-
re
cDonDavid T. Allen a
aCenter for Energy and Environmental Resources, The UbCollege of
Engineering, Center for Environmental Reseac ENVIRON International
Corporation, 773 San Marin Dr
t Reyes
d widecitly me and iscomme
Received 26 November 2011Predictions of ozone formation, due to
oxidation of alkenes in presence of NOx, generated by the
CarbonBond 2005 (CB05) and CB05-TU (CB05 with an updated toluene
scheme) condensed chemical mecha-
3through complex chemical reactions. Therefore, a reliable
chemicalmechanism is necessary to describe these complex
processesrelevant to O3 formation. Carbon Bond (CB; Whitten et al.,
1980,
U.S. for various air quality applications to model gas-phase
chem-istry related to O3 formation. All chemical mechanisms used in
airquality models are simplied and condensed to some
degree,especially those for use in 3-dimensional atmospheric
modelsneeded to represent the atmospheric reactions of various
volatileorganic compounds (VOCs) and nitrogen oxides (NOx)
underambient conditions (Dodge, 2000). For example, in CB05
(Yarwoodet al., 2005) and SAPRC-07 (Carter, 2010a), a few model
species(e.g., OLE and IOLE in CB05; OLE1 and OLE2 in SAPRC-07) are
used
* Corresponding author. Present address: College of Engineering,
Center forEnvironmental Research and Technology, University of
California, Riverside, CA92521, USA. Tel.: 1 951 781 5708; fax: 1
951 781 5790.
Contents lists available at
Atmospheric E
journal homepage: www.else
Atmospheric Environment 59 (2012) 141e150E-mail address:
[email protected] (G. Heo).AlkeneAlkene
chemistryOzoneChemical mechanismCB05
experiments and box modeling for four cases (1 for propene and 3
for isobutene) showed that theperformance of CB05 and CB05-TU in
simulating ozone formation from propene and isobutene can
beimproved by modeling propene and isobutene using the new
condensed reactions for propene andisobutene developed in this
work. The results of this study indicate that the capability of
condensedchemical mechanisms in simulating ozone formation can be
improved by (1) examining the relativeimportance of VOCs based on
their emissions and reactivity, (2) separately representing
relativelyimportant VOCs in the mechanism, (3) modeling less
important compounds using reactions of lumpedmodel species shared
by multiple compounds, and (4) evaluating mechanisms with
experimental datasuch as environmental chamber data.
2012 Elsevier Ltd. All rights reserved.
1. Introduction
Ozone (O ) is not directly emitted but formed in the
atmosphere
2010; Gery et al., 1989; Yarwood et al., 2005) chemical
mechanisms,along with Statewide Air Pollution Research Center
(SAPRC; Carter,1990, 2000, 2010a,b,c) chemical mechanisms, have
been used in theReceived in revised form21 May 2012Accepted 25 May
2012
Keywords:
nisms were tested by simulating 138 environmental chamber
experiments carried out in 7 differentenvironmental chambers and by
running box modeling with four cases. CB05 and CB05-TU
reasonablysimulated ozone formation from propene under typical
urban conditions and for cases with moderatelyelevated propene
concentrations. Chamber simulations of 47 propene e NOx and 5
isobutene e NOxd Smog Reyes, PO Box 518, 112 Mesa Road, Poin
h i g h l i g h t s
< The CB05 chemical mechanism is use< New reactions are
developed to expli
-
to represent various alkenes (Table 1), and, instead of
detailedreactions of real compounds, highly condensed reactions of
thesemodel species are used in air quality models. In this way, a
limitednumber of reactions represent the atmospheric reactions of
manycompounds (such as alkenes), reducing computational burdens
ofair quality modeling (e.g., computational time) and
allowingavailable resources to be used for improving other
components ofthe air quality modeling (e.g., representing diffusion
and transportin a better way).
Alkenes are a major contributor to O3 formation in many areasdue
to their relatively high reactivity (Calvert et al., 2000) and
largeemissions from various sources (Simon et al., 2010). For
example,propene and isobutene react with the hydroxyl radical (OH)
fasterthan most alkanes and aromatics commonly observed in
urbanareas (Atkinson and Arey, 2003; Calvert et al., 2000). In
southeastTexas, 7 alkenes (ethene, propene, 1,3-butadiene,
1-butene, iso-butene, trans-2-butene, cis-2-butene) are classied as
HighlyReactive Volatile Organic Compounds (HRVOCs), and their
emis-sions from point sources, including event emissions from
petro-chemical facilities (Murphy and Allen, 2005), are regulated
by TexasAdministrative Code, Title 30, Part 1, Chapter 115 (Texas
Commis-
G. Heo et al. / Atmospheric Envir142sion on Environmental
Quality (TCEQ), 2011). Since the rateparameters, products and their
yields in the CB05 and SAPRC-07condensed mechanisms are based on
assumptions about thetypical urban air composition of alkenes
(Seila et al., 1989), differ-ences between the assumed average
atmospheric compositionused during mechanism development and the
real-world atmo-spheric composition affect the performance of the
mechanisms. Forexample, in Houston, Texas, propene was found to
frequentlyexceed its typical concentrations observed in most urban
areas(Jobson et al., 2004; Gilman et al., 2009). As a result, in
Houston, theatmospheric composition of alkenes is often markedly
differentfrom the typical composition in most urban areas in the
U.S.(e.g., see Table 3 of Heo et al. (2010)). Therefore, using the
limitednumber of reactions based on the typical composition of
alkenescan lead to inaccurate predictions of air pollutants in
areas such asHouston. Heo et al. (2010) showed that using the
current reactionsfor OLE1 (a model species for most terminal
alkenes) in the xed-parameter version of SAPRC-07 (Carter, 2010a,c)
leads to under-predictions of O3 concentrations under conditions
where thehydrocarbon reactivity is dominated by propene. Using
environ-mental chamber simulations and box modeling, Heo et al.
(2010)
Table 1Examples of alkene speciation for CB05. Speciation for
SAPRC-07 is also shown forcomparison.
Compound CB05b SAPRC-07c
Ethene (CH2]CH2)a ETHa Ethenea
Propene (CH3CH]CH2) OLE PAR OLE11-Butene (CH3CH2CH]CH2) OLE 2
PAR OLE11-Pentene (CH3CH2CH2CH]CH2) OLE 3 PAR OLE13-Methyl-1-butene
(CH3CH(CH3)CH]CH2) OLE 3 PAR OLE11,3-Butadiene (CH2]CHCH]CH2) 2 OLE
OLE22-Butene (CH3CH]CHCH3) IOLE OLE22-Pentene (CH3CH2CH]CHCH3) IOLE
PAR OLE2Isobutene ((CH3)2C]CH2) FORM 3 PAR OLE22-Methyl-2-butene
(CH3CH]C(CH3)2) ALD2 3 PAR OLE22-Methyl-2-pentene
(CH3CH2CH]C(CH3)2) ALDX 4 PAR OLE2a ETHENE in SAPRC-07 and ETH in
CB05 are a model species only for ethene.b In CB05, OLE and IOLE
represent terminal C]C bonds and C-4 structures
(CeCCeC) having an internal C]C bond, respectively. PAR
represents parafniccarbons. FORM is formaldehyde (HCHO), ALD2 is
acetaldehyde (CH3CHO), and ALDXrepresents propanal (CH3CH2CHO) and
higher aldehydes. For details, see Yarwoodet al. (2005).
c In SAPRC-07, OLE1 represents alkenes with kOH < 7.0 104
ppm1 min1, and
OLE2 representsmore reactive alkenes except isoprene and
terpenes. For details, seeCarter (2010a).showed that separately
modeling reactions of individual alkenes(especially propene, for
southeast Texas) in the SAPRC condensedmechanism has the potential
to lead to more accurate simulationsof O3 formation in Houston.
