Development of a New Duct Leakage Test: Delta Q

Transcription

1 LBNL Development of a New Duct Leakage Tet: Delta Q Walke, I.S., Sheman, M.H., Wempen, J., Wang, D., McWilliam, J.A., and Dickehoff, D.J. Enegy efomance of Building Goup Lawence Bekeley National Laboatoy One Cycloton Road Bekeley, CA Thi wok wa uppoted by the Diecto, Office of Science, Office of Baic Enegy Science, of the U.S. Depatment of Enegy unde Contact No. DE-AC03-76SF Thi wok wa uppoted by the Aitant Secetay fo Enegy Efficiency and Renewable Enegy, Building Technologie, of the U.S. Depatment of Enegy unde Contact No. DE-AC03-76SF Thi wok wa uppoted by the Califonia Intitute fo Enegy Efficiency unde Contact No. S9902A though the U.S. Depatment of Enegy. The eeach epoted hee wa funded by the Califonia Intitute fo Enegy Efficiency (CIEE), a eeach unit of the Univeity of Califonia. ublication of eeach eult doe not imply CIEE endoement of o ageement with thee finding, no that of any CIEE pono. 1

2 Table of Content INTRODUCTION... 4 BACKGROUND... 4 OVERVIEW OF EXISTING DUCT LEAKAGE TEST METHODS... 4 DELTAQ DEVELOMENT... 7 DERIVATION OF DELTAQ RELATIONSHIS... 8 Figue 1. Example DeltaQ tet eult (l/ = cfm 0.47) FLOW ADJUSTMENTS FOR EXACT RESSURE MATCHING DELTAQ TEST ROTOCOL UNCERTAINTY ANALYSES Repeatability Table 1. Repeatability Reult Fo DeltaQ teting Table 2. DeltaQ Repeatability Tet in Two Long Iland Houe Figue 2: Repeatability Tet Reult (cm 2 =in 2 6.5) Senitivity to the peue ued in the DeltaQ calculation Figue 3. Adjuting the peue ued in the DeltaQ elationhip to bette fit meaued data (l/ = cfm 0.47) Table 3. Senitivity of DeltaQ eult to electing bet fit peue Uncetainty Etimate fo exponent and duct peue aumption FIELD EXERIENCE Initial ilot Tet Table 4. Compaion of duct leakage meauement pocedue Initial Field Evaluation Table 5. Initial Field Evaluation Figue 4. Compaion of DeltaQ to duct peuization tet fo total (upply plu etun) leakage. RMS diffeence = 9% of ai handle flow Figue 5. Compaion of DeltaQ to duct peuization tet fo upply leakage. RMS diffeence = 7% of ai handle flow Figue 6. Compaion of DeltaQ to duct peuization tet fo etun leakage. RMS diffeence = 8% of ai handle flow CSUC/LBNL Field Tet Compaion of DeltaQ and peuization tet eult Table 6. Compaion of DeltaQ to duct peuization eult in 87 Califonia houe (total leakage to outide) Figue 7. Compaion of DeltaQ total leakage (upply plu etun) to duct peuization leakage to outide at 25 a. RMS Diffeence = 12% of ai handle fan flow Table 7. Compaion of DeltaQ to duct peuization upply leakage eult in 87 Califonia houe Table 8. Compaion of DeltaQ to duct peuization etun leakage eult in 87 Califonia houe Figue 8. Compaion of DeltaQ upply leakage to duct peuization upply leakage to outide at 25 a. RMS Diffeence = 7% of ai handle fan flow Figue 9. Compaion of DeltaQ etun leakage to duct peuization etun leakage to outide at 25 a. RMS Diffeence = 8% of ai handle fan flow Compliance Teting

3 Table 9. Numbe of houe paing o failing the 6% total leakage tet, with Duct euization total leakage Table 10. Numbe of houe paing o failing the 6% total leakage tet, with Duct euization leakage to outide Table 11. Numbe of houe paing o failing the 10% total leakage tet, with Duct euization total leakage Table 12. Numbe of houe paing o failing the 10% total leakage tet, with Duct euization leakage to outide Time taken to pefom the DeltaQ Tet FUTURE WORK ASTM Duct Leakage Tet ocedue SUMMARY REFERENCES AENDIX. OTHER COMARISONS OF DUCT LEAKAGE TESTS Figue A1. Compaion of total duct leakage pediction uing flow captue hood and fan peuization technique in ingle family houe Figue A2. Compaion of duct leakage pediction uing flow captue hood and fan peuization technique in low ie apatment Figue A3. Compaion of Retun Leakage Flow uing fan peuization and houe peue tet method (uing meaued envelope-only leakage and contant 1.35 coection facto fo neutallevel hift due to attic duct leakage)

