6
The Effects of Reformulated Gasoline On Ozone and Its Precursors

The ability to distinguish the air-quality benefits of one reformulated gasoline (RFG) blend from that of another depends, to a substantial degree, on the overall magnitude of the effect of RFG on air quality. If the RFG effect is large, then the effect of two blends of RFG might be quite discernible. If on the other hand, the RFG has a lesser effect on air quality, it is likely to be very difficult to identify which of two RFG blends is preferable from an air-quality point of view, let alone to reliably quantify these effects. As a prelude to Chapter 7, in which an attempt is made to quantify and compare the ozone-forming potential of eight different RFG blends, this chapter assesses available information on the overall impact of the RFG program on ozone and its precursors as deduced from measurements.

The steps taken in the approach to make that determination are illustrated in the "Decision Tree" depicted in Figure 6-1. The chain of inference proceeds from the tests of the emissions from a limited sample of motor vehicles in the laboratory to determinations of the influences of the use of RFGs in light-duty vehicles (LDVs) on air quality. The figure also indicates the types of findings at each step along this chain, from considerations of the currently available and published observations to the reduction of ozone concentrations and other air-quality issues. The sequence of questions addressed are listed below.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



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 131
--> 6 The Effects of Reformulated Gasoline On Ozone and Its Precursors The ability to distinguish the air-quality benefits of one reformulated gasoline (RFG) blend from that of another depends, to a substantial degree, on the overall magnitude of the effect of RFG on air quality. If the RFG effect is large, then the effect of two blends of RFG might be quite discernible. If on the other hand, the RFG has a lesser effect on air quality, it is likely to be very difficult to identify which of two RFG blends is preferable from an air-quality point of view, let alone to reliably quantify these effects. As a prelude to Chapter 7, in which an attempt is made to quantify and compare the ozone-forming potential of eight different RFG blends, this chapter assesses available information on the overall impact of the RFG program on ozone and its precursors as deduced from measurements. The steps taken in the approach to make that determination are illustrated in the "Decision Tree" depicted in Figure 6-1. The chain of inference proceeds from the tests of the emissions from a limited sample of motor vehicles in the laboratory to determinations of the influences of the use of RFGs in light-duty vehicles (LDVs) on air quality. The figure also indicates the types of findings at each step along this chain, from considerations of the currently available and published observations to the reduction of ozone concentrations and other air-quality issues. The sequence of questions addressed are listed below.

OCR for page 131
--> Figure 6-1 The Decision Tree illustrates the steps taken in an effort to quantify and compare the ozone-forming potential of various RFG blends. The figure indicates the types of findings at each step that resulted from the committee's considerations of the currently available observations that are pertinent to the reduction of ozone concentrations and other air-quality issues. When comparing different RFG blends, such as a blend containing ethanol versus a blend containing MTBE, it is desirable to account for as many differences as possible between the RFG blends. What changes in motor-vehicle exhaust emissions of VOCs, NOx, CO, or air toxics are observed in laboratory tests when RFGs are used? Have the changes in emissions from RFGs indicated by laboratory studies been observed in emissions studies using tunnels and remote sensing of tailpipe exhaust?

OCR for page 131
--> Are there data to support meaningful analysis of atmospheric data to determine the effect of RFGs? Have changes in the concentrations of air toxics or oxygenates been observed in the atmosphere and can these changes be related to the use of RFGs? Have changes in the concentrations of CO been observed in the atmosphere and can these changes be related to the use of RFGs? Have changes in the concentrations of ozone been observed in the atmosphere and can these changes be attributed to the use of RFGs? This analysis proceeds from the information concerning the measurements of exhaust and evaporative emissions from individual vehicles to the observation of the effect of those emissions on atmospheric composition. When comparing two RFG blends, it is desirable to account for as many differences as possible between the RFG blends. What Changes in Motor-Vehicle Exhaust Emissions of Vocs, NOx, CO, or Air Toxics are Observed in Laboratory Tests When RFGs are Used? Probably the most extensive single data set on the emissions of motor vehicles using RFG blends is that compiled from the Auto/Oil Air Quality Improvement Research Program (AQIRP).1 This study included over 3,000 emissions tests. In Phase I of AQIRP, different sets of 26 reformulated fuels and 2 reference gasolines were tested in fleets composed of 20 then-current (1989) LDVs (cars and light-duty trucks) and 14 older vehicles (1983-1985). Further, two methanol blends (10% and 85% methanol in gasoline) and one industry-average fuel were tested in 19 flexible-fueled and 5 variable-fuel passenger vehicles. In Phase II of AQIRP, fuels were prepared in several sets, or matrices, to study the effects of individual fuel properties: (1) the composition set tested the effects of aromatic content, olefin content, T90 (temperature at which 90% of mass of the fuel has evaporated), T50 (temperature at which 50% 1   The complete set of data for all experiments is available in reports and on CD ROM from the Coordinating Research Council, 219 Perimeter Center Parkway, Suite 400, Atlanta, GA 30346.

