ARTICLE IN PRESS
Atmospheric Environment 42 (2008) 4498–4516 www.elsevier.com/locate/atmosenv
Emissions from light duty gasoline vehicles operating on low blend ethanol gasoline and E85 Lisa A. Graham, Sheri L. Belisle, Cara-Lynn Baas1 Emissions Research and Measurement Division, Environmental Science and Technology Centre, Environment Canada, 335 River Road, Ottawa, Ontario, Canada K1A 0H3 Received 19 September 2007; received in revised form 24 January 2008; accepted 29 January 2008
Abstract The results of two recent vehicle emission studies are described in this paper, along with a statistical analysis of the changes in tailpipe emissions due to the use of ethanol that includes the results from these two studies in combination with results from other literature reports. The first study evaluates the effect of two low blend ethanol gasolines (E10, E20) on tailpipe and evaporative emissions from three multi-port fuel injection vehicles and one gasoline direct injection vehicle at two different test temperatures. The second study evaluates the differences in tailpipe emissions and fuel consumptions of paired flexible fuel and conventional gasoline vehicles operating on California RFG Phase 2 and/or E85 fuels at 20 1C. The vehicles were tested over the four-phase FTP or UDDS and US06 driving cycles. Tailpipe emissions were characterized for criteria pollutants (CO, NOX, NMHC, NMOG), greenhouse gases (CO2, CH4, N2O), and a suite of unregulated emissions including important air toxics (benzene, 1,3-butadiene, formaldehyde, acetaldehyde, acrolein), and ozone reactivity. In the low blend ethanol study, evaporative emissions were quantified and characterized for NMHC. While contradicting, results can be seen among the various literature reports and with these two new studies, the statistical analyses of the aggregated data offers much clearer pictures of the changes in tailpipe emissions that may be expected using either low blend ethanol gasoline (E10) or E85. The results of the statistical analysis suggest that the use of E10 results in statistically significant decreases in CO emissions (16%); statistically significant increases in emissions of NMHC (9%), NMOG (14%), acetaldehyde (108%), 1,3-butadiene (16%), and benzene (15%); and no statistically significant changes in NOX, CO2, CH4, N2O or formaldehyde emissions. The statistical analysis suggests that the use of E85 results in statistically significant decreases in emissions of NOX (45%), NMHC (48%), 1,3-butadiene (77%), and benzene (76%); statistically significant increases in emissions of formaldehyde (73%) and acetaldehyde (2540%), and no statistically significant change in CO, CO2, and NMOG emissions. Crown Copyright r 2008 Published by Elsevier Ltd. All rights reserved. Keywords: Ethanol–gasoline blends; Mobile source air toxics; Greenhouse gas emissions; Evaporative emissions
1. Introduction Corresponding author. Tel.: +1 613 990 1270;
fax: +1 613 952 1006. E-mail address:
[email protected] (L.A. Graham). 1 Present address: Industry Canada, 50 Victoria St., Gatineau, Quebec, Canada, K1A 0C9.
Today, the motivations for blending ethanol with gasoline are different in Canada and the USA. However, these motivations have evolved over the last 15 years from a common starting point: reducing the negative impacts on local and regional
1352-2310/$ - see front matter Crown Copyright r 2008 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2008.01.061
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
air quality that arise from gasoline-powered vehicle emissions, specifically ground level ozone, toxic air pollutants, and carbon monoxide. In 1992, the US Clean Air Act implemented a wintertime oxygenated fuels program for cities with elevated ambient concentrations of CO during the cold months. This program required 2.7% by weight of oxygen in gasoline and the oxygenate of choice for this program was ethanol (US EPA, 2007a). In Canada, similar fuels were available, but were not mandated as in the USA. In 1995, the US Federal Reformulated Gasoline (RFG) regulations were introduced. RFG is required by the US Clean Air Act in cities with the worst smog pollution and other cities with smog problems may choose to use RFG. RFG is currently used in 17 states and the District of Columbia and about 30% of gasoline sold in the USA is reformulated (US EPA, 2007b). The RFG regulations established emissions performance requirements for gasoline and required an oxygen content of at least 2% by weight. This regulation also served to enhance energy security by extending the gasoline supply with the use of domestically produced and renewable energy sources. California has RFG regulations that are more stringent than the federal requirements and are intended to achieve similar goals. The oxygenate of choice in federal RFG was MTBE (methyl-t-butyl ether) (US EPA, 2007b) which accounted up to 87% of the oxygenate use. Ethanol was also used in RFG. Given the environmental concerns that have emerged concerning the detection of MTBE in groundwater and drinking water in the United States (McCarthy and Tiemann, 2001), many states have banned the use of MTBE in RFG (US EPA, 2007c). In