4 Intoduction Duct leakage i a key facto in detemining enegy loe fom foced ai heating and cooling ytem. Seveal tudie (Fancico and almite 1997 and 1999, Andew et al. 1998, and Siegel et al. 2001) have hown that the duct ytem efficiency cannot be eliably detemined without good etimate of duct leakage. Specifically, fo enegy calculation, it i the duct leakage ai flow to outide at opeating condition that i equied. Exiting tet method eithe peciely meaue the ize of leak (but not the flow though them at opeating condition), o meaue thee flow with inufficient accuacy. The DeltaQ duct leakage tet method wa developed to povide impoved etimate of duct leakage duing ytem opeation. In thi tudy we developed the analytical calculation method and the tet pocedue ued in the DeltaQ tet. A pat of the development poce, we have etimated uncetaintie in the tet method (both analytically and baed on field data) and deigned automated tet pocedue to inceae accuacy and educe the contibution of opeato eo in pefoming field tet. In addition, the tet ha been evaluated in ove 100 houe by eveal eeach team to how that it can be ued in a wide ange of houe and to aid in finding limit o poblem in field application. The tet pocedue i cuently being conideed by ASTM a an update of an exiting duct leakage tandad. Backgound ASTM ha long had a tandad on meauing duct leakage (ASTM E (1994)). That tandad ha had two method in it: duct peuization and blowe doo ubtaction. Both of thee wee method wee intended to quantify the leakage of the duct ytem unde fixed expeimental condition. The blowe-doo ubtaction method ha fallen into diue becaue the eo aociated with thi technique have been hown to be quite lage both in peciion and bia. Duct peuization method can povide both moe pecie and moe accuate meauement and ae often eaie to make. Neithe of thee method, howeve, can meaue the ai leakage unde actual opeating condition. Neithe of thee method can eaily epaate the upply fom etun leakage and the total leakage fom the leakage outide the conditioned pace. Becaue of the hot-coming of the method lited in the oiginal ASTM tandad, eeache have been developing impoved o altenative method of meauing duct leakage. Ue have ealized that thee ae eveal diffeent eaon fo wanting to meaue duct leakage eulting in diffeent et of taget citeia fo the tet method. No extant o popoed tet method can come cloe to meeting all of the popoed citeia; thu implying a need fo diffeent meauement appoache to achieve diffeent objective. Oveview of exiting duct leakage tet method The development of each of the following meauement technique wa ubject to a diffeent et of pioitie and hence compomie. Each one of them meaue a diffeent phyical quantity (e.g., hole ize, envelope peue change, etc.), although they all epot the ame paamete - duct leakage to outide at opeating condition. Detailed tep-by-tep pocedue fo thee tet ae given elewhee (Walke et al. 1997). 1. Duct euization. Meaue the ize of hole in the duct ytem and epeent them a a ingle hole. Leakage ai flow i then infeed fom aumed peue diffeence. 2. Houe eue Tet (HT). Meaue the peue diffeence aco a building envelope caued by leakage imbalance flow in thee configuation: ai handle on, ai handle off and ai handle on with the etun patially blocked. Leakage ai flow i then infeed fom the houe envelope leakage and a calculation pocedue to combine the thee tet eult. 3. Nulling eue Tet (NT). Simila to the HT, in that it ue the peue diffeence aco the building envelope caued by ai handle opeation. It eplace the houe envelope a a flow mete with a calibated flow meauement device. 4

5 4. Tace ga. Meaue tace ga concentation flowing into egite, out of egite, in the houe, outide and in all duct location. Change in ga concentation and the eulting tace ga ma balance on the houe and duct ytem ae ued to etimate the ai leakage flow. 5. Thee ae two method cuently ued in the ASTM E1554 Standad (ASTM 1994). The fit i Blowe Doo Subtaction. It ue the diffeence between blowe doo tet with egite coveed and blocked to detemine duct leakage. It i aely ued by pactitione due to the lage factional uncetaintie eulting fom detemining the diffeence between two elatively lage leakage aea. The econd method i imila to the duct peuization tet, but the whole houe and the duct ae peuized imultaneouly by a blowe doo and a flow mete i ued to meaue flow into the duct ytem. Thi econd method i alo aely ued due to uncetaintie in the duct ytem peue duing teting (including peue diffeence between the duct ytem and the houe). Duct peuization tet ae the mot common tet pefomed on duct ytem. They ae analogou to peue teting of building envelope (ASTM 1999) in that the tet meaue aiflow at pecified peue diffeence. All the egite in a ytem ae coveed and a meaued amount of ai i blown into the duct. The eulting duct peue indicate how leaky the duct ae. Thee tet meaue the ize of hole in the duct ytem, but not the flow to outide though them at opeating condition. To go fom hole ize to ai flow thee tet equie a duct ytem opeating peue to be aumed (often baed on meaued plenum and/o egite peue). Thi conveion fom hole ize to ai flow ha lage uncetaintie. Howeve, fo veifying that duct have little leakage, thee peuization tet ae ueful becaue if we etict ytem to only have mall hole, then the flow though them will neve be vey lage. Thi i why peuization teting i ued a a ceening tool in many utility and weatheization pogam and i gadually being adopted into code and tandad (e.g., popoed ASHRAE Themal Ditibution Efficiency Standad 152 (ASHRAE 2001), Califonia State Enegy code (CEC 1998), EA Enegy Sta pogam ( Thee ae eveal vaiation of duct leakage tet with inceaing time and equipment equiement (and complexity): Total Leakage. Thi i the implet tet and the mot ued. Both upply and etun duct ae teted at the ame time, o the plit between upply and etun leak mut be aumed. In addition, the faction of total leakage that leak to outide mut be aumed. Leakage to outide. Fo thee tet, the houe i peuized uing anothe fan to the ame peue a the duct o that any duct leakage i to outide. Thi tet ha the additional complication of equiing two fan and exta peue meauement. In addition, it equie moe time becaue the peue aco the duct and the building envelope mut be balanced. Supply/etun plit. To epaate upply and etun duct a phyical baie mut be intalled, uually inide the ai handle cabinet o etun plenum. The two ide of the duct ytem can then be teted epaately o that no aumption of the upply/etun plit i equied. The intallation of the epaating baie can be difficult and time conuming. If the baie i not intalled coectly the tet will oveetimate the duct leakage becaue the baie leakage will be included in both the upply and etun duct meauement. The HT meaue the peue change aco the building envelope and duct ytem due to ai handle opeation. It ue the houe envelope a a flow mete and theefoe equie envelope leakage to be meaued alo. Thi tet can be quick to do if envelope leakage i not meaued - although thi educe the accuacy depending on how well the envelope leakage can be etimated. The main poblem with thi tet ae the eult of being enitive to wind peue fluctuation aco the building envelope that ae the ame magnitude a the mall peue meaued duing the tet and the violation of aumption ued in the calculation pocedue. In addition, tet enitivity i tongly dependent on envelope leakage. Mot tudie have found the accuacy of thi tet to be too poo fo the eult to be ued in enegy analye (Walke et al. 1997, Andew et al and Fancico and almite 1997). The NT i imila to the HT except that a fan aited flowmete i ued to balance the duct leakage flow intead of uing the building envelope. Thi educe the uncetainty of the meaued flow, howeve thee i till the poblem of eliably meauing the mall envelope peue and the added difficulty of contolling a fan to exactly balance the duct leakage flow imbalance. The develope of thi technique have developed automated pocedue fo contolling the flow and taking long time aveage of peue 5

