LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings1 of 26Field Investigation of Duct System Performance in CaliforniaLight Commercial BuildingsWm. Woody Delp, Nance E. Matson, Eric Tschudy, Mark P. Modera, andRichard C. DiamondLawrence Berkeley National LaboratoryBerkeley, CaliforniaSynopsisThis paper discusses field measurements of duct system performance in fifteen systems located in eightnorthern California buildings.AbstractLight commercial buildings, one- and two-story with package roof-top HVAC units, make up approximately50% of the non-residential building stock in the U.S. Despite this fact little is known about the performanceof these package roof-top units and their associated ductwork. These simple systems use similar ductmaterials and construction techniques as residential systems (which are known to be quite leaky). This paperdiscusses a study to characterize the buildings, quantify the duct leakage, and analyze the performance of theductwork in these types of buildings.The study tested fifteen systems in eight different buildings located in northern California. All of thesebuildings had the ducts located in the cavity between the drop ceiling and the roof deck. In 50% of thesebuildings, this cavity was functionally outside both the building’s air and thermal barriers. The effectiveleakage area of the ducts in this study was approximately 2.6 times that in residential buildings.This paper looks at the thermal analysis of the ducts, from the viewpoint of efficiency and thermal comfort.This includes the length of a cycle, and whether the fan is always on or if it cycles with the coolingequipment. 66% of the systems had frequent on cycles of less than 10 minutes, resulting in non-steady-stateoperation.1. IntroductionLight commercial buildings, primarily one- and two-story buildings with individual HVAC package roof-topunits serving floor areas less than 10,000 ft2, make up a significant portion (50%) of non-residential buildingstock in the U.S. and California. Commercial retail strip-malls are among the largest percentage of lightcommercial buildings. This stock also consists of offices, restaurants and professional buildings.First-cost dominates construction practices in these buildings. This potentially leads to short-cuts inconstruction practices and/or using lower grade materials. Often resulting in buildings which appear visuallydistressed five to ten years after they are built; moisture damage due to leaky roofs, and uncontrolled
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 1
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings2 of 26infiltration being the most common visual indicators of problems. The buildings use package roof-top unitsfor HVAC, and as with the buildings, if not to a greater degree, first-cost dominates, with the same potentialproblems of poor construction practice and/or lower grade materials.Slowly the industry, and research community, is acknowledging that duct-work in residential HVAC systemsleak, sometimes by a very large amount. Roof-top units in commercial buildings use the same duct-work andinstallation techniques as residential systems (combinations of sheet-metal, duct-board, and flex-duct).Considering construction standards and practices, it would be a surprise if ducts in small commercialsystems did not leak. The industry acknowledges that the ducts “may” leak, but since, in commercialbuildings, the ducts are largely inside the building, there has been little interest in their performance, and inquantifying the extent of and the impact of duct leakage. While the ductwork may be physically inside thebuilding, inside the ceiling cavity, this cavity is often outside the building’s thermal and air barrier, thusducts in many light-commercial buildings are subject to the same loss mechanisms as residential ductslocated in attics.1.1 Other WorkResearchers have recently documented the leakage characteristics of residential ducts. This study uses dataobtained at LBNL for various studies (Jump, et al., 1996). Other than anecdotal evidence, the only significantwork in the area of small commercial systems is from the Florida Solar Energy Center (FSEC). FSEC lookedat the entire building envelope in a study titled “Uncontrolled Air Flow in Non-Residential Buildings”(Cummings, et. al., 1996). Their primary concern was with uncontrolled flow across the building envelope,and they did envelope leakage studies in 70 light-commercial buildings. Since ducts often dominate buildingleakage, they also performed duct leakage measurements in 43 of these buildings.1.2 GoalsThe goals for this current study fell in three basic areas: characterization of the building and HVAC systems,measurement of duct leakage area, and measurement system and register flows. Characterization involvedidentifying HVAC unit sizes, occupied areas, and the location of the thermal and air barriers. The goals ofduct leakage information were measurement of fan and register flows, along with direct leakage areameasurements.The goals for this current study fell in three basic areas: building and HVAC system characterization, ductleakage, and duct thermal losses. Characterization involved identifying unit sizes, occupied areas, thelocation of the thermal and air barriers, and the system-fan and register flows. Duct leakage came from directleakage area measurements. Single-day temperature monitoring yielded information on thermal losses.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 2
