Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 33
--> 3 The Concept of Ozone-Forming Potential and Its Quantification In any given airshed, it is common to find hundreds of different VOC species, each with its own unique chemistry. In the simplest approach to ozone mitigation based on VOC controls, emission reductions are implemented on a mass basis without any regard to the unique chemistry of each of the VOCs. The principle behind ozone-forming potential or reactivity1 is the notion that, in addition to the amount of a specific VOC species emitted into a given airshed, the difference in the chemistry of each of the VOCs needs to be considered when assessing the impact of those species on ozone formation. The utility of the concept of ozone-forming potential can be illustrated through a comparison of the impacts on ozone concentrations in an urban airshed of two ubiquitous VOC species: ethane and propene. 1 Because ozone-forming potential of a given VOC is dependent upon its propensity to react in the atmosphere, the term "reactivity" is often used to denote a species' ozone-forming potential. As discussed later in this chapter, terms such as kinetic reactivity (KR) and mechanistic reactivity (MR) are used to define specific processes that contribute to a species ozone-forming potential, whereas terms such as maximum incremental reactivity (MIR) are used w specify the method for calculating a species' ozone-forming potential.
OCR for page 34
--> If one were to increase the total mass of VOC emissions in a city, such as Los Angeles, by 20% through additional emissions of ethane, ozone levels would increase slightly. However, if the same amount of propene were added instead, there would be a large increase in ozone. Why the big difference between the two, given that both are rather simple hydrocarbons? The primary cause of the difference is the differing rates at which these two species react in the atmosphere. Ethane has an atmospheric lifetime of weeks. Little of the ethane emitted in an urban area reacts within that area before it is transported away. Its contribution to ozone formation within the urban area is therefore very small. Propene, on the other hand, has a lifetime of hours. Most of it will typically react near its source and thus be able to contribute to the photochemical production of ozone in the area in which it is emitted (or immediately downwind). A secondary, but smaller, cause for the differing impacts of the two species is the different number of ozone molecules formed in that environment for each molecule of ethane and propene that reacts. Differences in ozone productivity arising from the first effect are often expressed in terms of the kinetic reactivity (KR), and differences from the second are expressed in terms of the mechanistic reactivity (MR). Regulatory Application of VOC Ozone-Forming Potential There is, in fact, a significant historical precedent for accounting for VOC reactivity in U.S. regulatory policy, albeit to a limited extent (see Dimitriades 1996, for a history of VOC regulation in the United States). During the early years of ozone mitigation, it was recognized that there were some organics, for example ethane, that did not contribute significantly to smog formation on urban scales, whereas others, such as propene, did. Thus, two categories of organic gases were defined for regulatory purposes: unreactive and reactive.2 (see Table 3-1). However, the term 2 Reactive VOCs are operationally designated as reactive organic gases (ROG). However, because hydrocarbons make up most of the organic gas emissions, this category is also referred to as reactive hydrocarbons (RHC). Moreover, because methane dominates the unreactive category, nonmethane hydrocarbons or NMHC is another term that is often used. These and other
OCR for page 35
--> TABLE 3-1 Acronyms and Names Used for Classifying Organic Compounds Common Abbreviation Full Name Definition VOCa Volatile organic compound Organic compounds that are found in the gas phase at ambient conditions. Might not include methane. ROG Reactive organic gas Organic compounds that are assumed to be reactive at urban (and possibly regional) scales. Definitionally, taken as those organic compounds that are regulated because they lead to ozone formation. The term is predominantly used in California. NMHC Nonmethane hydrocarbon All hydrocarbons except methane; sometimes used to denote ROG NMOC Nonmethane organic compound Organic compounds other than methane NMOG Nonmethane organic gas Organic gases other than methane RHC Reactive hydrocarbon All reactive hydrocarbons; also used to denote ROG THC Total hydrocarbon All hydrocarbons, sometimes used to denote VOC OMHCE Organic material hydrocarbon equivalent Organic compound mass minus hetero-atom mass (i.e., carbon plus hydrogen mass only) TOG Total organic gas Total gaseous organic compounds, including methane. Used interchangeably with VOC a Unless noted otherwise, VOCs is the term used in this report to represent the general class of gaseous organic compounds. Source: U.S. Environmental Protection Agency at http://www.epa.gov/docs/OCEPAterms.7 unreactive is a misnomer, because even compounds such as ethane and methane do react and contribute to tropospheric ozone formation, terms are listed in Table 3-1. Unless noted otherwise, VOCs is the term used in this report to represent the general class of gaseous organic compounds.