6 that educe thee uncetaintie. The poblem with leaky envelope eulting in mall envelope peue diffeence that ae had to meaue till exit. The tace ga tet inject tace gae into the houe and the duct ytem at vaiou location. Sample ae than taken fom the houe and duct ytem (and outide) and the change and diffeence in tace ga concentation ae ued to calculate duct leakage flow. Thee tace ga tet equie conideable time and equipment expenditue. In addition, the tet can only be eliably pefomed by highly killed technician with many yea of expeience in tace ga teting and analyi. Taditionally, tace ga tudie of ai flow in building have been the tandad againt which othe diagnotic ae evaluated. Howeve, fo duct leakage, the poo mixing of gae in the vaiou duct location outide conditioned pace and the inability to ample coectly at etun leak mean that the tace ga meauement can have lage uncetaintie and cannot be ued a a efeence. Fo ceening of low leakage level fo compliance teting the cuent duct leakage diagnotic ued in mot pogam i the fan peuization tet of total duct leakage. The eaon fo thi ae: Robutne. The fan peuization tet ha almot no etiction on the type of ytem it can be ued on, o the weathe condition duing the tet. Repeatability. Combining the eult of eveal eeach poject togethe (Walke et al and 1998) with the field expeience of othe ue howed that the epeatability of peuization teting i vey good. eciion. The uncetainty in leakage flow will be mall if the allowable leakage i et to a low numbe. The tet i good at meauing hole ize but not at extapolating to leakage flow at opeating condition. If we etict the application to mall hole ize, even lage eo in etimating the peue diffeence co the hole will not eult in lage leakage flow uncetaintie. Simplicity. It i eay to intepet the eult of fan peuization without having to pefom many (o any with the appopiate hadwae) calculation. Thi allow the wok cew pefoming the tet to evaluate the duct duing the tet and alo allow the wok cew to enue that the tet ha been pefomed popely becaue they can ee if the eult make any ene. Familiaity. Wok cew that have pefomed envelope leakage tet ae familia with the tet method fo duct, becaue envelope teting ue a imila appaatu and calculation/intepetation method. Becaue the peuization tet i uually implemented to meaue the total leakage of the duct and not jut the leakage to outide it will oveetimate the leakage equied fo enegy lo etimate. Howeve, fom a code compliance teting point of view, thi eo i in the ight diection becaue it mean that the tue loe will be le than thoe indicated by the tet. In othe wod, a ytem whoe total leakage pae a leakage pecification i guaanteed to have the leakage to outide be le than the pecification. Fo enegy ating (uing ASHRAE 152 o imila method) of home with leaky duct the imple peuization tet can be poo due to the aumption about peue aco the duct leak (Walke et al and almite and Fancico 1999). Fo thee leaky ytem, the othe tet method may give bette eult becaue leaky duct tend to poduce a lage peue ignal fo the DeltaQ, NT and HT method that educe thei uncetaintie. In addition to the accuacy equiement, diffeent leakage tet application have diffeent time/cot equiement. eviou tudie and field expeience have hown that the bigget dawback with the peuization tet i the equiement of coveing all the egite and attaching the flow and peuization equipment. In addition, moe detailed veion of the tet equie ineting a blockage to epaate the upply and etun and uing a blowe doo to match the duct and houe peue both of which can be time conuming. Significant advance have been made ecently in the ue of mall and lightweight data acquiition and contol ytem that can be ued in duct leakage teting. Thee ytem ae highly ecommended fo all the tet that meaue low peue: DeltaQ, NT and HT. In addition, the ability to contol a fan to maintain a et peue diffeence o flow (a equied by the DeltaQ and NT) i made much eaie with an automated data acquiition and fan peed contol ytem. The pionee of both tet ecommend the ue 6

7 of automated ytem. In addition to impoving the accuacy and peciion of the tet, automation can ignificantly educe the time equied fo the tet and the dependence on an individual opeato. DeltaQ Development In Januay of 1999, a Duct Leakage Wokhop Subcommittee of ASHRAE S152, met at the ASHRAE Winte Meeting in Chicago, to dicu altenative to exiting meauement technique. Chuck Gaton (enn State Univeity) peented an idea that would ue the diffeence between houe peuization tet with the ai handle fan on and off to detemine duct leakage. Thee wee eveal poblem aociated with thi popoed appoach. A enviaged, a complex mathematical invetigation would be equied to olve the "invee" poblem eulting fom knowing the eult (the flow diffeence) without knowing the functional fom of the elationhip that detemined the tet eult. The expeimental pocedue popoed elied on a fom of blowe doo ubtaction that would have unacceptably lage eo fom the ame ouce a the oiginal blowe-doo ubtaction method. aallel wok by LBNL developed a veion of the tet that ovecame thee poblem. LBNL ued ome implifying aumption to olve the modeling poblem analytically (dicued in detail in thi epot) and developed a tet potocol that would ue exiting equipment and pocedue. Thi tet potocol equie meauing the diffeence in flow at the ame envelope peue diffeence, with the ai handle fan on and the ai handle fan off. The tet get it name fom thi diffeence in flow (the common ymbol fo flow being Q), hence the name DeltaQ. Thi tet epeent the next tage in development fo blowe-doo ubtaction method. The objective of developing the DeltaQ tet wa to develop fate, bette, and cheape eidential duct leakage tet pocedue. Whee: fate=minimize labo hou; bette=maximize epoducibility and accuacy; cheape=equie leat capital outlay fo tet equipment. The DeltaQ tet addee thee citeia a follow: Fate. The DeltaQ tet equie little equipment etup only a blowe doo and ome peue tubing need to be intalled. The NT can be pefomed by uing a mall fan/flowmete (which take imila time a a blowe doo). Howeve, the NT ometime equie eveal configuation of ducting and duct peuization fan location to be et up depending on the location of the duct ytem. Latly if the envelope leakage i to be meaued anyway in a building (a i often the cae in weatheization pogam, home enegy ating, etc.), then the blowe doo i aleady et-up and the only time equied fo the DeltaQ tet i the time fo the meauement. Bette. Cuent duct leakage tet tade off epoducibility and accuacy. Duct peuization tend to be highly epoducible (Walke et al and 1998 and Andew 1998) but it doe not meaue duct leakage flow to outide diectly, eulting in le accuacy. The othe tet that meaue envelope peue hould have highe accuacy becaue they ae cloe to meauing the duct leakage to outide at opeating condition, but they have vey poo epoducibility due to meauing vey mall peue diffeence aco building envelope (Walke et al and Andew 1998) whoe peue fluctuation ae of the ame magnitude a the peue ignal that the tet ae uing. The DeltaQ tet aim to be moe epoducible by meauing at elatively high envelope peue whilt till concentating on meauing duct leakage to outide at opeating condition faily diectly. Cheape. The DeltaQ tet ue appaatu (a blowe doo) that i widely available and that almot all people inteeted in duct leakage aleady have, e.g., weatheization pogam. Thi minimize the capital cot in pepaing to do the tet. Secondly, becaue the DeltaQ tet take little time, the labo cot ae minimized. Howeve, thee can be a tadeoff between inceaed capital cot fo automating the tet (that equie the ue of data acquiition ytem and compute to ecod the data) and the time aving eulting fom the automation. In pactice, the time aving, ecod keeping ability, and opeato eo eduction eulting fom automating the tet outweigh the highe capital cot of thi option. 7

