1、556 2010 ASHRAEABSTRACTChemical reactions between ozone and terpenoids canyield secondary organic aerosol (SOA), which are potentially alarge source of indoor particles that are harmful to humanhealth. The mass of SOA formed in a building is influenced bythe operation of the heating, ventilation, an
2、d air-conditioning(HVAC) system. This investigation models the influence ofHVAC systems on SOA concentrations in residential and com-mercial buildings. A parametric analysis explores the role ofventilation and recirculation rates, filtration efficiency andloading, and the operation of heat exchanger
3、s. In a rural setting,the median residential and commercial SOA concentrationsfor all simulations were 17.4 g/m3(1.09 109lb/ft3), with arange of 2.47 to 27.0 g/m3(1.54 1010 1.68 109lb/ft3),and 10.6 g/m3(6.61 1010lb/ft3), with a range of 1.81 to26.3 g/m3(1.13 1010 1.64 109lb/ft3), respectively. In an
4、urban setting, the median predicted residential and commercialSOA concentrations were 68.0 g/m3(4.24 109lb/ft3), witha range of 14.7 to 108 g/m3(9.17 1010 6.74 109lb/ft3),and 44.8 g/m3(2.80 109lb/ft3), with a range of 11.6 to105 g/m3(7.24 1010 6.55 109lb/ft3), respectively. Themost influential HVAC
5、parameters are the flow rates throughthe system, particle filtration efficiency, and indoor temperaturefor the residential and commercial models, as well as ozoneremoval on used filters for the commercial model. The resultspresented herein can be used to estimate the effects of alteringHVAC system c
6、omponents and operation strategies on indoorSOA concentrations and subsequent exposure.INTRODUCTIONParticulate matter (PM) diameter spans many orders ofmagnitude, from a few nanometers to tens of micrometers, andexposure to particles has been associated with harmful effectson human health. Fine part
7、icles (4 g/m3(2.5 1010lb/ft3) then Yg,sris constant at 0.197. Thesize-resolved mass yield, Yg,sr, is modeled as a lognormaldistribution, and its parameters were fitted to the POC-Seedexperiment by converting the tri-modal lognormal distributionof the steady-state SOA number concentration into a uni-
8、modal lognormal mass distribution, with GM = 0.37 m andGSD = 1.52. The ozone and terpenoid reaction rate constant,k, was calculated as follows. Neglecting ozone decay due toirreversible wall deposition (experiments were in Teflon-lined chamber), a steady-state mass balance with ozone andterpenoid co
9、ncentrations yields k = 0.05 ppb1h1. Yg,srandk are identical in both models.HVAC Parameter (i): HVAC FlowThe air exchange rates used in the models are listed inTable 2. The HVAC system directly controls the ventilationand recirculation rates, and these were varied to explore theirinfluence on SOA fo
10、rmation. The Flow cases used infiltrationand recirculation air exchange rates from Riley et al. (2002)and Waring and Siegel (2008). The residential Duty caseassumed cycling of conditioning equipment, and thereforerecirculation for one-sixth of the total time, and the Continu-ous case considered the
11、air handler fan to be running the entiretime. For the commercial HVAC Flow cases, all operation wascontinuous and three air makeup cases were considered, withassumed air exchange rates based on engineering judgmentthat were also used in Waring and Siegel (2008). The 100%outside air (OA) case represe
12、nts a building for which air recir-culation is undesirable. The 50% OA/50% recirculated airTable 1. For the Rural and Urban Ambient CasesThe outdoor ozone concentrations and particle number distributions, including the total number concentrations (#/cm3or multiply by 2.83 104for #/ft3) and the geome
13、tric mean diameters (GM) and log of geometric standard deviations log(GSD) for each mode.Ambient CaseOzone Conc.(ppb)Particle DistributionsMode 1 Mode 2 Mode 3Number (#/cm3)GM(m)log(GSD) ()Number (#/cm3)GM(m)log(GSD) ()Number (#/cm3)GM(m)log(GSD) ()Rural* 25 6650 0.015 0.225 147 0.054 0.557 1990 0.0
14、84 0.266Urban* 100 99,300 0.013 0.245 1100 0.014 0.666 36,400 0.05 0.337*Ozone concentrations were assumed, and particle number distributions are from Jaenicke (1993). 2010, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transact
15、ions 2010, Vol. 116, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission. 560 ASHRAE Transactions(RA) case represents a heavily occupied building, and the 10%OA/90% RA case rep
16、resents a lightly occupied building.In the residential model, the Duty and Continuous flowcases also cause different airflow regimes in the modeledindoor space. Indoor spaces without continuous recirculationare assumed to have average air flows of lower velocity thanthose indoor spaces with continuo
17、us recirculation. Sincehigher velocity flows lead to a decreased thickness of theboundary layers adjacent to surfaces, the deposition parame-ters of pand O3are expected to increase with higher velocityflows. Thus, the residential model uses different values of pand O3for the Duty and Continuous case