Since similar results may be expectedfor CB05, this study assesses
the capability of CB05 and CB05-TU(CB05 with an updated toluene
scheme; Whitten et al., 2010) insimulating O3 formation from
oxidation of propene against envi-ronmental chamber data, and
presents newly developedcondensed reactions that can be used to
separately model propenein CB05 and CB05-TU. Note that CB05-TU is
also tested in this studybecause CB05-TU showed better performance
in simulating ozoneformation from aromatics than CB05 against
toluene e NOxchamber experiments (Whitten et al., 2010).
Isobutene ((CH3)2C]CH2) is often the most abundant
branchedterminal alkene found in the atmosphere of Houston (Jobson
et al.,2004; Gilman et al., 2009) and is one of the 7 HRVOCs
regulated byTCEQ (TCEQ, 2011). How isobutene is mapped intomodel
species inCB05 can also inuence the accuracy of ozone predictions
by CB05(Table 1). For a description of speciation of alkenes in
CB05, refer tothe Supplementary Material. The current modeling
approach forisobutene in CB05, modeling isobutenes reactions with
reactions ofFORM and PAR (Table 1; Yarwood et al., 2005), could
lead to inac-curate model predictions. Therefore, isobutene was
selected fromthese terminal alkenes having a branch at their C]C
bond, andcondensed reactions to separately model isobutene were
devel-oped and evaluated against environmental chamber data to
betterrepresent O3 formation from isobutene.
In this study, four tasks were performed. First, the capability
ofCB05 in modeling O3 formation from oxidation of propene
wasevaluated against 47 propeneeNOx chamber experiments. Second,new
condensed reactions for propene and isobutene were devel-oped for
use in CB05 and CB05-TU, and these new reactions weretested against
47 propene e NOx experiments and 5 isobutenee NOx experiments.
Third, CB05-TU with the new reactions forpropene and isobutene were
tested against 86 surrogate VOCmixture e NOx chamber experiments
where at least 8 differentVOCs including propene and toluene were
injected to examine theoverall impact of the updated alkene
chemistry (this study) andtoluene chemistry (Whitten et al., 2010)
on the mechanismperformance in modeling O3 formation. Fourth, box
modeling wascarried out using four cases to examine the impact on
simulated O3of using the newly developed reactions for propene and
isobuteneunder more realistic conditions compared to chamber
conditions.
2. Data and methods
2.1. Chemical mechanisms used
Model simulations were conducted using four versions of theCB05
chemical mechanism, two that were developed previously(CB05 and
CB05-TU) and two that were developed in this work(CB05-OLE and
CB05-TU-OLE). CB05 is the standard CB05 mecha-nism as documented by
Yarwood et al. (2005). CB05-TU is a modi-ed version of CB05
developed by Whitten et al. (2010) with anupdated representation of
the reactions of toluene and otheraromatics. CB05-TU is the same as
CB05 for alkenes and other non-aromatic species, but was found to
show improved performance insimulating O3 formation against chamber
experiments wheretoluenewas injected (Whitten et al., 2010).
CB05-OLE and CB05-TU-OLE were developed in this work to assess the
effects of repre-senting the reactions of propene and isobutene
separately, ratherthan using lumped model species as shown in Table
1.
Reactions for propene and isobutene for use with CB05 andCB05-TU
were developed using only two new model species, PRPE
onment 59 (2012) 141e150for propene and IBTE for isobutene, and
are given in Table 2.
-
Table 2Reactions for propene (PRPE) and isobutene (IBTE) for use
with CB05 and CB05-TU. Note: Reactions of OLE (for terminal C]C
bonds) in CB05 are also shown for comparison.
Reactants Productsa Rate constantb Note
1 PRPE OH 0.985 (ALD2 FORM XO2 HO2) 0.015 XO2N 0.045 PAR
FALLOFF: F 0.50, N 1.13 cko: 8.00E-27$(T/300)3.50
kinf: 3.00E-11$(T/300)1.00
2 PRPE O3 0.33 OH 0.38 HO2 0.54 ALD2 0.73 FORM 0.528 CO 0.0425
MEOH 0.075 CH4 0.15 XO2 0.04 (H2O2 AACD) 0.185 FACD 0.28 PAR
5.50E-15$Exp(1888/T) d
3 PRPE NO3 0.5 (NO2 NTR) 0.96 XO2 0.48 HO2 0.04 XO2N 0.5 FORM
0.625 ALD2 4.60E-13$Exp(1155/T) e4 PRPE O 0.2 ALD2 0.3 ALDX 0.3 HO2
0.1 OH 0.2 (XO2 CO FORM)
0.01 XO2N 1.2 PAR1.02E-11$Exp(280/T) f
5 IBTE OH 0.90 (FORM HO2 XO2) 0.10 XO2N 2.70 PAR
9.47E-12$Exp(504/T) g6 IBTE O3 0.594 OH 0.048 HO2 0.546 (XO2 C2O3)
0.238 MEO2 1.246 FORM
1.005 PAR 0.153 CO 0.111 FACD 0.035 H2O22.70E-15$Exp(1632/T)
h
PAR
00.3
LD2(XO
G. Heo et al. / Atmospheric Environment 59 (2012) 141e150 1437
IBTE NO3 0.96 (FORM XO2 NO2) 0.04 XO2N 2.888 IBTE O 0.5 ALD2 0.5
ALDX 2.0 PARe OLE OH 0.8 FORM 0.33 ALD2 0.62 ALDX 0.80 XO2e OLE O3
0.1 OH 0.44 HO2 0.74 FORM 0.18 ALD2
0.22 XO2 PARe OLE NO3 NO2 0.91 XO2 0.09 XO2N FORM 0.35 Ae OLE O
0.2 ALD2 0.3 ALDX 0.3 HO2 0.1 OH 0.2
0.01 XO2N 0.2 PARFootnotes to this table provide detailed
information on how thesecondensed reactions were derived. For more
details of CB05 modelspecies used in Table 2, refer to Yarwood et
al. (2005). Most rateparameters for PRPE in Table 2 are from IUPAC
(2010). The reactionsof IBTEwith OH, O3 and NO3 in Table 2 are
based on rate parametersfrom Atkinson and Arey (2003).
a FORM, ALD2 and ALDX are HCHO, CH3CHO and higher aldehydes; PAR
and NTR repreand acetic and higher carboxylic acids, respectively.
The general framework used to approXO2 and XO2N are an operator for
representing NO to NO2 conversion from alkylperoxy (RRO2,
respectively.
b Rate constants are in units of molecule$cm3$second1.c For
reaction PRPE OH, rate parameters are from IUPAC (2010) and organic
nitrate y
are major products (Calvert et al., 2000), for which a yield of
0.985 (1.0e0.015) was gid Rate parameters are from IUPAC (2010).