8 Deivation of DeltaQ Relationhip The DeltaQ tet i baed on meauing the change in flow though duct leak a the peue aco thoe leak i changed. The change in duct leak peue diffeence ae ceated by peuizing and depeuizing the whole houe (including the duct) uing a blowe doo ove a ange of peue (including both peuization and depeuization). The blowe doo i ued to both ceate and meaue the flow occuing though the duct leak and the building envelope. In the ame way a envelope leakage tet, the peue and flow ae meaued at eveal fixed peue tation, with time aveaged peue and flow ecoded at each peue tation. The peuization (and depeuization) tet ae pefomed twice: once with the ai handle on and again with the ai handle off. The duct opeating peue ued in the calculation pocedue ae detemined by meauing the tatic peue in the upply and etun plenum elative to the conditioned pace. It i aumed that the envelope, upply and etun leak ae each epeented by powe law peue-flow elationhip. In addition, it i aumed that the additional peue developed by the blowe doo will be the ame eveywhee fo the envelope and the duct. Thi aumption i affected by envelope peue vaiation due to wind and tack peue (a with tandad blowe doo tet of building envelope). If thee ae lage leak in the duct ytem and coeponding lage flow, thee may alo be peue dop though the duct ytem again violating thi aumption of unifom peue. In ode to geneate ignificant peue dop catatophic leak ae equied uch a diconnected duct. Howeve, at thi level of leakage, the effect of violating of the aumption of unifom peue will not be elevant becaue the leakage tet will till indicate that thee i catatophic leakage. With the ai handle on the duct ytem peue ae not aumed to be unifom, but the peue aco the building envelope i aumed to add to the duct ytem opeating peue. Note that thi duct ytem peue aumption i not the ame a auming a peue aco leak fo duct peuization tet. The final eult fo the DeltaQ tet doe not cale diectly with peue a in the peuization tet. In geneal we have an unknown numbe and location of upply and etun leak. Each of i upply and j etun leak ha it own flow coefficient, C and peue exponent, n. The um of the flow though the building envelope and duct leak i the flow though the blowe doo. With the ai handle fan off at a given envelope peue diffeence () we have: nenv n,i ( ) + C ( ) + C ( ) n j Q off ( ) = Cenv,i i,j j, (1) allupplyleak alletunleak whee: Q off = meaued flow though blowe doo with ai handle fan off = peue diffeence aco envelope (in-out) i = peue diffeence aco upply leak i j = peue diffeence aco etun leak j C env = flow coefficient fo building envelope C,i = flow coefficient fo upply leak i C,j = flow coefficient fo etun leak j n env = envelope peue coefficient n,i = upply leak i peue exponent n,j = etun leak j peue exponent Fo the ai handle off tet it i aumed that the peue aco each individual leak i the ame a the envelope peue diffeence. Becaue it i not poible in any pactical way to know the numbe, ize and location of the duct leak it i aumed that all the upply leak can be combined and all the etun leak can be combined uch that: whee C = (2) C,i allup plyleak C = C, j (3) alletunleak 8

9 C = flow coefficient fo all of the upply duct leak C = flow coefficient fo all of the etun duct leak A well a gouping togethe all the leakage coefficient, the peue fo individual leak ae alo educed to ingle value. Latly, it i aumed that a ingle peue exponent can alo be ued. Then Equation 1 can be witten a: Q off env n ( ) ( ) ( ) env n n + C + C ( ) = C (4) Note that pefoming ai handle off meauement ove a ange of peue diffeence i the ame a pefoming a egula blowe doo tet fo envelope leakage. One of the advantage of the DeltaQ tet method i that it alo povide the leakage of the building envelope. Uing the ame model of the houe, upply and etun duct, with the ai handle fan on we have: Q on n ( ) ( ) ( ) ( ) env n n = C + C + + C env (5) whee: Q on = meaued flow though blowe doo with ai handle fan on = peue diffeence between upply and houe (houe a efeence). = peue diffeence between houe and etun (etun a efeence). Uing the etun a the efeence make a poitive numbe. When the building i peuized and the magnitude of etun peue i geate than the impoed blowe doo envelope peue the etun tem in Equation 5 i negative (i.e. flow into houe). Becaue thee i a peue dop between a plenum and the attached egite, the two meaued peue ( and ) do not neceaily epeent peue aco all duct leak--except in the pecial cae of a ingle leak in each of the upply and etun. Moe geneally, they epeent the chaacteitic peue that each et of leak i ubject to; in tem of ou analyi we can ee thi a inflection point in the DeltaQ v. elationhip. Since the leak that matte mot ae the one at o nea the plenum peue, we will aume ue thoe peue in the analyi and examine the enitivity of that aumption a we do o. In futue wok we will be examining the poibility of detemining thee peue by thoe that give the bet fit to the meaued data. Howeve, fo implicity and conitency, at the peent time we ue the plenum peue. In addition to wok by LBNL, John Andew of Bookhaven National Laboatoy (BNL) ha pefomed ome example calculation howing that inceaing thee efeence peue doe not neceaily inceae the eulting calculated leakage flow (peonal communication 11/2000). Thi i unlike peuization leakage tet method in which highe ytem peue would imply highe leakage flow. The DeltaQ i the diffeence between the ai handle on and ai handle off meauement: n ( ) = Q ( ) Q ( ) = C ( + ) ( ) n n n ( ) + C ( ) ( ) ) Q on off (6) Defining the upply and etun leakage flow: Q n = C ( ) n ; Q C ( ) = (7 and 8) whee: Q = upply leak flow at opeating condition to outide Q = etun leak flow at opeating condition to outide Equation 7 and 8 can be eaanged: 9

