Model-Based Benchmarking with Application to Laboratory Buildings

Transcription

1 Mdel-Based Benhmarking with Appliatin t Labratry Buildings Cntat persn: Cliffrd Federspiel Center fr Envirnmental Design Researh 390 Wurster all, #1839 Berkeley, CA Tel (510) FAX (510) liff_f@ulink.berkeley.edu Cliffrd Federspiel, Ph.D. Qiang Zhang Edward Arens, Ph.D. Center fr Envirnmental Design Researh University f Califrnia Berkeley, Califrnia 1

2 ABSTRACT The mst mmn methd f benhmarking energy use in buildings is t mpare the energy use f the building under nsideratin with the energy use f a ppulatin f like buildings. Usually there is sme empirial mpensatin fr features and fatrs that affet energy use suh as the size f the building and the weather nditins. Tw fundamental limitatins f this apprah are: 1) nly similar kinds f buildings an be mpared, and 2) the entire ppulatin may be ineffiient, whih wuld ause many ineffiient buildings t be rated as effiient. The first limitatin is imprtant when benhmarking labratry buildings beause there is n publi database f energy use and building features that an be used t nstrut empirial benhmarks fr labratries. The send limitatin is als imprtant beause there is evidene that energynsuming presses in labratry buildings, espeially VAC systems, are ineffiient beause f highly nservative design praties. This paper desribes a benhmarking methd that is fundamentally different than the methd desribed abve. The priniple f the new methd is t nstrut a benhmark that represents the minimum amunt f energy required t meet a set f basi funtinal requirements f the building. These requirements inlude de-mpliant envirnmental ntrls, adequate lighting, et. The benhmark is mputed based n idealized mdels f equipment and system perfrmane. Using idealized mdels prdues a benhmark that is independent f design and easy t mpute. One the benhmark has been mputed fr a single building, an effetiveness metri is mputed by dividing the mdel-based benhmark by the atual nsumptin. This metri, r its inverse, an be mpared with the metris f ther buildings. Sine funtinal requirements have been inrprated int the benhmark, it is pssible t mpare the perfrmane f dissimilar buildings, r buildings that have rare r unique funtinal requirements. The perfrmane f the mdel-based benhmarking methd was mpared with tw alternative methds based n the ability t predit atual energy use. Using building energy data frm the UC Berkeley ampus, it was shwn that the mdel-based benhmarking methd was mre aurate when a mbinatin f labratry and nn-labratry buildings was analyzed. 2

3 INTRODUCTION Benhmarking is ne f the first ativities in the press f deiding whether r nt t invest in energy-nservatin measures in buildings. Cnsequently, imprvements in benhmarking methds uld have a large impat n energy use and the prfitability f mpanies that use energy r prvide energy servies. Mst energy servie mpanies (ESCOs) and ther rganizatins respnsible fr energyeffiieny f buildings use the mean r median value f the energy use intensity (EUI) fr the kind f building being investigated as a benhmark fr determining whether r nt the building is a gd andidate fr energy nservatin measures. The EUI is the average pwer nrmalized by grss plan area f the building. In the U.S., it is typially expressed in kw-h/ft 2 /year r MBTU/ft 2 /year. The EUI aunts fr nly ne building feature that affets energy nsumptin: plan area. T aunt fr the effet f ther features that affet energy nsumptin, benhmarks have been nstruted by using statistial methds t rrelate ther features with energy use [1, 2]. Sharp s methd is based n an analysis f the 1992 Cmmerial Buildings Energy Cnsumptin Survey (CBECS) database [3]. Linear regressin mdels were used t rrelate building harateristis with energy nsumptin. Seven f the 75 harateristis investigated were fund t be statistially signifiant indiatrs f energy nsumptin. Sharp s methd has been mdified slightly and used as the basis f the Energy Star benhmark [4]. Rather than using ensus latin as a prxy fr weather, the Energy Star benhmark expliitly mpensates fr weather. The Energy Star benhmark is the 25th perentile f the EUI distributin beause this level is expeted t be the level required fr mpliane with energy des. An energy analysis ativity that is related t benhmarking is baselining. The key differene between benhmarking and baselining is that benhmarking generally invlves a mparisn f energy perfrmane with ther buildings while baselining generally invlves a mparisn f past energy perfrmane f a single building with urrent energy perfrmane. The mst mmn methds f baselining are similar t the methds desribed abve fr benhmarking. Statistial methds are typially used t rrelate weather data and ther imprtant variables f a single building with measured energy use. Examples f this kind f baselining are desribed in [5, 6]. Labratry buildings nsume nsiderably mre energy per square ft than ther kinds f mmerial buildings, and they are beming inreasingly energy intensive. In [7] it is estimated that energy use intensities in labratry buildings are fur t five times higher than thse fund in nn-labratry buildings, suh as ffies, and that energy nsumptin in labratry buildings in Califrnia is grwing expnentially at a rate f 3.9% per year. In [8] it was shwn that the energy intensity f labratry buildings n the UC Berkeley ampus is three times greater than that f 3

