat PumpClothes Dryer (HPCD). The estimated median energy savings were 34% (264 kWh/year or 0.72kWh/day),andaverageannual savingsare36%(308kWh/yearor0.85kWh/day).Estimatedelectricaldemandreductionsduringutilitycoincidentpeaksummersystemhourwere0.09kW(or48%ofdryercontributed peak demand) as shown in Table 1. Although unvented HPCDs use less electricity thanstandard resistance dryers, they release a significant amount of heat during operation. The interior-locatedunventedunitsledtoveryhighutilityroomtemperaturesandincreasesinspace-coolingenergythat likely compromise identified savings. Given the heat issue, these unvented appliances areappropriate in a cooling-dominated climate only if installed outside the conditioned space. Weanticipate another technologymarketed by LG—ventedHPCD—may be themost appropriate systemtypeforFlorida.Furtherresearchiswarranted(seeMartin,Sutherland,andParker2016).

VariableSpeedPoolPumps

A third of Florida homes have pool pumps, which often use more than 3,500 kWh/year.Replacing standard pool pumps in five Central and South Florida homes with variable-speed pumpsresultedinlargeenergyanddemandsavings.Pre/postenergysavingsaveraged68%(7.3kWh/day)and


rangedfrom49%–80%(4.9–10.3kWh/day).Meanannualenergycostsavingsamountedto$320(2,665kWh/year)withanexceedinglyrapidsimplepaybackof2.7years.Electricaldemandreductionsduringpeaksummersystemhourwereverylarge:1.08kW(86%)asshowninTable1.

Table1.SummaryofPeakSystemHourDemandImpacts

RetrofitPackageorMeasure

PeakSummerDemandPre(kW)

PeakSummerDemandPost(kW)

PeakSummerDemandDelta(kW/%)

PeakWinterDemandPre(kW)

PeakWinterDemandPost(kW)

PeakWinterDemandDelta(kW/%)

ShallowRetrofit 3.41 2.74 0.67/20% 3.72 3.47 0.25/7%DeepRetrofit 4.97 3.01 1.96/39% 4.51 1.80 2.71/60%Min-SplitHeatPump 3.12 2.61 0.50/16% 3.71 1.65 2.06/56%LearningThermostat 2.40 2.23 0.18/7% 1.78 1.54 0.25/14%HeatPumpClothesDryer 0.18 0.09 0.09/48% n/a n/a n/aVariable Speed PoolPump

1.26 0.18 1.08/86% n/a n/a n/a

Conclusions

Fifty-sixallelectricFloridahomeswererecruited,audited,andinstrumentedtocaptureenergyconsumption on up to 18 end-uses. Detailed,monitored end-use datawere collected pre- and post-retrofit. The resolution of end-use data enabled the identification of phenomena and unexpectedbehavior that would have gone unnoticed using algorithm-based disaggregation methods or othermethodsoftenusedforformalizedmeasurementandverificationpurposes.

Baselinemeasurementsfromtheprojectalloweddevelopmentofseasonalend-useloadprofilesfor the utility service territory. The sample then served as a testbed to evaluate energy and peakdemand reductions from a variety of packaged retrofits (shallow and deep) and individual emergingtechnologies. Shallow retrofit energy savings of 8-10% were demonstrated as well as annual energyreductionsaveraging38%fordeeperretrofits.Individualtechnologyevaluationsshowedlargepotentialfor mini-split heat pumps with 34% HVAC energy use reduction, and learning thermostats with 10%HVACreduction.Variable-speedpoolpumpsdemonstrated68%energyusereductionswhileheatpumpclothes dryers showed 34% lower dryer energy use versus conventional resistancemodels. Potentialimprovementstoboththeshallowanddeepretrofitsegmentswereidentified.Refrigeratorcoilcleaningcan be dropped and smart power strips could be optional for the shallow retrofit segment, whileroutinelyinstalledlearningthermostatscouldsignificantlybolsterperformance.

As shown, utilities could potentially offer programs to capture the described project savings.Scaled-up programs, marketed and incentivized, could help utilities meet their energy use and peakreduction goals. Barriers to large-scale implementation include Florida’s reliance on Rate ImpactMeasure(RIM)evaluationsthatfocusonrevenuelossesaswellashomeowneraversiontolargercapitalinvestments.Thesemightbeaddressedthroughrebatesandefficientprogramdesign.