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings3 of 262. MethodsBuilding selection consisted of buildings with package roof-top cooling systems, whose owners/occupantswere willing to cooperate with the study. All of the buildings in this study were occupied, which meantworking around the schedules of the occupants. This required the tests to be as non-obtrusive as possible, andconsisted of three distinct parts: walk-through characterization, leakage and flow measurements, and thermalmeasurements.2.1 BuildingsThere were eight buildings involved in the current study, three of which were LBNL office spaces. Theremainder were: a Stockton area office building, an office space located in a Sacramento area industrial park,a shoe repair store located in a Sacramento area strip-mall, a health food store in Marin county, and a Marincounty gymnastics facility. In total, we tested a total of fifteen HVAC systems in these eight buildings.2.2 Walk-Through InformationA simple walk-through with the occupants yielded most of the characterization information. Major items ofimportance were the name plate information on the HVAC equipment, duct material and location, buildingthermal barrier, and building air barrier. Other items such as occupancy schedules, internal loads, etc. wereobtained by filling out a questionnaire with the building occupants. Appendix A contains the questionnaireand the protocol for this study.2.3 Flow MeasurementsFan-flow measurements were measured with the tracer-gas method outlined by Delp, et al. (Delp, et al.1996). Due to the restrictions of working in an occupied building, register-air flows were measured using aflow hood only.2.4 Leakage MeasurementsThis study measured effective leakage areas using a modified duct pressurization method, as shown inAppendix B. The method uses a single set-up to measure the combined leakage area of both the supply andreturn duct systems. By using the HVAC unit as a flow meter, it is possible to determine the breakdown ofreturn and supply leakage. Leakage area calculations are proportional to the flow into the system. Therefore,uncertainties in leakage area are proportional to the uncertainties in measuring the flow into the system.Since the uncertainties in measuring the flow are within 5%, the same range applies to the leakage arevalues.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 3
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings4 of 262.5 Thermal MeasurementsThis study used small, battery-operated self-contained temperature loggers for all the thermal measurements.These loggers have a resolution and accuracy of approximately 0.2OC, and store 1,800 data points. Acollection interval of 12 seconds allows six hours of data storage. The loggers have a delayed start feature,allowing them to be left in place to start simultaneously at a pre-determined date. We collected the followingtemperatures: outside air, ceiling cavity, room, supply plenum, and at least one supply register.3. ResultsResults are presented in four primary sections: building and HVAC characteristics, duct leakage area,thermal issues, and occupant interactions.3.1 Building and HVAC CharacterizationFigure 1 shows the floor area versus the unit size, for both the LBNL and the FSEC data sets. Since the FSECdata set is larger, whenever the appropriate data is available, we use it for comparison. The important pointhere is the floor area served by each unit. This figure shows that the California (LBNL) buildings are similarto those in Florida (FSEC). Light-commercial buildings frequently have a greater load density (ton/ft2) thansingle-family residential homes, due to internal loads such as equipment, lights, and people. Unfortunatelywith most light-commercial buildings accurate load information is not available during design, andcontractors/engineers resort to a rule-of-thumb approach to equipment selection, often resulting in oversizedequipment. It is worth noting the values in the figures are installed capacities, and do not necessarilycorrespond to actual space loads.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 4
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings5 of 26Unit Size (ton)0246810121416Floor Area (ft2)050010001500200025003000350040004500FSECRegression Line(r2 =.45)LBNL Comm.FSEC Comm.LBNL Comm. 360 140FSEC Comm. 340 110Residential 570 90(ft2/ton) Avg. Std. Dev.Figure 1. Floor area -vs- unit size: using the current LBNL and FSEC (Cummings, et al., 1996)commercial data along with residential (Jump, et al., 1996) summary information. The FSEC unit sizeis derived from the total installed capacity in the building divided by the number of units.The fifteen systems had an average unit size of 3.9 tons, this compares with the FSEC data of 4.5 tons, andthe residential of 2.9 tons.The average floor area served by each unit was 1,500 ft2 for the current study, 1,400 ft2 for the FSECbuildings, and 1,800 ft2 for the residential buildings. Since the area served by each unit is similar among allthree data sets, the light commercial units are 30 to 50% larger than those found in single family residentialhouses.Figure 2 shows the total number of registers (supply and return) versus the unit size. Since FSEC data on thenumber of registers was not available, the comparison is only with the residential data set. In both cases thereis a widespread range in the number of registers for any given unit size, e.g., the number of registers found ona four-ton commercial unit ranged from 4 to 16. In general, the commercial units have fewer registers per tonthan the residential units, because commercial spaces are usually open plan, with a few large rooms andfewer, but larger, registers.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 5