10 10 ( ) n Q C = ; ( ) n Q C = (9 and 10) Subtituting C and C into the DeltaQ equation, we get: ( ) + + = n n n n Q Q Q (11) Thi equation can be olved fo Q, Q, n and n given the meaued plenum peue, Q and. Howeve, it i eaie (and moe obut) if we fix the duct leakage peue exponent. Expeiment to chaacteize the peue exponent in a wide ange of duct configuation have hown that a value of 0.6 i uitable fo mot duct ytem (Walke et al and Siegel et al. 2001). Thi aumption would not be valid fo the cae of maive duct failue (e.g. a big hole o a diconnected duct). It i not uually too had to identify uch cae and to pefom the calculation with a lowe exponent. If we fix the value of n=0.6, and do a little algebaic manipulation we get a fom that give DeltaQ in tem of a diffeence between the upply and etun leak and i a little cleae to intepet (e.g., it i eaie to ee that when =0, then Q i the diffeence between upply and etun leak). ( ) + + = Q 1 Q Q (12) The DeltaQ can be meaued ove a ange of envelope peue (both poitive and negative). If n i fixed at 0.6 and and ae meaued at ytem opeating condition, then we only have two unknown the upply and etun leakage. Thi implie that only two peue tation ae equied in ode to detemine the leakage. Howeve, expeimental and analytical wok ha hown that uncetaintie ae educed if moe than thi minimum of two peue tation ae ued. Leak ae uually ditibuted thoughout a duct ytem, with ome at high peue diffeence at the plenum, ome at intemediate peue at connection in the duct and othe leak at low peue at the egite. Thi ange of hole with diffeent peue mean that at ome envelope peue diffeence ome leak contibute moe than othe to the eulting leakage flow and theefoe to the meaued DeltaQ.

11 Ai Handle Off Ai Handle On DeltaQ DeltaQ at zeo 120 peue = 65 cfm (31 l/) Ai Flow, cfm DeltaQ Flow, cfm eue Diffeence, a Figue 1. Example DeltaQ tet eult (l/ = cfm 0.47) Figue 1 i an example DeltaQ tet howing that the DeltaQ can have multiple maxima and minima, and be non-monotonic and non-linea. To account fo thi we popoe that a wide ange of envelope peue between zeo and 25 a be ued. Alo, the houe hould be both peuized and depeuized. The final DeltaQ (Equation (12)) i then leat quae fitted to all the data point o that the effect of all the leak ae included in the final olution. The intepolated value of the aiflow at a zeo peue diffeence diectly give u the diffeence between the upply and etun leakage unde nomal opeating condition (in thi cae 65 cfm (31 l/)). Flow Adjutment fo Exact eue Matching The mot difficult apect of the DeltaQ tet i the equiement that the ai handle on and ai handle off meauement be pefomed at identical indoo-outdoo peue diffeence. In pactice thi i not poible due to fluctuation in meaued envelope peue caued pimaily by wind peue and the difficulty in contolling blowe doo in uch a pecie manne. One of the key advantage of the DeltaQ method i that it make thee two meauement cloe in time; the hote the time diffeence between the two meauement the malle the piece of the wind pectum that the meauement i ubject to. Automatic meauement and contol ytem can ait in thi tak by adjuting the blowe doo fan peed to attempt to maintain the et peue diffeence, thu educing the fit ode effect. In ou field teting expeience, thi appoach ha poven to be bette than uing manual adjutment to the blowe doo, but thee ae till eidual on-off diffeence. Ideally the fan-on and fan-off flow would be meaued at exactly the ame peue tation, but noie (e.g. fom the wind) often equie that the data be coected fo mall diffeence in the meaued peue. At a given peue tation we have the envelope peue diffeence when the ytem i on ( on ), it coeponding flow (Q on ), the envelope peue when the ytem i off ( off, meaued ) and it coeponding flow (Q off,meaued ). Auming a powe law elationhip between peue diffeence and flow, and uing the meaued envelope peue exponent, the meaued flow with the ytem off can be coected to the flow at the ame peue a when the ytem i on uing: 11

12 n env Q on off Q = off,meaued (13) off,meaued Thi coection can be ued to intepolate the data to give flow etimate fo exactly matched peue. Becaue the meaued off and on ae cloe to begin with, any uncetaintie in auming that the peue flow elationhip i a powe law and in evaluating the peue exponent ae mall. In othe wod, becaue the flow coection will be mall anyway, the eo in thi intepolation pocedue will not be ignificant. Fo example, if n env i fixed at 0.67 but i eally 0.6, then the eo in the flow coection i only about 1% at low peue (5 to 10 a) and about 0.2% at 25a. DeltaQ tet potocol The following tep-by-tep pocedue ha been developed in ode to take the data equied by the DeltaQ tet. Note that all envelope peue ae meaued elative to outide i.e. in out, o that peuization of the houe i a poitive peue. Similaly, flow into the houe though the blowe doo ae alo poitive. 1. Connect the blowe doo aembly to the building envelope uing a window o doo opening. Seal o tape opening to avoid leakage at thee point. 2. Intall the envelope peue diffeence eno. The outide peue meauement location hould be helteed fom wind and unhine. Both the inide and outide peue meauement location hould be a fa away a poible fom the localized ai flow induced by the ai moving appaatu. All the envelope peue ue the outide peue a the efeence. 3. With the blowe doo opening blocked, blowe doo off and ytem off, meaue the peue diffeence aco envelope zeo. 4. With the ai handle fan on, meaue the upply ( ) and etun ( ) plenum opeating tatic peue elative to the conditioned pace. Note that both peue ae ecoded a poitive numbe fo ue in the analyi, i.e., the etun peue i NOT negative. 5. Tun on the blowe doo and adjut the flow until thee i 5 a (0.02 inche of wate) envelope peue diffeence, with the houe at a highe peue than outide (fo peuization teting). Recod the envelope peue diffeence ( env ) and flow (Q on ) though the ai-moving device at thi peue tation. Only ecod peue and flow eading when the peue eading i within 0.5 a (0.002 inche of wate) of the 5 a (0.02 inche of wate) opeating point. It i ecommended that multiple peue and flow eading ae ecoded at each opeating point and aveaged fo ue in the calculation pocedue. Note that all the blowe doo flow ae poitive out of the houe and negative if into the houe. 6. Repeat tep 5, but with the envelope peue diffeence, env, incemented by 5 a each time until the envelope peue diffeence i 25 a. At each env peue tation the peue diffeence mut be within 0.5 a (0.002 inche of wate) of the equied opeating point. Recod the envelope peue diffeence with the ai handle fan on, on, fo each peue tation. Becaue the capacity of the ai handling equipment, the tightne of the building, and the weathe condition affect leakage meauement, the full ange of the highe value may not be achievable. In uch cae ubtitute a patial ange encompaing at leat five data point, with the ize of peue incement uitably adjuted. At each peue tation, the ai handle fan on and off condition mut both have the ame taget peue. 7. Tun off the ai handle fan and epeat tep 5 and 6, ecoding Q off and off at each peue tation. 8. Repeat tep 5, 6 and 7, but with the houe depeuized, i.e., fo the fit point, adjut the flow though the blowe doo until thee i a -5 a envelope peue diffeence, with the houe at a lowe peue than outide. The magnitude of the envelope peue diffeence, env, i then incemented by 5 a each until the envelope peue diffeence i -25 a. 9. With blowe doo opening blocked, ai-moving device fan off and ai handle fan off meaue peue diffeence aco envelope with blowe doo off zeo. 10. Subtact the aveage of the zeo meauement fom each env to obtain. 11. Fit the ai handle off peue and flow data to the powe law elationhip in ode to obtain n env. 12. Adjut the flow fo coect peue matching uing Equation Calculate Q i at each peue tation, i, by ubtacting Q off,i fom Q on,i. 12