4 nn-labratry buildings. Fr labratry buildings nstruted after 1980 it is six times that f nn-labratry buildings. One f the reasns that the energy intensity f labratries is s high is beause f the VAC requirements that are speifi t labratries. Due t the nature f wrk in labratries, the air hange rate must be higher than in ther kinds f mmerial buildings, and they are usually supplied with 100% utside air. Large quantities f air are exhausted frm the labratry either thrugh the exhaust frm the upied spae r frm fume hds r ther lal exhaust devies. The mvement f large quantities f air auses the fan pwer used by labratry buildings t be high. Cnditining large quantities f air auses the hiller pwer t be high. Cnerns fr upant safety and reliable press peratin mbined with nsiderable unertainty abut the magnitude and variatin f heating and ling lads ften leads t deisins whih result in the ineffiient peratin f labratry buildings. This prblem is amplified by the fat that the energy intensity f labratry buildings is high and the energy nsumptin is grwing expnentially. Cnsequently, there is a need fr tls that will allw peratins staff t determine hw well labratry buildings are perating s that design and peratinal prblems an be addressed. One prblem with existing benhmarking methds is that they d nt suffiiently aunt fr differing funtinal requirements f buildings. This prblem is partiularly aute in labratry buildings, where the funtinal requirements are unique and vary nsiderably frm ne labratry t anther. It als makes it diffiult t apply existing benhmarking methds t dissimilar buildings. Fr example it is nt pssible using existing benhmarking methds t mpare the perfrmane f labratry buildings with ffie buildings. The ability t d s is imprtant beause nearly all labratry buildings ntain nn-labratry spae. Althugh Sharp s methd des aunt fr sme funtinal requirements, many f the funtinal requirements that have a signifiant impat n energy use are nt inluded. Fr example, temperature ntrl, humidity ntrl, ventilatin rate, filtratin effiieny, and plug and press lads are nt expliitly treated as funtinal requirements. Anther prblem with existing benhmarking methds is that all urrent benhmarks are based n the perfrmane f ther buildings. They d nt reflet the extent t whih the energy effiieny uld be imprved beause the entire ppulatin uld be making ineffetive use f energy. Therefre, existing benhmarking methds annt be used by an energy engineer t determine the energy-saving ptential that exists even in buildings that are nsidered t be energy-effiient. METOD There are numerus perfrmane metris in existene fr engineered systems. Tw imprtant kinds are effiieny metris and effetiveness metris. Effiieny metris are used t mpare utput with input. Metris f this type inlude the thermdynami effiieny f a heat engine, whih is shaft pwer divided by fuel pwer, and the mehanial effiieny f a fan, whih is 4

5 aerdynami pwer divided by shaft pwer. Effiieny metris are nt appliable t the develpment f a whle-building energy nsumptin benhmark beause it is diffiult t define the utput f a building and beause it is diffiult t quantify the utput even if it an be defined. The utput is nt the energy nsumptin. It might be the mfrt prvided t the upants, r it might be the wrk utput f the upants. Effetiveness metris invlve a mparisn with a benhmark, and are therefre relevant t the develpment f a whle-building energy nsumptin benhmark. An example f an effetiveness metri is heat exhanger effetiveness, whih is defined as the atual heat transfer divided by the maximum pssible heat transfer [9]. Engineering effetiveness metris d nt always use the theretially best perfrmane as a benhmark. Fr example, ventilatin effetiveness is ften defined as the measured age aumulatin f air in a building divided by the age aumulatin fr a perfet-mixing system, whih has twie the age aumulatin f the mst effetive system (a plug-flw system) [10]. The key differene between effiieny and effetiveness is that effiieny is a mparisn f input and utput while effetiveness is a mparisn f a key system variable (nt neessarily the utput) with a well-defined, alulable, and ften theretially ideal benhmark. The mst mmn perfrmane metri fr whle-building energy nsumptin is EUI. This metri is nt partiularly useful by itself beause many ther fatrs besides plan area affet energy nsumptin. This fat is evident frm the range f values in Table 1 f [8]. In this set f buildings, whih are all labratry buildings lated n the UC Berkeley ampus (and therefre expsed t the same weather), the standard deviatin is 70% f the mean. This illustrates that EUI is nt a disriminating metri. Part f the reasn that there is a large variatin in this metri fr this set f buildings is beause sme f the buildings are nt air-nditined, beause lighting effiieny varies, beause plug and press lads vary, and beause the design f the air distributin systems vary. In this setin, a benhmark that mpensates fr weather differenes, design differenes, and usage differenes is desribed. The bjetive is fr the benhmark t be the energy nsumptin f an ideal building that nsumes the minimum amunt f energy required t ahieve the same indr temperature, humidity, lighting, and ventilatin nditins as the atual building. The energy nsumptin benhmark derived frm the ideal building is determined using mathematial mdels, s the methd is alled mdel-based benhmarking. Cmpliatins that arise frm defining and mputing the theretial minimum are addressed by using simplifying assumptins. The result is a benhmark that represents a highly effetive use f energy. Mdel-based benhmarking has tw parts. First, the benhmark is mputed and the atual energy nsumptin is mpared with the benhmark. The rati f the benhmark t the atual nsumptin is an effetiveness metri analgus with ther engineering effetiveness metris suh as heat exhanger effetiveness. The send part f mdel-based benhmarking invlves a mparisn f the effetiveness f a partiular building with that f a set f buildings, and with the past perfrmane f that same building. This part f mdel-based benhmarking invlves statistial mparisns. Sine the benhmarking alulatins mpensate fr funtinal 5

6 requirements, it is pssible t use mdel-based benhmarking t mpare the perfrmane f buildings with dissimilar features and funtinal requirements. Defining the Benhmark The perfrmane f the ideal building is mre diffiult t quantify than the perfrmane f the ideal heat exhanger. Therefre, the ideal building is seleted with features that make the alulatin f the minimum energy nsumptin a tratable prblem with sme simplifying assumptins. The fllwing is a list f the imprtant features and assumptins: N energy strage This definitin als implies that the struture f the building is nt used fr thermal strage. Defining the benhmark building as having n energy strage signifiantly simplifies the alulatins. Mst labratry failities have little r n energy strage. Sine the benhmark is defined as having n strage, labratry buildings with energy strage may, in thery, use less energy than the benhmark. N ndutin r transmissin The benhmark building has perfet insulatin and allws n transmissin f slar energy int the building. This assumptin als signifiantly redues the mplexity f mputing the benhmark. Labratry energy nsumptin is dminated by ventilatin rather than by heat transfer thrugh the shell, s there wuld be little benefit t inluding ndutin and transmissin in the alulatins. If this benhmark were used t analyze the energy nsumptin f nn-labratry buildings, the results wuld indiate that the nn-labratry buildings were less effetive at using energy beause heat transfer thrugh the shell is a larger mpnent f the lad in nn-labratry buildings. Maximum use f daylight The lighting pwer benhmark is zer between sunrise and sunset. When the building is in use between sunset and sunrise, the benhmark is the average speifi lighting pwer reprted in [8], whih is 0.04 W/m 2. Empirial benhmark fr plug and press lads Fr the labratry spae, the default speifi plug and press pwer is the average value reprted in [8], whih is 0.11 W/m 2. Fr the nn-labratry spae, the default speifi plug and press pwer is 140 W/persn, whih is derived frm pwer requirements f ffie mputers. 6

