United States Environmental Protection Agency Atmospheric Sciences Research Laboratory Research Triangle Park, NC 27711 Research and Development EPA/600/S3-88/012 Apr. 1988 .''•4 > v°/EPA Project Summary Development and Testing of the CBM-IV for Urban and Regional Modeling M. W. Gery, G. Z. Whitten, and J. P. Killus A new chemical kinetics mechanism for simulating urban and regional photochemistry based on the carbon-bond method of hydrocarbon condensation was developed and evaluated in this project. The Carbon-Bond Mechanism-IV (CBM-IV) is a condensed version of an expanded mechanism (CBM-X) that was ini- tially developed using currently available laboratory and smog chamber data. In addition to a general updating of the mechanism to include the most recent kinetic, mechanistic, and photolytic information, the CBM-IV comprises extensive improvements to the chemical representations of aromatic species, blogenic hydrocarbons and peroxyacetyl nitrate (PAN). CBM-IV performance in predicting ozone, formaldehyde, and PAN con- centrations was evaluated against the results of approximately 160 experiments from four different smog chambers. Both the maximum predicted concentrations and the time to the maximum were compared. Other parameters such as hydrocarbon and NOX decay rates were also addressed. The results of these evaluations indicate substantial improvement in the ability of the CBM-IV to simulate aroma- tic and isoprene systems. For- maldehyde predictions for the isoprene experiments were also very good. The CBM-IV overpredicted maximum ozone concentrations by 2% and underpredicted for- maldehyde by 9% for 68 different hydrocarbon/NOx mixture experi- ments from three different smog chambers. This Project Summary was developed by EPA's Atmospheric Sciences Research Laboratory, Research Triangle Park, NC, to announce key findings of the research project that is fully documented in a separate report of the same title (see Project Report ordering information at back). Introduction The CBM-IV was developed for use in EPA's EKMA procedure. It can also be used in large air quality simulation models (AQSMs) such as the Urban Airshed Model (UAM) and the EPA Regional Oxidant Model (ROM). In the following discussion, the authors present an overview of the development of the CBM-IV and summarize the evaluation of the mechanism. Development of the CBM-IV The core of the CBM-IV consists of a set of explicitly represented inorganic reactions in addition to the reactions of common carbonyl species and their products (e.g., formaldehyde, acetaldehyde, and PAN). Because the large number of organics involved in tropospheric photochemistry precludes a fully explicit chemical treatment ot individual organic compounds condensation techniques must be employed to represent larger organic species. In the CBM-IV, reactive organics are usually treated within <. ------- surrogate approximation; however, some specific compounds that (1) cannot easily be included in surrogate schemes and (2) are common constituents of polluted atmospheres (i.e., ethene, isoprene, and to a lesser degree, toluene, xylene, and acetaldehyde) are treated individually. Several major improvements that were made to the CBM to develop the CBM- IV have led to significant improvement in the performance of the mechanism in smog chamber evaluation exercises. The most important improvements are as follows: (1)New chemical kinetics and product data have been included to update both the inorganic and organic components. (2)The temperature-dependence of acetylperoxy radical reactions with NO and NOg has been improved, and the kinetics of peroxy- peroxy radical reactions has been updated to more closely represent experimental evidence available from both chemical kinetics and smog chamber studies (3) The chemistry of ethene was altered to account for glycolaldehyde "formation. (4)The reaction of acetaldehyde with HC>2 was eliminated because there is no substantial proof of its oc- currence. (5)The chemical equilibrium between HC>2, NOa and peroxynitric acid (PNA) was updated. (6)The use of formaldehyde (FORM) as a surrogate for glyoxal was eliminated, making the species FORM an explicit representation of formaldehyde and allowing the CBM-IV to be used for for- maldehyde-specific simulations. (7) The chemistry of aromatic species (toluene and xylene) has been significantly altered to account for the sensitivity of ozone formation to NMHC/NOX. (8)A new condensed isoprene (ISOP) mechanism was developed and evaluated. (9)New carbon-bond fractions for pinene were tested and incorporated in accordance with the desire to utilize that species as a biogenic hydrocarbon surrogate. It is beyond the scope of this summary report to directly discuss all of the enhancements to the CBM-IV. Therefore, a focus is made on some significant improvements in the aromatic and isoprene reaction schemes. Improvements to the CBM-IV Aromatic Reaction Scheme In the CBM-IV, two surrogate species are utilized to represent the chemistry of all aromatics. These surrogate species are TOL for toluene (mono-alkylbenzenes) and XYL for xylene (di- and tri-alkylbenzenes). These surrogate species were selected because the differentiation between surrogate chemistries must represent the widest possible range of aromatic reactivities if they are to provide the most appropriate representation of specific compounds. The most important chemical features to differentiate in the surrogate chemistries are the OH reaction rate and the secondary reaction scheme. Of these two, appropriate representation of the secondary reaction scheme is the most difficult to achieve because this chemistry is complex and varies among aromatic species. One critical difference between the product yields of