OCR for page 36
--> though at much lower rates, on a per mass basis, than other compounds. Such low-reactivity compounds, particularly carbon monoxide (CO) and methane, do contribute to ozone formation, because emission rates of those compounds are very large. (The contribution of CO to ozone-forming potential is discussed further in Chapters 6 and 7.) A complication in this two-category approach is deciding where to place the dividing line between unreactive and reactive VOCs. Somewhat arbitrarily, that dividing line has been chosen to be at the level of reactivity of ethane. In the United States, but outside of California, species with reactivities equal to or less than that of ethane are placed in the unreactive category. California has been a leading force in the application of reactivity assessment to ozone mitigation efforts. For example, California uses ozone-forming potential in its Low Emission Vehicles and Clean Fuels Program (LEV/CF) to adjust and regulate the amount of emissions from vehicles (CARB 1991). A fuel with higher VOC emissions, but a lower net reactivity than the reference fuel, is permitted in the program, thus providing an incentive to develop fuels with less-reactive emissions. (The current CARB program is limited, however, to exhaust emissions, and, as discussed in Chapter 4, evaporative emissions can be quite important.) The use of reactivity in California's regulatory air-quality programs has been a major catalyst for continuing research on ozone-forming potential and its application to policy-making. As the understanding of how to define ozone-forming potential operationally has grown substantially in recent years, the use of ozone-forming potential to other regulatory issues (e.g., emissions from consumer products) is now under consideration. Operational Definition of Ozone-Forming Potential Using Reactivity The photochemical degradation of most VOC species is initiated by reaction with the OH radical (i.e., Reaction 2.1 in Chapter 2). Therefore, for most VOCs, the KR of a specific VOC is greater if its OH-radical reaction rate constant is greater. As seen in Table 3-2, these rate constants can vary by many orders of magnitude. A relatively simple type of reactivity scale, sometimes referred to as the OH-reactivity or the kOH scale, expresses the relative contribution of VOCs in terms of their rates of reaction with OH (e.g., Darnall et al. 1976; Chameides et al. 1992).
OCR for page 37
--> TABLE 3-2 OH Rate Constants (kOH) and Maximum Incremental Reactivity (MIR)a for Selected Compounds Compound 1012 × kOHb (cm3·molecule-1 s-1) MIRc O3 formed/g VOC emittedb Carbon monoxide 0.21 0.065 Methane 0.0062 0.016 Ethane 0.25 0.32 Propane 1.1 0.57 n-Butane 2.4 1.18 n-Octane 8.7 0.69 2,2,4-Trimethylpentane 3.6 1.34 Ethene 8.5 8.3 Propene 26 11.0 trans-2-Butene 64 13.2 Isoprene 101 9.3 α-Pinene 54 3.9 Benzene 1.2 0.81 Toluene 6.0 5.1 m-Xylene 24 14.2 1,2,4-Trimethylbenzene 32 5.3 o-Cresol 42 2.5 Formaldehyde 9.4 6.6 Acetaldehyde 16 6.3 Acetone 0.22 0.49 2-Butanone 1.1 1.4 Methanol 0.94 0.65 Ethanol 3.3 1.7 Methyl tert-butyl ether 2.9 0.73 Ethyl tert-butyl ether 8.8 2.2 tert-Butyl formate 0.75 No value cited a MIR combines kinetic and mechanistic reactivities of a specied compound for conditions that maximize the predicted reactivity of VOCs by making the reactive systems very NOx rich. b Rate constants at 298 K are taken from Atkinson (1994, 1997) and Le Calve et al. (1997) c From Carter (1997), http://www.cert.ucr.edu/˜carter/bycarter.htm. The MIR of the assumed urban mix used in the calculations was 4.06 g of O3 per gram of VOC emitted.