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings6 of 26Unit Size (ton)0246810Number of Registers (Supply and Return)024681012141618LBNL Comm.ResidentialLBNL Comm. 2.2 1.0Residential 3.5 1.0 (Reg/ton)Avg. Std. Dev.Figure 2. Number of registers -vs- unit size: using the current LBNL commercial andresidential (Jump, et al., 1996) data.In order to understand the dynamics of duct losses, details of the building need to be addressed. Figure 3summarizes many of the characterization details pertaining to the buildings. Each of the buildings had a dropceiling with the ducts run in the ceiling cavity. Because of this, two critical building details are the location ofthe thermal and the air barrier. 25% of the buildings had insulation placed both at the roof deck and on top ofthe ceiling tiles. The remainder of the buildings were divided between roof-only and ceiling-only insulation.40% of the buildings had a directly vented ceiling cavity. In these buildings, the lay-in acoustical ceiling tilesformed the major air barrier. In 50% of the buildings the primary thermal barrier was at the ceiling tiles,which implies that the ducts are entirely outside the conditioned space. In 25% of the buildings the ceilingcavity acted like a buffer zone, with the temperature floating between the room and outside temperatures.With these buildings, the thermal barrier is in-between the roof and ceiling. In the remainder of the buildingsthe thermal barrier was at the roof, however even in these buildings, the ceiling cavity temperature wasslightly higher than the room.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 6
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings7 of 26Insulation DetailsCeiling CavityThermal BarrierCeilingOnlyCeilingDirect-VentedUn-ventedRoofOnlyIn-betweenRoof25%50%40%37%Both25%Figure 3. Building thermal and air barrier characterization for the current LBNL commercialbuildings.Figure 4 summarizes HVAC unit characterization details. Duct material fell into two basic types: all metaltrunk-and-branch, and flex-octopus. 60% of the systems had all metal ducts, while the remainder had someform of flex-octopus. There are two types of basic ductwork configurations found with the typical light-commercial package roof-top unit: bottom discharge, and side discharge. Bottom-discharge eliminatesductwork exposed outside since it penetrates the roof directly under the unit. The typical side-dischargeinstallation includes 900 elbows directly off the unit, ideally cutting down on the amount of duct exposed onthe roof. Economics and local practice govern which method is used. Bottom discharge units require the useof a special curb to support the unit, while side discharge units typically use a field-fabricated platform for theunit. 33% of the HVAC units had bottom-discharge ductwork, while the remainder used a side-dischargearrangement.Air side economizers minimize cooling energy use when it is cooler outside than inside. 47% of the units hadsome sort of economizer; however, they were not checked for functionality. Only one unit had functioningminimum outside air (an intentional opening in the return duct directly to outside). All of the others eitherhad no outside air provisions, or had the dampers permanently shut.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 7
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings8 of 26Unit DischargeEconomizerMinimum Outside AirNoneNone or Permanently ShutDuct MaterialAll MetalSome FlexBottom60%67%53%93%SideFigure 4. Characterization of HVAC unit details for the current LBNL commercial buildings.Figure 5 shows system fan flow versus unit size. Fan flow is often used as an indicator of installation quality.The general rule-of-thumb is a flow of 400 cfm/ton. This corresponds to the rated flow, required to obtain thepublished efficiency ratings, for the HVAC system. When the flow drops below approximately 250 cfm/ton,the coil operates with a temperature conducive to frost formation. Even in dry climates, flow rates this lowimpair system efficiency. Most of the units had adequate flow, which makes sense with units with fewregisters, hence low pressure drop systems. Two systems had flows close to the 250 cfm/ton range.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 8