toluene and those of higher molecular weight aromatic species is the inability of toluene to form high yields of reactive products (such as dicarbonyl species) promptly after OH reaction. Such products are stable but thermally and photolytically reactive. Therefore, their presence perpetuates the reactive nature of the system even after the initial hydrocarbon is exhausted. Available smog chamber data indicate that toluene systems rapidly consume NO, but that toluene reaction continues well after the period during which reactive product formation would be severely limited by diminished NO concentrations. Ozone formation in these toluene experiments initially appears to be very fast but then rapidly terminates, and ozone is often consumed in later stages of an experiment. This dichotomous behavior is somewhat unique for a "reactive" hydrocarbon and may indicate that, unlike other aromatics and olefins, the product mixture formed in the initial portion of toluene experiments may form less reactive species as the experiment progresses and NO is consumed. Prior to development of the CBM-IV, most kinetics simulation mechanisms represented the addition of OH to aromatics with the following simplified form: OH + ARO (+ 02) -» ARO-02. ARO-02. + NO->NO2 + reactive products where the aromatic R02 radical reacted exclusively with NO to form reactive organic products and NO2. This reactio effectively produces highly reactiv products in a rather prompt fashion. On the basis of smog chamber and product yield data for toluene, we believe that this prompt formation of products, especially highly reactive and conjugated dicarbonyls, incorrectly describes the toluene oxidation reaction when NO concentrations become limited. Possible pathways that had not been considered were alternate reactions of the aromatic R02 radicals that do not involve oxidation of NO. The aromatic R02 radicals probably do not have long half-lives, and if they cannot react with NO, they must be lost by an alternate reaction pathway. Our methodology in developing the CBM-IV has been to estimate the mechanism, products, and kinetics of an alternate reaction of the aromatic R02 radical. Such a reaction becomes important only at low NO concentrations and produces less reactive products than the NO-to-N02 conversion process. This reaction has been included in the CBM-IV and has led to significantly better predictions. In the development and performance evaluation of the new CBM-IV aromatics reaction schemes, an attempt was made to simulate all usable toluene and xylene experiments from the UNC and UCR- EC chambers. Also a number of the hydrocarbon-reactivity-mixture experiments performed at UNC were utilized, in which aromatic species were substituted into otherwise identical mixtures. For toluene, the UNC outdoor smog chamber experiments were the basis of the development efforts. On the average, maximum ozone was over predicted for all UNC days by only 4 ± 8%. Similar improved results were obtained for the UCR-EC experiments and the xylene simulations. The average maximum ozone overprediction for the combined toluene and xylene simulations from both chambers was 1 ± 12%. The standard deviation error is much smaller than in earlier work. Isoprene and Pinene The objective in developing this portion of the CBM-IV was to provide a chemical representation of biogenic hydrocarbons by deriving a condensed mechanism for isoprene and estimating valid carbon bond splits for pinene. In the former task, the product species considered in the condensed isoprene mechanism were limited to those already included in the CBM-IV. In the ------- svelopment of the isoprene mechanism Re following significant phenomena to be associated with isoprene oxidation were found: Highly reactive products, including methacrolein and methylvinyl ketone. Some reaction of these species with ozone was also indicated as would be expected from the olefinic character of these two products; A high rate of radical production, presumably from photolysis of isoprene products; A high yield of PAN as well as a PAN-like compound, probably from methacrolein; and A high yield of formaldehyde. The general methodology for the formulation of the condensed isoprene mechanism was based on the following rationale: The olefinic nature of isoprene products would best be simulated by ethene, since the alkene bonds of methacrolein and methylvinyl ketone are partly deactivated and resemble ethene in their reactivity to OH and 03. This representation also gives a high yield of formaldehyde from the secondary oxidation of ethene, which resembles the formaldehyde yield of isoprene. The " rmation of PAN and PAN-like com- ounds from isoprene was simulated by assuming that both acetaldehyde and acetylperoxy radicals are formed during isoprene oxidation. Inevitably, this can result in an overprediction of measured PAN, since simulated PAN represents PAN analogues as well. The radical yield for products was simulated by the formation (and photolysis) of methyl- glyoxal. The product yields were adjusted to give the best fits to the mid- range of hydrocarbon-to-NOx-ratio isoprene experiments, in which notable double ozone peaks become evident. The double ozone peaks in isoprene/NOx experiments are caused by ozone reaction with isoprene and isoprene products, and continued ozone formation due to PAN decomposition at elevated temperatures. Simulation of the double peak effect gives some indication that a reasonable balance has been achieved for the multiple processes occurring in isoprene oxidation. In addition to the isoprene simulations, pinene experiments were used to develop a new carbon-bond pa- rameterization for pinene. The new "epresentation is 0.5 OLE, 1.5 ALD2, and .0 PAR, which is a slight