The product side was constructed as follows:
CH2OO* and CH3CHOO* were represented as 0.37 HCOOH 0.16 (OH HO2)
0.12 H(CH3OH CO) 0.15 CH4, respectively, based on IUPAC (2010).
CH3CHOO, a stabilized Crits dominant reaction with H2O (Sauer et
al., 1999; IUPAC, 2010). CH2CHOwas represente(2005) and Fig. S1 of
Heo et al. (2010), and CHOeCHO (glyoxal) was represented as
FORMCH3CHO, CH3OH, HCOOH and CH3OOH were renamed as FORM, ALD2,
MEOH, FACD and
e Rate parameters are from IUPAC (2010), and product yields were
derived asHCHO NO2 XO2)} 0.04$XO2N. There are various uncertainties
in product distributirepresented as 0.5 (NTRNO2) 0.48 (HO2 2 XO2)
0.5 FORM 0.625 ALD2 0.04XO2
f From OLE O reaction of Yarwood et al. (2005); rate constant
was slightly modiedg Rate parameters from Atkinson and Arey (2003);
organic nitrate yield (10%) from Tua
(Calvert et al., 2000), for which a yield of 0.90 (1.0e0.10) is
given. Acetone is relative uh Rate parameters from Atkinson and
Arey (2003); product yields from Neeb and M
IBTE O3 0.7 {(CH3)2COO* HCHO} 0.3 {CH2OO* ACET}. (CH3)2COO* was
represenCH3 CO2} 0.05 (ACET H2O2) 0.78 {XO2 FORM C2O3 OH} 0.34
MEO2. (dominant products after reaction with H2O (Sauer et al.,
1999; IUPAC, 2010), and A(OH HO2) 0.12 H2 0.51 CO based on IUPAC
(2010).
i Rate parameters are from Atkinson and Arey (2003), and
products were formulated asbased on Berndt and Bge (1995).Acetone
was represented as 3 PAR.
j Rate parameters are from Calvert et al. (2000). Due to lack of
experimental data, thebalancing carbon.
k Directly from Yarwood et al. (2005).
Table 3An overview of indoor environmental chambers at UCR and
TVA used for mechanism ev
Chamber Chamber ID Reactor type Reactor
Evacuable Chamber at UCR EC Single w5.8Indoor Teon Chamber at
UCR ITC Single w6.4Ernies Teon Chamber at UCR ETC Single
w3.0Dividable Teon Chamber at UCR DTC Dual w5.0 (XCE-CERT Teon
Chamber at UCR CTC (11e82a) Single w5.0CE-CERT Teon Chamber at UCR
(rebuilt) CTC (>82a) Dual w2.5 (XUCR EPA chamber EPA Dual w90
(XTVA indoor chamber TVA Single w28
References: Dodge (2000), Carter (2000, 2010a), Carter et al.
(2005).a Run number of the chamber experiment.3.44E-13 i
1.14E-11$Exp(130/T) j
.95 HO2 0.70 PAR 3.20E-11 k2 ALDX 0.33 CO 6.50E-15$Exp(1900/T)
k
0.56 ALDX PAR 7.00E-13$Exp(2160/T) k2 CO FORM)
1.00E-11$Exp(280/T) k2.2. Environmental chamber data
A database of chamber experiments continually expanded
andevaluated for w35 years by William Carter at the University
ofCalifornia at Riverside (UCR) was used for the chamber
simulationsin this study (Carter, 2000, 2010a). Most UCR chamber
data (w2000
sent parafnic carbons and organic nitrates; MEOH, FACD, AACD are
CH3OH, HCOOHximate the product side is the same as in Gery et al.
(1989) and Yarwood et al. (2005).O2) radicals, and an operator for
representing NO to organic nitrate conversion from
ield (1.5%) is from OBrien et al. (1998). Under high NOx
conditions, ALD2 and FORMven. 0.45 PAR was added for balancing
carbon (i.e., 3 carbons on each side).PRPE O3 0.5 (HCHO CH3CHOO*)
0.5 (CH3CHO CH2OO*). Criegee biradicals2 0.51 CO and 0.16 CH3CHOO
0.4 (OH CH2CHO) 0.06 CH2]C]O 0.085iegee biradical, was represented
as 0.5 ALD2 0.5 CH3COOH 0.5 H2O2 by assumingd as 0.25 (OH FORM CO)
0.75 {XO2 HO2 CHOeCHO} based on Kuwata et al.COHO2 (Whitten et al.,
2010); CH2]C]Owas represented as FORM CO. HCHO,AACD, respectively,
and 0.28 PAR was added for balancing carbon.follows: PRPE NO3
(1e0.04)$0.5 {(NTR XO2 HO2) (CH3CHO ons for reactions of alkenes
with NO3 (Calvert et al., 2000), so the product side wasN to be
consistentwith the updated OLENO3 reaction for CB6 (Yarwood et al.,
2010).based on Calvert et al. (2000); 1 PAR was added for balancing
carbon.zon et al. (1998). Under high-NOx conditions, acetone and
FORM are major productsnreactive based on its k(OH) (IUPAC, 2010)
and was represented as 3 PAR.oortgat (1999) and IUPAC (2010). The
product side was constructed as follows:ted as follows: (CH3)2COO*
0.05 (CH3)2COO 0.78 {CH3C(O)CH2 OH} 0.17 {2CH3)2COO, a stabilized
Criegee biradical, was represented as acetone and H2O2, itsCET was
represented as 3 PAR. CH2OO* was condensed as 0.37 HCOOH 0.16
follows: IBTE NO3 0.04$XO2N (1 0.04)$(CH3C(O)CH3 HCHO NO2
XO2)
product side was tentatively given based on Carter (2010a). 2.0
PAR was added for
aluation (Heo, 2009).
volume (m3) Light source Relative humidity Operation period
Xenon arc w50% 1975e84Blacklight w50% 1982e86Blacklight dry
(
-
against (1) 47 propene e NOx experiments to examine
theircapability in simulating O3 formation from propene, (2) 5
isobutenee NOx experiments to examine their capability in
simulating O3formation from isobutene, and (3) 86 surrogate
VOCmixtureeNOxexperiments wheremajor components of urban
atmospheres otherthan propene were also injected.
Running chamber simulations with a chemical mechanismrequires
wall mechanisms that characterize chamber-dependenteffects such as
chamber-dependent radical sources and NOx off-gasing from the
chamber walls (Jeffries et al., 1992; Carter et al.,2005). For
chamber simulations to evaluate the CB05 variantsused in this
study, wall mechanisms used for evaluating CB05 byYarwood et al.
(2005) and CB05-TU by Whitten et al. (2010) wereused. Chemical
systems such as CO e NOx lack chemical processesgenerating new
radicals (e.g., hydroperoxy radical (HO2) formationfrom the
photolysis of formaldehyde) and are sensitive to chamber-dependent
radical sources. However, chemical systems (such aspropene e NOx,
isobutene e NOx, and surrogate VOC mixturee NOx) that have a strong
internal radical source are relativelyinsensitive to
chamber-dependent radical sources.
2.4. Comparing mechanisms using box modeling
nvironment 59 (2012) 141e150experiments, not including very
recent experiments) and chamberdata of the Tennessee Valley
Authority (TVA,w60 experiments) areavailable in this chamber
database that is publicly available
athttp://www.cert.ucr.edu/wcarter/SAPRC/SAPRCles.htm
(Carter,2010a). The chamber data used in this study were extracted
fromthis large UCR database (April 23, 2010 version), and
includechamber experimental data from 138 experiments carried out
in 6different chambers at UCR and in one chamber at TVA. Table
3provides an overview of these 7 environmental chambers.