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings9 of 26Unit Size (ton)0246810Fan Flow (cfm)05001000150020002500300035004000Target Value forNormal Design:400 cfm/tonLBNL Comm.250 cfm/tonLower Limit:Likely Frost on the CoilLBNL Comm. 370 80Residential 370 86 (cfm/ton)Avg. Std. Dev.Figure 5. Fan flow -vs- unit size for the current LBNL commercial buildings along with residential(Jump, et al., 1996) summary information.3.2 Leakage Area of the Duct SystemsThe main emphasis of the current study was to measure the leakage area of the ducts. There are several waysto compare the systems to each other, and to other data sets. The goal of comparison is to find a way tonormalize the data, making direct comparison of different systems possible.Figure 6 shows the combined leakage area (ELA25) versus the unit size for both commercial data sets. Thedata have a large spread in values. A linear regression on the FSEC data only had a r2 of 0.29. The LBNLand the FSEC leakage values fall in the same general range for any given unit size. Normalizing leakage areawith the unit size (cm2/ton) does not yield a constant due to the large spread in values. However, theresidential and FSEC data sets had similar average values for leakage are per ton (cm2/ton), while the LBNLsmall commercial buildings had a 40% higher average value, due to outliers.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 9
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings10 of 26Unit Size (ton)0246810121416Combined ELA25 (cm2)02505007501000125015001750LBNL Comm.FSEC Comm.FSECRegression Line(r2 =.29)LBNL Comm. 110 66FSEC Comm. 78 48Residential 76 51Combined ELA (cm2/ton) Avg. Std. Dev.Figure 6. Combined leakage area (ELA25) -vs- unit size using the current LBNL and FSEC (Cummings,et al., 1996) commercial data along with residential (Jump, et al., 1996) summary information.Combined leakage area includes both supply and return leakage. The FSEC unit size is derived fromthe total installed capacity in the building divided by the number of units.Figure 7 shows the combined leakage area (ELA25) versus the floor area for both commercial data sets.Again, the data show a large spread in values. A linear regression on the FSEC data only had a r2 of 0.26.The LBNL data grouping is similar to, and slightly higher than, the FSEC data. It is common to presentbuilding envelope leakage results by normalizing leakage area with floor area (cm2/m2). The average cm2/m2in the LBNL data set was over 2.5 times that of the residential data, while the FSEC data was just over 2times the residential. These data suggest that light-commercial duct systems leak air at a rate much greaterthan residential systems, for any given floor area.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 10
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings11 of 26Floor Area (m2)0100200300400Combined ELA25 (cm2)02505007501000125015001750LBNL Comm.FSEC Comm.FSECRegression Line(r2 =.26)LBNL Comm. 3.4 1.6FSEC Comm. 2.7 1.7Residential 1.3 .65Combined ELA (cm2/m2) Avg. Std. Dev.Figure 7. Combined leakage area (ELA25) -vs- floor area using the current LBNL and FSEC(Cummings, et al., 1996) commercial data along with residential (Jump, et al., 1996) summaryinformation. Combined leakage area includes both supply and return leakage.Figure 8 shows the combined leakage area (ELA25) versus the total number of number of registers for theLBNL and residential data sets (the number of registers was not available from FSEC). For the same numberof registers the commercial buildings consistently have higher leakage areas. The average cm2/register amongthe LBNL data is 2.3 times that of the residential buildings. This makes sense, since the likely leakage site isat any connection, and as commercial buildings use larger ducts than residential, the larger connection siteshave a greater potential for leakage. A linear regression on the LBNL data had a r2 of 0.80. This suggeststhat leakage area normalized by the number of registers can be used as a metric for identifying problematicsystems.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 11
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings12 of 26Number of Registers024681012141618Combined ELA25 (cm2)020040060080010001200LBNL Comm.Regression Line(r2 =.80)LBNL Comm.ResidentialLBNL Comm. 51 15Residential 22 13cm2/Reg Avg. Std. Dev.Figure 8. Combined leakage area (ELA25) -vs- number of registers using the current LBNL commercialand residential (Jump, et al., 1996) data. Combined leakage area includes both supply and returnleakage.3.3 Thermal IssuesFigure 9 shows typical temperature profiles for an office conference room over a six-hour period, and itillustrates several points concerning thermal issues. The change in temperature between the supply airplenum, down-stream of the coils, and the register affects system efficiency. It also impacts thermal comfort,the longer the duct the greater the temperature rise, which leads to uneven temperature distribution by thesystem. The figure illustrates this, as register 1 is closer to the plenum than register 2. The length of cycleduration also effects the temperature rise. Energy from the air stream cools the ducts until they reach asteady-state temperature. The figure shows that air temperature in this particular system never reached steadystate, the longer the cycle the lower the temperature rise from the plenum. Finally, this figure shows that thefan does not cycle on and off with the cooling equipment. The cooling equipment shuts off at the bottom ofeach spike on the plot. The plenum and both registers continue to rise to approximately the same temperaturebefore the beginning of the next cycle. This recovers the energy used to cool the ducts, but it also startsheating the room, as the unit is on the roof in the sun. With the fan running, and the cooling equipment off,