13 14. Do a leat quae fit of the and Q pai to Equation 12 to find upply leakage: Q, and etun leakage: Q. A expeience wa gained with the tet, thi pocedue wa efined. Initially we attempted to match the ai handle on and off peue conecutively at each peue tation. Howeve, mot ai handle incopoate a delay between the witch being activated and the ai handle being tuned on and the ai handle itelf hould be allowed to un fo about 30 econd to each teady opeation. Many ytem have no independent ai handle witch and the ai handle mut be activated though themotat etpoint manipulation. All of thee combine to poduce ignificant time delay between each ai handle on and ai handle off meauement. The tet wa teamlined by taking all of the ai handle off data followed by all the ai handle on data o that thee i only a ingle wait fo the ai handle to be opeating. Given that the ai handle on and ai handle off peue ae neve exactly matched anyway taking the data futhe apat in time (between on and off meauement) doe not have a ignificant effect o long a indoooutdoo tempeatue o mean wind peed to not change geatly (tempeatue change geate than 18 F (10 C) o wind peed change geate than 7 m.p.h. (3 m/)) duing the tet. Lage change would eult in ome biaing of the meaued envelope peue between the ai handle on and off tet. To futhe educe the time equiement and to educe any opeato inconitencie, a compute pogam ha been developed that ecod all the neceay data and contol the blowe doo in ode to cloely match the ai handle on and off peue. Thi oftwae alo pefom ome imple uncetainty analye. Fo example, the ai handle off data i taken twice and if the diffeence between thee two tet ae above a cetain citeia (e.g., if the fitted data yield envelope flow coefficient that diffe by moe than 2%), then the oftwae tell the opeato that the tet need to be epeated. Thee diffeence between ai handle off tet have eveal main caue: wind peue fluctuation on the building envelope, poo placement of indoo and/o outdoo peue tap, o peue tubing that i moved duing meauement. The data acquiition ytem take 120 data point at each peue tation and the mean value ae ued in the calculation. The fit veion of the oftwae ued the tandad deviation of thee 120 point to detemine when the blowe doo flow and envelope peue ae teady enough to ecod the eading. A tandad deviation of 0.5 a wa found to give a eaonable compomie between meauement peciion and the time taken to do the tet. A new oftwae veion i cuently being teted that ue a 0.1a limit of the tandad eo in the mean of the 120 point. Thi change ha little effect on the opeation of the meauement and data acquiition ytem. The oftwae alo check to ee if thee wa an unuual occuence (uch a individual point many tandad deviation fom the mean) duing a et of 5 peue point. Fo each et (ai handle on and ai handle off) the data ae fitted to a powe law to detemine a flow coefficient, C, and a peue exponent, n. The oftwae aume that the meaued data ae valid if the peue exponent i between 0.5 and 0.8, and that the coelation coefficient fom the leat quae fit i bette than In the futue a moe appopiate meaue of goodne of fit will be ued. Fo example, uing 95% confidence inteval fo flow coefficient Uncetainty analye Repeatability eliminay epeatability teting ha been completed uing multiple tet in a tet building at LBNL. The tet building wa located in a coatal hillide canyon at LBNL in Bekeley Califonia. The tet taile ha a 24 ft (7.3 m) by 50 ft (15.2 m) flooplan, with the long axi aligned with the pevailing wind. Inteio floo to ceiling height wa 8 ft (2.4 m). Thee ae no inteio wall o patition. The ame duct ytem wa teted 20 time ove eveal day. The maximum ecoded windpeed duing the tet wa 5 m.p.h. (2 m/). Thi i the windpeed meaued on ite at eave height and i lowe then typically meteoological windpeed due to the helteing effect of the canyon wall, tee, and adjacent imilaly-ized taile. The indoo to outdoo tempeatue diffeence duing the tet wee typically le than 15 F (8 C). Thee eult howed that the epeatability eo wee quite mall. Table 1 ummaize the epeatability teting eult. Both the tandad deviation and 95% confidence inteval ae given. 13

14 Aveage Supply Leakage, cfm (l/) 19 (9) Table 1. Repeatability Reult Fo DeltaQ teting Aveage Retun Leakage, cfm (l/) 66 (31) Standad deviation of upply leakage, cfm (l/) 11 (5) Standad deviation of etun leakage, cfm (l/) 16 (8) Supply Leak 95% Confidence Inteval, cfm (l/) 5 (2) Retun Leak 95% Confidence Inteval, cfm (l/) 7 (3) The ai handle flow of appoximately 1000 cfm (472 l/) wa meaued uing flow hood and tace ga meauement. Theefoe, the duct leakage and the uncetainty in leakage ae mall compaed to the ai handle flow. The epeatability eo wee quite mall in tem of flow ate, but the mall leakage flow lead to lage factional eo. Fo both uncetainty in enegy loe fom the duct ytem and low leakage pecification teting, the uncetainty a a faction of fan flow i the mot impotant facto. In thi cae the 95% confidence inteval wee between 0.5% and 1.0% (fo upply and etun epectively) of the fan flow what we would conide to be an excellent eult. A imila eult (mall abolute eo at low leakage) wa alo found by Andew (2000). In the futue, thee tet will be epeated with geate upply and etun leakage and changed taile envelope leakage. If poible tet will be pefomed at highe windpeed and tempeatue diffeence to inceae the potential tet to tet vaiability due to changing envelop peue. We expect that the fluctuation in envelope peue duing the tet (motly due to fluctuating wind peed and diection) could lead to inceaed tet eult vaiation. Howeve the epeatability eult howed that the tet vaiability doe not inceae vey much with the meaued envelope peue vaiability (hown in Figue 2). The vaiability i the tandad deviation of the offet peue meaued with the ai handle and blowe doo off. Thi i a good eult, becaue we would like the tet to be elatively inenitive to thee peue fluctuation o that it will give good eult in a wide ange of weathe condition. Figue 2 alo how the meaued equivalent leakage aea (ELA) fo each tet that wa calculated fom the ai handle off data. The ELA value alo how no tend with the offet tandad deviation that indicate the tet pocedue i unbiaed (fo thi building) by fluctuation in meaued peue. John Andew of BNL (Andew 2000), ha pefomed the DeltaQ tet thee time each in two houe. Fo each houe the aveage leakage and the aveage abolute diffeence fom the mean wee calculated fo the thee tet. Table 2 how that the aveage diffeence ae imila in magnitude to the above LBNL tet, but ae popotionally le a a faction of the duct ytem leakage becaue the leakage wa much highe in the BNL ytem. Table 2. DeltaQ Repeatability Tet in Two Long Iland Houe Houe Supply Leakage Retun Leakage Aveage, cfm (l/) Aveage Diffeence fom the mean, cfm (l/) Aveage, cfm (l/) Aveage Diffeence fom the mean, cfm (l/) (101) 8 (4) 46 (22) 14 (7) 2 72 (34) 19 (9) 286 (134) 9 (4) 14

15 Q, Q, (% Fan Flow) Q Q Building ELA ELA (in^2) Offet Standad Devation (a) Figue 2: Repeatability Tet Reult (cm 2 =in 2 6.5) Senitivity to the peue ued in the DeltaQ calculation The implifying aumption that allow all the upply leak to be epeented by a ingle leak and all the etun to alo be epeented by a ingle etun leak eult in a elationhip with a chaacteitic fom (Equation 12). Obevation of thi elationhip ove many DeltaQ tet have hown that it ha imila chaacteitic fo evey tet, and that ometime the individual data pai (that epeent the tue um of all the leakage flow) do not coepond well to thi functional fom. Figue 3 illutate how the hape of the deived DeltaQ function i changed by alteing the chaacteitic peue ued in Equation 12. The tandad DeltaQ tet ue the plenum peue a the leak peue motly becaue they ae eay to meaue with the leat factional uncetainty and becaue they ae elatively lage compaed to the othe peue in the ytem. Changing the peue to be one half the plenum peue give an indication of the enitivity of the pedicted leakage flow to uing thi peue. The figue how that halving the peue make only a mall change to the DeltaQ function. A key obevation hee i that the peue ued ae both outide the ange of the meaued data (±25 a). Ou obevation have hown that the peue indicate inflection point in the DeltaQ elationhip that can be intepeted a peue at which lage individual leak have the flow though them change diection if thee ae ingle dominant leak. We can ue thi infomation to develop bet fit DeltaQ elationhip. The bet fit peue wee elected to make the DeltaQ elationhip fit the meaued data cloe. In Figue 3, thee i a clea jump in the DeltaQ eult at aound minu 15 a. If we ue thi a one efeence peue and 25 a a the othe we get the bet fit line hown in the figue that epoduce the inflexion point at 15 a equied to fit the meaued eult. Note that the inflexion point ae only clealy een fo ytem with lage individual leakage ite whee the DeltaQ meaued data ha lage tep change. In many tet thi i not the cae, and the mooth, non-inflected (ove the tet peue ange) elationhip give vey good fit. Thi i alo the cae if the peue aco lage individual leak ae geate o le than the peue ued in the tet. E.g., when the dominant leak ae at the plenum. 0 15

16 We ae cuently developing a calculation pocedue that extend the fitting pocedue to include the efeence peue. Thi ha two clea advantage: the fitted elationhip will be cloe to the meaued data and the tet pocedue will be quicke becaue we no longe need to meaue plenum peue. + + = QQ Q ( ) n n n n Figue 3. Adjuting the peue ued in the DeltaQ elationhip to bette fit meaued data (l/ = cfm 0.47) Unlike duct peuization tet, the DeltaQ eult ae not vey enitive to the chaacteitic peue a long a the peue ae within a eaonable ange, e.g., within a facto of two. Table 3 give numeical value of upply and etun leakage fo the data ued in Figue 3 baed on uing diffeent chaacteitic peue. Uing the tandad DeltaQ appoach of uing the plenum peue i within one cfm of the bet fit eult. Thi eult i eauing becaue it indicate that the aumption of implifying the DeltaQ pocedue by uing jut the plenum peue doe not intoduce lage eo. The eaon fo thi i that the functional fom of the DeltaQ elationhip combine with the leat quae fitting method to poduce an extemely obut calculation pocedue. Table 3. Senitivity of DeltaQ eult to electing bet fit peue lenum eue =63 a, =100 a Half lenum eue =31 a, =50 a Bet Fit eue =15 a, =24 a Q, cfm (l/) 28 (13) 21 (10) 28 (13) Q, cfm (l/) 56 (26) 49 (23) 55 (26) R Uncetainty Etimate fo exponent and duct peue aumption In addition to the fit of meaued data dicued above, the uncetainty aociated with fixing the value of n (peue exponent) and uing plenum peue ha been invetigated paametically by uing DeltaQ tet meauement and vaying n and the upply and etun peue. By expeimenting with a lage numbe of field tet and analytically examining the functional fom of the DeltaQ elationhip we found that the 16