OCR for page 38
--> This approach has some significant advantages. OH-rate constants for a large number of VOCs have already been well characterized by laboratory experiments, and many others can be estimated with a fair degree of reliability (e.g., Kwok and Atkinson 1995; Atkinson in press). Moreover, these constants are defined by the VOC species themselves and not the environment in which the VOCs are emitted (other than minor temperature dependencies). Thus, the OH reactivities for a wide range of VOC species can be readily calculated and compared. Combining these OH reactivities with data on the ambient concentrations of these VOCs provides a measure of the rate at which the various VOC species are oxidized and produce peroxy radicals (e.g., via Reaction 2-2 and Reaction 2-4 in Chapter 2), and thus provides a rough estimate of their relative potential roles in ozone-formation (Chameides et al. 1992). There are, however, significant limitations to using the OH-reactivity scale to characterize the roles of VOCs: The method does not account for the potentially different yields of peroxy radicals formed from different VOCs, the different reactive pathways these peroxy radicals can take once they are produced, and the varying tendency of VOCs to enhance or inhibit radical levels, and thus influence the contribution of other VOC species to ozone formation. All of these factors can have a significant effect on the amount of ozone formed from the oxidation of a VOC (Carter and Atkinson 1989; Bowman and Seinfeld 1994; Carter 1994). For this reason, the OH-reactivity scale does not always correlate well with other measures of ozone-forming potential, particularly for the more rapidly reacting VOCs (e.g., Dodge 1984; Bergin et al. 1998a). For example, aromatics, which have strong NO x sinks and radical sources in their mechanisms, can have relatively high reactivities under conditions with low ratios of VOC to NOx, but negative values of reactivities when the VOC to NOx ratio is sufficiently high, because NOx, which (as NOx) would otherwise photolyze to form ozone, is removed from the system. MR is used to account for this second influence on the ozone-forming potential of VOCs (Carter and Atkinson 1989). In general, the variability in mechanistic reactivities is substantially less pronounced than that of kinetic reactivities, and thus the species-to-species variability of reactivity scales that combine KR and MR tend to follow the variability in KR but not exactly (see Table 3-2). If KR is defined as the number of molecules of a specific VOC that react within a given airshed (by photolysis, reaction with the OH radical, reaction with NO3 radical, or reaction with ozone) and MR is the number of ozone molecules that are formed for each VOC molecule in the system
OCR for page 39
--> that reacts, the total number of ozone molecules formed from a given VOC molecule is equivalent to the product of the two quantities, that is, This way of dissecting the ozone-forming potential of a compound, although remarkably simple, is also quite powerful and instructive. However, it also has its limitations. For example, neither KR nor MR is a property inherent in a compound. Instead, both are dependent upon the protocol established to calculate them (e.g., the type of environments in which the VOC exists and the length of time used to assess the amount of the VOC that reacts and the ozone that is formed). Thus, the use of the relationship expressed in Equation 3-1 requires an operational definition for quantifying reactivities. Quantifying Ozone-Forming Potential Using Reactivity If ozone-forming potential is to be used in ozone mitigation programs, it is necessary to develop an operational definition for ozone-forming potential, and a protocol for quantifying it. One such definition is the incremental reactivity (IR) proposed by Carter and Atkinson (1989) and Carter (1994).3 IR is defined as the number of additional grams of ozone formed per gram of VOC compound added to a base mixture (the VOC compound could be present in the base mixture): where IRi is the incremental reactivity of species i; ∆[O3] is the change in some ozone metric used to assess the impact of VOCs on air quality (e.g., the 1-hr peak or 8-hr averaged ozone concentration in an airshed) or the total human exposure to ozone above some threshold concentration); and ∆[VOCi] represents a change in the emissions of species i 3 Another scale, developed by Derwent and Hov (1979), is the photochemical ozone creation potential (POCP) scale. It is used to quantify the ozone-forming potential of VOC emissions. In general, the IR approach and the POCP approach produce qualitatively similar results.
OCR for page 40
--> (e.g., from an RFG blend). Defining IR in this way takes into account both the KR and MR of a given VOC species, and, in principle, the incremental reactivity can be broken into kinetic and mechanistic reactivities: where KRi and MRi are, respectively, the kinetic and mechanistic reactivities of the species i. The IR, as defined by Equation 3-2, is an absolute measure of ozone-forming potential (e.g., the number of grams of ozone per gram of VOC). A somewhat more useful quantity for developing ozone mitigation strategies is the relative incremental reactivity (RIR). RIR is defined as the reactivity of one compound normalized to the reactivity of a base mixture: where IRi is the incremental reactivity of species i, fj is the fraction of species j in a base mixture containing n different VOCs so that the denominator in the above expression