Experiments of single test compound e NOx are useful intesting
each component of a chemical mechanism (e.g., propenechemistry), at
least for reactive compounds such as propene andisobutene. On the
other hand, for testing a chemical mechanism asawhole, surrogate
VOCmixture e NOx experiments are useful. Ina surrogate mixture
experiment, a mixture simulating a targetatmospheric composition
(e.g., an urbanmixture) is injected. In thisstudy, two types of
chamber experiments were used: single testVOC e NOx and surrogate
VOC mixture e NOx experiments.
Chamber effects (e.g., wall reactions such as NOx
offgasing,which are difcult to accurately describe (Jeffries et
al., 1992;Dodge, 2000; Carter et al., 2005; Heo, 2009), must be
considered inselecting chamber experiments to be used in evaluating
mecha-nisms. In order to minimize the impact of chamber effects
onmechanism evaluation, experiments that are expected to have
beensignicantly inuenced by chamber effects in general should not
beused. In selecting UCR and TVA experiments, criteria for the
initialNOx level ([NOx]0), the ratio of O3 formed toNO oxidized
(Max(O3)/[NO]0), and the chamber light source were used to makea
compromise between increasing the total number of
selectedexperiments and minimizing uncertainty due to chamber
effectsand other artifacts (e.g., high oxygen atom concentration
due topresence of high NOx; Paulson et al., 1992) (Whitten et al.,
2010).Criteria applied for selecting single test VOCeNOx
experiments forpropene and isobutene are (1) 0.010 [NOx]0 0.500
parts permillion (ppm), and (2) Max(O3)/[NO]0 1.0. Criteria applied
forselecting surrogate VOC mixture e NOx experiments are (1)0.010
[NOx]o 0.200 ppm, and (2) Max(O3)/[NO]0 1.0. Black-light
experiments were not used when sufcient data fromcomparable
experiments with arc lights are available because thearc light
source is considered to be the better representation ofnatural
sunlight (Carter et al., 2005). However, we are not aware ofstrong
direct evidence showing that there are systematic differ-ences in
mechanism performance between evaluation usingblacklight
experiments and evaluation using experiments with anargon arc light
or natural sunlight. After applying these criteria, 47experiments
of propene e NOx and 86 experiments of surrogateVOC mixture e NOx
were selected. For the surrogate VOC mixturee NOx experiments, in
most cases, 8 different VOCs (n-butane,n-octane, ethene, propene,
trans-2-butene, toluene, m-xylene, andformaldehyde) were injected.
For the ve TVA experiments, 54different VOCs simulating ambient
VOCmixtures were injected. Fortesting isobutene chemistry, chamber
data for only 2 isobuteneeNOx blacklight experiments and 3
isobutenee other VOCseNOxblacklight experiments were available.
Thus, all 5 experimentswere used. Tables S1, S2 and S3 in the
Supplementary Materialprovide additional information on the
experiments for propene,isobutene and surrogate mixtures,
respectively.
2.3. Mechanism evaluation using environmental
chambersimulations
Chamber simulations were performed using the SAPRC softwarethat
has been used for evaluating various versions of the SAPRC andCB
mechanisms (Carter, 2000 and 2010a; Heo et al., 2010; Yarwood
G. Heo et al. / Atmospheric E144et al., 2005; Whitten et al.,
2010). Each mechanism was evaluatedBoxmodeling was also carried out
to characterize the impact onmodeled ozone of using the new
reactions for propene and iso-butene (Table 2) under conditions
based on atmosphericmeasurements. Four cases were selected (Table
4). LP1.Base isa modeling case for the morning of August 25, 2000
at the La Portemunicipal airport, 30 km southeast of downtown
Houston, Texas.On this date,103 parts per billion (ppb) ethene and
90 ppb propenewere measured at around 7:30 CST and winds were
relativelystagnant during the morning hours (Heo et al., 2010; Heo,
2009).LP1.Base was used to examine the impact on simulated ozone
ofusing the new PRPE reactions (Table 2) while LP1.IBTE,
LP1.IB-TEPRPE and LP2.IBTE were used to examine the impact of
usingthe new IBTE reactions (Table 2) under various conditions.
LP1.IBTEand LP1.IBTEPRPE are variants of LP1.Base where the
initialethene and propene concentrations ([ethene]0 and
[propene]0)were reduced and the initial isobutene concentration
Table 4Four box modeling cases used to examine the impact
onmodeled ozone of using thenew propene and isobutene
reactions.
Case ID Description
LP1.Base The La Porte August 25, 2000 case used byHeo et al.
(2010). Ethene (103 ppb) and propene(90 ppb) dominated the
reactivity. Initial VOCconcentrations were set based on
measurementsat the La Porte airport (Jobson et al., 2004) and
noadditional emissions were added. For details,see Heo et al.
(2010).
LP1.IBTE LP1.Base but with adjusted initial concentrationsfor
ethene, propene and isobutene: ethene(decreased from 103 ppb to 5
ppb), propene(decreased from 90 ppb to 5 ppb), and
isobutene(increased from 0.7 ppb to 100 ppb).
LP1.IBTEPRPE LP1.Base but with adjusted initial
concentrationsfor ethene, propene and isobutene: ethene(decreased
from 103 ppb to 70 ppb), propene(decreased from 90 ppb to 70 ppb),
and isobutene(increased from 0.7 ppb to 50 ppb).
LP2.IBTE A modeling case based on Faraji et al. (2008).
Initialconcentrations and composition of VOC emissionswere based on
aircraft measurements, and additionalVOC and NOx emissions were
added over themodeling period (Faraji et al., 2008). However,
isobutene was increased from 0.5 ppb to 22.5 ppb.
-
Fig. 1. Comparison of mechanism performance between CB05 and
CB05-OLE against47 propene e NOx experiments: (a) Max(O3), (b)
Max(D(O3eNO)), and (c) NOx
nvironment 59 (2012) 141e150 145([isobutene]0) was increased
(Table 4). LP2.IBTE is a variant ofthe box modeling case used by
Faraji et al. (2008) (Table 4). TheVOC composition used for
LP2.IBTE is based on aircraft datameasured on multiple early
mornings in August and September of2000 near the La Porte airport
(Faraji et al., 2008). For this study,the fraction of VOC for
isobutene was increased from0.0011 ppm ppmC1 VOC to 0.050 ppm ppmC1
VOC; as a result,the initial isobutene concentration was increased
from 0.5 ppb to22. 5 ppb. However, the initial VOC/NOx ratio was
kept at 10 (450ppmC VOC/45 ppm NOx). Box model simulations were
performedusing the same simulation software package as used by Heo
et al.(2010) (Carter, 2000 and 2010a). Additional information
regardingthe box model simulations is available in the
SupplementaryMaterial.
3. Results and discussion
This section presents (1) chamber simulation results for
testingthe four versions of CB05 against measured data of the
138chamber experiments listed in Tables S1 to S3 in the
Supple-mentary Material and (2) box modeling results for the four
cases(Table 4). The three performance metrics that were used
forquantifying the performance of a chemical mechanism in
thechamber simulations are the maximum ozone
concentration(Max(O3)), maximum D(O3eNO) (Max(D(O3eNO)), and
NOxcrossover time. In this work, Max(O3) is dened as the highest
O3concentration by the end of the experiment but no later than 6
hsince the start of chamber irradiation (i.e., t 0) because inmost
cases chamber data after hour of 6 were not gatheredand are not
quality-assured. The metric D(O3eNO), dened as([O3] [NO])tt ([O3]
[NO])t0, quanties the amount of O3formed and NO oxidized since t 0
and is useful even when thereis no signicant O3 production (Carter
and Atkinson, 1987).Max(O3) and Max(D(O3eNO)) are useful because a
primary goal ofcondensed chemical mechanisms for urban/regional
photo-chemical models is accurate prediction of maximum O3
concen-trations; however, these metrics do not provide information
onthe rate of O3 formation. The NOx crossover time, dened as
thetime elapsed since t 0 before the NO2 concentration becomesequal
to the NO concentration, contains information on the rate ofNO
oxidation into NO2, which accompanies O3 formation. There-fore, the
NOx crossover time is a useful performance metricand was also used
in this work. Model errors of Max(O3) andMax(D(O3eNO)) were
calculated as {(model e experimental)/experimental} in units of
percentages (%); model errors of NOxcrossover times were calculated
as (model e experimental) inunits of minutes (min). Means and
standard deviations of thesemetrics were used to summarize
performance over multipleexperiments.