February 2006, the federal RFG regulations were amended to remove the requirement for oxygenate content (US EPA, 2007d). With the US Energy Policy Act of 2005, energy security (reducing the dependence on foreign suppliers) has become the primary motivator for the use of ethanol in gasoline, as it can be produced domestically (US EPA, 2007f). Canada has regulations controlling the level of sulphur and benzene in gasoline (Environment Canada, 2007a). In December 2006, the government announced that it intends to mandate a 5% by volume renewable fuel content based on the gasoline pool, starting in 2010 (Environment Canada, 2007b). The primary environmental motivation for the use of renewable fuels in gasoline in Canada is to achieve a reduction in lifecycle greenhouse gas emissions. Economic motivators for Canadian
4499
farmers and rural communities are equally as important as the Canadian bio-economy grows. Low blend ethanol gasoline results in changes in some vehicle tailpipe emissions. While individual studies often showed contradictory results, generally, emissions of CO are reduced, effects on NOX, NMHC and some air toxic emissions are generally minimal, while emissions of formaldehyde and acetaldehyde are increased compared to traditional gasoline. Higher ethanol blends, up to 85% ethanol in gasoline, have been available on a limited basis in both Canada and the USA for many years. Since these fuels cannot be used in conventional vehicles for several reasons, specially equipped flexible-fuel vehicles are required (US EPA, 2007e). Interest in these higher blends has increased very recently with the new US Energy Policy Act of 2005, and the Renewable Fuel Standard which took effect in September 2007. This program is designed to significantly increase the volume of renewable fuel that is blended into gasoline with the primary motivation of energy security. Renewable fuels also offer lifecycle greenhouse gas emission reduction opportunities. As seen with low blend ethanol studies, individual E85 studies show conflicting results, but generally, E85 results in emission reductions for NOX, 1,3butadiene, benzene, and NMHC, emission increases for formaldehyde and acetaldehyde and little change (sometimes positive, sometimes negative) in CO and NMOG emissions. This paper begins by presenting selected results of two recent studies conducted at Environment Canada’s Emissions Research and Measurement Division. The first study (Baas and Graham, 2006a, b, c) was conducted over 2 years (2004–05) and involved four light duty gasoline vehicles of different technologies operating on low blend ethanol gasolines at two test temperatures (20 and 10 1C). The tailpipe emissions were characterized for criteria pollutants (CO, NOX, NMHC, NMOG), greenhouse gases (CO2, CH4, N2O), and particulate matter (PM2.5). Detailed speciation was completed for non-methane hydrocarbons, carbonyl compounds, vapor phase organic acids, sulphur dioxide, ammonia, polycyclic aromatic hydrocarbons, and particulate matter (organic and elemental carbon, particle phase organic and inorganic ions). Evaporative emissions at 20 1C were also quantified and characterized for NMHC. Only the criteria, greenhouse gas, and selected air toxic emissions are
ARTICLE IN PRESS 4500
L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
presented in this paper. The second study (Belisle and Graham, 2006), conducted in November and December of 2005, involved paired flexible fuel and conventional vehicles with the objective to evaluate changes in emissions due to flexible fuel vehicle (FFV) operation on gasoline and E85 as compared to similar conventional vehicles operating on gasoline. Emissions characterization included criteria pollutants (CO, NOX, NMHC, NMOG) and greenhouse gases (CO2, CH4, N2O). Detailed speciation was completed for NMHC and carbonyl compounds. Since the results of individual studies often contradict one another, we also present statistical analyses of the results of these two studies, combined with literature results, in order to present a clearer picture of potential emissions changes due to use of either low ethanol gasolines or E85. 2. Methodology Many of the test procedures used in these two studies are common and are summarized below. Differences between the two test programs, where they exist, are noted. 2.1. Tailpipe emissions Emissions measurements were conducted using the chassis dynamometer testing facility at the Emissions Research and Measurement Division of Environment Canada. The procedures used are detailed in the US EPA Federal Code of Regulations, Schedule 40 Part 86. For the low blend ethanol study, a four-phase implementation of the FTP was used rather than the usual three-phase FTP in order to facilitate particulate matter sample collection. The FTP is representative of a nondemanding style of urban driving and allows for comparison of cold- and hot-start emissions. Emissions were also characterized on the US06 cycle to capture emissions during aggressive and high speed driving. For the E85 study, emissions were characterized for each phase of the standard three-phase FTP cycle, with a 10-min soak between Phases 2 and 3. Samples for determining emissions of CO, CO2, NOX, and total hydrocarbons (THC) were collected on a per phase basis. For each dilute exhaust sample collected, a corresponding dilution air sample was collected. Dilute exhaust samples for determining methane, nitrous oxide, ethanol, and for speciation of NMHC were collected on a per phase basis. A single dilution air sample