is the total incremental reactivity of a base mixture, such as an RFG blend. The advantages of working with relative incremental reactivities are threefold. First, in a policy-making context, comparisons of reactivities between species or VOC sources are often of most interest. Second, RIP, tends to be less sensitive to variations in ambient conditions and thus provides a more robust measure of reactivity. Third, RIR is often easier to develop from three-dimensional models, because there is no apparent absolute scale (e.g., the location and timing of how ozone changes is not uniform) (see McNair et al. 1994). The two dominant methods that have been used to assess species' reactivities (IR and RIR) are via direct experimental measurement, for example, in an environmental (or smog) chamber, and numerical simulation using computer-based, air-quality models (Carter and Atkinson 1989; Carter 1994; Derwent and Jenkin 1991; Bowman and Seinfeld 1995; McNair et al. 1992; Yang et al. 1995; Bergin et al. 1996). Both methods have serious limitations. Smog chambers do not realistically represent the physics of pollutant transport and the impact of fresh emissions. Moreover, most do not operate over the full range of NOx
OCR for page 41
--> concentrations and VOC to NOx ratios typically encountered in the polluted atmosphere. Thus, the conditions inside a smog chamber do not reflect those of the ambient air. Given the sensitivity of many VOC reactivities to environmental conditions, smog chamber experiments, by themselves, provide reactivity estimates that are less applicable to atmospheric conditions than those derived from air-quality models. Furthermore, smog chambers have artifacts (e.g., chamber wall and background effects) that can affect the results, particularly if the compound reacts slowly or has radical sinks in its mechanism (Carter and Lurmann 1991). However, chamber experiments are necessary to develop (parameterized) chemical mechanisms for those VOCs for which product and mechanistic data are not yet available from laboratory studies. Data from those chemical mechanisms can then be included in chemical mechanisms for the assessment of their ozone-forming potentials. Because models can be run for conditions that more accurately reflect actual atmospheric conditions, they can, in principle, provide a more appropriate measure of a species' reactivity than that obtained from a smog chamber. However, virtually all photochemical mechanisms used in current air-quality models are based on data from smog chambers. Thus, the ability of models to accurately simulate air quality depends critically upon reliable extrapolation of smog chamber data to atmospheric conditions and elimination of chamber wall and background effects. This has proven to be a very difficult task (Dodge in press). For these reasons, a level of uncertainty is inherent in any assessment of ozone-forming potential, A variety of approaches has been adopted that attempt to characterize and minimize this uncertainty and thus provide a foundation if reactivity were to be implemented in a policy-making context. Chemical Mechanisms and Their Development Because of the aforementioned limitations of smog chambers, air-quality models have played a central role in the quantification of VOC reactivity. Of the various components within air-quality models, the chemical mechanism, which attempts to reproduce the VOC-NOx-air photooxidation process discussed in Chapter 2, is perhaps the most critical component when these models are used to quantify reactivity. This section briefly reviews how these mechanisms are developed and discusses principal mechanisms currently in use.
OCR for page 42
--> Any chemical mechanism used in an air-quality model must be designed so that it can, at a minimum, reproduce the major features of the VOC-NOx-air photooxidation process. The principal chemical mechanisms used in current air-quality models, along with representative airshed modeling applications and their key attributes, are listed in Table 3-3. With the exception of the Harwell Master Chemical Mechanism, all the chemical mechanisms in use today include various kinds of parameterizations, approximations, and condensations to simplify the very complex chemical processes that actually occur when VOCs are oxidized in the atmosphere. There are hundreds of different organic compounds in the atmosphere, and from a numerical point of view, it is often impractical to explicitly follow each species. If this were attempted, the chemical mechanisms would be huge (e.g., the Harwell Master Chemical Mechanism (Jenkin et al. 1997) that has over 7,000 reactions) and would be computationally burdensome in three-dimensional models. TABLE 3-3 Commonly Used Chemical Mechanisms for Air-Quality Modeling and Reactivity Studies Mechanism Description Reference Statewide Air Pollution Research Center 1990 (SAPRC-90/93/97) Explicit for a large number of organics, but uses a lumped representation for reactive products. Designed, in part, for reactivity applications. Carter 1990, 1995, 1997 Carbon Bond IV (C84) Lumped by number of carbon bonds in compounds. Specified by EPA for regulatory purposes. Gery et al. 1989 Lurmann, Carter, Coyner (LCC) Earlier and more-condensed version of SAPRC-90. Used for the earlier CIT grid-model reactivity-assessment calculations. Lurmann et al. 1987 Regional Acid Deposition Model, version 2 (RADM-2) Developed for use in regional acid-deposition modeling. Similar to LCC in detail, except more detailed model for peroxide formation. Stockwell et al. 1990 Harwell Used in Europe. Very large number of compounds represented explicitly. Derwent and Hov 1979 Harwell Master Chemical Mechanism Detailed, explicit mechanism with over 7,000 reactions. Jenkin et al. 1997