OCR for page 131
--> has evaporated), and the addition of methyl tert-butyl ether (MTBE)); (2) the RVP-oxygenate set tested the effects of Reid vapor pressure (RVP), as well as the addition of ethanol, ethyl tert-butyl ether (ETBE), and MTBE; (3) the methanol set tested various methanol-gasoline mixtures; and (4) the sulfur-series set tested effects of varying the sulfur content of the fuel. The properties of the RFGs used in the AQIRP compositional and sulfur tests (and those used in MTBE and ethanol blends discussed in Chapter 7 of this report) are summarized in Table 6-1. Exhaust emissions were measured from the various vehicles as they ran on a dynamometer under the Federal Test Procedure (FTP) protocol. Gas chromatographic and high-performance liquid chromatographic analyses of the exhaust emissions were made for all measurable components, including 140 structurally different hydrocarbons with from I to 12 carbon atoms, as well as ethers, methanol, ethanol, and 12 different aldehydes and ketones. Samples of exhaust emissions were segregated according to the point in the cycle of engine operation (cold start, hot stabilized, hot start, and composite) to reconstruct the emissions inventories for various vehicular operating scenarios. For some fuel-vehicle combinations, evaporative emissions were tested (modes of operation, hot soak, diurnal, and running loss). Emissions of Toxics Many of the RFG blends used in the AQIRP studies showed significantly lower total mass emissions of toxics than the industry-average gasoline. This is illustrated in the comparisons shown in Figure 6-2 for industry-average gasoline (A) and one of the RFG blends studied (C2). The comparison is made for the older fleet, the current fleet, federal Tier 1 vehicles, and vehicles with "advanced technology." With the exception of formaldehyde,2 the RFG blends showed significantly lower toxic emissions for every class of vehicle when compared to emissions resulting from the industry-average gasoline. 2   Many RFG blends appear to result in an increase in formaldehyde exhaust emissions. That is attributed to the presence of MTBE in the fuel, which can generate formaldehyde during combustion.

OCR for page 131
--> TABLE 6-1 Properties of Some of the Research RFG Blends Used in AQIRP and California Studies Codea Composition Identifierb Aromatics (vol %) Oxygenates (vol %)c Olefins (vol %) T50 (°F) T90 (°F) RVP (psi) Sulfur (ppm by wt) AQIRP Phase I A Industry average 32.0 0 9.2 218 330 8.7 339 B Certified 29.9 0 4.6 220 309 8.7 119 C AMot 43.8 15.4 (M) 3.3 213 288 8.7 284 D amOT 20.7 0 22.3 218 357 8.5 316 E AMOT 43.7 14.8 (M) 17.2 220 357 8.7 267 F* amot 20.0 0 3.2 197 279 8.8 290 G AmOt 44.3 0 17.4 214 286 8.8 317 H aMOt 20.2 14.6 (M) 20.2 168 286 8.5 312 I AmoT 42.9 0 4.1 239 353 8.9 261 J aMoT 21.4 14.9 (M) 4.0 208 356 8.6 297 K Amot 45.7 0 4.9 208 294 8.8 318 L AmOT 47.8 0 17.7 236 357 8.5 266 M aMOT 18.0 14.5 (M) 21.8 193 356 8.7 301 N aMot 21.4 13.9 (M) 5.7 164 292 8.8 294 O AMOt 46.7 14.6 (M) 19.3 204 283 8.6 288 P amOt 20.3 0 18.3 190 284 8.5 333 Q amoT 21.5 0 4.8 234 357 8.6 310 R AMoT 46.0 15.2 (M) 4.0 225 354 8.4 279 S   21.2 0 3.8 199 280 8.0 297 T   18.1 9.7 (E) 3.6 174 276 9.8 246 U   19.1 9.7 (E) 3.1 171 278 9.6 278

OCR for page 131
--> Codea Composition Identifierb Aromatics (vol %) Oxygenates (vol %)c Olefins (vol %) T50 (°F) T90 (°F) RVP (psi) Sulfur (ppm by wt) AQIRP Phase I MM   22.2 14.8 (M) 5.4 167 289 8.0 345 AQIRP Phase II C1   22.7 0 4.6 208 297 6.9 38 C2   25.4 11.2 (M) 4.1 202 293 6.8 31 Y4   24.9 10.9 (M) 1.2 201 298 9.1 44 Y5   24.3 11.1 (M) 1.3 200 299 9.0 138 Y6   24.6 10.7 (M) 1.1 200 297 8.9 258 Y7   24.9 10.6 (M) 1.1 201 299 8.8 350 Y8   24.6 10.7 (M) 1.0 201 300 8.8 443 B2   26.7 0 2.5 220 318 8.9 49 Y2   26.1 0 2.3 220 316 8.8 466 California Studies 63   23.4 11.6 (M)d 5.0 196 296 6.8 32 64   23.3 11.2 (E)d 4.8 188 297 7.8 34 a Fuel mixtures A-R are the compositional matrices for RFGs used in AQIRP Phase I; Y4-Y8 are sulfur matrices (with MTBE) from AQIRP Phase II; B2 and Y2 are from sulfur-varied fuels used in AQIRP Phase I with no added MTBE. b Composition indicator: A/a, high/low aromatics; M/m, high/low MTBE; O/o, high/low olefins; T/t, high/low T90. c Oxygenates added are indicated with letters: MTBE (M), ethanol (E). d For these two fuels, the oxygenate composition is given in mass %.

OCR for page 131
--> Figure 6-2 Comparison of the mass of exhaust toxics: acetaldehyde, formaldehyde, 1,3 butadiene, and benzene (mg/mi) from the industry-average fuel (A) and an RFG (C2), using the FTP composite. Chemicals are displayed from top to bottom as follows: acetaldehyde, formaldehyde, 1,3-butadiene, and benzene. On the x-axis, the results are divided into those for older, current, federal Tier-1-control, and advanced-technology cars. Source: Adapted from AQIRP Technical Bulletin No. 17, 1995. Emissions of VOCs The specific and total reactivity (using the MIR scale) of VOCs in exhaust, evaporative (i.e., diurnal and hot soak), and running-loss emissions from current-fleet vehicles using several of the AQIRP-tested reformulated gasolines as well as the industry-average gasoline are shown in Figure 6-3. Speciation and reactivity data on exhaust emissions were obtained from Hochhauser et al. (1992); data on evaporative emissions and running losses were obtained from Burns et al. (1992). In the case of each type of emission, the ordering of the fuels has been adjusted to show the progression of emissions from the lowest-emitting fuel to the highest-emitting fuel. In viewing these figures, it should also be borne in mind that the ozone-forming potential of VOC emissions is determined by the total mass of the emissions as well as the reactivity of the species that are emitted. The relative contribution of each of these factors can be inferred by comparing the specific and total reactivities of the emissions because the specific reactivity is a measure of the amount of ozone

OCR for page 131
--> formed per unit mass of VOC emitted and the total reactivity is the product of the specific reactivity and the mass of VOC (and CO) emitted per mile traveled (see Table 3-9). Finally, it should noted that in addition to emissions data for current fleet vehicles, AQIRP data exists for emissions from older fleet vehicles. Although the older-fleet data differ somewhat from that of the current fleet (e.g., the ordering of the fuels with increasing reactivity), the basic conclusions concerning the nature and magnitude of the emissions reductions that might be obtained from RFG do not. Inspection of Figure 6-3 indicates that rather substantial changes in the reactivity of VOC emissions can result from variations in gasoline formulation. In the case of exhaust emissions for example, the specific reactivities of the fuels tested vary by a factor of 1.4, and total reactivities by a factor of about 2 (Figure 6-3A). The variability in the reactivities of diurnal and hot soak emissions are of a similar magnitude, although the ordering of the fuels changes significantly (Figure 6-3 B and C). By comparison, the range of running-loss reactivities is considerably larger (i.e., factor of 2 variability in specific reactivity and a factor of almost 70 in total reactivity) (Figure 6-3D).3 However, the maximum reduction in the reactivity of the VOC missions obtained by switching from the industry-average formulation to the most favorable of the RFGs tested is, in each case, considerably smaller. For exhaust, diurnal, and hot soak emissions, the reduction in specific and total reactivity from the industry average is about 25% or less. In the case of running losses, the reduction is more substantial; i.e., a factor of about 2 for the total reactivity. Of course the most important parameter to consider here is the composite reactivity of all the LDV emissions; i.e., the reactivity obtained from the gases emitted by all exhaust, evaporative, and other loss processes. An example of such composite reactivities is given in Figure 6-4. In this case, composite, specific reactivities were calculated for each fuel using the AQIRP measurements of exhaust emissions (weighted for all cycles of operation), evaporative emissions, running losses, resting losses, and refueling losses from LDVs using the EMFAC-7E emissions model and the measured vapor pressures of the fuels. The relative 3   According to Burns et al. (1992), running loss emissions were measured at less than 0.2 g/test on all but two vehicles in each fleet. In the vehicles which had higher running losses, differences could be seen between fuels, but fuel effects could not be determined because of the limited data and its variability.

OCR for page 131
--> Figure 6-3 AQIRP current fleet vehicles using various RFGs (see Table 6-1) and industry-average fuel. Reactivities are expressed using the MIR scale. (A) Exhaust emissions; (B) diurnal evaporative emissions; (C) hot-soak evaporative emissions; and (D) running-loss emissions. Source: Bums et al. 1992 and Hochhauser et al, 1992.

OCR for page 131
--> Figure 6-4 Comparison of the specific reactivity (potential g O3/g VOC for the total VOC emissions) with the contribution of using industry-average fuel A and various RFGs. (TOG (total organic gas) is considered to be interchangeable with VOC.) The properties of the fuels and the compositional abbreviations shown on the x-axis are described in Table 6-1. Emissions are displayed in the bars from top to bottom as follows: refueling and storage, running losses, evaporative, and exhaust. For data represented by circles, the mass of CO emissions is not included in the denominator of the specific reactivity values plotted. The addition of CO reflects the importance of a very low reactivity compound that is emitted along with the VOCs. Source: Adapted from AQIRP Technical Bulletin No. 12, 1993. weighting of the various emissions to produce a composite emission was made to simulate the conditions present in Los Angeles, California in 1995. Here again we find substantial differences in the reactivities resulting from the fuels tested. The fuel range of reactivities from the least reactive fuel (F) to the most reactive fuel (L) is a factor of about 1.5. However, the reactivity resulting from the least reactive fuel is only about 20% less than that obtained with the use of the industry-average fuel.

OCR for page 131
--> Another interesting facet of the reactivities in Figure 6-4 relates to the role of CO. Note in the figure that the circle above each of the bars represents the specific reactivity for the appropriate fuel when the reactivity of the CO emissions is included. The average increase in the reactivities from the inclusion of CO is 18 ± 2%; the ordering of the fuels is also changed somewhat. These results dearly demonstrate the need to include CO emissions when assessing the ozone-forming potential of LDV emissions. Exhaust Emissions of NOx The AQIRP data suggest that the effect of RFG on exhaust emissions of NOx will vary depending upon the specific properties of the blend. For example, NOx emissions were lowered by 6 ± 1.9%4 by reducing olefin content from 20 to 5%, while reducing T90 from 390°F to 280°F increased NOx emissions by 5 ± 2.4%, and the impact of lowering aromatic VOC content did not have a statistically significant effect (i.e., NOx emissions were lowered by 2.1 ± 2.0%). The effect of adding oxygenates to the fuel tended to produce a small increase in NOx emissions. For example, increasing ethanol from 0 to 10% gave rise to a 5 ± 4.1% emissions increase. On the other hand, while adding 15% MTBE and 17% ETBE also resulted in an emissions increase, the increase was not statistically significant (i.e., 3.6 ± 5.4% for MTBE and 5.5 ± 6.4% for ETBE). The average of experiments with added oxygenates was a statistically significant increase of 4.8 ± 2.9%. By far the largest decrease in NOx emissions were achieved by lowering the sulfur content of the fuel. This effect is discussed in more detail in the next section. Effect of Fuel Sulfur Content of RFGs on Exhaust Emissions Dramatic changes in exhaust emissions of all ozone precursors (i.e., VOC, CO, and NOx) were obtained from the sulfur set of AQIRP tests (see Table 6-1). In these tests, the fuel's sulfur content was varied while the 4   All uncertainties are twice the standard deviations of the mean expressed as 2σ or 95% confidence levels.

OCR for page 131
--> 95%) confidence interval. Thus, while this analysis suggests that the oxygenated fuels program probably has had some small ameliorative effect on CO concentrations, its impact does not appear to be spatially uniform and in many cases is too small to discern with a high degree of statistical confidence. A very similar conclusion was reached in a report of the National Science and Technology Council (NSTC 1997). The NSTC report reviewed various studies relating to the ambient air-quality effects of oxygenated fuels. It concluded that CO concentrations in urban areas have been decreasing at a rate of 2.8% per year for the last 10 years. This decrease is attributable primarily to EPA-mandated motor-vehicle emissions standards and improved vehicular emissions control technology. However, the NSTC report concluded that the benefits of oxygenated fuels on ambient air quality in cold climate areas could not be confirmed. (See Anderson et al. (1994) for additional information on the influence of oxygenated fuels on ambient CO.) Have Changes in the Concentrations of Ozone Been Observed in the Atmosphere and Can These Changes be Related to the USE of RFGs? Assessing the effects of RFG on ambient ozone air quality involves challenges similar to those discussed above for CO. For example, Larsen and Brisby (1998) attempted to assess the effect of California's cleaner-burning gasoline program on ozone concentrations. In that study, for the Sacramento, South Coast, and San Francisco Bay areas, Larsen and Brisby reported ozone decreases of 14%, 17%, and 4%, respectively. However, the contribution of cleaner-burning gasoline to this decrease is uncertain because of the presence of many other ongoing ozone-mitigation efforts. To address this problem, Larsen and Brisby assumed that the contribution of the cleaner-burning fuels program to the overall ozone decrease was proportional to the estimated percent reduction in the precursor emission inventory resulting from the program. Thus, even though the Larsen and Brisby study was based on ambient ozone concentrations, the attribution of a portion of the observed ozone decrease to the use of cleaner-burning gasoline was derived from an emission inventory and does not constitute empirical verification of program effectiveness.

OCR for page 131
--> To further illustrate some of the difficulties with applying trend analysis to ambient ozone data, consider the log-transformed ozone concentrations from Riverside, California, presented in Figure 6-11.6 As in the CO analysis, the data are decomposed into its long-term, seasonal, and short-term components. Because the information from the moving-average filter (Zurbenko et al. 1996) used here is not reliable at the beginning and at the end of the time-series, data for the first and last years are not included in these figures. At this site, the long-term, seasonal, and short-term components contribute about 2%, 63%, and 34%, respectively, to the total variance of the ozone data. To examine whether the use of RFGs in California had an impact on ambient ozone concentrations, data during the 1980-1997 period from several locations in the Los Angeles Air Basin of California were also analyzed. As was the case for the CO analysis in the previous section, an overall downward trend in ozone over the past 15-year period is evident in the long-term component at Riverside (Figure 6-11D). Between 1981 and 1996, ozone has decreased by about 30% at Riverside; the largest decrease of about 20% in ozone concentrations occurred between 1989 and 1993. Ozone then increased slightly in 1994, and then decreased again in 1995. Whereas the oxygenated fuels program was implemented in California in 1992, the RFG program was implemented in 1996. Figure 6-11 indicates the presence of a strong downward trend in ozone before these programs were implemented. Unfortunately, data for the time period after the RFG program was implemented are not yet available for this type of analysis to dearly discern the impact of this control strategy on ozone air quality. For example, if an abrupt change of 10% in the middle of ozone time-series data illustrated in Figure 6-11 were introduced, it would contribute only about 0.5% to the total variance. This illustrates that the detection of any abrupt change of the order of 10% or less and its attribution to a specific control of an emission is a formidable task. These results demonstrate the difficulty in linking a particular emissions-control policy to a change in ozone concentrations. Clearly, the problem of assessing the effectiveness of a particular air-pollution control program requires further development. 6   The rationale for using the log-scale for ozone was discussed by Rao et al. (1997).

OCR for page 131
--> Figure 6-11 Daily maxima of ozone concentrations at Riverside, CA, from 1980 to 1997 (A). Three components of the overall trend are seasonal (B), short term (C), and long term (D). Documentation of RFG Effects in a Future Observational Program On January 1, 2000, federal Phase II reformulated gasoline (RFG) will be required in commercially available LDVs operating in areas classified as being in severe nonattainment of the National Ambient Air Quality Standard (NAAQS) for ozone. On the basis of estimates from the Com-

OCR for page 131
--> plex Model, EPA expects that this action will result in reductions in both exhaust and evaporative emissions of VOCs and some air toxics from LDVs, as well as LDV exhaust emissions of CO and NOx. It is further believed that these emissions reductions will help alleviate the severity of the ozone pollution in the severe nonattainment areas where the program is to be implemented, although, for the reasons discussed above, these effects are not expected to be large or even observable. Will the projected air-quality benefits of Phase II of the federal RFG program be met? As with any regulatory program, the committee recommends that a complete and comprehensive RFG program should include—part and parcel—a plan for documenting the impact of the program and assessing to what extent the expected benefits are realized. The committee further recommends that this plan be organized around addressing a progression of three scientific questions7 that attempt to document the effect of Phase II RFG on ozone precursor compounds and their ozone-forming potential. (Ideally, such a plan would include a fourth question that addresses the effect of the Phase II RFG on ozone concentrations. However as discussed above, it is unlikely that such a signal in ambient ozone concentrations could be discernible given the relatively large variability in ozone, the myriad factors that affect ozone concentrations, and the rather small overall impact RFG is projected to have on ozone.) The three questions recommended here for consideration are briefly discussed below. Question 1: Do in-use Phase II RFG blends decrease the emissions from LDVs? This first question can be addressed in much the same way that the potential air-quality benefits of RFG were initially assessed in studies such as the AQIRP and California Ethanol Testing Program (see Chapter 7). Representative vehicles can be selected and then subjected to emissions tests using dynamometers, etc. In this case, however, actual, in-use Phase II RFG would be used instead of prospective RFG formulations. Fungibility issues, such as that related to in-tank blending of RFGs, could then, in principle, be directly tested and assessed. 7   These questions tend to mirror the progression of questions included in the Decision Tree in Figure 6-1.

OCR for page 131
--> Question 2: Are changes in emissions resulting from the use of Phase II RFG blends observable under driving conditions? Although measurements of LDV emissions in a laboratory setting are informative, they do not necessarily represent the emissions of LDVs in operation under actual driving conditions. Confirmation that laboratory-measured emission reductions also occur on the road can be obtained through tunnel studies and remote sensing of tailpipe emissions. As noted earlier in this chapter, these measurements characterize LDV emissions under a limited set of conditions and, as such, do not comprehensively quantify LDV emissions. Nevertheless, they do provide a real-world test of the emissions and as such are an important step in linking laboratory-measured LDV emissions to an ambient concentrations signal. Question 3: Are changes in emissions resulting from the use of Phase II RFG blends observable as a signal in the ambient concentrations of ozone precursor compounds? Establishing the connection between changes in LDV emissions and the ambient concentrations of the compounds contained in those emissions is a more-formidable task. The most-straightforward approach for accomplishing such a task is through the use of time-series analyses of a long-term record of ambient concentrations of VOC, CO, and NOx to isolate a signal that can be associated with Phase II RFG. However, this approach presents a variety of challenging problems. The time-series record must encompass a period significantly before as well as after initiation of Phase II RFG and the data set must include highly accurate and precise measurements. Even under those circumstances, identification of a shift in the time series of the quantity of interest due to RFG can be obscured by other transient factors (e.g., meteorological variations or. implementation of other emissions control programs). Therefore, there is a need to develop and evaluate techniques for detecting ambient effects of a control program separately from the effects of meteorological variability. For those reasons, it is recommended that an alternative approach be taken to document the effect of Phase II RFG usage on ambient precursor concentrations. This alternate approach would be to use measurements of various tracers in conjunction with measurements of

OCR for page 131
--> VOC, CO, and NOx to (1) characterize the contributions of LDV emissions to the concentrations of ozone-precursor compounds; (2)estimate the ozone-forming potential of these compounds through the application of various observation-based methods (e.g., Cardelino and Chameides 1995); and (3) document the change in this contribution that can be attributed to the use of RFG. Tracer species that would be useful in this regard include those that could be used to identify LDVs emissions (e.g., acetylene for LDV exhaust), as well as those that could serve as a fingerprint of emissions from LDVs using RFG (e.g., MTBE). These measurements would ideally be made in a variety of locations within and surrounding each severe nonattainment area to document effects occurring on regional scales as well as local or urban scales. Especially important in this regard would be the enhancement of monitoring capabilities in rural areas of the United States. Summary The first investigation in this chapter focused on determining if changes in ozone precursors (NOx or VOCs, CO, air toxics, and oxygenates) have been observed in the emissions studies done on individual vehicles tested under controlled conditions in the laboratory. The most comprehensive study undertaken to date of the effects of varying gasoline properties, the Auto/Oil Air Quality Improvement Research Program (1989-1995), indicated that substantial ozone-precursor emissions reduction benefit should be achieved by RFG. Decreases in the ozone-forming potential (as measured by the MIR scale) of emissions from LDVs of as much as 20% appear to be possible. The most dramatic effects on ozone-precursor exhaust emissions seen in the various gasoline compositional matrices studied were those due to lowering the fuel's RVP and the amount of sulfur-containing compounds. Only slight reductions, less than 10%, in the CO and VOC emissions can be ascribed to the addition of either MTBE or ethanol. The second investigation focused on determining if changes in NOx or VOCs, CO, air toxics, and oxygenates, have been observed in the emissions studies done in tunnels or from remote sensing of exhaust. From a qualitative point of view, these studies appear to be consistent with the laboratory tests. Reductions were observed in the LDV emissions of NOx, VOCs, CO, and various toxics, and they appear to be at

OCR for page 131
--> least partially attributable to the introduction of RFGs. Formaldehyde emissions were found to increase—most likely from the combustion of MTBE. These studies also indicated that high-emitting vehicles are responsible for a disproportionate share of the VOC and CO emissions. The tunnel studies and remote-sensing measurements also indicated that the addition of oxygenates to fuel substantially reduced the emissions of CO and VOCs from these high emitting vehicles, perhaps because these high-emitters are operating with faulty or nonfunctioning catalytic converters. However, the data from these studies could not be used to discern the relative air-quality benefits of fuels using MTBE or ethanol. The third and final investigation sought to identify RFG effects in the atmosphere by analyzing ambient data. Such an undertaking is easily confounded by competing and offsetting interferences (e.g., meteorological variations and the existence of other contemporary control programs), and statistically significant trends specifically attributable to the RFG program could not be identified. Several areas of the country have seen significant improvements in air quality, including reductions in ambient CO and ozone concentrations. In the case of CO, it appears that some portion of the decrease can be attributed to the addition of oxygenates to fuels but the magnitude of the oxygenate effect is not spatially uniform and in some areas is too small to discern with statistical confidence. In the case of ozone, it is not clear if any portion of the concentration decrease can be directly associated with the addition of oxygenated compounds to motor fuel or the development and use of RFG. Thus, it would appear that RFGs have an impact on ozone-precursor emissions from LDVs by reducing both the mass and ozone-forming potential of these emissions. However, discerning a statistically significant effect of RFGs on ambient ozone concentrations has thus far proven to be quite difficult. This is most likely because ambient ozone concentrations tend to be quite variable from year to year and the RFG program is but one of a multitude of ozone-mitigation programs underway in the nation whose impact on ozone is of a similar or larger magnitude. Thus, air-quality models—which are themselves subject to significant uncertainty—present the only avenue for estimating the magnitude of the effect of RFG on ozone concentrations. As described in Text Box 6-1, simulations using these models indicate that the overall reduction in ozone from the implementation of the RFG program is likely to be a few percent. This finding should not be interpreted to mean that RFG use is

OCR for page 131
--> Text Box 6-1 Model Predicted Effects of RFG On Ground-Level Ozone Laboratory tests and tunnel studies suggest that the use of RFGs in LDVs lowers the ozone-forming potential (as measured by the MIR scale) of an individual vehicle's emissions using an RFG blend with the lowest MIR by about 20% (see Rigure 6-4). Yet, analysis Of ambient data is unable to identify a discernible impact on ground-level ozone concentrations. Does that indicate an inconsistency or gap in our understanding of the processes that lead to the formation and accumulation of ozone pollution? Not necessarily. In the first place, ozone concentrations generally do not respond in a linear fashion to decreases in VOCs (see discussion in chapter 2). Moreover, emissions from LDVs represent only a fraction of the total VOC emissions in an airshed, Thus, it might be expected that the effect on ambient ozone of a ˜20% decrease in the reactivity of motor-vehicle. emissions would be considerably less than 20%. A more quantitative assessment of the probable impact Of RFGs on ozone can be made using air-quality models. One could ask, Are changes in emissions resulting from the use of RFG blends observable in air-quality models, and has the performance of those models been evaluated? A version of the gridded Urban Airshed Model was exercised as part of the AQIRP Study to do just such an assessment (AQIRP 1997a). In this study, the Urban Airshed Model was used to simulate ozone Concentrations when different RFG fuels were used for conditions typical of Los Angeles, New York, and Chicago-Milwaukee. Simulations were first carried-out for a historical ozone episode in each metropolitan area (Los Angeles, August 26-28, 1987; New York, July 9-11, 1988; and Chicago-Milwaukee, June 24-28, 1991). RFG effects on ozone were then estimated using the same meteorological conditions that occurred during the historical episode and emissions projections for 2000 and 2010 that included the emissions reductions for motor vehicles predicted by the data from the Auto/Oil study. Table 6-6 lists the predicted change in peak ozone for each simulation for changes in T50, T90, and sulfur content of the fuels. As might be expected, lowering these fuel properties does in fact lead to a decrease in peak ozone concentrations. However, the ozone decrease is quite small—about 1 part per billion by volume (ppb) or less—although in many cases still

OCR for page 131
--> An independent model assessment of the impact of the federal RFG program was carried out by the New York State Department of Environmental Conservation using the emissions inventory prepared by the Ozone Transport Assessment Group (OTAG). The study involved a regional-scale application of the Urban Airshed Model (UAM-V) with a model. domain covering much of the eastern third of the continent. The regions where the RFG program was implemented during 1995 is presented in Figure 6-12. A comparison of model simulations of a multi-day ozone episode during July 7-18, 1995, with and Without the RFG program indicates ozone decreases up to 3 ppb over Chicago, Lake Michigan, and along the northeastern corridor (see Figure 6-13). Of course it should be recognized that air-quality models simulations are themselves uncertain because of the uncertainties in both the algorithms (e.g., the chemical mechanisms) and the input data (e.g., the emission inventories) used to run the models. Even recognizing these Uncertainties, it seems unlikely that the RFG program could result in ozone decreases of more than 10 ppb. For example, even if the mobile source emissions used in the model simulations were underestimated by a factor of 2, the maximum ozone decrease would probably be less than 10% at most because peak ozone concentrations generally respond nonlinearly to Changes in ozone precursor concentration. Thus, model simulations predict that RFG has a beneficial effect of a few percent On Overall Ozone concentrations. It is therefore not surprising · that discerning an RFG-signal in the ambient ozone data has proven to be. difficult. It also suggests that it will be difficult to discern the impacts of two RFG blends with subtle differences in their properties. This issue is addressed as a case study in Chapter 7. ineffective. As noted earlier, reduction of RVP in gasoline prior to the RFG program is thought to have had a significant air-quality benefit. As discussed in the next chapter, such a reduction size limits the ability to document the benefits of RFGs and to reliably distinguish between the ozone-forming potentials of different RFG blends.

OCR for page 131
--> TABLE 6-6 Predicted Effects on Peak Hourly Ozone Concentrations Expected Due to Changes in Certain Fuel Composition Variables in Three Cities As Estimated by Using the Urban Air Shed Modela   Change in Fuel Variableb City, Year, Episode Dayc T50d(215ºF to 185ºF) T90(325ºF to 280ºF) Sulfur (320 to 35 ppm)   Change in Peak Ozone (ppb) from That of the Historical Episode Los Angeles, August 28, 1987       2000 -0.3 ± 0.3* -0.9 ± 0.3*   2010 -0.1 ± 0.2 -0.1 ± 0.2   New York, July 11, 1988       2000 -0.1 ± 0.1 -0.4 ± 0.1* -0.4 ± 0.1* 2010 0.0 ± 0.1 0.1 ± 0.1 -0.4 ± 0.1* Chicago-Milwaukee, June 26, 1991       2000 -0.8 ± 0.7* -1.2 ± 0.9* 0.0 ± 0.9 2010 -0.2 ± 0.7 -1.0 ± 0.8* 0.4 ± 0.8 Chicago-Milwaukee, June 27, 1991       2000 -0.6 ± 0.5* -0.9 ± 0.7* -0.2 ±0.7 2010 -0.1 ±0.4 -0.5 ± 0.4* 0.1 ± 0.4 Chicago-Milwaukee, June 28, 1991       2000 -0.3 ± 0.2* -0.5 ± 0.3* -0.2 ± 0.3 2010 -0.1 ± 0.2 -0.3 ± 0.2* 0.0 ± 0.2 a The predicted effects may not be reflective of the greatest change in gasoline composition such as changes from the late 1980s to when California Phase 2 RFG began to be used. b Main effects are shown with 95% confidence intervals. An* denotes statistically significant effects. c Data from the location and date that was used to establish meteorological conditions employed in each simulation. d The effects of T50 on ozone may be underestimated because only the effects on emissions from lower exhaust emitters are included. The effect of T50 on emissions from higher emitters could not be estimated from the available data and are assumed to be zero. Source: AQIRP Technical Bulletin No. 21, 1997a.

OCR for page 131
--> Figure 6-12 The areas where the RFG program was implemented during 1995. Figure 6-13 Maximum change in ozone from RFG as predicted by the UAM-V model for July 7-18, 1995 episode.