7 Fan pwer In thery, the minimum fan pwer required t mve air is zer beause it uld be mved at an arbitrarily lw stati pressure (i.e., with arbitrarily lw resistane). This is nt a reasnable benhmark beause duts must have a finite size. Therefre, the speifi fan pwer speified by the Califrnia energy de (Title 24), fr nstant vlume systems (1700 W/(m 3 /s)) is used as the benhmark fr fan pwer. Transprtatin systems Effiient elevatr systems use unterweights and energy revery s that they d nt ntribute substantially t the ttal energy nsumptin f a building. Cnsequently, the benhmark fr transprtatin systems is zer pwer. This benhmark will penalize buildings with hydrauli elevatrs mre than buildings with unterweighted elevatrs. Effiient air distributin VAV labratries will use nsiderably less energy than nstant vlume labratries as lng as the design air hange rate is suffiiently lw and as lng as fume hds are lsed when nt in use. Arding t [11], sashes f fume hds n the UC Berkeley ampus are generally fund t be in the lsed psitin. This indiates that the need fr upants t be wrking at the hds is intermittent, and that seleting the benhmark as nstantly lsed wuld nly penalize failities where fume hds were left pen unneessarily and persistently. With sashes in the lsed psitin, the ventilatin rate will usually be dependent n the air-hange requirement and nt n the fume hd exhaust flw rate. Cntrl f waste heat It is assumed that the ideal building an use waste heat when heat is needed, and that it an rejet waste heat when it is nt needed. Fr the labratry spae, the benhmark is based n ntrlling heat frm lighting and plug and press lads. Fr the nn-labratry spae, the benhmark is based n ntrlling heat frm lighting. Fr equipment, this uld be ahieved by lating the equipment in a ventilated abinet, whih was exhausted when heat was nt needed, but whih was reyled when heating was required. Cmputing the Benhmark It is pssible t mpute the energy nsumptin f the benhmark frm first priniples with relatively little infrmatin abut the building. Inputs The inputs fr the benhmarking alulatins are shwn in Table 1. The required inputs have n defaults; the user must prvide these. There are 12 ther inputs with defaults that may be hanged by the user. The defaults are shwn in Table 1 in brakets. 7

8 Table 1: Inputs fr the benhmark. Required Inputs Inputs with defaults 1 Plan area f lab spae Lab air-hange rate [6 per hur] 2 Plan area f nn-lab spae Speifi ventilatin rate fr nn-lab spae [ (m 3 /s)/persn] 3 Linear feet f fume hds Spae temperature [22.2 C] 4 Fratin f lab spae that is Spae relative humidity [50%] air-nditined 5 Fratin f nn-lab spae that is air-nditined Shedule f peratin [24/7 fr lab; 7am - 9pm, 7 days fr nn-lab] 6 Latin Number f lab upants [1/(71 m 2 )] 7 Eletrial nsumptin Number f nn-lab upants [1/(71 m 2 )] 8 Fuel nsumptin Speifi fan pwer [1700 W/(m 3 /s)] 9 Time duratin Speifi pump flw rate [ (m 3 /s)/w] 10 Speifi pump pwer [99868 W/(m 3 /s)] 11 Plug and press lad fr lab spae [0.11 W/m 2 ] 12 Plug and press lad fr nn-lab spae [140 W/persn] The default fr the design air hange rate is derived frm [12]. The default fr the shedule is derived frm the peratin f labs n the UC Berkeley ampus. The defaults fr speifi pump flw rate, speifi pump pwer, and average plug and press lad pwer fr the lab is derived frm measurements made by [8]. The default fr the number f upants is derived frm the CBECS database [3]. The default fr the plug and press pwer fr nn-lab spae is based n ne mputer per persn. Arding t [13], the pwer nsumptin f a mputer, mnitr, and laser printer perating in idle mde is 56 W, 60 W, and 24 W, respetively. Calulatins Table 2 shws the initial alulatins and hurly variables that are alulated. Details regarding these alulatins are inluded in Appendix B. Table 2: Calulated nstants and variables. Initial alulatins Prperties alulated hurly Lad-related alulatins 1 Indr pressure Outdr humidity rati Minimum utdr air flw rate 2 Indr vapr pressure Outdr speifi enthalpy Maximum utdr air flw rate 3 Indr humidity rati Outdr air density Cling lads (lab and nn-lab) 4 Indr enthalpy System status (n r ff) Fan pwer (lab and nn-lab) 8

9 5 Indr density Pump pwer (lab and nn-lab) 6 Fume hd flw rate eating lad (lab and nn-lab) 7 Exhaust flw rate f lab 8 Oupant lads Outputs After the hurly pwer alulatins are mpleted, the results are aumulated fr the time perids f interest. Tw perfrmane metris are mputed. They are the eletrial nsumptin effetiveness and the fuel nsumptin effetiveness. The eletrial nsumptin effetiveness, dented as ε e, is defined as the eletrial nsumptin f the benhmark divided by the atual eletrial nsumptin fr the same time perid. Similarly, the fuel nsumptin effetiveness, dented as ε f, is defined as the fuel nsumptin f the benhmark divided by the atual fuel nsumptin fr the same time perid. Cnsequently, higher values are better than lwer values, and the values shuld range between zer and ne. Statistial Cmparisn After the eletrial energy and fuel nsumptin effetiveness metris have been mputed fr a partiular building they are mpared with the metris fr a set f buildings. The mean and variane f eah metri in the mparisn set is mputed s that the user an determine if the perfrmane f the test building is abve r belw the nrm, and by hw muh. If the perfrmane is signifiantly prer than average, then the prtls desribed in [8] uld be used t investigate the ause. Benhmarking Tl and Database Desriptin In rder t mpare the energy nsumptin f ne r mre labratry buildings with that f thers, a database has been reated using Mirsft Aess. The database ntains tables fr the building statistis prvided by the users as well as tables fr weather data. Frms fr entering data, initiating alulatins, and displaying the results have been reated s that the database is easy t use. The building statistis inlude the data neessary fr alulating the benhmarks as well as data that will be useful fr filtering. These additinal inputs inlude perfrmane metris that may have been determined frm a mre detailed audit f the building using the prtls desribed in [8], as well as design infrmatin that is relevant t energy nsumptin analysis (e.g., VAV r nstant vlume air distributin). A detailed desriptin f the benhmarking tl and database an be fund in [14]. 9