OCR for page 43
--> Moreover, if it were practical, laboratory data are available for only a small subset of the relevant reactions, and for all others their rate constants and the products they form would have to be estimated by extrapolation or by analogy from the simpler, better-studied systems. Thus, preserving the full complexity of the atmospheric VOC chemical system in a model might not necessarily increase the reliability of the model's predictions. Chemical mechanisms in air-quality models, therefore, are typically based on the assumption that the atmospheric oxidation of complex VOCs can be simulated by analogy to simpler ones or by using parameterizations to describe the full suite of elementary reactions. To ensure that these simplifying assumptions are capable of adequately simulating the real world, chemical mechanisms should be, and generally are, tested against experimental data from smog chambers in which the relevant chemical processes are monitored under controlled and well-characterized conditions. These data are then used to tune the various parameterizations contained in the mechanism or to test whether model predictions using the mechanism match experimental results. Various types of chamber experiments are used to test different aspects of the chemical mechanisms. Irradiations of single VOCs in the presence of NOx are used to test the mechanism's ability to simulate the oxidation of and ozone production from an individual VOC; NOx-air irradiations of more complex VOC mixtures test the performance of the model as a whole; and experiments in which the effect of adding single VOCs to irradiations of NOx and complex mixtures test model predictions of the VOC's incremental reactivity. Evaluation of chemical mechanisms with smog chamber data is complicated by uncertainties in chamber effects, and separate characterization experiments are needed to evaluate those effects. Chamber data are currently available to test the mechanisms for only a subset of the many types of VOCs emitted into the atmosphere. For the other species, reactions are either derived by analogy with mechanisms for compounds that have been studied, or they are represented in the model as if they reacted in the same way as some other chemically similar species. Mechanisms are further simplified or extrapolated using an approach referred to as "lumping." In this approach, a single hypothetical (or pseudo) species is used in the model to represent a larger number of compounds assumed to react in the same way, or a group of model species is used to represent aspects of the reactions of various chemical compounds. The lumping approaches, and the approxi-
OCR for page 62
--> An alternate approach, adopted by Bergin et al. (1995, 1998a), assessed the impact of environmental variability by comparing reactivities calculated using a three-dimensional, grid-based model for the Los Angeles area with those derived from box-model simulations for 30 cities. Because of their large spatial domain, three-dimensional models cover domains with a wide range of environmental conditions and the reactivities derived from these models represent composite averages over this domain. Reactivities from box models, on the other hand, focus on a single set of environment conditions corresponding to a specific air mass following a specific trajectory. In Figure 3-4 the relative reactivities for a variety of VOCs calculated on the basis of peak ozone concentrations, population-weighted ozone exposures, and spatially weighted ozone exposures from a three-dimensional-model simulation are compared along with box-model-derived results. Here again, while significant variability is seen (and in the cases of toluene and ethylbenzene a change in the sign of the relative reactivity), the general trend in the reactivities from one species to another tends to be reasonably consistent. Another relevant study is that of Khan et al. (1999), who conducted a reactivity study on eight VOC solvents having a wide range of reactivities in three different airsheds: Los Angeles, the Swiss Plateau, and Mexico City. Although the relative reactivities for the eight compounds were found to be similar between the Los Angeles and Switzerland domains, the very high VOC loadings found in Mexico City led to more substantial differences, one being that the aromatic species could have negative reactivities. The reactivity of the aromatics being greatly reduced in regions with lower NOx and higher VOCs was discussed earlier. Uncertainties in Relative Reactivities of Motor-Vehicle Emissions The previous discussion pertains to the reactivities of individual compounds and their attendant uncertainties. However, the charge to this committee (see Chapter 1) is to look at the use of relative reactivities as applied to motor-vehicle emissions, which are composed of hundreds of compounds. This complexity introduces some extra issues, in particular emission-composition uncertainty. A variety of modeling studies, listed in Table 3-8, have examined the reactivity of source emissions. In large
OCR for page 63
--> Figure 3-4 Comparison of three-dimensional- and trajectory-modeled relative reactivities. Source: Bergin et al. 1998a. Reprinted with permission from Encyclopedia of Environmental Analysis and Remediation; copyright 1998, John Wiley & Sons, New York.