18、s. The commercialmodel has continuous flow for all cases, so it uses one constantvalue for both pand O3.Similar to in Riley et al. (2002) and Waring and Siegel(2008), we used the model of Lai and Nazaroff (2000) to deter-mine specific values of p. One input in their model for pisthe friction velocit
19、y, u* (cm/s or ft/h), which is an empiricalparameter that describes the level of turbulence intensity neara surface. This parameter thus represents the air flow condi-tions in a space, with higher values for u* associated withhigher velocity flows. Typical values of u* for indoor environ-ments are 0
20、.3 to 3 cm/s (35.4 to 354 ft/h) (Lai and Nazaroff,2000). For the residential model, the Duty case was assignedthe pfor u* = 1 cm/s (118 ft/h) and the Continuous case foru* = 3 cm/s (354 ft/h). For the commercial model, all threeflow cases were assigned pfor u* = 3 cm/s (354 ft/h).Sabersky et al. (19
21、73) described O3for two residentialcases in the same home, without and with the forced air systemoperating, at 2.9 and 5.4 h1, respectively. In the residentialmodel, our Continuous case assumes the forced air system isalways on, so it was assigned as O3= 5.4 h1. For the Duty case,we assumed a value
22、of 5.4 h1when the system was on and avalue of 2.9 h1when the system was off, for an overall valueof O3= 3.3 h1. For the commercial model, O3was assignedfor all flow cases as 4.2 h1, which is an average of the officeozone deposition loss rates summarized in Weschler (2000).HVAC Parameter (ii): PM Fil
23、trationFive removal devices were used in the residential andcommercial models: four porous-media filters and one elec-trostatic precipitator (ESP). The efficiency curves for all fivefilters are displayed in Figure 2. We assumed that each filterretains the efficiency shown in Figure 2 and is constant
24、 overtime, though filter removal efficiency typically changes withloading (Hanley et al. 1994; Wallace et al. 2004). Filterefficiency data for the four porous-media HVAC filters wereobtained from ASHRAE Standard 52.2 tests (ASHRAE 2007)provided by filter manufacturers. The ASHRAE Standard52.2 proced
25、ure challenges filters with particles from 0.3 to10 m, so the fibrous filtration theory described by Hinds(1999) was used to extend the data into the full range used inthis study, following the procedure in Riley et al. (2002). Theseare the same filter curves used in Waring and Siegel (2008).The fif
26、th filter used in our models was an in-duct ESP, andits efficiency curve was derived from Wallace et al. (2004),who reported size-resolved mean deposition rates in a town-home with the central house fan operating continuously, bothwithout and with an in-duct ESP operating. The size-resolvedefficienc
27、y of the in-duct ESP was calculated for each reportedparticle diameter in Wallace et al. (2004) with the relationship,ESP= (r,townhome)(ESP), where ESPis the difference indeposition loss rates with and without the ESP operating,r,townhomeis the rate of recirculated air in the townhome(reported by th
28、e authors as 5.4 h1), and ESPis the calculatedsize-resolved efficiency of the in-duct ESP. Wallace et al.(2004) reported size-resolved deposition rates for the particlediameter range of 0.0181 to 1.843 m, so the efficiencies ofmodeled particle diameters that were lower than this rangewere assigned t
29、he ESPfor the particle diameter of 0.0181 mand those higher were assigned the ESPfor the particle diam-Table 2. Summary of Air Exchange Rates Used in the Residential and Commercial ModelsAir Exchange Rate (h1)ResidentialCommercialDuty Continuous100% OA50% OA/50% RA10% OA/90% RAi0.75 0.75 0.25 0.25 0
30、.25v0 0 4 2 0.4r0.67 4 0 2 3.6Figure 2 Filter efficiency curves for the MERV 5, 6, 11,and 15 filters (M5, M6, M11, and M15) and theelectrostatic precipitator (ESP) used in thismodeling study. 2010, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Publ
31、ished in ASHRAE Transactions 2010, Vol. 116, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission. ASHRAE Transactions 561eter of 1.843 m. The same efficiency curve for ESPwas u
32、sedin the residential and commercial models due to the lack ofefficiencies reported for ESPs in commercial systems. The useof an in-duct ESP produces ozone, so our models coupled anindoor ozone emission rate, EO3, to the residential and com-mercial scenarios that employed the ESP. Wallace et al. (20
33、04)did not report ozone emission rates for the studied ESP, so ourresidential and commercial models used the rate of a commer-cially available unit in Viner et al. (1992) of EO3= 21.6 mg/h(4.76 105lb/h) for the continuous HVAC Flow cases in theresidential and commercial models. For the Duty case in