10 Features The tl has been designed t handle time intervals f arbitrary duratin. The start date fr all intervals but the first is the end date fr the previus interval. The tl has als been designed t handle multiple meters s that data frm eletrial bills an be entered diretly int the database. Results are displayed with a set f graphs. Tw f the graphs are used t shw hw the eletrial and fuel nsumptin effetiveness values f a single building mpare with a ppulatin f buildings. The mputed effetiveness is shwn as a vertial line n a smth distributin that is derived frm the statistis f the ppulatin. A third graph shws the satter plt f the eletrial and fuel nsumptin effetiveness f the targeted building and the ppulatin f buildings with whih it is being mpared. The furth graph is a time series f the eletrial nsumptin effetiveness f the target building fr the set f time intervals that the effetiveness was mputed. This time series an be used t establish a baseline fr the building, and an als be used t detet unusual perfrmane. The benhmarking alulatins are als transfrmed int metris that are familiar t energy engineers fr eah f the subsystems in the building. Fr example, the energy alulatins fr the lighting pwer are transfrmed int a metri suh as average Watts per unit f plan area. The benhmark is als transfrmed int a whle-building average pwer density (e.g., average ttal Watts per unit f plan area). This metri, mbined with a target fr the effetiveness uld be used as a design-intent target. RESULTS OF APPLICATION OF MODEL-BASED BENCMARKING Statistial Analysis Methds We used a set f parametri and nn-parametri statistial methds t analyze the perfrmane f the mdel-based benhmarking methd and mpare it with existing benhmarking methds. All f the nn-parametri statistis used here are desribed in Siegel and Castellan [15]. The parametri statistis used here are desribed in all intrdutry statistis texts. We used the Pearsn prdut-mment rrelatin effiient and the Spearman rank-rder rrelatin effiient t test fr assiatin between atual and predited energy nsumptin. If the differenes between the atual nsumptin and the nsumptin predited by the mdels were nrmally distributed, then the square f the Pearsn effiient is the perentage f the variane explained by the mdel. As nted by Sharp [1] and thers, the residuals fr building energy-use data are frequently nt nrmally distributed. The Spearman effiient is a nnparametri measure f assiatin that has a similar interpretatin as the Pearsn effiient. It an be used t measure assiatin when the underlying distributin is nt knwn. We used the rbust rank-rder test t mpare the effetiveness f buildings with and withut mehanial ling. The rbust rank rder test is a nn-parametri equivalent t the tw-sample t-test, whih is the standard parametri test fr a differene between means. 10

11 Effet f Cling Equipment n Effetiveness Due t strit energy requirements fr state buildings in Califrnia, sme f the labratry buildings n the UC Berkeley ampus are nt mehanially led. This fat allws us t mpare the effet f air-nditining n energy nsumptin effetiveness. Of the 19 labratry buildings studied, ne had n mehanial ling, eight had negligible ling apaity, and five were mpletely led by vapr mpressin. The effetiveness f these 14 buildings is shwn in Table 3. The mean and median f the eletrial nsumptin effetiveness fr the unled buildings were 0.60 and 0.71, respetively. The mean and median f the eletrial nsumptin effetiveness fr the led buildings were 0.29 and 0.27, respetively. Sine the benhmark mpensates fr the funtinal requirement f ling, the effetiveness f the led buildings shuld be equal t the effetiveness f the unled buildings if the ling system were as effiient as the ther systems in the building. The rbust rank rder test [15] was used t determine whether r nt the differene between the median values f the tw sets is signifiant. The prbability f bserving a differene this large r larger with this sign by hane is 2.8%. Therefre the differene is statistially signifiant. This des nt mean that mehanial ling aused the differene, but it des prvide evidene fr a ausal relatinship. little ling 100% ling Table 3: Effetiveness f led and unled buildings U P % Perfrmane Cmparisn Cmparing ne benhmarking methd with anther is mpliated by the fat that the true energy-use effetiveness annt be measured. wever, we an measure hw well different benhmarking methds mpensate fr features r funtinal requirements that affet energy use. This an be amplished by mparing the degree f assiatin (rrelatin) between the mdel assiated with eah methd and the atual energy nsumptin. In this setin, the mdel-based benhmarking methd desribed in this reprt is mpared with benhmarking based n the EUI metri and Sharp s methd using buildings lated n the UC Berkeley ampus. All f these buildings nminally have the same upant density, mputer density, and shedule, and all f them are wner-upied. Nne f the labratries have enmizers; all are 100% utside air systems. Therefre, Sharp s methd differs frm the EUI methd nly by mpensating fr whether r nt the building has a hiller. Fr eah methd, tw different rrelatin effiients were mputed. They were the Pearsn prdut-mment rrelatin effiient and the Spearman rank-rder rrelatin effiient. 11