OCR for page 64
--> TABLE 3-8 Summary of Source Emissions Reactivity Modeling Studies Reference Model Type Mechanism Application Trijonis and Arledge (1976) Calculated (not modeled) Trajectory EPA smog chamber data Estimated major source reactivities for metropolitan Los Angeles. Chang et al. (1989) Trajectory LCC Methanol-fueled vehicle impacts with respect to conventionally fueled vehicles. Russell et al. (1990) Three-dimensional (CIT) LCC Potential methanol-fueled vehicle impacts for the SCAQS episode (compared with equal mass emissions from conventional vehicles). McNair et al. (1994) Three-dimensional (CIT) LCC Calculations of RAFs for four fuels. Simulations were performed for the SCAQS episode. Yang et al. (1996) Trajectory SAPRC-90 Rate constant and exhaust composition uncertainty calculations for the RAFs from reformulated gasolines and methanol. Bergin et al. (1996) Trajectory and three-dimensional (CIT) SAPRC-90 Report on box model study described above and a three-dimensional study of the effects of rate constant and product yield uncertainties on predicted ozone impacts of five alternative fuel RAFs. Russell et al. (1995) Trajectory and three-dimensional (CIT) SAPRC-90 Evaluation of combined results of most previous studies. An economic analysis was also performed. Dunker et al. (1996) Three-dimensional (UAM) CB4 Extensive evaluation of how reformulated and alternative fuels would affect ozone formation in Los Angeles, New York, and Dallas. Tied directly to program to assess how fuel blends affect both emissions composition and emissions rates. Guthrie et al. (1996) Three-dimensional (UAM) CB4 Modeling of potential impacts of the use of three alternative fuels (CNG, M85, and RFG) in two urban areas.
OCR for page 65
--> part, because of the alternative fuel regulations promulgated in California (see Chapter 5), this issue has been explored in most detail for motor-vehicle exhaust emissions. When CARB implemented regulations for the LEV/CF6 program, it introduced the concept of reactivity adjustment factors (RAFs) to provide a mechanism for manufacturers who build vehicles powered by alternative fuels (including reformulated gasoline) to take advantage of the lower ozone-forming potential of the emissions from these vehicles. An RAF is defined as the ratio of the specific exhaust reactivities of two fuels (per gram of emission of an alternatively-fueled vehicle to that of a conventionally fueled vehicle). The specific reactivity of fuel i (SRi) is where FAi is the fraction of species i in fuel A and Ri is the MIR of species i. The RAF for fuel A is defined as the ratio of the exhaust reactivities: where FBi, is the fraction of species i in the base (reference) fuel. If the alternative fuel's RAF is less than 1, then a proportionally greater amount of VOCs can be emitted, such that the RAF times the mass of emissions meets some total emissions standard. In practice, the appropriateness of RAF—calculated using MIR values—was tested using a grid model and adjustments were made as necessary. The sources and magnitude of the uncertainties in RAFs have been investigated by a variety of investigators, including Yang et al. (1996), McBride et al. (1997), and AQIRP (discussed later). The studies of Yang 6 Aspects of this program are discussed in Chapters 4 and 5.
OCR for page 66
--> et al. (1995) and McBride et al. (1997) revealed that although the 2-σ uncertainty in the relative reactivity of individual species due to uncertainties in chemical mechanisms generally range from about 20% to 40%, that range grossly overstates the uncertainty in the composite relative reactivity of a specific emissions source. An example would be reactivities from a fleet of motor vehicles using one type of fuel versus another. In this case, much of the chemical uncertainty tends to cancel out (provided one is using relative reactivities instead of absolute reactivities), leaving an uncertainty of only a few percent. A much larger uncertainty arises from the variability and difficulty in characterizing how different vehicles respond to fuel composition changes. This is largely due to the limited amount of test data and the limited knowledge of how well a vehicle fleet is characterized by the data. This leads to substantial uncertainties in the composition of the emissions, which feed directly into the calculation of the source reactivity. The result is an uncertainty (95% confidence level) in relative reactivities for source categories such as motor-vehicle emissions of about 15-30% (Yang and Milford 1996; Bergin et al. 1998a). Reactivity for 1-HR Peak and 8-HR Averaged ozone Concentrations Another specific question under consideration is whether reactivity scales developed for a peak 1-hr ozone concentration (i.e., in accordance with the current form of the National Ambient Air Quality Standards (NAAQS)) would be significantly different from a similar scale developed for a peak 8-hr ozone concentration (i.e., the new form of NAAQS). At present there is little information to assess this issue. of relevance is a study of Khan et al. (1999) in which the authors compared the reactivities of several compounds based on their impact on the peak 1-hr and the average 8-hr ozone concentrations. The comparison is shown in Figure 3-5. Major differences were only found in the halogenated aromatics that had very small reactivities to begin with. The relative reactivities of the other species did not change appreciably. This result, albeit limited, appears to suggest that reactivity scales derived for peaks of 1-hr averaged ozone concentration should largely apply to peaks of 8-hr averaged ozone concentrations in urban areas.
