United States
Environmental Protection
Agency
.
Atmospheric Sciences ^ \
Research Laboratory -x
Research Triangle Park NC 2771 1 ,"
-
Research and Development
EPA/600/S3-87/014 June 1987
&EPA Project Summary
A Surrogate Species Chemical
Reaction Mechanism for
Urban-Scale Air Quality
Simulation Models
Frederick W. Lurmann, William P. L Carter, and Lori A. Coyner
During the second year of a two-year
program, a surrogate species chemical
mechanism was refined, evaluated, and
adapted for use in air quality simulation
(AQS) models. The purpose of the pro-
gram was to develop an improved
chemical mechanism for use in the AQS
models that are used to develop ozone
control strategies.
The updated chemical reaction mech-
anism was evaluated against data from
491 environmental chamber experi-
ments conducted in indoor and outdoor
facilities. The results of the evaluation
indicate the mechanism's predictions
are qualitatively and quantitatively con-
sistent with the data from a large num-
ber of single organic-NOx and multi-
organic NOX experiments.
The mechanism was adapted for use
in the single-cell and multi-cell AQS
models. Guidelines were developed for
using the mechanism. These include
procedures for assignment or individ-
ual organic species to the chemical
classes in the mechanism and default
organic speciation profiles. Sensitivity
analysis was performed to identify the
AQS model inputs that strongly influ-
ence predicted volatile organic com-
pound (VOC) control requirements.
This Project Summary was devel-
oped by EPA's Atmospheric Sciences
Research Laboratory, Research Triangle
Park, NC, to announce key findings of
the research project that is fully docu-
mented in two separate volumes of the
same title (see Project Report ordering
information at back).
Introduction
Atmospheric simulation models are
essential planning tools for the develop-
ment of emission control strategies for
regions that presently exceed the ambi-
ent air quality standard for ozone. The
models are used to estimate the emis-
sion control requirements needed to
prevent exceedances of the air quality
standards in the future. One of the most
important components in air quality
simulation (AQS) models is the chemi-
cal mechanism that describes the for-
mation of ozone from volatile organic
compounds (VOC) and nitrogen oxides
(NOX). As part of a coordinated research
program to develop reliable ozone con-
trol strategies, the U.S. Environmental
Protection Agency sponsored this re-
search study to develop and test an im-
proved chemical mechanism.
Many chemical mechanisms have
been developed in the last 10 years for
simulating ozone formation from VOC
and NOX. All of the mechanisms in-
tended for atmospheric applications in-
corporate approximations and conden-
sations of species and reactions
because it is currently impossible to ex-
plicitly include reactions for the hun-
dreds of organic compounds present in
ambient air. Several approaches are
available for lumping the organic spe-
cies into a manageable number of
chemical classes in the mechanisms.
The approach adopted in this study was
to use the surrogate species approxima-
tion where the explicit chemistry of se-
lected compounds is used to represent
-------
the chemistry of all similar compounds.
For example, reactions for propene and
trans-2-butene are used as surrogates
for the reactions of terminally double-
bonded and internally double-bonded
alkenes. Provided the mechanism is for-
mulated with a sufficient number of sur-
rogate species (-10), the surrogate spe-
cies approach is quite capable of
representing the majority of organic
species present in urban air.
The surrogate species approach has
several advantages over other schemes
such as the carbon bond lumping ap-
proach. Surrogate species mechanisms
can easily be updated and expanded
since the reactions for each surrogate
species are independent of other parts
of the mechanism. Also, since whole
molecules rather than lumped bond
groups are used in the surrogate mech-
anism, they are not reliant on the as-
sumption that different parts of the
molecules react independently.
Prior to this study, surrogate species
chemical mechanisms had been sub-
jected to only limited testing against en-
vironmental chamber data. A key pur-
pose of this study was to extensively
test the predictive abilities of the surro-
gate species mechanism against cham-
ber data for a broad range of conditions.
Testing of the New Mechanism
The new mechanism was tested
against data from 491 experiments. The
experiments were carried out in four dif-
ferent environmental chambers: the
University of North Carolina's (UNC)
150,000-liter dual outdoor chamber, the
Statewide Air Pollution Research Cen-
ter's (SAPRC) 50,000-liter outdoor
Teflon chamber (OTC), SAPRC's 6,400-
liter indoor Teflon chamber (ITC), and
SAPRC's 5,800-liter evacuable indoor
chamber (EC). Procedures were devel-
oped to represent the light intensities
and spectral distributions in the differ-
ent chambers. Appropriate methods
were also developed to characterize the
major chamber effects, such as NOX off-
gassing, ozone deposition, and the
chamber free radical sources in a con-
sistent manner for all four chambers.