was collected over each test (FTP or US06). For the carbonyl compound analysis, dilute exhaust samples were collected on a per phase basis and one dilution air sample was collected over each sampling day. 2.2. Evaporative emissions For the low blend ethanol study, evaporative emissions were measured over the 1 h diurnal heat build and hot soak cycles. The diurnal heat build cycle simulated non-running emissions released as fuel in the vehicle expands as a result of increases in ambient temperature. The hot soak cycle simulates non-running emissions released after the vehicle has been running for a period of time. Although the vehicle has been turned off, residual heat from the engine continues to heat the fuel system components, causing evaporative emissions. Samples for determining evaporative emissions were automatically taken from the SHED at the end of each cycle. These samples were immediately directed to the automated analyzer. Separate samples for determining evaporative emissions of NMHC and ethanol were drawn from the SHED and collected in TedlarTM bags. 2.3. Analytical methods All analytical methods used in these studies are accredited to ISO 17025 standards. Concentrations of CO, CO2, NOX, and THC were determined with standard test cell analyzers (non-dispersive infrared, chemiluminescence, and flame ionization). Concentrations of ethanol were determined using an Innova Model 1312 Photoacoustic Multi-Gas Analyzer following procedures similar to Loo and Parker (2000). Samples for determining carbonyl compounds were collected on Sep-Pak silica cartridges coated with 2,4-dinitrophenylhydrazine (DNPH). The samples were extracted and analyzed by reverse phase high performance liquid chromatography using an Agilent 1100 Series Liquid Chromatograph with an ultraviolet–visible (UV–vis) diode array detector. A total of 18 carbonyl compounds (C1–C8) were determined using this method. Approximately, 160 NMHC were determined using a Hewlett-Packard 6890 gas chromatograph (GC) with a flame ionization detector (FID). An Entech M7000 cryogenic concentrator was used for sample concentration and introduction.
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
4501
Methane was determined and confirmation of the C2 and C3 hydrocarbons was accomplished by simple gas loop injection onto a capillary column. A Hewlett-Packard 6890 gas chromatograph equipped with a gas sampling valve and a FID was used for the analysis. Nitrous oxide was determined using a HewlettPackard 5890A Series II GC with an electron capture detector. The detection limits for each method and sample type are summarized in the Supplementary material along with details of the analytical methods.
have matching octane number, fuel sulphur content, and vapor pressure within each seasonal grade. For each seasonal grade, the test fuels included a base fuel containing no ethanol, a 20% ethanol tailor blend, a 10% ethanol tailor blend, and a 10% ethanol splash blend. The splash blend fuels were made by simply ‘‘splash’’ blending a volume of ethanol with the base fuel, resulting in lower sulphur, higher octane, and higher vapor pressure than the base fuel. Selected fuel properties are presented in Table 1. Complete fuel specifications are given in the Supplementary material.
2.4. Low blend study
2.4.3. Test procedure As the vehicles used for testing were in-use and potentially exposed to gasoline with sulphur content higher than the test fuels, it was necessary to perform a conditioning sequence on each vehicle to remove residual sulphur from the catalytic converter. The procedure involved running the vehicle at a rich air/fuel ratio and at a high catalyst temperature to facilitate the formation of hydrogen sulphide from the residual sulphur on the catalyst and is described in more detail in Durbin et al. (2003). The charcoal canister of the vehicle collects evaporative hydrocarbon emissions during the SHED tests, which are then purged into the engine while driving. These canisters are never fully purged but maintain a fixed amount of trapped vapor called the canister ‘‘heel’’. This presented a problem because of the possibility of carryover of fuel vapors. To mitigate this problem, two new OEM canisters were purchased for each vehicle at the beginning of the program and seasoned using the summer grade E0 fuel. The 20 1C testing began using the first canister, and the fuels were tested in ascending ethanol content starting with the base