OCR for page 67
--> Figure 3-5 Ratio of 8-hr average peak ozone relative reactivity to 1-hr average peak ozone relative reactivity for six solvents. Results are for a 3-clay simulation in Los Angeles. Source: Adapted from Khan et al. 1999. On the other hand, a number of caveats should be borne in mind before this result is used to justify the application of trajectory-model-derived reactivity scales based on VOC impact on peak 8-hr averaged ozone concentration. In the first place, recall that Eulerian-model-derived reactivities based on the model's predicted peak ozone concentration did not compare well with the trajectory-model-derived MIPs. Second, reactivities derived from trajectory models are typically based on very limited simulation times, and thus the use of those models to derive a peak 8-hr averaged ozone-reactivity scale is questionable. Finally, the promulgation of the new 8-hr NAAQS for ozone is likely to extend nonattainment into larger geographical regions that include rural as well
OCR for page 68
--> as urban and suburban areas (Chameides et al. 1997). Thus far, little work has been done to assess reactivities at these large, regional scales. Moreover, ozone chemistry at the regional scale and in rural areas has generally been found to be NOx-limited (OTAG 1997), where implementation of VOC emission controls and using a VOC-reactivity scale might prove to be less effective. Outstanding Technical Issues in Quantifying Reactivity The scientific and policy-making communities have made significant advances in understanding and implementing methodologies for quantifying VOC ozone-forming potential using the concept of incremental reactivity. Nevertheless, key issues remain. Among these are the uncertainties in the understanding of the atmospheric chemistry of specific VOCs, and thus in the ability to quantify their ozone-forming potential, and the variability in reactivity between different environments. It was earlier stated that ozone sensitivity to VOC can, in general, vary from place to place within a given airshed and from episode to episode. Thus, environmental variability is not limited solely to one city versus another, but also to different locations within a city and also from one time to another. Further, it is not apparent that a reactivity scale developed for high-ozone episodes will be the same as one developed for more typical conditions. Also, as was found in Los Angeles, the impact on the peak ozone concentration is not likely to be the same as the impact on ozone exposure surrogates. Another important issue relates to the role of NOx. VOC reactivity, and its use in control strategies, is of much less relevance in a system and in locations that are strongly NOx-limited. Thus, VOC reactivity should be viewed as a way of providing extra benefits to a strategy based on the implementation of VOC emissions controls. A major complication can arise, however, when a given control measure affects NOx emissions as well as VOC emissions, especially if the emission changes for the two sets of precursors are directionally different, which might be the case for reformulated gasoline using ethanol versus MTBE. Under these circumstances, one can, in principle, derive reactivities for NOx as well as VOCs to assess the net impact of the control measure on ozone. However, little
OCR for page 69
--> research has been undertaken on the derivation and application of NOx reactivities. Moreover, as implied earlier, NOx reactivities would likely be even more dependent upon location and episodic conditions than VOC reactivities. Application of NOx reactivities for a national ozone mitigation program would therefore be problematic. Finally, consideration should be given to the future use of reactivity scales for particulate matter (PM) and ozone formation. Similar to ozone, different VOCs can lead to a substantial variation in the formation of secondary particulate matter; many VOCs will form no extra secondary organic particles, but others can lead to a substantial amount. In some cases, the compounds that lead to little ozone formation lead to little PM formation, and those that have a high ozone-forming potential also can form a large amount of particles. In other cases, the opposite is true. Models for simultaneously assessing PM reactivity and ozone reactivity are still under development. Summary Ozone atmospheric chemistry involves many thousands of reactions and a similar number of compounds. The two primary precursors to ozone formation are VOCs (and Co) and NOx, although this, alone, is an oversimplification. There are hundreds of different VOCs emitted into the atmosphere, and no two have the same chemistry; thus, they each have a different impact on ozone. Further complexity comes from the fact that the atmosphere is highly variable, both in its physical and chemical make-up. Thus, not only does ozone formation respond differently to different VOC species, but it will often respond differently to the same compound in different locations or during different episodes at the same location. A variety of metrics or scales have been proposed to quantify the ozone-forming potential of an individual VOC or a mixture of VOCs arising from a specific source or type of emission. The reactivity paradigm is but one of a number of approaches that have been developed for this purpose. It is based on scientifically sound concepts and can provide a useful approach for policy-makers attempting to decide which VOCs or types of emissions to regulate and to what degree. Indeed, the state of California has already applied the reactivity paradigm to its regulation
OCR for page 70