The types and number of environ-
mental chamber experiments used in
the testing program are listed in Table 1.
The N0x-air, NOx-CO-air, and n-butane
NOX runs were used to test the inor-
ganic chemistry and refine the chamber
characterization procedures. Single or-
ganic compound-NOx experiments
were employed to test the reactions for
Table 1. Summary of Environmental Chamber Runs Used for Mechanism Evaluation
Number of Runs*
Type of Environmental Chamber Run
EC
ITC
OTC
UNC
Total
Characterization
Single Organic-NOx
Known Mixtures
Auto Exhaust
Dynamic Injection
Totals
NOx-air and
NOx-Co-air
Oxygenates
Ethene
Propene
Butenes
Toluene
Other Aromatics
n-Butane
C5+
Simple Mixtures
Surrogate Mixtures
Catalyst and
Noncatalyst
Simple Mixtures
10 14 10
37 71
7
6
15
6
13
7
14
6
22
11
1
2
7
5
2
13
5
8
45
2
5
62
117
102
80
15
6
22
5
5
4
7
6
18
33
25
9
192
25
14
49
16
20
24
27
20
40
151
25
9
491
*UNC = UNC outdoor chamber, EC = SAPRC evacuable indoor chamber,
ITC = SAPRC indoor Teflon chamber, OTC = SAPRC outdoor Teflon chamber.
each of the organic precursor species
included in the mechanism. Experi-
ments with organic mixtures were used
to test the predictive ability of the mech-
anism for conditions representative of
the real atmosphere. These ranged
from simple mixtures like propene/n-
butane to complex mixtures including
more than 15 compounds and automo-
bile exhaust.
The average bias and error in the
mechanism's predictions for maximum
ozone concentrations are listed in Table
2. The evaluation data indicate the
mechanism underpredicts ozone yields
in carbonyl-NOx experiments by 5% on
the average. The average error in the
maximum ozone predictions is ±25%
for carbonyl-NOx systems. Mechanism
performance for formaldehyde is better
than for higher aldehydes and ketones.
The performance data show that the
mechanism overpredicts ozone yields
in alkene-NOx experiments by 3% on
the average. The average error in the
maximum ozone predictions is ±21%
for these systems. Mechanism perform-
ance for ethene and propene is signifi-
cantly better than that for butenes.
The evaluation data indicate the
mechanism overpredicts ozone yields
in aromatic-NOx experiments by 1% on
the average. The average error in the
maximum ozone predictions is ±21%
for aromatic-NOx systems. The mecha-
nism tends to overpredict ozone yields
in benzene and toluene-NOx systems
and underpredict ozone yields in m-
xylene, o-xylene, and mesitylene-NOx
systems on the average. This level of
performance on aromatic runs is
surprisingly good considering that the
aromatic mechanism is highly parame-
terized and that the identity and subse-
quent chemistry of half or more of the
aromatic photooxidation products are
unknown.
The evaluation results for alkane-NOx
simulations are not satisfactory. The
mechanism's maximum ozone predic-
tions show large errors (±69% on the
average) and a bias toward overpredic-
tion. The poor mechanism performance
for alkane-NO,, runs is due to uncer-
tainty in the alkane chemistry, espe-
cially for the C6+ alkanes, and uncer-
tainty in the chamber radical source
strength. Alkanes are less reactive than
the other compounds employed in the
testing program, and simulations of
alkane-NOx experiments are extremely
sensitive to the assumed chamber radi-
cal source strength. Because of the un-
certainty and variability of the chamber
radical source, the alkane mechanism
cannot be evaluated without ambiguity
using chamber data.
The mechanism's performance simu-
lating mixtures of organic compounds
is good. Overall, the mechanism pre-
dicted the- maximum ozone in 225 mix-
ture experiments with an average bias \
-------
of +4% and an average error of ±24%.
The predicted rates of NOX oxidation
and timing of the ozone maximum also
have little bias and less than 30% error.
The mechanism's performance for sim-
ple mixtures was not quite as good as
its performance for the surrogate mix-
tures and auto exhaust. However, over-
all these results show the mechanism is
qualitatively and quantitatively consis-
tent with the chamber data. Good per-
formance in testing against organic
mixture experiments is important be-
cause the mechanism will primarily be
used to simulate mixtures in atmos-
pheric modeling.
Adaptation of the Mechanism
Condensed versions of the mecha-
nism employed in the testing program
were developed for use in AQS models.