alteration to ihe representation used earlier. Mechanism performance was evaluated using 12 isoprene/NOx smog chamber experiments conducted by UNC. The average of the difference between predicted and measured maximum ozone was about 6 ± 23%. We consider this very good evidence that the new condensed isoprene chemistry successfully represents the photochemical processes in these systems. The simulation results indicate that the characteristic second ozone peak caused by thermal decomposition of PAN or PAN-like species under high afternoon temperatures was successfully simulated for most experiments. In addition, because isoprene is often used to represent the photochemistry of biogenic hydrocarbons, it was also the authors' intention to develop a condensed isoprene representation that would predict formaldehyde concentrations as accurately as possible. Maximum formaldehyde concentrations were overpredicted by only 5 ± 16% for the isoprene/NOx systems. Performance Evaluation Using Complex Hydrocarbon Mixtures Many individual hydrocarbon/NOx experiments from the UNC and UCR- EC smog chambers were used to evaluate individual portions of the CBM- IV. In addition to the aromatic and biogenic hydrocarbon systems, individual species experiments for formaldehyde, acetaldehyde, ethene, and larger olefins were also simulated. To test CBM-IV performance in simulating different reactive hydrocarbon mixtures, some well characterized smog chamber data sets from the UNC dual outdoor chambers and the UCR-EC and indoor Teflon chamber (ITC) were selected. The authors simulated 21 synthetic automobile exhaust and urban surrogate hydrocarbon-mixture experiments from UNC, along with a few experiments of authentic automobile exhaust. These experiments were complemented by a set of 28 hydrocarbon reactivity experiments to focus more closely on the chemical effects of substituting aromatics and some other species in surrogate hydrocarbon mixtures. Eleven EC seven-component-mixture experiments and five ITC multiday experiments from the UCR chambers were simulated, resulting in a total of 65 mixture experiments from these chamber groups. Including the less complex tests, slightly fewer than 200 experiments were simulated during CBM-IV evaluation. For the hydrocarbon mixture experiments, the average overprediction of maximum 03 concentration (excluding one point) was 2 ± 22%. Formaldehyde was underpredicted by 9 ± 34 %. Fi- nally, for all experiments simulated (including those for all individual hydrocarbon/NOx systems), maximum ozone overprediction was 4 ± 23%. Conclusions The development of the CBM-IV began with the gathering of recent chemical kinetic and mechanistic data for tropospheric gas-phase chemistry. The reaction rates and product yields were then updated and tested. In particular, large portions of the inorganic section were altered to reflect current knowledge. The authors also improved a few specific portions of the organic chemistry section for which there were ample data to test the assumptions. Mechanism evaluation and demonstration of the chemical dynamic characteristics of the CBM-IV were performed using approximately 170 smog chamber experiments from the different UNC and UCR smog chambers. The authors compared maximum experimental ozone, PAN, and formaldehyde concentrations with predicted values and provided plots of these comparisons for each experiment so that the dynamic processes could be discussed and the goodness-of-fit over the entire experimental period could be compared. The improvements to the isoprene condensation and the representation of toluene (and other aromatics) oxidation processes provided much better predictions of ozone and formaldehyde than those of previous mechanisms. For the isoprene tests, the mechanism overpredicted maximum ozone concentrations by 6 ± 22%; for the aro- matic experiments, the overprediction was 1 ±12%. These results indicate that the new mechanistic representations significantly diminish the uncertainty associated with these calculations. Less obvious in the overall results were the improvements to predictive capabilities provided by the enhanced organic radical and peroxyacetyl reaction chemistry. These changes appear to have increased the accuracy of the radical concentration predictions during the midday period, resulting in better agreement between PAN formation and decay rates and hydrocarbon decay rates over a large range of temperatures. These improvements and others just described led to a slight overprediction (2 ± 22%) of the maximum ozone ------- concentration averaged for simulations of 68 different experimental mixtures in three different smog chambers. Maximum formaldehyde concentrations were underpredicted by 9 ± 34% for these mixture experiments. Given the larger uncertainties in this calculation, due to both ambiguous organic oxidation processes and less precise measurement methods (than for ozone), these results can be considered to be good. M. W. Gery, G. 2. Whitten, and J. P. Killus are with Systems Applications, Inc., San Rafael, CA 94903. Marc/a C. Dodge is the EPA Project Officer (see below). The complete report, entitled "Development and Testing of the CBM-IV for Urban and Regional Modeling," (Order No. PB 88-180 039/AS; Cost: $38.95, subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA22T61 Telephone: 703-487-4650 The EPA Project Officer can be contacted at: Atmospheric Sciences Research Laboratory U.S. Environmental Protection Agency Cincinnati, OH 45268 United States Environmental Protection Agency Center for Environmental Research Information Cincinnati OH 45268 HO .25; Official Business Penalty for Private Use $300 EPA/600/S3-88/012 0000529 PS •ft-U.S. GOVERNMENT PRINTING OFFICE: 1988—548-013/870 ------- |