3.1. Evaluating CB05 and CB05-OLE against propene
chamberexperiments
CB05 reasonably simulated O3 formation from oxidation ofpropene
for experiments with relatively low and moderately highpropene
concentrations (EPA and TVA experiments listed inTable S1) as shown
in Fig. 1. Note that results for CB05-TU andCB05-TU-OLE are not
shown in Figs. 1 and 2 because their repre-sentation of alkene
chemistry is the same as in CB05 and CB05-OLE, respectively. CB05
simulated Max(O3) with model errorswithin 25% for most of the 47
experiments while simulating betterfor experiments with their
measured Max(O3) lower than0.250 ppm than for experiments with
their measured Max(O3)larger than 0.250 ppm (Fig. 1a). The average
model errors
G. Heo et al. / Atmospheric Ecommitted by CB05 for the 47
experiments were 13% (16%) forcrossover time.
-
ne b7, et, ethCB05refe
G. Heo et al. / Atmospheric Envir146Max(O3), 10% (10%) for
Max(D(O3eNO)), and 18 min (17 min)for the NOx crossover time (Table
5). Assumptions used in devel-oping the OLE reactions of CB05
(Table 2) can explain these resultsdemonstrating that ozone
formation from propene is reasonably
Fig. 2. Comparison of performance in simulating O3 formation
from oxidation of isobute3 experiments of isobutene e other
VOCseNOx. Note: For ETC253, ETC255 and ETC25ethene (0.678),
m-xylene (0.096), n-hexane (0.407); for ETC255, isobutene (0.195
ppm)ethene (0.745), m-xylene (0.092), and n-hexane (0.389).
CB05(s1:OLE 2 PAR) isCB05(s2:IOLE) is CB05 but with speciation of
isobutene into IOLE. For more detail,well simulated by the OLE
reactions (which are reactions not justfor propene but also for
many other terminal alkenes (moreexactly, terminal C]C bonds)). The
reaction parameters such asrate constants for the OLE reactions in
CB05 are heavily based onthe oxidation mechanisms of propene
although two model speciesfor aldehydes (ALD2 and ALDX) are used as
products in the OLEreactions (Table 2, Yarwood et al., 2005; Gery
et al., 1989). Note thatALD2 (CH3CHO) is produced from the reaction
of propene and OHbut ALDX (model species in CB05 for higher
aldehydes such aspropanal (CH3CH2CHO)) is not and that propene is
assumed toconstitutew35% (0.33/(0.33 0.62)) of OLE in CB05 as
implied bythe ALD2 and ALDX yields in reaction OLE OH in Table 2.
Previousstudies on evaluating the OLE chemistry of CB05 with
chamberdata also heavily relied on propene e NOx experiments,
andevaluation of the OLE chemistry against chamber experiments
forterminal alkenes other than propene was limited (Yarwood et
al.,2005; Gery et al., 1989), in part due to lack of such chamber
dataavailable for mechanism evaluation. In principle, the reactions
ofOLE could be re-derived to exclude propene. In practice,
thechemistry of terminal alkenes is dominated by the double
bond
Table 5Comparison of model errors between CB05 and CB05-OLE for
47 propene e NOxexperiments. CB05TU is the same as CB05 for
propene.
Max(O3) [%] Max(D(O3eNO) [%] NOx crossovertime [min]
CB05 &CB05TU
CB05-OLE CB05 & CB05TU CB05-OLE CB05 &CB05TU
CB05-OLE
Average 13 2 10 2 18 5Std. dev. 16 13 10 9 17 13allowing one
model species (OLE) to represent the family ofcompounds. Therefore,
it is both reasonable and practical to leavethe OLE reactions
unchanged regardless of whether propene istreated using the
explicit species PRPE.
etween CB05 and its variants: (top) for 2 experiments of
isobutene e NOx; (bottom) forhene, m-xylene and n-hexane were also
injected. For ETC253, isobutene (0.211 ppm),ene (0.737), m-xylene
(0.0943), n-hexane (0.388); for ETC257, isobutene (0.108 ppm),but
with speciation of isobutene into OLE 2 PAR instead of FORM 3
PAR.
r to section 3.2.
onment 59 (2012) 141e150In comparison, overall, CB05-OLE
performed better than CB05(Fig. 1, Table 5). For example, the
average model error of Max(O3)was reduced from 13% (16%) to 2%
(14%), which resultedfrom the reduced model errors for experiments
with theirmeasured Max(O3) larger than 0.250 ppm (Fig. 1a).
CB05-OLEshowed performance comparable to SAPRC-07 (adjustable)
insimulating O3 formation from propene for these 47
chamberexperiments (Heo et al., 2010). For example, the average
modelerrors in simulating Max(O3) were 2% (14%) for CB05-OLE and4%
(16%) for SAPRC-07 (adjustable), which is the adjustable-parameter
version of SAPRC-07 described in Carter (2010a,b).The results conrm
that CB05-OLE has potential to better simulateO3 formation from
propene than CB05 under conditions that theemissions of terminal
alkenes dominate the hydrocarbon reac-tivity and the relative
contribution of propene to those terminalalkenes is more than
assumed during development of a simpliedand lumped chemical
mechanism (e.g., 80% (Table 3 of Heo et al.(2010)) instead of 35%
(Yarwood et al., 2005)). Similar resultswere seen in chamber
simulations and box modeling usingdifferent representations for
propene in SAPRC-07 (Heo et al.,2010).
3.2. Evaluating CB05 and CB05-OLE against isobutene
chamberexperiments
Fig. 2 shows the time proles of O3 concentrations simulated
byCB05 and CB05-OLE against measurements of 5 chamber experi-ments
for isobutene: 2 experiments of isobutene e NOx and 3experiments of
isobutene e other VOCs e NOx. Although onlyblacklight experiments
were used, the results show that usinga separate model species for
isobutene (IBTE) and using the IBTE
-
Fig. 3. Comparison of mechanism performance between CB05,
CB05-TU andCB05-TU-OLE against 86 surrogate VOC mixtureeNOx
experiments: (a) Max(O3),
nvironment 59 (2012) 141e150 147reactions listed in Table 2
could improve the performance of CB05and CB05-TU in simulating O3
formation from oxidation of iso-butene. Based on chamber
simulations of the two isobutene e NOxexperiments (DTC052B and
ITC694), modeling isobutene oxidationwith FORM 3 PAR in CB05 leads
to prediction of faster O3formation in the early stage but slower
O3 formation later,compared to observations (Fig. 2). This early O3
over-prediction ismore apparent in the O3 time series for the three
VOC mixturee NOx experiments (ETC253, ETC255 and ETC257) for which
VOCsother than isobutene were also initially injected (Fig. 2). The
over-predicted O3 formation in the early stage can be explained
byarticial radical production from FORM (formaldehyde) due tousing
FORM 3 PAR to represent isobutene in CB05. When thefour reactions
listed in Table 2 for IBTE were used to model theoxidation of
isobutene, this problem of early O3 over-prediction andmid- and
late-stage O3 under-prediction for DTC052B and ITC694was mitigated
as shown by time proles of O3 for CB05-OLE inFig. 2.
In comparison, representing isobutene by either OLE 2 PARor IOLE
instead of FORM 3 PAR (default speciation in CB05)did not result in
improved model performance comparable tothat shown by CB05-OLE in
simulating O3 for these isobuteneexperiments (Fig. 2). In terms of
the carbon bond concept (Whittenet al., 1980), the branched C]C
bond of isobutene is not wellmodeled by either OLE (for terminal
C]C bond without a branch)or IOLE (for internal C]C bond without a
branch). Representingisobutene with OLE 2 PAR was not comparable to
modelingwith CB05-OLE in simulating O3 but showed overall
somewhatbetter performance in simulating O3 formation from
isobutenethan that with FORM 3 PAR (CB05) and that with IOLE(Fig.
2). The new reactions for isobutene (Table 2) could allowcomparison
of measured andmodeled concentrations of isobuteneandmore accurate
estimation of the impact of isobutene emissionson O3 formation in
areas inuenced by industrial isobuteneemissions.
3.3. Evaluating CB05 variants against surrogate mixture
chamberexperiments
CB05 and CB05-TU were evaluated to test their capability
insimulating O3 formation from surrogate VOCs e NOx mixturesusing
data from 86 chamber experiments where at least 8 differentVOCs
including propene and toluene were injected. Becausearomatics were
injected for these experiments, CB05 and CB05-TUare expected to
give different results, unlike the simulations of thealkene
experiments discussed above. Then, CB05-TU-OLE wasevaluated using
those chamber data to examine the overall impacton model
performance in simulating O3 formation of the updatedalkene
chemistry described above combined with the updatedaromatic
chemistry in CB05-TU (Whitten et al., 2010). For the 86chamber
experiments (Table S3), isobutene was not injected exceptfor the 5
TVA chamber experiments where a small amount of iso-butene was
injected.
In general, CB05-TU performed better than CB05 in simulatingO3
formation for these experiments (Fig. 3, Table 6), which
isconsistent with the results previously reported by Whitten et
al.(2010) and attributable to the improved toluene chemistry.
Theaverage model errors for CB05-TU vs. CB05 are as follows:
23%(11%) vs.31% (15%) for Max(O3),20% (10%) vs.27% (13%)for
Max(D(O3eNO)), and 3 min (8 min) vs. 4 min (11 min) forthe NOx
crossover time. Despite the more explicit approach tomodeling
propene in CB05-TU-OLE, CB05-TU and CB05-TU-OLEshowed very similar
results (Fig. 3 and Table 6) because propenedid not dominate the
VOC reactivity in these experiments due to
G. Heo et al. / Atmospheric Epresence of other reactive
compounds injected. For the 86(b) Max(D(O3eNO)), and (c) NOx
crossover time.
-
experiments, propene was injected in a smaller amount on a
molarbasis than toluene, on average, by about 30% (10%), and
otherreactive VOCs such as ethene, trans-2-butene and m-xylene
werealso injected. However, based on the results in section 3.1,
underatmospheric conditions that are dominated by propene
emissions,for example, industrial propene emissions from
petrochemicalfacilities, the effect on modeled O3 of the updated
propene chem-istry could be larger, which is demonstrated by box
modelingresults in section 3.4.
Table 6Comparison of model errors between CB05, CB05-TU and
CB05-TU-OLE for 86 surrogate
Max(O3) [%] Max(D(O3eNO)
CB05 CB05-TU CB05-TU-OLE CB05 CB
Average 31 23 22 27 2Std. dev. 15 11 11 13 1
G. Heo et al. / Atmospheric Envir148Fig. 4. Comparison of
modeled O3, OH and HO2 between CB05-TU and CB05-TU-OLEfor a
modeling case for propene, La Porte 8/25/2000 Base case (LP1.Base):
(a) O3, and(b) OH and HO2.3.4. Comparing CB05-TU and CB05-TU-OLE
using box modelingcases
CB05-TU and CB05-TU-OLE were compared by carrying out
boxmodeling with the four cases based on ambient measurements(Table
4). Results for CB05 and CB05-OLE were not shown in thissection
because (1) differences in modeled ozone between the twomechanisms
were nearly the same as differences between CB05-TUand CB05-TU-OLE
and (2) CB05-TU showed better O3 performancein the chamber
simulations (see section 3.3).
Comparison of CB05-TU and CB05-TU-OLE for case LP1.Base
(forwhich [propene]0 is 90 ppb) showed that using new reactions
forpropene (PRPE) listed in Table 2 increased the modeled 1-h
O3concentrations up to 7.9 ppb and the 1-h Max(O3) by 4.6 ppb
(3.8%when calculated by ([O3]CB05-TU-OLE e
[O3]CB05-TU)/[O3]CB05-TU)(Fig. 4a). These results can be explained
by higher OH and HO2concentrations with CB05-TU-OLE than with
CB05-TU, on average,by about 10% over the period of 7:30 to 14:00
CST (Fig. 4b), and themagnitude of impact on modeled O3 of
separately modeling pro-pene with CB05-TU-OLE is w50% of that shown
for SAPRC-07 byHeo et al. (2010). Mao et al. (2010) reported decits
in modeled OHand HO2 sources relative to HO and HO2 sinks during
the Texas AirQuality Study 2000. Thus, increases in OH and HO2
shown by CB05-TU-OLE compared to CB05-TU are consistent with OH and
HO2measurements.
Simulations with 3 modeling cases for isobutene,
LP1.IBTE([propene]0 5 ppb, [isobutene]0 100 ppb),
LP1.IBTEPRPE([isobutene]0 100 ppb, [propene]0 70 ppb), and
LP2.IBTE([isobutene]0 22.5 ppb, [propene]0 1.7 ppb) (Table 4),
showedthat using the new reactions for isobutene (IBTE) listed in
Table 2increased the 1-h Max(O3) by 20.3 ppb (21.6%), 9.3 ppb
(7.7%) and3.4 ppb (3.1%), respectively (Fig. 5). The magnitude of
impact washighest for LP1.IBTE for which isobutene dominated the
reactivityand was lowest for LP2.IBTE for which [isobutene]0 was
relativelymoderate compared to 100 ppb for LP1.IBTE and 60 ppb
forLP1.IBTEPRPE. For case LP1.IBTE, representing IBTE as OLE 2
PARresulted in Max(O3) closer to that predicted by CB05-TU-OLE
thanusing the default speciation of FORM 3 PAR (Fig. 5a).
However,speciation into OLE 2 PAR resulted in 1-h Max(O3) higher by
only1.4 ppb (1.1%) than that with CB05-TU for case
LP1.IBTEPRPE(Fig. 5b) and resulted in 1-h Max(O3) over-predicted by
10.3 ppb(9.1%) compared to that with CB05-TU-OLE for case
LP2.IBTE(Fig. 5c). Based on the results presented in Figs. 2 and 5,
speciatingisobutene into OLE 2 PAR is not a completely satisfactory
solutionto the under-prediction of O3 formation from isobutene by
CB05and CB05-TU, compared to using the isobutene (IBTE) reactions
in
VOC mixture e NOx experiments.
[%] NOx crossover time [min]
05-TU CB05-TU-OLE CB05 CB05-TU CB05-TU-OLE
0 20 4 3 30 10 11 8 8
onment 59 (2012) 141e150Table 2.In combination with chamber
simulation results described in
sections 3.1 and 3.2, the box modeling results for propene (1
case)and isobutene (3 cases) indicate that modeling propene and
iso-butene using the reactions listed in Table 2 can improve
theperformance of CB05 and CB05-TU in simulating O3 under
ambientconditions inuenced by industrial propene and isobutene
emis-sions. However, further studies under ambient conditions
(e.g., 3-dimensional air quality modeling) are needed to clarify
conditionsunder which propene or isobutene needs to be separately
modeled.
-
Fig. 5. Comparison of modeled O3 between CB05-TU, CB05-TU-OLE
andCB05-TU(OLE 2 PAR) for 3 box modeling cases for isobutene: (a)
LP1.IBTE,(b) LP1.IBTEPRPE, and (c) LP2.IBTE.
G. Heo et al. / Atmospheric Envir4. Conclusions
In CB05 and CB05-TU, ozone formation from alkenes isdescribed by
highly condensed chemical reactions of a few modelspecies. These
mechanisms were tested against environmentalchamber data for their
capability in simulating O3 formation fromoxidation of propene and
isobutene. The chamber simulation andbox modeling results show that
(1) CB05 and CB05-TU reasonablysimulate O3 formation from propene
under typical urban condi-tions and for cases inuenced by
moderately elevated propeneconcentrations, (2) separately modeling
O3 formation from pro-pene and isobutene using condensed reactions
for propene andisobutene (e.g., reactions for PRPE (propene) and
IBTE (isobutene)listed in Table 2) has potential to improve the
performance of CB05and CB05-TU in modeling O3 under conditions
where the hydro-carbon reactivity is dominated by propene and/or
isobutene, (3)CB05-TU is generally better than CB05 in simulating
O3 formationfor the 86 surrogate VOCs e NOx experiments, which is
consistentwith results reported by Whitten et al. (2010) for
toluene e NOxexperiments, and (4) the characteristics of chamber
experiments(e.g., VOC composition) inuence the magnitude of impact
onmodeled ozone of using more explicit alkene reactions.
Historically, developers of chemical mechanisms for use
in3-dimensional air quality models have represented
atmosphericreactions of many different VOCs with highly condensed
reactionsof a limited number of model species by using assumptions
on theaverage air compositions of major VOC classes (e.g., alkenes
andaromatics) or major carbon bond types (e.g., terminal
carbondouble bonds). However, the air composition changes over
timeand is spatially different. Certain compounds may need to get
moreattention due to changes in emissions over time, and
certaincompounds may warrant more attention for some regions than
forother regions (e.g., HRVOCs in southeast Texas).
Developingcondensed chemical reactions (e.g., OLE reactions in
CB05) opti-mized only for one region may result in inaccurate model
predic-tions because inmany cases the air quality modeling domain
coversmultiple states or much of the entire continental U.S.
Therefore,a practical but more scientically robust alternative
approach is(a) ranking the relative importance of compounds based
on theiremissions and reactivity, (b) separately modeling more
importantcompounds (e.g., ethene and propene), (c) lumping other
lessimportant compounds into model species shared by
multiplecompounds and developing condensed reactions for those
modelspecies. For example, propene is separately modeled in the
toxicsversion of the SAPRC-07 mechanism (SAPRC-07T, Hutzell et
al.,2012). As demonstrated in this work, chamber data are crucial
todeveloping and evaluating chemical mechanisms because chamberdata
produced under well-characterized and well-controlledconditions
allow testing of chemical mechanisms withoutinvolving uncertainties
in emissions and meteorology. Lack ofuseful chamber data for
mechanism evaluation is a critical obstacleto developing reliable
mechanisms. For example, among the7 alkenes (ethene, propene,
1,3-butadiene, 1-butene, isobutene,trans-2-butene, cis-2-butene)
regulated as HRVOCs in southeastTexas, robust chamber data for
mechanism evaluation are availableonly for ethene and propene. The
reliability of chemical mecha-nisms used in air quality models can
be strengthened by conduct-ing targeted mechanism improvements
using experimental data asillustrated by improving mechanisms for
propene and isobutene inthis work.
Acknowledgments
This study was in part funded by the Texas Air Research
Center
onment 59 (2012) 141e150 149(TARC) through TARC project
079UTA0102A, Implementation of
-
Modied Carbon Bond Mechanisms in CAMx. The authors thankall
investigators who contributed to producing the chamber dataused in
this study.
Appendix A. Supplementary material
Supplementary material associated with this article can befound,
in the online version, at
http://dx.doi.org/10.1016/j.atmosenv.2012.05.042.
References
Atkinson, R., Arey, J., 2003. Atmospheric degradation of
volatile organic compounds.Chemical Reviews 103, 4605e4638.
Berndt, T., Bge, O., 1995. Products and mechanism of the
reaction of NO3 withselected acyclic monoalkenes. Journal of
Atmospheric Chemistry 21, 275e291.
Calvert, J.G., Atkinson, R., Kerr, J.A., Madronich, S.,
Moortgat, G.K., Wallington, T.J.,Yarwood, G., 2000. The Mechanisms
of Atmospheric Oxidation of Alkenes.
Hutzell, W.T., Luecken, D.J., Appel, K.W., Carter, W.P.L., 2012.
Interpreting predictionsfrom the SAPRC07 mechanism based on
regional and continental simulations.Atmospheric Environment 46,
417e429.
IUPAC, 2010. IUPAC Subcommittee for Gas Kinetic Data Evaluation
Web site. http://www.iupac-kinetic.ch.cam.ac.uk/ (Accessed in
February, 2010).
Jeffries, H.E., Gery, M.W., Carter, W.P.L., 1992. Protocols for
Evaluating OxidantMechanisms for Urban and Regional Models. Final
Report to the U.S. EPA,Cooperative Agreement 815779,
EPA-600/R-82/112, June 1992. Available
at:http://www.cert.ucr.edu/wcarter/pubs/mechprot.pdf.
Jobson, B.T., Berkowitz, C.M., Kuster, W.C., Goldan, P.D.,
Williams, E.J., Fesenfeld, F.C.,Apel, E.C., Karl, T., Lonneman,
W.A., Riemer, D., 2004. Hydrocarbon sourcesignatures in Houston,
Texas: Inuence of the petrochemical industry. Journalof Geophysical
Research-Atmospheres 109, D24305.
http://dx.doi.org/10.1029/2004JD004887.
Kuwata, K.T., Hasson, A.S., Dickinson, R.V., Peterson, E.B.,
Valin, L.C., 2005. Quantumchemical and master equation simulations
of the oxidation and isomerizationof vinoxy radicals. Journal of
Physical Chemistry A 109, 2514e2524.
Mao, J., Ren, X., Chen, S., Brune, W.H., Chen, Z., Martinez, M.,
Harder, H., Lefer, B.,Rappenglck, Flynn, J., Leuchner, M., 2010.
Atmospheric oxidation capacity inthe summer of Houston 2006:
comparison with summer measurements inother metropolitan studies.
Atmospheric Environment 44 (33), 4107e4115.
Murphy, C.F., Allen, D.T., 2005. Hydrocarbon emissions from
industrial release
G. Heo et al. / Atmospheric Environment 59 (2012)
141e150150Oxford University Press, New York, 552 pp.Carter, W.P.L.,
1990. A detailed mechanism for the gas-phase atmospheric
reactions
of organic compounds. Atmospheric Environment 24A,
481e518.Carter, W.P.L., 2000. Documentation of the SAPRC-99
Chemical Mechanism for VOC
Reactivity Assessment. Report to the California Air Resources
Board, Contracts92-329 and 95-308.
http://www.cert.ucr.edu/wcarter/absts.htm#saprc99.
Carter, W.P.L., 2010a. Development of the SAPRC-07 Chemical
Mechanism andUpdated Ozone Reactivity Scales. Final Report to the
California Air ResourcesBoard, Contract No. 03-318, 06-408, and
07e730, Revised January 27,
2010.http://www.cert.ucr.edu/wcarter/SAPRC/.
Carter, W.P.L., 2010b. Development of the SAPRC-07 chemical
mechanism. Atmo-spheric Environment 44 (40), 5324e5335.
Carter, W.P.L., 2010c. Development of a condensed SAPRC-07
chemical mechanism.Atmospheric Environment 44 (40), 5336e5345.
Carter, W.P.L., Atkinson, R., 1987. An experimental study of
incremental hydrocarbonreactivity. Environmental Science &
Technology 21, 670e679.
Carter, W.P.L., Cocker, D.R., Fitz, D.R., Malkina, I.L.,
Bumiller, K., Sauer, C.G.,Pisano, J.T., Bufalino, C., Song, C.,
2005. A new environmental chamber forevaluation of gas-phase
chemical mechanisms and secondary aerosol forma-tion. Atmospheric
Environment 39, 7768e7788.
Dodge, M.C., 2000. Chemical oxidant mechanisms for air quality
modeling: criticalreview. Atmospheric Environment 34,
2103e2130.
Faraji, M., Kimura, Y., McDonald-Buller, E., Allen, D., 2008.
Comparison of the carbonbond and SAPRC photochemical mechanisms
under conditions relevant tosoutheast Texas. Atmospheric
Environment 42, 5821e5836.
Gery, M.W., Whitten, G.Z., Killus, J.P., Dodge, M.C., 1989. A
photochemical kineticsmechanism for urban and regional scale
computer modeling. Journal ofGeophysical Research 94 (D10),
12925e12956.
Gilman, J.B., Kuster, W.C., Goldan, P.D., Herndon, S.C.,
Zahniser, M.S., Tucker, S.C.,Brewer,W.A., Lerner, B.M.,Williams,
E.J., Harley, R.A., Fehsenfeld, F.C.,Warneke, C.,de Gouw, J.A.,
2009. Measurements of volatile organic compounds during
the2006TexAQS/GoMACCScampaign: Industrial inuences, regional
characteristics,and diurnal dependencies of the OH reactivity.
Journal of Geophysical Research-Atmospheres 114, D00F06.
http://dx.doi.org/10.1029/2008JD011525.
Heo, G., 2009. Condensed chemical mechanisms and their impact on
radical sourcesand sinks in Houston. Ph.D. dissertation, University
of Texas at Austin, August,2009. Available at:
http://repositories.lib.utexas.edu/handle/2152/9702.
Heo, G., Kimura, Y., McDonald-Buller, E., Carter, W.P.L.,
Yarwood, G., Allen, D.T., 2010.Modeling alkene chemistry using
condensed mechanisms for conditions rele-vant to southeast Texas,
USA. Atmospheric Environment 44 (40), 5365e5374.events in the
Houston-Galveston area and their impact on ozone
formation.Atmospheric Environment 39, 3785e3798.
Neeb, P., Moortgat, G.K., 1999. Formation of OH radicals in the
gas-phase reaction ofpropene, isobutene, and isoprene with O3:
yields and mechanistic implications.Journal of Physical Chemistry A
103, 9003e9012.
OBrien, J.M., Czuba, E., Hastie, D.R., Francisco, J.S., Shepson,
P.B., 1998. Determina-tion of the hydroxyl nitrate yields from the
reaction of C2eC6 alkenes with OHin the presence of NO. Journal of
Physical Chemistry A 102, 8903e8908.
Paulson, S.E., Flagan, R.C., Seinfeld, J.H., 1992. Atmospheric
photo-oxidation ofisoprene: part I. The hydroxyl radical and ground
state atomic oxygen reactions.International Journal of Chemical
Kinetics 24, 79e101.
Sauer, F., Schfer, C., Neeb, P., Horie, O., Moortgat, G.K.,
1999. Formation of hydrogenperoxide in the ozonolysis of isoprene
and simple alkenes under humidconditions. Atmospheric Environment
33, 229e241.
Seila, R.L., Lonneman, W.A., Meeks, S.A., 1989. Determination of
C2 to C12 AmbientAir Hydrocarbons in 39 U. S. Cities from 1984
through 1986. Report to the U.S.Environmental Protection Agency,
Washington, D. C., EPA/600/3e89/058.
Simon, H., Beck, L., Bhave, P.V., Divita, F., Hsu, Y., Luecken,
D., Mobley, J.D.,Pouliot, G.A., Reff, A., Sarwar, G., Strum, M.,
2010. The development and uses ofEPAs SPECIATE database.
Atmospheric Pollution Research 1 (4),
196e206.http://dx.doi.org/10.5094/APR.2010.026.
TCEQ, 2011. TCEQ web site.
http://www.tceq.texas.gov/airquality/stationary-rules/hrvoc.html
(accessed in August 2011).
Tuazon, E.C., Aschmann, S.M., Arey, J., Atkinson, R., 1998.
Products of the gas-phasereactions of a series of
methyl-substituted ethenes with the OH radical. Envi-ronmental
Science & Technology 32 (14), 2016e2112.
Whitten, G.Z., Hogo, H., Killus, J.P., 1980. The carbon-bond
mechanism: a condensedkinetic mechanism for photochemical smog.
Environmental Science & Tech-nology 14, 690e700.
Whitten, G.Z., Heo, G., Kimura, Y., McDonald-Buller, E., Allen,
D.T., Carter, W.P.L.,Yarwood, G., 2010. A new condensed toluene
mechanism for Carbon Bond:CB05-TU. Atmospheric Environment 44 (40),
5346e5355.
Yarwood, G., Rao, S., Yocke, M., Whitten, G.Z., 2005. Updates to
the Carbon BondMechanism: CB05. Final Report to the U.S. EPA.
Available at:
http://www.camx.com/publ/pdfs/CB05_Final_Report_120805.pdf.
Yarwood, G., Jung, J., Heo, G., Whitten, G.Z., Mellberg, J.,
Estes, M., 2010. Updates tothe Carbon Bond Mechanism for Version 6
(CB6). Presented at the 9th AnnualCMAS Conference, Chapel Hill, NC,
October 11-13, 2010. Available at:
http://www.cmascenter.org/conference/2010/agenda.cfm.
Modeling ozone formation from alkene reactions using the Carbon
Bond chemical mechanism1. Introduction2. Data and methods2.1.
Chemical mechanisms used2.2. Environmental chamber data2.3.
Mechanism evaluation using environmental chamber simulations2.4.
Comparing mechanisms using box modeling
3. Results and discussion3.1. Evaluating CB05 and CB05-OLE
against propene chamber experiments3.2. Evaluating CB05 and
CB05-OLE against isobutene chamber experiments3.3. Evaluating CB05
variants against surrogate mixture chamber experiments3.4.
Comparing CB05-TU and CB05-TU-OLE using box modeling cases
4. ConclusionsAcknowledgmentsAppendix A. Supplementary
materialReferences