Mechanisms were developed for use in
single-cell models that can accommo-
date large chemical mechanisms and
for use in multi-cell models that require
fairly small chemical mechanisms. Very
little mechanism condensation was re-
quired for the mechanism designed for
use in single-cell models such as the
021PM AQS model. Significant mecha-
nism condensation assumptions were
implemented in the mechanism de-
signed for use in the multi-cell models
such as the Urban Airshed Model.
Predictions from the condensed ver-
sions of the mechanism were compared
to predictions of the detailed mecha-
nism for a range of mixtures and
NMOC/NOX ratios. The results showed
the single-cell model mechanism's pre-
dictions are almost identical (i.e., within
±2%) to the detailed mechanism's pre-
dictions for all of the key species. Pre-
dictions from the multi-cell model
mechanism agree with those from the
detailed mechanism within ±10% for all
key species.
Information on speciation of organics
for the classes in the mechanism was
developed. First, a master list showing
the assignment if individual organic
compounds to organic classes in the
mechanism was compiled. The uncer-
tainty of each assignment was ranked,
based on whether or not the surrogate
species employed for the assigned class
could represent the reactivity of the in-
dividual species well. Second, ambient
speciated NMOC data collected at the
ground and above the mixed layer in
the mornings in urban areas were ana-
lyzed. A default NMOC composition
I profile for emissions and ambient con-
centrations near the surface were devel-
Table 2. Average Model Performance
for Maximum Ozone
Run Type
Bias (%) Error (%)
Formaldehyde
Acetaldehyde
Other Carbonyls
All Carbonyls
Ethene
Propene
Butanes
All Alkenes
-1
-26
+4
+2
+3
+4
-5
+3
19
26
44
18
18
34
25
21
Butane +31 67
Branched Alkanes +34 49
Long-chain Alkanes +83 84
All Alkanes +46 69
Benzene
Toluene
Xylenes
Mesitylene
All Aromatics
+3
+ 77
-9
-77
5
24
16
21
+ 1
19
All Single HC Runs +12 33
Simple Mixtures +10 35
Mini Surrogates +10 22
Full Surrogates +3 23
Auto Exhaust -11 15
All HC Mixtures +4 24
All Run Average +7 28
"Positive bias indicates model overpredic-
tion.
oped from ambient data collected in 25
cities using a consistent measurement
and speciation protocol. A default com-
position profile for NMOC aloft was
compiled from aircraft data collected
upwind of four cities. These default pro-
files can be used in atmospheric model-
ing applications where site-specific data
are not available.
Sensitivity analysis was carried out
using the updated chemical mechanism
in the OZIPM AQS model. The sensitiv-
ity analysis was designed to identify the
input parameter that most strongly in-
fluences the NMOC control require-
ments in EKMA analyses. Almost all of
the sensitivity runs were performed at
several NMOC/NOX ratios and dilution
rates since the sensitivity of the model
to parameter variations is known to de-
pend on these parameters. The results
of the analysis confirmed the impor-
tance of the following input parame-
ters:
• NMOC/NOX ratio
• NMOC composition
• Post-8 a.m. emission rates along
the trajectory
• Future changes in NOX emission
rates
• Ozone and NMOC concentrations
aloft
Other relatively important parameters
include the mixing height, radiation,
and initial PAN concentrations. The re-
sults of the sensitivity analysis are in-
tended to help air quality planners prior-
itize efforts for obtaining input data for
the photochemical models used to de-
velop control strategies.
A set of sample problems and instruc-
tions for implementing both versions of
the mechanism in AQS models were de-
veloped. These will allow users of the
mechanism to ensure the mechanism is
properly implemented.
Frederick W. Lurmann and Lori A. Coyner are with ERT, A Resource Engineering
Co., Inc., NewburyPark. CA 91320.
William P. L. Carter is the EPA Project Officer (see below).
The complete report, consists of two volumes, entitled "A Surrogate Species
Chemical Reaction Mechanism for Urban-Scale Air Quality Simulation
Models:"
"Volume I. Adaptation of the Mechanism," (Order No. PB 87-180 592/AS;
Cost: $24.95)
"Volume II. Guidelines for Using the Mechanism," (Order No. PB 87-180
600'/AS; Cost: $18.95)
The above reports will be available only from: (costs subject to change)
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
J'JNl.''8/ j ^v.rp
; ' • - • -'• ~ 0 ?
.- - , ,„ u .i,
' s. ; i * - rv <• - ' ''
Official Business
Penalty for Private Use S300
EPA/600/S3-87/014
0000329 PS
U S EHVIR PROTECTION AGENCY
REGION 5 LIBRARY
Z30 S DEARBORN STREET
CHICAGO IL 60604
l.li..li,,,,l!,,ll,.. .lul.ii.it
------- |