12 Table 4 shws the square f the tw effiients fr eah f the three methds when applied t 19 labratry buildings n the UC Berkeley ampus. The Siegel-Tukey test was used t test whether r nt the size f the residuals was signifiantly different frm ne methd t anther. This test indiates that the differene between Sharp s methd and the mdel-based methd is nt statistially signifiant. Table 4: Cmparisn f methds. Pearsn, 2 R Mdel-based methd 41% 55% Sharp's methd 46% 63% EUI methd 40% 52% Spearman, Inspetin f the residuals shws that ne f the 19 buildings is an utlier. This labratry ntains a lass 100 leanrm, s the filtratin requirements are signifiantly different. The benhmarking alulatins used here are designed fr typial filtratin requirements, s the leanrm uses signifiantly mre energy than the benhmark. This prblem uld be eliminated if filtratin requirements were inluded as an input t the benhmarking alulatins. If this building were eliminated frm the data set, then the rrelatin effiients are as shwn in Table 5. Again, the Siegel-Tukey test indiates that the differenes are nt signifiant. 2 R s Table 5: Cmparisn with utlier remved. Pearsn, 2 R Mdel-based methd 73% 53% Sharp's methd 54% 58% EUI methd 51% 45% Spearman, 2 R s Table 6 shws the tw effiients fr eah f the three methds when applied t 19 labratry buildings and 9 nn-labratry buildings n the UC Berkeley ampus. This table illustrates that mdel-based benhmarking is better at mparing the perfrmane f dissimilar building types than are empirial methds f benhmarking. In this ase the differene between the mdel-based methd and Sharp s methd is statistially signifiant. The single-sided prbability f bserving a larger differene by hane is just 3.2%. wever, the differene between the mdel-based methd and the EUI methd is still nt statistially signifiant at the 95% level f nfidene. The single-sided prbability f bserving a larger differene by hane is 6.4%. Table 6: Cmparisn with dissimilar building types. Pearsn, 2 R Mdel-based methd 43% 41% Sharp's methd 16% 19% Spearman, 2 R s 12

13 EUI methd 22% 18% DISCUSSION AND CONCLUSIONS The benhmarking methd desribed in this paper will penalize buildings that use ineffiient systems fr energy-nsuming funtinal requirements. Fr example, buildings with the fllwing design subsystems r features will be penalized: 1. Oversized systems if part-lad effiienies are pr 2. Cnstant vlume systems 3. Any systems that use reheat 4. ydrauli elevatrs 5. Ineffiient lighting systems 6. Ineffiient fans and air distributin systems Additinally, the fllwing peratinal fatrs will be penalized: 7. Fume hds left pen 8. Pr ntrller tuning, if it indues sequential heating and ling 9. Faulty ntrl lgi, if it indues simultaneus heating and ling The benhmarking methd will als penalize buildings in whih fume hds must be used (i.e., pened) ntinually. Based n the experiene f UC Berkeley failities staff and the fat that pen fume hds are a safety hazard, it is expeted that this will be a rare requirement. The benhmarking methd will als penalize buildings that d nt have labratry spae r that have very little labratry spae beause ndutin and transmissin heat transfer is a larger fratin f the heating and ling lad in thse buildings. This will nly be a prblem fr the ase where a small labratry is nneted t a large nn-labratry building. The magnitude f these unfair penalties is diffiult t quantify and it will be different fr eah ase. The analysis desribed in the Results setin indiates that air-nditined buildings may use energy less effetively than nn-air-nditined buildings even after the funtinal requirement f ling has been nsidered. This result may indiate that typial mehanial ling designs are ineffiient relative t the effiieny f ther systems suh as lighting systems, r that the benhmarking methd unfairly penalizes mehanial ling. It is pssible that simplifying assumptins used t mpute the benhmark, whih inlude the use f mpnent effiienies, may unfairly penalize mehanial ling and ventilatin relative t lighting r plug and press pwer. Mre researh is needed t determine the ause f the finding, but it is nsistent with the pereptin that VAC systems in mdern labratries are signifiantly versized, whih auses ineffiieny. When applying the mdel-based benhmarking methd t nn-lab buildings in additin t lab buildings, the mdel-based apprah was learly better than the tw alternatives. This fat demnstrates ne f the advantages f mdel-based benhmarking ver empirial methds. A 13

14 single mdel an be used t benhmark and mpare a wide variety f buildings. It shuld be nted, hwever, that althugh the rrelatin between energy use and predited energy use was muh higher using the mdel-based methd than either f the tw empirial methds, it was muh lwer than when applied t just labratry buildings. This uld be due t a large variatin in effiieny when nsidering a larger, mre diverse ppulatin f buildings, r it uld be that the mdels are less aurate when applied t nn-lab buildings. Future researh is needed t determine the effiay f using the existing mdel-based benhmarking methd fr analyzing the energy perfrmane f nn-lab buildings. ACKNOWLEDGEMENTS This prjet was funded by the Califrnia Institute fr Energy Effiieny (CIEE) with transitin funding frm the Publi Interest Energy Researh (PIER) prgram f the Califrnia Energy Cmmissin (CEC). Publiatin f researh results des nt imply CIEE endrsement f r agreement with these findings, nr that f any CIEE spnsr. The authrs thank Paul Blak, Phil Maynard, Patriia Mead, and Venzi Nikifrv frm the UC Berkeley Department f Failities Management fr their assistane with aquisitin f energyrelated data frm labratry buildings n the UC Berkeley ampus, and Pat Thrsn frm the Envirnment, ealth, and Safety Divisin f Lawrene Berkeley Natinal Labratry fr help with aquisitin f weather data. The authrs als thank Karl Brwn frm CIEE and Dale Sartr frm LBNL fr their enuragement and supprt f this prjet. This artile was published in Energy and Buildings, 34(3), by Cliffrd C. Federspiel, Qiang Zhang, and Edward Arens as Mdel-based benhmarking with appliatin t labratry buildings, pp , 2002, and is psted with permissin frm Elsevier Siene. 14

15 REFERENCES [1] Sharp, T., 1996, Energy Benhmarking in Cmmerial Offie Buildings, Preedings f the ACEEE 1996 Summer Study n Energy Effiieny in Buildings, 4, [2] Birtles, A. B., 1997, Energy Effiieny f Buildings: Simple Appraisal Methd, Building Servies Engineering Researh and Tehnlgy, 18(2), [3] U. S. Department f Energy, 1995, Cmmerial Building Charateristis 1995, Washingtn D.C.: U. S. Department f Energy. [4] iks, T. W. and D. W. Clugh, 1998, The Energy Star Building Label: Building Perfrmane thrugh Benhmarking and Regnitin, Preedings f the ACEEE 1996 Summer Study n Energy Effiieny in Buildings, 4, [5] Reddy, T. A., N. F. Saman, D. E. Claridge, J. S. aberl, W. D. Turner, and A. T. Chalifux, 1997, Baselining Methdlgy fr Faility-Level Mnthly Energy Use Part 1: Theretial Aspets, ASRAE Transatins, 103(2), [6] Snderegger, R., 1998, Baseline Mdel fr Utility Bill Analysis Using Bth Weather and Nn-Weather-Related Variables, ASRAE Transatins, 104(2), [7] Mills, E., G. Bell, D. Sartr, A. Chen, D. Avery, M. Siminvith, S. Greenberg, G. Martn, A. de Almeida, L. E. Lk, 1996, Energy Effiieny in Califrnia Labratry-Type Failities, LBNL reprt n. LBNL [8] uizenga, C. W. Van Liere, and F. Bauman, 1998, Develpment f Lw-Cst Mnitring Prtls fr Evaluating Energy Use in Labratry Buildings Center fr Envirnmental Design Researh, UC Berkeley, reprt n. CEDR [9] Inrpera, F. P. and D. P. DeWitt, 1996, Fundamentals f eat and Mass Transfer, Furth Editin, Wiley, New Yrk, [10] Federspiel, C. C., 1999, Air-Change Effetiveness: Thery and Calulatin Methds, Indr Air, 9, [11] Blak, P., 1998, Department f Failities Management, University f Califrnia, Berkeley, Persnal mmuniatin. [12] Bell, G.C., D. Sartr, E. Mills A Design Guide fr Energy-Effiient Researh Labratries. LBNL Reprt N. 777, CIEE. [13] Brwn, W. K Labratry Design Lads - What We Knw, Dn t Knw, and Need t Knw. ASRAE Transatins, 102(1), [14] Federspiel, C. C., Q. Zhang, and E. Arens, 1999, Labratry Field Studies/Perfrmane Feedbak, CEDR [15] Siegel S. and N. J. Castellan, 1988, Nnparametri Statistis fr the Behaviral Sienes, New Yrk: MGraw-ill. 15

16 [16] List, R. L., 1971, Smithsnian Meterlgial Tables, Washingtn, D.C.: Smithsnian Institutin, [17] ASRAE, 1997, Chapter 6: Psyhrmetris, 1997 ASRAE Fundamentals andbk, Atlanta: ASRAE. 16

17 APPENDIX A: NOMENCLATURE AND DEFINITIONS 17

18 Figure 1 shws the flw rates used in the alulatins. The first letter f the subsripts refers t the spae that the air is ming frm. The send letter refers t the spae that the air is ging t. r F a r l F al r l F la,e r F a Air-nditined ffie spae Air-nditined Lab spae r l F la,h (1-r ) F a nn-air-nditined ffie spae r l F l Nn-air-nditined Lab spae (1-r l ) F l Fume hds (1-r l ) F la,h (1-r ) F a (1-r l ) F al (1-r l ) F la,e Figure 1: Shemati diagram f the labratry building. Cnstants λ : speifi lighting pwer (default = 0.04 W/m 2 ) π : speifi plug and press pwer fr lab (default = 0.11 W/m 2 ) pp,l π pp, : speifi equipment pwer fr ffie (default = 140 W/persn) π f : speifi fan pwer, (default = 1700 W/(m 3 /s)) π p : speifi pump pwer, (default = W/(m 3 /s)) a : standard temperature lapse rate ( K/m) : eiling height (3.05 m) C 1 : saturated vapr pressure nstant ( ) C 2 : saturated vapr pressure nstant ( ) C : saturated vapr pressure nstant ( ) 3 C 4 : saturated vapr pressure nstant ( ) C : saturated vapr pressure nstant ( ) 5 C : saturated vapr pressure nstant ( ) 6 C : saturated vapr pressure nstant ( ) 7 C : saturated vapr pressure nstant ( ) 8 C : saturated vapr pressure nstant ( ) 9 C : saturated vapr pressure nstant ( ) 10 C 11: saturated vapr pressure nstant ( ) C 12 : saturated vapr pressure nstant ( ) C : saturated vapr pressure nstant ( ) 13 18

19 COP : effiient f perfrmane f hiller used fr benhmark alulatins (default = 5) d : height f the pening belw the sash f a lsed fume hd (0.076 m) f : speifi pump flw rate (default = (m 3 /s)/w) p g : gravitatinal nstant ( m/s 2 ) h : heat generatin per persn (100 W/persn) m : mleular weight f dry air ( kg/mle) a m : mleular weight f water ( kg/mle)= 0 w P : standard atmspheri pressure at sea level (101.3 kpa) R : gas nstant f dry air ( J/(kg K) a R : gas nstant f water vapr ( J/(kg K) w T : nditined spae temperature (default = 22.2 C) T : standard abslute temperature at sea level (288 K) 0 V : fume hd fae velity (0.5 m/s) Variables ε e : eletrial nsumptin effetiveness ε f : fuel nsumptin effetiveness φ : relative humidity φ : relative humidity f nditined air ρ : density f air ρ a : density f utdr (ambient) air ρ : density f nditined air Ω : air hange rate δ : slar delinatin, degree f ar A Lab : Calulated labratry spae area ' A Lab : Reprted labratry spae area A l : grss plan area (e.g., square ftage) f the lab A : grss plan area f the nn-lab spae ET : equatin f time, minutes f time F al : flw frm the utdrs t the lab F a : flw frm the utdrs t the ffie F a : maximum flw frm the utdrs t the ffie F a : minimum flw frm the utdrs t the ffie F : flw frm the ffie t the utdrs a F la, e : flw frm the lab t the utdrs thrugh the general spae exhaust 19

20 F la, h : flw frm the lab t the utdrs thrugh the fume hds h : speifi enthalpy f air h : speifi enthalpy f nditined air, : ling lad f ffie spae h : speifi enthalpy f transfer air frm ffie t lab : speifi enthalpy f nn-air-nditined ffie air h, u, : lad (heating r ling) f air-nditined ffie spae, l : heat generated by upants in the lab area, : heat generated by upants in the nn-lab (ffie) area sunset : sunset hur sunrise : sunrise hur L : linear quantity f fume hds LAT : lal latitude, degree f ar LON : lal lngitude, degree f ar LSM : lal standard time meridian, degree f ar M : ttal mass f labratry air N : number f labratry upants l N : number f ffie upants p : vapr pressure f water in air w p w, p ws p ws, : vapr pressure f water in nditined air : saturated vapr pressure f water in air : saturated vapr pressure f water in nditined air P : atmspheri pressure P, : lighting pwer in the lab spae l l P l, : lighting pwer in the nn-lab (ffie) spae P pp, l : plug and press pwer in the lab spae P pp, : plug and press pwer in the nn-lab (ffie) spae q : minimum utdr air vlume flw per persn fr the ffie a q a : maximum utdr air vlume flw per unit f plan area fr the ffie r : fratin f lab spae that is air-nditined l r : fratin f ffie spae that is air-nditined R : gas nstant f air R : gas nstant f nditined air W : humidity mass rati W : humidity mass rati f nditined air 20

21 APPENDIX B: ENERGY CONSUMPTION CALCULATIONS 21

22 This Appendix ntains a list f the alulatin predures fr mputing the benhmarks. Unit nversins are nt shwn here. Prealulatins Lab Area: Spae Pressure: A ' = ( A * L) / 2 (A-1) Lab Lab + Spae pressure is mputed based n the NACA standard atmsphere. Belw an altitude f meters (35332 feet) abve sea level, the pressure f the standard atmsphere is given by the fllwing equatin [16]: P g a ar a P 1 0 Z T (A-2) = 0 Minimum utdr air vlume flw rate fr ffie Q = q N (A-3) a a Maximum utdr air vlume ffie Q = q A (A-4) a a Saturated vapr pressure f nditined air: Arding t [17], the saturated vapr pressure is given by the fllwing empirial relatin when the air temperature is less than 0 C: p ws C , = exp + C 2+ C3T + C4T + C5T + C6T + C7 ln (A-5) T ( T ) When the temperature is greater than 0 C, the saturated vapr pressure is given by the fllwing empirial relatin: p ws C8 2 3, = exp + C 9+ C10T + C11T + C12T + C13 ln (A-6) T ( T ) 22

23 The values fr the nstants are given in Appendix A. Using the nstants in Appendix A requires that the temperature in Equatins A-5 and A-6 is in K, and results in Pasal pressure units. Vapr pressure f nditined air: The vapr pressure is equal t the prdut f the relative humidity and the saturated vapr pressure. p w, = pws, φ (A-7) umidity mass rati f nditined air: The humidity mass rati is the rati f the mass f water vapr in the air t the mass f dry air in the air. It is mputed based n the mleular masses f dry air and water, the pressure, and the vapr pressure as fllws: W m m w w, = (A-8) a p P p w, Gas nstant f nditined air: The gas nstant f the air is the mass-weighted average f the gas nstants f dry air and water vapr. It is mputed as fllws: R = R a + R w 1+ W W (A-9) Speifi enthalpy f nditined air: The speifi enthalpy f air is mputed as fllws: h T = + W ( T ) 1+ W (A-10) Nte that Equatin A-10 is different than that published in [17] beause Equatin A-10 is the energy per unit mass f mist air rather than per unit mass f dry air. Density f nditined air: The density is alulated as fllws: 23

24 P ρ = (A-11) R T Fume hd mass flw rate The ttal mass flw rate thrugh the fume hds is mputed as fllws: Ttal labratry exhaust air mass flw rate F LVd la, h = ρ (A-12) The ttal mass flw rate f air exhausted frm the labratry thrugh fume hds and spae exhaust is mputed as fllws: M = A (A-13) F ρ l ( F ) = max, MΩ (A-14) la la h, Equatin A-13 indiates that the ttal exhaust flw rate may either be determined by the design air hange rate r by the number f fume hds. Labratry supply air mass flw rate The supply air mass flw rate t the labratry is mputed as fllws: Labratry upant lad The labratry upant lad is mputed as fllws: F al = F la (A-15) = h, l Nl (A-16) Offie upant lad The ffie upant lad is mputed by substituting the number f ffie upants fr the number f labratry upants in Equatin A-15. Lab lighting lad The lab lighting lad is mputed as fllws: P l, l = λal (A-17) 24

25 Offie lighting lad The ffie lighting lad is mputed by substituting the grss plan area f the ffie fr the grss plan area f the lab in Equatin A-16. urly Calulatins Saturated vapr pressure f utdr air: The saturated vapr pressure f utdr air is mputed using Equatin A-5 r A-6 with the utdr air temperature substituted fr the nditined air temperature. Vapr pressure f utdr air: The vapr pressure f utdr air is mputed using Equatin A-7 with the utdr air temperature substituted fr the nditined air temperature, and the relative humidity f the utdr air substituted fr the relative humidity f the nditined air. umidity mass rati f utdr air: The humidity mass rati f utdr air is mputed using Equatin A-8 with the utdr air vapr pressure substituted fr the nditined air vapr pressure. Gas nstant f utdr air: The gas nstant f utdr air is mputed using Equatin A-9 with the utdr air humidity mass rati substituted fr the nditined air humidity mass rati. Speifi enthalpy f utdr air: The speifi enthalpy f utdr air is mputed using Equatin A-10 with the utdr air temperature substituted fr the nditined air temperature, and the humidity mass rati f the utdr air substituted fr the humidity mass rati f the nditined air. Density f utdr air: The density f utdr air is mputed using Equatin A-11 with the utdr air temperature substituted fr the nditined air temperature, and the gas nstant f the utdr air substituted fr the gas nstant f the nditined air. Is system n? If the shedule indiates that the system is n, then O = 1. Otherwise, O = 0. Minimum utdr air mass flw rate fr the ffie: 25

26 The minimum utdr air mass flw rate fr the ffie is the maximum f the flw rate required fr ventilatin and the makeup airflw rate fr the lab. F a = ρ Q (A-18) a a Maximum utdr air mass flw rate fr the ffie: The maximum utdr air mass flw rate fr the ffie is mputed as fllws: Lighting Energy Use: F a = ρ Q (A-19) a a Fllwing equatins uld alulate sunrise and sunset hur fr eah mnth. Sine we assume that there is n differene between sunset r sunrise times amng the days in ne mnth, we nly alulate twelve sunset and twelve sunrise times in ne year. sunrise sunset = [ ars( tan LAT tanδ ) ET 4 ( LSM LON)]/ 60 (A-20) = [ ars( tan LAT tanδ ) ET 4 ( LSM LON)]/ 60 (A-21) Table 7: Mnthly effiients fr sunrise and sunset alulatins. Mnth J F M A M J J A S O N D ET, min δ, degrees Cmpute the thermal lad n the ffie spae The fllwing alulatins are made assuming that the entire ffie spae is air-nditined. Cmpensatin fr partially air-nditined ffie spaes is made later. The ffie lad alulatin is based n the assumptin that fan pwer des nt ntribute t the lad. The lgi fr mputing the lad when there is an enmizer and ntrl f waste heat frm lights is as fllws. If h h 0, then a F a = F a (A-22) = F h h P (A-23) ( a ) pp a,, Otherwise, mpute the lad with the maximum flw rate and n lighting lad as fllws: ( h ha ) Ppp, 1 F a,, = (A-24) 26

27 If 0, then, 1 F a = F a (A-25) = (A-26),1 Otherwise, mpute the lad with the minimum utdr airflw rate and the maximum lighting lad as fllws: ( h ha ) Ppp Pl, 2 F a,,, = (A-27) If, 2 0, then = 0. The utdr airflw rate under this nditin is mputed as fllws. Cmpute the lad assuming a minimum flw rate and n lighting lad as fllws: If 0, then, 3 Otherwise, ( h ha ) Ppp, 3 F a,, = (A-28) F a = F a (A-29) F a =, h + P h a pp, (A-30) If, 2 > 0, then F a = F a and =, 2. If the system is ff, then the ffie lad is zer. Cmpute ffie ling lad: If the ffie lad is negative, then the ling lad equals the magnitude f the ffie lad. Otherwise it equals zer. This is mputed as fllws: Cmpute the ffie heating lad: The ffie heating lad is mputed as fllws: = min(0, r ) (A-31), = max(0, ) (A-32) h, 27

28 Cmpute ffie fan pwer: The ffie fan pwer is mputed as fllws: Faπ f Pf, = (A-33) ρ a Cmpute ffie pump pwer: The ffie pump pwer is mputed as fllws: P p, max,, Cmpute ffie eletrial pwer: The eletrial pwer is mputed as fllws: ( h ) f pπ p = (A-34), Cmpute thermal lad f the lab spae = (A-35) COP, P + Ppp, + Pl, + Pf, + Pp, The fllwing alulatins are made assuming that the entire lab spae is air-nditined. Cmpensatin fr partially air-nditined lab spaes is made later. The lab lad alulatin is based n the assumptins that fan pwer des nt ntribute t the lad, and that waste heat frm lights, and plug and press lads is used fr heating but rejeted when ling. The thermal lad with waste heat frm lights and plug and press pwer is as fllws: ( h ha ), l Pl, l Ppp l l, h Fla, = (A-36) The lgi fr mputing the lab lad t effetively use waste heat is as fllws. First, mpute the lad arding t Equatin A-38. Then mpute the lad assuming that the heat frm lights and equipment is exhausted. ( h ha ) l l, e Fla, The lab lad with effetive use f waste heat is the fllwing: = (A-37) If the sign f l, h is nt the same as the sign f l, e, then the lad is zer. Otherwise, the lad is equal t the value f l, h r l, e with the smallest magnitude. 28

29 Cmpute the lab ling lad This alulatin is similar t the ling lad alulatin fr the ffie. = min(0, r ) (A-38), l l l Cmpute the heating lad fr the lab This alulatin is similar t the heating lad alulatin fr the ffie. = max(0, ) (A-39) h, l l Cmpute the lab fan pwer This alulatin is similar t the fan pwer alulatin fr the ffie. Flaπ f Pf, l = (A-40) ρ a Cmpute the lab pump pwer This alulatin is similar t the pump pwer alulatin fr the ffie. P p, l max, (, l h, l ) f pπ p = (A-41) Cmpute the ttal eletrial pwer fr the lab This alulatin is similar t the eletrial lad alulatin fr the ffie. = + P + P + P + P (A-42) COP, l P l pp, l l, l f, l p, l Cmpute the eletrial nsumptin benhmark The eletrial nsumptin benhmark is the sum f the eletrial lads times the time interval rrespnding t eah lad (ne hur). Cmpute the fuel nsumptin benhmark The fuel nsumptin is the sum f the heating lads times the time interval rrespnding t eah lad (ne hur). Cmpute the eletrial nsumptin effetiveness 29

30 The eletrial nsumptin effetiveness is the eletrial nsumptin benhmark divided by the atual eletrial nsumptin. Cmpute the fuel nsumptin effetiveness The fuel nsumptin effetiveness is the fuel nsumptin benhmark divided by the atual fuel nsumptin. 30