--> of motor-vehicle emissions and the fuels used to power those vehicles. Exactly what metric should be chosen is, in part, a question of policy reflecting a set of priorities of the relevant stakeholders. Within the reactivity paradigm, a number of different scales can be used. Each one provides a measure of the ozone-forming potential of a VOC or mixture of VOCs under a specific set of conditions. In this report, the maximum incremental reactivity (MIR) scale is used as the primary quantitative measure of ozone-forming potential. That scale reflects the ozone-forming potential of VOCs under conditions where ozone control is most sensitive to decreases in VOCs and is also the scale that the state of California has proposed using for its regulatory applications. For simplicity and in the interest of brevity, the term "reactivity" is used to denote the MIR, unless stated otherwise. Moreover, reactivity is expressed in a variety of ways. The specific reactivity, derived from box modeling, is the reactivity normalized to the change of mass of VOC emissions and has units of grams of ozone formed per grams change of VOC emitted or grams of ozone per grams change of VOC. The total reactivity is obtained by multiplying the specific reactivity by the mass of VOC emissions per mile driven and has units of grams of ozone per mile. The relative reactivity is a unitless quantity which is derived by dividing the (specific or total) reactivity of a compound or class of compounds by the (specific or total) reactivity of some reference VOC, standard VOC, or VOC mixture. Sometimes the term absolute reactivity is used in this report to denote either the specific or total reactivity as a way of distinguishing them from the relative reactivity. Each of these terms is listed in Table 3-9. There are a number of limitations to the reactivity approach that should be borne in mind. Because the ozone-forming potential of VOCs can vary from locale to locale, it should not, in principle, be uniformly applied to the entire nation, except to facilitate regulatory application. Ideally, its use as a certification tool on a nationwide basis would allow for regionally-specific applications and, potentially, the development of regionally-tailored control strategies. Assessing the economic viability of implementing regionally-specific rules for certifying RFGs is beyond the scope of this report. In its current state of development, a limitation of the use of a reactivity approach beyond full certification is that it only considers the ozone-forming potential of VOCs and CO. Thus it is of less use for
OCR for page 71
--> TABLE 3-9 Terms Used in the Report to Denote Reactivitya Term Definition Units Specific reactivity Reactivity (as MIR) normalized to the change in mass of VOC emissions g O3/g change VOC Total reactivity Product of specific reactivity (as MIR) and the mass of VOC emissions per mile driven g O3/mile Absolute reactivity Either the specific or total reactivity g O3/g change VOC or g O3/mile Relative reactivity Ratio of the specific or total reactivity (as MIR) of a compound or class of compounds to that of some reference or standard VOC or VOC mixture Unitless a In this report, the term reactivity is used to denote the maximum incremental reactivity (MIR). MIR reflects the ozone-forming potential of VOCs under conditions that are most sensitive to these VOCs. developing VOC-based control strategies in areas where only NOx emission controls are needed. The reactivity approach is also of limited utility in assessing the impacts of control strategies that increase (decrease) VOCs emissions, while decreasing (increasing) NOx emissions. As it turns out, this might occur in the case of motor-vehicle emissions using specific types of RFG blends. It is also important to note that the determination of reactivities for VOCs is a computational process that requires the application of a numerical model. The types of models that can be used for this purpose range from rather simplistic trajectory or box models to very complex, three-dimensional grid-based or Eulerian models. All of those models rely on a chemical mechanism for simulating the ozone-forming process, and a variety of algorithms for representing this chemistry have been adopted. Although differences between model results do occur (for example, in the case of the reactivities of the aromatics), in general, the relative reactivity of VOCs derived from different models and models using different chemical mechanisms tend to be reasonably consistent. For this reason, it is believed that the uncertainties (or potential errors) in reactivities can be minimized by focusing on relative as opposed to absolute reactivities.
OCR for page 72
--> In general, the 2-σ (or 95% confidence level) uncertainty in the relative reactivities in most of ubiquitous VOCs (that have been studied extensively) is about 20-40%. The relative reactivity of a composite set of VOCs arising from a single source, such as motor vehicles, tends to be somewhat smaller (i.e., about 15-30%). Much of the uncertainty in this later case arises from potential errors in defining the speciation of the emissions as opposed to those associated with the chemistry of the species. For this reason, the use of relative reactivity to assess the ozone-forming potential of different sources is best suited to situations where the reactivity of the emissions is quite different. As will become apparent in later chapters, this tends to not be the case for emissions from motor vehicles using slightly different RFGs. That will, in turn, limit the ability to use reactivity to distinguish robustly between the air-quality benefits of various RFG blends.
Representative terms from entire chapter: