A COMPARISON OF FIELD AND LABORATORY MEASUREMENTS OF RESIDENTlAL REFRIGERATOR ENERGY USE

Transcription

1 HEINEMEIER ET AL. A COMPARISON OF FIELD AND LABORATORY MEASUREMENTS OF RESIDENTlAL REFRIGERATOR ENERGY USE Kristin E. Heinemeier and Alan K. Meier Lawrene Berkeley Laboratory ABSTRACT The laboratory-rated energy use of refrigerators is of interest to onsumers utility foreasters and poliymakers. The laboratory test proedure does not attempt to simulate atual onditions so it must be periodially validated as refrigerator onfigurations and tehnologies hange. The test proedure has not been field-validated for at least 15 years in spite of signifiant improvements in refrigerator energy effiieny. In this study the field energy use of 393 refrigerators were ompared to these rated values (or "labels"). In this omparison the laboratory test overestimated the typial refrigerator's annual field use by about 14 perent. The peak onsumption is also important for foreasting; The peak monthly use was about 16 perent higher than the laboratory test expressed as average monthly use. Post-1982 refrigerators used about 22 perent less annual energy than their labels indiate while the labels aurately predit the onsumption of pre-1982 refrigerators. 4.5

2 HEINEMEIER ET AL. A COMPARISON OF FIELD AND LABORATORY MEASUREMENTS OF RESIDENTlAL REFRIGERATOR ENERGY USE Kristin E. Heinemeier and Alan K. Meier Lawrene Berkeley Laboratory INTRODUCTION Residential refrigerators in the United States onsume the equivalent eletrial output of 2s large eletrial power plants. They are also the largest end use of eletriity in most Amerian homes. As a result the energy use of refrigerators is important to both onsumers and utilities. A standard energy testing proedure was developed for U.S. refrigerators in the In this test the refrigerator is plaed in an environmental hamber for 24 hours while the ambient temperature is maintained at 32 C. The refrigerator doors are kept losed during the test and no food loads are inserted. Adjustments and additional tests are required for refrigerators with speial features. 1 The laboratory test proedure does not attempt to dupliate typial operating onditions. The higher ambient temperatures are used in plae of heat gains from food loading and door opening. For this reason it is essential to ompare test proedure results with field measurements. This was in fat undertaken informally by the National Bureau of Standards about futeen years ago. (The doumentation has sine then disappeared.) Other test proedures have been developed with very different onditions in Europe and Japan.2 Japanese test proedures take plae at lower ambient temperatures and inlude door openings and food loading. However these ompliations gready inrease the ost of the test and ability to dupliate results. The test proedure results are used by the Federal Trade Commission for preparation of energy use labels whih are affixed to every refrigerator sold in the U.S. It is hoped that the onsumers will make better-informed purhasing deisions based on the label information. The reendy-passed energy effiieny standards are also based on energy use as determined by this test proedure. The results are used in an aggregate form by energy demand foreasteis. Trends in refrigerator energy use as refleted in the tests are inorporated in projetions of future eletriity demand. In general the foreasters assumed that a model using the equivalent of 12 kwh/year in the test would also use 12 kwh/year in a home. Refrigerators have hanged signifiandy sine the test proedure was developed. Insulation levels and ompressor effiieny have gready inreased. Evaporator designs and onttois have also improved. These modifiations should make the refrigerators more dependent on ambient temperature beause standby losses have beome a smaller fration of total energy use. As a resuit it is no looger ertain that the test proedure provides areliabie estimate of field use nor even that a model whih uses less energy than another in the lab test will use less in the field This ould have impliations for onsumers utility foreasters and standards-setters. In addition the peak eletriity use of refrigerators has ome under srutiny. The relationship between the rated energy use and the field use during peak periods has not been examined. We report here an updated orrelation between the test values and field onsumption for residential refrigerators. The test energy use was also orrelated with the measured peak energy use. These orrelations will assist utilities in making more aurate assessments of refrigerator energy use on baseload and peak demand. 1 FI'C and DOE regulationl allow manufaturen to "rate" produts onservllive1y; therefore some manufaturen have givm a "olletion" of units the worst "rating" of all units in the group. This would bdon most of ten forlogistial rea SOIlI Le. to!"edue the number of 1ab1s.) 2 A. Meier "F.nrgy Use Test Proedures for Applianes: A ail Study ouapanse Rfrigerators". ASHRAE TrtJlLfatiOM 93 (1987). 4.51

3 HEINEMEIER ET AL. TECHNICAL APPROACH Measured refrigerator energy use for individual refrigerators was obtained from six different field monitoring programs aross the United States and numerous individual measurements (see Table 1). Most were utility loadresearh programs in whih the refrigerators were monitored for over one year. Measurements were taken between 1981 and The minimum data requirement was annual energy onsumption and model number for eah refrigerator although nearly all soures provided at least monthly energy use and some studies measured energy use as frequently as eah hour. The rated (or "label") energy use for eah refrigerator was taken from the Assoiation of Home Appliane Manufaturers (AHAM) diretories.3 These diretories have been published annually sine 1975; however publiation lapsed for four years between 1977 and The diretories also provided the age and some features of eah refrigerator. Despite the great amount of refrigerator field energy onsumption data being olleted in the United States we found only six projets with data suitable for this projet Only these soures reorded the refrigerator model number. Even then there were gaps due to meter failure or lapses in monitoring. If less than a year of data were available the onsumption was annualized for the annual omparisons. However refrigerators monitored for less than nine months were rejeted. sine seasonal dependenies would make an aurate extrapolation impossible. About half of the reorded refrigerator model numbers were not listed in the AHAM diretories. The low mathing rate was due to three fators: The refrigerator was purhased bef ore 1975 (hene predating the diretori.es) or during the gap in the diretories. The refrigerator model number was inorretly transribed (suh as a '' for a '6') by the energy auditor and therefore ould not be loated in the diretories. The diretory model number either laked or differed in the terminal digits or letters. Sine there is no standard nomenlature methodology aross manufaturers these digits ould either refer to irrelevant features suh as olor ol' door opening diretion or to presene of an iemaker or other features whih have a signifiant effet on energy use. Refrigerators with model number transription errors were deleted if the ambiguity led to a signifiant range in possible onsumption As a result of these problems only a fration of the refrigerators that have been monitored in the United States ould be inluded in this study. Table II summarlzes how the potentially huge dataset was narrowed down to a muh smaller number of units qualifying for this type of analysis. COMPARISONS OF LABELED ENERGY USE WITH FIELD MEASUREMENTS The annual energy data and their soures are presented in Figure 1. The data represent 259 refrigerators taken from six major studies. and a few individual measurements. (A ommentary desribing eah data set will be published in a forthoming report 4) Eah data point represents one refrigerator exept the PSE&O refrigerators where eah point represents the average onsumption of from 1 to 14 idential refrigerators. In the ORNL data 41 refrigerators are shown although they represent only two different models (so that all the points are in two vertical olumns).5 The data from a midwest refrigerator manufaturer represents up to five years of data for eah of 35 3 Assoiation of Home AppIiane Manufaturen (AHAM) "1986 Diretory of Certified Refrigeraton and Freezen " 2 North Waker Drive Chiago IL. (Similar diretories were used for earlier yean.) 4 AJan Meier and Kristin Heinemeier A Comparison óf Rated and Atual Energy Consumption in of Residential Refrigeraton' Lawrene Berkeley Laboratory Report (fonhoming). 5 The ORNL data set is IUlUSUal beause the refrigeraton COIIsistently used more than premted by the lab tesl The design feature responsible for this behavior ould not be detennined by the researhen responsible for the data. 4.52

4 HEINEMEIERETAL. refrigerators. A line of equal test and field energy onsumption is drawn in this figure. If the two measurements agreed perfedy Ihen the points would lie along this line. The average annual onsumption of the 259 refrigerators in the ompilation was 1254 kwh per year with a standard deviation of 434 kwh. The measured annual onsumption was ompared to the manufaturer's rating (based on the DOE laboratory test). The results are presented as a satterplot in Figure 1. Most of the points lie lose to the perfet-agreement line indiating that the laboratory test proedure predited field onsumption quite wed. In general annuai field energy use was lower than the laboratory test but rarely by more than 2 perent. The equation of the best-fit lioe is: AnnuaI Field Energy Use = 1. x (Annual Label Use) The rrelation oeffiient (.62). however. was low. (See Table lli). For a typial refrigerator. with a rated onsumption of 12 kwh per year. the label overpred.its field energy use by 14 perent MOfllhly Conswnption Monthly data for 213 refrigerators were omplled. A satterplot -of the data is shown in Figure 2. The highest lowest and average monthly field CODsumption are shown for eah refrigerator. The monthly variation in onsumption was high: it always exeeded 2 perent of the meao and often exeeded 4 perent The peak monthly value for eah refrigerator - whih ourred most frequendy in August - was signifiandy higher than the meao. In general the peak month onsumpoon was higher than the laboratory test onsump OOn. The line best fitting the results was: DOE AnllUQl Label Use Peal Monthly Energy Use = 1.14 x [ 12 ] + 2. The fit however. was poor:the R2 was.46. For a typial refrigerator with a rated annual onsumption of 12 kwh per year (averaging 1 kwh per month) the peak month orresponds to roughly a 16 perent inrease over the label. Daily VariDlion For those few refrigerators where daily data are available it is interesting to observe how the daily onsumption varies aross the year. Figure 3 shows the average daily onsumption for eah day of the year as wed as the average daily onsumption represented by the label annual onsumption (= LabeI/36S). These values are averaged aross the 24 refrigerators for whih this information is available. There is learly a seasonal variation in demand;. the average daily demand during mid-june to mid-august is 26% greater than that during the rest of the year. For this subset of refrigerators (and in ontrast to our onlusion based on the larger set of refrigerators) the rated use is greater than the field use on all but a few days. Vintage RefrigeralOr effiieny has improved gready in the last fifteen yeats. The annual energy data were examined to determine if the improved effiieny was apparent in field performane. Disussions with manufaturers indiated that the greatest hanges ourred around so the data were partitioned into units sold before 1982 and

5 HEINEMEIER ET AL. those sold in 1982 or later. Fi~ 4 is a satterplot of the refrigerators divided into the two age groups. The newer refrigerators used m average less energy than the older units 144 versus 1374 kwh per year. No new units used more than 166 kwh per year although the sample size is smaller (only 69 units). Lines best fitting the two groups were alulated and plotted in Figure 4. Based m the present data it appears that new and old refrigerators have a different relationship between the rated and fleld use. Whereas an old unit's rated and field use will be almost idential new refrigerators will use signifiantly less energy in the field than indiated by their label. For an old and new typial refrigerator both having rated use of 12 kwh per year the old unit will use about 1185 kwh in the field while the new unit will use about 923 kwh (or about 22 perem less). SOURCESOFUNCERT~Y There are numerous soures of unertainty in both the field and test energy data whih ould affet the validity and generalizability of the results. Some of these fators are desribed below. The refrigerators in this ompilation are not statistially representative. Some of the studies did seek to obtain a statistially representative group but others used a group-of-onveniene (usually ompany employees). Few refrigerators were monitored in the Southern part of the ountry. Few studies made any temperature measurements in the refrigerator the kithen or even noted if the house was air onditioneel It was not DOwn if the household had a seond refrigerator (or if the mmitored one was the primary unit) or a freezer. Refrigerators are often used in suh unonventional ways that they have aurated the attention of anthropologists.6 Lak of information regarding the oupants and the homes in whih the units were loated preludes any ompensalory adjustments. Many refrigerators have automati iemakers. The presene of an iemaker does not always show m the model number beause it is of ten an af ter-sale modifiation. One study measured the inreased energy use from iemakers.7 Energy use inreased as muh as 2 perent when the iemaker was inluded while onduting the DOE test However automati iemakers are usually not operating so this must be onsidered an upper limit This error is usually less than 15 perent but an be as great as 3 perent for ertain units.8 This study does not address this soure of error. We found that the results ontinue to flutuate as new refrigerators are added. For example in a previous version on this work whih examined fewer refrigerators and a higher perentage of pre-1982 refrigerators we found a muh smaller differene between field and laboratory energy onsumption.9 Thus we believe that the sample size is still too small and further data olletion will be needed before we will be able to make onlusions with a high degree of onfidene. IMPLICATIONS FOR FORECASTING AND CONSERVATION PROGRAMS Most eletriity demand foreasting models have treated the eletriity used for residential refrigeration as onstant or only slightly varying. This study suggests that refrigerator energy use varies signifiantly aross the seasons. Indeed refrigerators are signifiant ontributors to the summer peak. While the inremental summer inrease II Bru:e Hal:tt and Lorm Lutzenheiser "Shelf Life: An Inquiry Into Whll-and Who - an Be Found in Your Re frigerator" Ener" Auditor & Retrofitter MayfllDle BR Laboratories FilllJl Report " l.tjboratory Tutill ufcertified RefrigeratonlFreeer8. HWllington Bea:h CA: Prepared for the California Energy Commission Agreement No BR Laboratorlea (see previous refenm:e). Alan Meier and Kristin Heinemeier "Energy Use of Residential RefrigeratOl'l: A Compari8Clll of LaboraIory and Fiell Use". ASHRAETr_tioM 1988 (in press) "<"=~u"...~~.~"'nu~'u~~~'~~~<.<."'n"v"~.. m.~ ~~~~mm~.#_... '.'.""N==N=~~mNN'.'.N..."""»"'Y"."... ==m~n.vm"nn~v~~"<">"''''''n'' 4. ~

6 HEINEMEIER. ET AL. of one refrigerator is not a major load the aggregate impat of 1.3 refrigerators per household may represent a signifiant fration of the air onditioning load. (Indeed some of ihe air onditioning peak is probably a refrigeration peak.) To our knowiedge load management tehnologies for refrigerators have never been seriously onsidered in the United States. Modifiations of existing refrigerator designs to redue peak energy use would ertainly be worth investigating. The orrelation presented in this paper allows foreasters to more aurately estimate the ontribution of new refrigerators to residential eletriity demand without extensive monitoring. In its plae however the utility must ollet sales data on the shipment-weighted energy use of refrigerators in its servie area. Many utilities are onsidering rebaie programs to enourage the purhase of energy effiient refrigerators. Most programs rely on the label to alulate the energy savings and ost effetiveness of the rebate programs. This study suggests that the labels overestimate refrigerator energy use for new refrigerators. on the other hand the programs have rarely inluded the peak power savings. This study suggests that peak power savings will be substantially greater than indiated by the label. CONCLUSIONS The measured data on energy use of 259 refrigerators in six separate studies have been ompiled and analyzed. Despite the large number of refrigerators being monitored surprisingly few had suffiient data to permit a omparison with the label onsumption. The DOE energy test proedure appears to be a remarkably good preditor of field annual energy use. The best-fit line indiated that the label slightly overestimated field use but by a onstant amounl At a typial onsumption of 12 kwh per year the field use was 14 perent less than the label. This suggests that onsumer deisions and appliane standards an be based the DOE energy test proedure without introduing a large error. There is signifiant month-to-month variation in energy use. August onsumption was typially the highest and the peak-month energy use of a typial unit was about 16 perent alxjve the label. Utility foreasters should make this adjustment when prediting the refrigerator's ontribution to the peak energy use. New refrigerators have more insulation higher-effiieny ompressors and better evaporators. These fators were expeted to hange the relationship between laboratory and field performane. Indeed post:'1982 refrigerators used about 22 perent less energy than their labels while older models onformed loselyto the labels. Again the sample sizes are small so the results must be onsidered tentative. ACKNOWLEDGEMENTS This work was supported by the Assistant Seretary for Conservation and Renewable Energy Offie of Building and Community Systems Building Systems Division of the U.S. Department of Energy under Contrat No. DE-AC3-76SF98. We also atnowledge the assistane of Arthur D. Little Company Consumers Power Company Paifi Gas & Eletri Company Publi Servie Gas & Eletri Co. Sierra Paifi Power the Whirlpool Corporation and many other individual data ontributors. 4.55

7 HEINEMEIER. ET AL. Table 1. Soures of field monitored data. Soure Lo. Units Interval Notes Consumer's Power Co. MI 31 Monthly Oak Ridge National Lab VA 5 Monthly 2 models: effiient and standard - 4 Monthly 2 ompressor designs: effiient and standard Paifi Gas & Eletri Co. CA 21 3 minutes Publi Servie Eletri & Gas NI 1 Monthly 8 models: mean onsumption for eah model Siena Paifi Power Co. NV III Daily Midwest manufaturer IN 1 Monthly Up to 4 years of data Misellaneous - 4 Annual Misellaneous - 2 Monthly 25 Quarterly Up to 5 years of data TOTAL 394 Some units not inluded due to insumlent data. Table n. Narrowing down of potential dataset Data Status Number of Refrigerators Unit was monitored 5+ (estimated) and model number reorded 394 and model number mathed 259 and at least 9 months data 239 and monthly data 213 and daily data 24 and hourly data. 8 Table m. Regression data. Best Fit Standard Number of Standard EnergyUse Formula R2 Error Units Mean Deviation Maximum Annual Field = 1. x Label (kwh/year) Peakmonth Field = 1.14X(~') (kwh/month) Pre-82 annual Field =.99 x Label (kwh/year) Post-82 annual Field = 1.8 x Label (kwh/year) 4.56

8 HEINEMEIER. ET AL >-.s:::. ~.::.t E :::l rj) U >- l... Q) C LU o :::l «" Q) il: ~ ~----~ o~ \; n \- )( )( ' ' ' ' ' ' ' " ' "" ' ' '''" ' " " ' " " " ' " 3 O~~~_--_--r_~~_~~~~_.~~_~~_.~~_~~_~~~ o Label Annual Energy Consumption (kwh/yr) Consumers 6. ORNL (N = 35) X Sierra Paifi Power MI Power Company (N = 1) NV (N = 16) Misellaneous (N = 5) Publi Servie + V ORNL. VA (N - 33) zz Manufaturer. IN (N = 75) Peifi Ges Eletri and Eletri BEST FIT and Gas. NJ (N = 8) CA (N = 8) Figure 1. A satterplot of DOE laboratory test ("label") onsumption versus the measured annual field onsumption. A line of perfet agreement is drawn for referene; units to the right of the line represent those with the label onsumption greater than the field onsumption. The field onsumption of neady all of the refrigerators was within 3 perent of the label. In general the field onsumption was less than the label. The olumns of units (ORNL PSE&G) represent groups of idential refrigerators. 4.57

9 HEINEMEIER. ET AL. ê E -~ ~.:ti.... l. E ::; I/) U.. > ) Q) w > ~... :E :5! Q) u:: 276~ ~--~--~ o " a~ a\': /"1;) ~O x. x. "" ' ''~:///~ ""' "" " Label Monthly Energy Consumption (kwh/mo) Puk Month IN - 13) o Average Month Minimum Month Figure 2. A satterplot of DOE laboratory test ("label") monthly onsumption versus the measured minimum mean and peak monthly field onsumption. The range in monthly onsumption was of ten greater than 3 perent. Peak onsumption - most frequendy in August - was generally above the label. 4.58

10 HEINEMEIER ET AL 5~ ~------~ :: o... CO 3 E ;: l) :: o u > l "- l) :: LU 2 > t! l l) l t! "- l) > < o+-~~~~.-~-~~.-~-r~'-~-'-r~'-~~'-'-~-'~'-'-~ o Day of the Year Field Measured Figure 3. The average daily onsumption for eah day of the year and the average daily onsumption represented by the label annual onsumption averaged aross the '24 refrigerators for whih daily data are available. There is learly a seasonal variation in demand; the average daily demand during mid-june to mid-august is 26% greater than that during the rest of the year. 4.59

11 HEINEMEIER. ET AL 33~ ~ 3 27 >- -"- '? 24 ~ ~... a. E :::I I/) ~ 21 1eoo : o U >- l 16 "- l) : w ~ 12 : : < " l) Li: / Label Annual Energy Consumption (kwh/yr) o Pre-1982 (N - 121)!2!..!2..fiL o Post-1982 (N - 69) Post 82 Fit Figure 4. A satterplot of DOE laboratory test ("label' ') annual onsumption versus the measured annual field onsumption for pre-1982 and post-1982 models. In this small sample the post-1982 refrigerators (inluding 1982) used less energy than those sold before Moreover the relationship between the test and field onsumption is different for the two groups. 4.6