34、theresidential model, the ozone emission rate is one-sixth theContinuous case and is EO3= 3.6 mg/h (7.93 106lb/h).HVAC Parameter (iii): O3FiltrationAs ozone-laden air passes through a porous-media HVACfilter, ozone can be removed by the filter, predominately dueto reactions with loaded particles (Hy
35、ttinen et al. 2003; Beket al. 2006, 2007; Zhao et al. 2007). We modeled New andUsed filter cases for ozone removal. Both models assumed thatO3= 0% for their New cases. The models assumed Usedvalues of O3= 10% and 41% for residential and commercialbuildings, respectively (Zhao et al. 2007). The resid
36、entialvalue of 10% is the mean of eight particle-laden filters takenfrom actual residences, and the commercial value of 41% isthe mean of five particle-laden filters from commercial envi-ronments. Though the ozone removal efficiency of a porous-media filter is likely associated with its particle rem
37、ovalefficiency, our model does not link the two since there areinsufficient data to make such an association. All scenarioswith an ESP are assigned the ozone removal value of O3= 0%.HVAC Parameter (iv): TemperatureThe heating or cooling coil operation influences the airtemperature in a modeled space
38、. Three different indoor temper-atures, 18.3, 23.9, and 29.4C (65, 75, and 85F) were consid-ered. The total mass of SOA formed increases as temperaturedecreases (Leungsakul et al. 2005; Sarwar and Corsi 2007).Though the reaction rate of ozone and terpenoids decreases,this decrease is surpassed by th
39、e increase in gas-to-particlepartitioning that occurs as the vapor pressures of condensingproducts decrease. For reactions between ozone and d-limo-nene (the primary SOA forming reactant in the POC), Leung-sakul et al. (2005) report that the mass of SOA formed changesat a rate of 0.016C1 (0.0089F1)
40、and Sarwar and Corsi(2007) report a rate of 0.04C1(0.022F1). The experi-ments reported in Coleman et al. (2008) were conducted at atemperature of 23C (73.4F), and temperature formationfactors, FT, were calculated with the averages of the two ratesfrom Leungsakul et al. (2005) and Sarwar and Corsi (2
41、007).These FTadjust the mass of SOA formed at the experimentaltemperature to represent that which would occur at themodeled temperatures. For both models, FTequals 1.13, 0.98,and 0.82 for the cases of 18.3, 23.9, and 29.4C (65, 75, and85F), respectively.HVAC Parameter (v): Relative HumidityThe heati
42、ng or cooling coil also influences the relativehumidity (RH) in a space. Both models utilize three differentvalues for indoor RH of 25, 50, and 75% to model the range ofRH that occurs in buildings in different climates. As RHdecreases, the water available for reactions becomes limited.Some products
43、of the ozone and d-limonene reaction (stabilizedCriegee intermediates) that can react with water instead reactwith other products of the ozone and d-limonene reactions (alde-hydes) to form less volatile compounds, increasing total SOAmass formed (Leungsakul et al. 2005). Leungsakul et al. (2005)repo
44、rt that the mass of SOA formed changes at a rate of 0.0009%1. The experiments reported in Coleman et al. (2008)were conducted at an RH of 50%, and RH formation factors,FRH, were calculated with the rate from Leungsakul et al. (2005).These FRHadjust the mass of SOA formed at the experimentalRH to rep
45、resent that which would occur at the modeled RH. Forthe residential and commercial models, FRHequals 1.02, 1.0, and0.98 for the cases of 25, 50, and 75%, respectively.Base Case Definitions and Number of Reported ScenariosFor the residential and commercial models, the variedHVAC parameters, as well a
46、s the literature sources used forinput values, are summarized in Table 3. A Rural and Urbanbase case for both the residential and commercial models wasselected based on typical values for each parameter. Within theRural and Urban distributions, the residential base caseconsisted of a Duty flow cycle
47、, a MERV 6 filter the require-ment for new homes in ASHRAE Standard 62.2 (ASHRAE2004), a Used filter with an ozone removal efficiency of 10%,and indoor temperature of 23.9C (75F) and an RH of 50%.Within the Rural and Urban distributions, the commercial basecase consisted of a 10% OA/90% RA flow cycl
48、e, a MERV 6filter, a Used filter with an ozone removal efficiency of 41%,and indoor temperature of 23.9C (75F) and an RH of 50%.Each combination of the parameters was modeled. The resi-dential model had 324 unique scenarios (162 each of Rural andUrban) and the commercial model had 486 unique scenari
49、os(243 each of Rural and Urban). RESULTS AND DISCUSSIONUsing Equations (1) through (4), the resulting SOAconcentrations, CSOA, varied over an order of magnitude,depending on the HVAC input parameters. SOA yield andother size-resolved parameters are lognormally distributed, sothe median was used as a descriptive statistic. For the residen-tial model, the median CSOAover all 162 Rural scenarios was17.4 g/m3(1.09 10-9lb/ft3), with a range of 2.4