FutureEffort

Using the lessons learned fromthis study,a similarproject isbeingplanned forCalifornia. Inadditiontoefficiencyretrofits in itsvariedclimates, theCaliforniastudyhasgoalsofcreatinga legacysample forbaselineend-use loadprofilesandhow theynaturally changeover time. Itwouldalsobepossible to closely evaluate impacts from roof-basedphotovoltaic (PV) panels byorientation, tilt and


locationaswellashowshadingfromPVarraysinfluencecoolingloads.Theprojectplanalsoincludesanevaluationofhowon-sitebatterystorage andelectricvehiclechargingcaninfluencetheloadshapes.

Detailed end-use monitoring allows evaluation of end-use disaggregation schemes usingAdvanced Meters Infrastructure (AMI) data. Utility time series data could be run through existingdisaggregation schemes to compare estimates to actually measured energy-end-uses.Moreover, theaccuracyofthedisaggregationproceduresconvertingAMIdataintoend-usescouldlikelybeimprovedbyevaluationofhowtoreduceerrorsinestimation(Mayhornetal.2015).

Acknowledgements

The PDR project represented collaborative research between Florida Power& Light Companyand the U.S. Department of Energy, Building America Program. Special thanks to Craig Muccio forprojectdirection.Thanksalso toprojectpartners:CarrierCorporation,PanasonicCorporation,PentairAquaticSystems,WhirlpoolCorporationandNestLabs improved the reachof theproject.Finally,ourappreciationtothepatienthomeownersofthe56researchsitesoverthefour-yearproject.

References

Cetin,K.,M.Siemann,andC.Sloop.2016.“DisaggregationandFuturePredictionofMonthlyResidentialBuildingEnergyUseDataUsingLocalizedWeatherDataNetwork.”Proceedingfrom2016ACEEESummerStudyonEnergyEfficiencyinBuildings,Washington,DC.

Haberl,J.,C.Culp,andD.Claridge,2005.“ASHRAE’sGuidelines14-2002forMeasurementofEnergyand

DemandSavings:HowtodetermineWhatWasReallySavedbytheRetrofit.”PaperpresentedattheFifthInternationalConferenceforEnhancedBuildingOperations,Pittsburgh,PA,Oct.11–13.

Martin,E.,K.Sutherland,andD.Parker.2016.“MeasuredPerformanceofHeatPumpClothesDryers.”

Proceedingfrom2016ACEEESummerStudyonEnergyEfficiencyinBuildings,Washington,DC.Mayhorn,E.,G.Sullivan,R.Butner,H.Hao,andM.Baechler,2015.CharacteristicsandPerformanceof

ExistingLoaddisaggregationTechnologies,PNNL-24230,PacificNorthwestNationalLaboratory,Richland,WA.

Parker, D., K. Sutherland, and D. Chasar. 2016. “Evaluation of the Space Heating and Cooling Energy

SavingsofSmartThermostats inaHot-HumidClimateusingLong-termData.”Proceedingfrom2016ACEEESummerStudyonEnergyEfficiencyinBuildings,Washington,DC.

Parker, D., K. Sutherland, D. Chasar, J.Montemurno, and J. Kono. 2016. “Phased Retrofits in Existing

Homes in Florida Phase I: Shallow and Deep Retrofits (February 2016).” National RenewableEnergyLaboratory(NREL),Golden,CO.

SutherlandK.,D.Parker,E.Martin.2016.“EvaluationofMini-SplitHeatPumpsasSupplementalandFullSystem Retrofits in a Hot Humid Climate.” Proceeding from 2016 ACEEE Summer Study onEnergyEfficiencyinBuildings,Washington,DC.

Sutherland,K.,D.Parker,E.Martin,D.Chasar,B.Amos.2016.“"PhasedRetrofits inExistingHomes in

FloridaPhaseII:ShallowPlusRetrofits."NationalRenewableEnergyLaboratory,Golden,CO.

Phased Deep Retrofit Project: Real-Time Measurement … · 2017 International Energy Program Evaluation Conference, Baltimore, MD Phased Deep Retrofit Project: Real-Time Measurement

Documents