17 eult become moe enitive at lowe meaued ytem peue. In the vat majoity of cae the upply and etun leakage only change by a few cfm with eaonable change in peue and exponent. Futue wok on optimizing the chaacteitic peue will look at thee change in moe detail. Andew (2000) ha pefomed futhe uncetainty etimate uing both analytical technique and meaued field data. He found the theoetical and field-tet eult to be encouaging fo DeltaQ. The following point wee found to be mot impotant: The pedicted eult wee only weakly dependant on the aumption about duct opeating peue and duct leakage location (a we found in ou expeimental data alo). A monte-calo imulation technique that poduced vaiation in the flow and peue ued in the analyi wa ued to examine fou cae: balanced leakage, upply dominant, etun dominant and low leakage. 30 Simulation wee un fo each cae and analyi of the eult howed that the 3% tandad deviation ued on the meaued data howed how DeltaQ i about fou time bette at etimating the diffeence between upply and etun leakage than at meauing the um of upply and etun leakage. Field Expeience Initial ilot Tet The pilot tet of the DeltaQ pocedue wa pefomed in a houe whee we have aleady made eveal duct leakage meauement and the duct ytem chaacteitic wee well known. Fo thi fit houe the meauement wee pefomed manually, without compute contol o data acquiition. The following table ummaize the tet eult fo compaion pupoe. The ageement between the NT, DeltaQ and Tace ga eult indicate the duct peuization eult ae oveetimating the upply leakage (and undeetimating the etun) - motly becaue of uncetainty in etimating the peue aco the duct leak. Thi ytem had elatively low ai handle flow (about 330 cfm (155 l/)) o the etun leakage in thi cae i a lage faction of the ai handle flow. Supply Leakage cfm (l/) Retun Leakage cfm Table 4. Compaion of duct leakage meauement pocedue DeltaQ 5 (2) 181 (85) Duct euization 1 Duct euization 2 NT Tace ga n/a (24) (14) (8) 116 (55) 95 (45) (l/) 1- Conveted to opeating peue uing peue pan egite peue meauement 2- Conveted to opeating peue uing plenum peue 151 (71) 160 (75) Initial Field Evaluation In an effot to detemine the pactical limit of uing the DeltaQ tet eveal eeache have applied the tet pocedue to ome eidential houe. Thiteen houe have been teted by LBNL, BNL (Andew 2000) and Davi Enegy Goup (DEG). In mot cae, peuization tet and ai handle flow tet wee alo pefomed. The BNL tet wee pefomed manually but the DEG tet wee pefomed uing the oftwae developed by LBNL, and LBNL taff tained DEG taff on how to pefom the tet. Table 5 how the eult of thee tet. Thee houe cove a ange fom new to old (zeo to 100 yea), a lage ange of ize (up to about 3700 ft 2 (344 m 2 )) and a lage ange of duct ytem, a hown by the ange of ai handle flow. The ytem mateial include heet metal, duct boad and platic flex duct. The majoity 17

18 of the duct in the Califonia and Nevada houe wee in the attic, with ome ytem having duct in cawlpace. The Long iland houe had the majoity of thei duct in unheated/uninulated baement. Houe Location Ai Handle Flow, cfm (l/) Long Iland, NY 679 (319) Long Iland, NY 912 (423) leaanton, CA 1600 (752) Walnut Ceek, CA 879 (413) Folom, CA 1200 (564) Tacy, CA 1782 (834) Tacy, CA 1525 (717) Tacy, CA 2494 (1172) La Vega, NV 1551 (729) La Vega, NV 1900 (893) La Vega, NV 2114 (993) Alameda, CA 1265 (595) San Fancico, CA 515 (242) Table 5. Initial Field Evaluation Supply Leakage, cfm (l/) 214 (101) 72 (34) 61 (29) 136 (64) 52 (24) 109 (51) 70 (33) 102 (48) 81 (38) 18 (9) 3 (1) 48 (23) 58 (27) Retun Leakage, cfm (l/) 46 (22) 286 (134) 47 (22) 58 (27) 46 (22) 61 (29) 64 (30) 53 (25) 14 (7) 23 (11) 11 (5) 64 (30) 106 (50) DeltaQ Duct Leakage Supply Leakage, % of Ai Handle Flow Retun Leakage, % of Ai Handle Flow 32% 7% 8% 31% 4% 3% 15% 7% 4% 4% 6% 3% 5% 4% 4% 2% 5% 1% 1% 1% 0% 1% 4% 5% 11% 21% euization tet wee alo pefomed in thee houe fo leakage to outide. The peuization calculation ued half the plenum peue a the opeating peue to which the leakage flow wee conveted. On aveage, total leakage (upply plu etun) fo the peuization tet wee 2% of fan flow highe than the DeltaQ tet (about 10% of the meaued total). The RMS diffeence wa conideably highe at 9% of fan flow. Figue 4 illutate thee eult togethe with an equality line. 18

19 Duct euization (% Ai Handle Flow) Equality Line DeltaQ (% Ai Handle Flow) Figue 4. Compaion of DeltaQ to duct peuization tet fo total (upply plu etun) leakage. RMS diffeence = 9% of ai handle flow. Looking at the upply and etun leakage eult epaately how that the upply leakage aveaged 8% fo DeltaQ and 12% fo duct peuization. The peuization upply duct leakage wa conitently highe than DeltaQ value, a illutated in Figue 5, with an RMS diffeence of 7%. Figue 6 i a imila illutation compaing the etun leakage meauement. The etun leakage eult how a cloe ageement between the two tet, with mot of the diffeence occuing in thee houe. Fo the etun leak the aveage eult ae the oppoite of the upply leak: the DeltaQ tet aveaged 12% and duct peuization 8%. The RMS diffeence wa 8%, but thi i almot entiely diven by the thee houe with lage diffeence. Ignoing thee houe educe the RMS diffeence to 1%. 19