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 12
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings13 of 26the unit acts as a heat exchanger heating the room below. For the 6 hour period represented in this figure, thehot air delivered during the off-cycle accounted for 33% of the total space cooling load.19 July 19969:3010:0010:3011:0011:30Noon12:301:001:302:002:30oC8101214161820222426Register 1Register 2CeilingRoomSupply PlenumOffice Conference RoomEquipmentshuts offEquipmentturns onFigure 9. Typical temperature data for 6 hours of operation3.3.1 Temperature Rise From Supply Plenum to RegisterAs stated above, the temperature rise from the supply plenum to a supply register affects both systemefficiency and thermal comfort. In order for the temperature to rise after leaving the plenum, there needs to bea potential difference between the temperature in the plenum and the temperature of the ambientsurroundings (i.e., it has to be hotter outside the duct than inside the duct). Figure 10 shows the temperaturerise from the supply plenum to the register plotted versus this temperature difference, between ambient andplenum. With the ducts located, in all cases, in the ceiling cavity, the ceiling cavity serves as the ambienttemperature. All points represent conditions in the mid-afternoon (~ 2 to 4 p.m.), and as close to a steady-state operation as possible. Systems that never reach steady-state are labeled as transient, and thetemperatures are from the end of the longest on-time cycle available. The temperature rise ranged from 0.5OCto almost 6OC. Both extremes, high and low, occurred near the greatest potential difference; however, thetemperature difference between ambient and the plenum alone does not correlate well with the temperaturerise.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 13
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings14 of 26Temperature Difference betweenAmbient and Supply Plenum (oC)12131415161718192021Temperature RiseFrom Plenum to Register (oC)0123456XXXXXSemi-Steady State(on-time > 10min.)Transient at end of on-cycle(on-time < 10min.)XFigure 10. Temperature rise from supply plenum to register -vs- temperature difference betweenambient and supply plenum using the current LBNL commercial data. All data taken in mid-afternoon,at the end of an on-cycle.The length of the duct impacts the temperature rise in addition to the ambient conditions. Figure 11 plots thetemperature rise from the plenum versus the length of duct. Again the data shows a high degree of scatter.The longest duct has one of the lowest temperature rises, illustrating that the length does not properlycorrelate with the temperature rise.There are several factors not taken into account by either of these two previous attempts at correlating thetemperature rise. Among these factors are, the effective temperature of the ambient, taking into accountradiation effects, differing amounts of insulation on the ducts, different duct sizes, and finally, different airvelocities within the duct. Unfortunately, not enough data was taken to easily compare the temperature risefrom one system with another.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 14
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings15 of 26Length of DuctFrom Plenum to Register (m)051015202530Temperature RiseFrom Plenum to Register (oC)0123456XXXXXSemi-Steady State(on-time > 10min.)Transient at end of on-cycle(on-time < 10min.)XFigure 11. Temperature rise from supply plenum to register -vs- the length of the duct the from thesupply plenum to register using the current LBNL commercial data. All data taken in mid-afternoon,at the end of an on-cycle.3.3.2 Cycle DurationFigure 9 showed that on-cycle time impacted the plenum to register temperature rise. The system shown inthis figure had no period of steady operation with an average on-time of slightly over 5min. Figure 12 showsa typical temperature rise versus time. This figure is for a different system. It shows that the ducts effectivelyreached a steady-state condition in approximately ten minutes. This particular register was far from theplenum (24 m). Registers closer to the plenum reach steady-state operation sooner, leading to periods ofuneven temperature distribution.66% of total number of systems tested operated a significant portion of the time with the on-time cycles lessthan 10 minutes. 33% of total number of systems had no single on-cycle longer than 10 minutes. This resultsin non-uniform (from register to register) and constantly changing temperatures (at all registers) in thedistribution system, and the space, leading to thermal comfort issues.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 15
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings16 of 26Time Into On-Cycle02468101214161820Temperature RiseFrom Supply Plenum to Register (oC)01234GymStart of Cycle~2:30pmLength of Duct fromPlenum to Register »
`24mFigure 12. Typical temperature rise from supply plenum to register -vs- time after start of on-cycle.Plot is for a single cycle for a single system.3.3.3 Fan OperationFigure 13 shows two hours of temperature data for two different systems. With one of the systems, the store,the fan cycles with the compressor. While the other system, the gym, regardless of whether or not thecompressor is on, the fan stays on all of the time. Keeping the fan on serves two functions (1) maintains airmovement and (2) (assuming provisions exist) provides outside air. During the off-cycle, in the store, theplenum temperature rises dramatically, while the register temperature approaches the room temperature. Atthe beginning of the on-cycle the plenum temperature is more than 15OC higher than the registertemperature. This is because the fan is off, and the unit and plenum are on the roof, the effective temperature,including solar-radiation, warms the air in the plenum and unit considerably. In the gym, the fan stays onduring the equipment off-cycle, and both the plenum and register temperatures warm approximately the samebefore the beginning of the next on-cycle, approaching and slightly exceeding the room temperature.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 16
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings17 of 26Gym2:00pm2:30pm3:00pm3:30pm4:00pmoC12162024283236oC12162024283236Supply PlenumRegisterCeilingStoreCeilingRoomRegisterSupply PlenumRoomOutsideFan On ConstantFan CyclesFigure 13. Two systems: in the store the fan cycles with cooling equipment, and in the gym the fan stayson all the time.3.3.4 Conduction LossesThe basic definition of the conduction effectiveness concerns the fraction of the capacity lost. In the simplestterms, neglecting any leakage, the term is simply (1)
`
`st():Conduction effectiveness at time tTtreg():Register temperature at time tTtroom():Room temperature at time tTtplenum():Supply plenum temperature at time tEquation (1) yields an instantaneous value for effectiveness at any given time t. It includes both conductionand impacts of thermal cycling. By summing each of the quantities in the above equation one arrives at acumulative effectiveness for any given time t’ (2)
`
`- -(1)Where:
`
`=
`sregroomplenumroomtTtTtTtTt()(()())(()())
`
`e
`
`e
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 17
`
`

`

`)
`(
`- -(cid:242)(cid:242)00(2)Where:
`sregroomtplenumroomttTtTtdtTtTtdt'()()()()''
`( )
`)
`(
`
`=
`
`e
`
`e
`
`( )
`
`st':Cumulative conduction effectiveness up to time t’The cumulative effectiveness in equation (2) gives us a running total of the ratio of the energy delivered at theregister to the potential at the plenum. The uses of the above formulations have subtle differences between thefan cycles with unit and fan-on constant cases.3.3.4.1 Fan Cycles with UnitFigure 14 shows a single cycle for a case where the fan cycles with the cooling unit (fan on/off case). Thebuilding is the same store shown in Figure 13. This cycle was the first of the day, and both the plenum andregister temperatures started off high. The instantaneous and cumulative effectiveness started low since theducts were warm and had to be cooled. An important point here is the fact that the cumulative lagged behindthe instantaneous effectiveness; at the end of the on-cycle, the instantaneous was over 90% while thecumulative was around 80%. The difference between the two effectiveness values is a consequence of theenergy stored in the ducts as a fraction of the energy supplied to the plenum. Nothing was known about theflow out of the registers after the fan shut off, as a result, what happens to this stored energy is not clear fromthis information. The overall analysis for the fan-off cases only includes on-cycle information.
`
`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings18 of 26
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 18
`
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings19 of 26Time Into Cycle (min)-101234567891011oC121416182022242628Conduction & Cycle
`s (%)30405060708090100Room (oC)Plenum (oC)Register (oC)Instantaneous es (%) Cumulative es (%)StoreStart of Cycle~ 2:10pmFigure 14. Analysis of an individual cycle with a fan on/off case.Figure 15 shows the cumulative effectiveness versus time for all four of the cycles captured in the store. Witheach consecutive cycle the ducts start off a little cooler. This accounts for the fact that the effectiveness startsat a higher value for each consecutive cycle. The final three cycles end near the same value (~85%); however,due to the short on-times they have not reached a steady-state.
`
`Sleep Number Corp. - EXHIBIT 2021
`IPR2019-00514 - Page 19
`
`e
`

`

`LBNL# 40102Field Investigation of Duct System Performance in California Light Commercial Buildings20 of 26Time Into Cycle (min)0123456789Cumulative
`s (%)30405060708090100First cycle of the daySecond cycleThird cycleFourth cycleStoreFigure 15. Cumulative supply effectiveness ( ees) -vs- time for four different cycles in a fan on/off case.3.3.4.2 Fan Stays OnFigure 16 shows three hours of temperature information for a fan-always-on case. This is the same gym inFigure 13, only with different time and temperature scales. The length of the off-cycle is important since aftera period of