2.4.1. Test vehicles Three multi-port fuel injected (MPFI) vehicles representing different technologies (emission standards) and one gasoline direct injection (GDI) vehicle were tested. These vehicles were a 1998 Ford Escort ZX2 (US EPA Tier 1 emission standard, approximately 80,000 km), a 2001 Nissan Sentra CA (California SULEV zero evaporative emission standard, approximately 12,000 km), a 2003 Dodge Caravan (US EPA LEV emission standard, flexible fuel, approximately 25,000 km), and a 2000 Mitsubishi Dion Exceed (Japanese LEV emission standard, not currently sold in North America, approximately 25,000 km). The Escort and the Sentra were tested at 20 and 10 1C. The Caravan and the Dion were tested at 20 1C only. Tests were performed using four summer grade fuels (for tests at 20 1C) and four winter grade fuels (for tests at 10 1C). 2.4.2. Test fuels The test fuels were blended for this project by Halterman Fuels of Texas and were designed to Table 1 Low blend ethanol gasoline study fuel properties Summer grade fuels
1
Specific gravity (kg L ) Net heating value (BTU lbm1) Fuel fraction oxygen Sulphur content (ppm) Benzene (vol%) Total aromatics (vol%) Motor octane number RVP (psi)
Winter grade fuels
S-E0
S-E10
S-E10-Spl
S-E20
W-E0
W-E10
W-E10-Spl
W-E20
0.705 18,927 0 34 0.1 7.9 86.0 8.8
0.725 18,127 0.036 34 0.5 11.0 85.0 8.6
0.717 18,182 0.036 31 0.7 8.1 89.0 9.4
0.734 17,319 0.073 35 0.8 14.1 85.7 8.7
0.693 18,975 0 33 0.5 8.3 85.0 13.4
0.726 18,096 0.036 33 0.5 12.6 84.3 13.1
0.705 18,200 0.037 26 0.4 7.3 89.5 13.8
0.714 17,494 0.073 27 0.3 6.3 90.0 13.2
ARTICLE IN PRESS 4502
L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
fuel. Before the repeat 20 1C tests were performed, the first canister was replaced with the second seasoned canister. The repeat base fuel tests were therefore performed with identical canister conditions as the initial base fuel tests. To prepare for the 10 1C testing, the first canister was purged continuously for approximately 4 weeks alternately with pressurized clean air and under vacuum to remove as much of the canister heel as possible. The 10 1C testing began using the second canister and the fuels were tested in ascending ethanol content starting with the base fuel. Before the repeat 10 1C tests were performed, the first canister was re-installed. Although it had been exposed to ethanol fuels from the 20 1C testing, it was hoped that the purging process removed most of the ethanol contamination making the initial and repeat base fuel tests as similar as possible with regards to canister conditions. The vehicle preparation and test sequence that was followed is provided in the Supplementary material. 2.5. E85 study The objective of this project was to determine whether FFVs, when run on regular gasoline, produced increased emissions as compared to operation on E85 and as compared to conventional vehicles of the same make and model operating on gasoline. 2.5.1. Test vehicles Two 2002 Chrysler Caravans meeting California LEV 1 LDT emission standards, each with approximately 50,000 km; and two 2004 Chrysler Sebrings meeting California ULEV 1 emission standards, each with approximately 50,000 km were selected for the study. These vehicles were obtained from federal fleet operators in the Ottawa, Canada region. Upon receipt of the test vehicles, it was determined that the conventional Sebring did not meet the same emission standards as its flexible fuel counterpart and therefore would not render valid comparisons. A suitable replacement vehicle could not be found, so the project was continued using only three vehicles. 2.5.2. Test fuels The three vehicles were tested on the current certification gasoline (California RFG Phase 2 certification fuel). The two FFVs were also tested on a commercial E85 blend obtained from a local
Table 2 E85 study fuel properties
Specific gravity (kg L1) Net heating value (BTU lb1) Fuel fraction oxygen Sulphur content (ppm) Benzene (vol%) Total aromatics (vol%) Research octane number RVP (psi)
E85
California RFG Phase 2
0.784 13,867 0.294 17 n/a 6.6 104 7.3
0.743 18,132 0.018 (MTBE) 37 0.8 26.2 96.8 5.7
distributor. The fuel parameters are summarized in Table 2. 2.5.3. Test procedure The vehicles were fuel exchanged to the certification fuel and the evaporative emissions control systems were purged with butane as prescribed in the FTP procedure. The vehicles were driven over two repeats of the LA4 cycle, fuel exchanged again and driven over two more repeats of the LA4 cycle. The conventional vehicle was ready for emissions testing at this point. Being designed to be capable of running on ethanol–gasoline blends of up to 85% ethanol, the FFV fuel system incorporates a fuel composition sensor that measures ethanol content in the fuel. This information is used to adjust the engine parameters to best suit the fuel blend. The status of this sensor can be surveyed through the OBD II system. It was observed in other testing at the laboratory that the fuel sensors took longer than expected to report the correct ethanol reading, although according to one vehicle manufacturer representative, the sensor should nearly instantly sense the fuel composition. At the beginning of the study, additional mileage accumulation was performed on the two FFVs until the fuel sensor read the correct fuel composition (zero ethanol). After a final fuel exchange and overnight soak, three repeats of the FTP were conducted with emissions characterization over consecutive days. The two FFVs were then fuel exchanged to E85 and immediately tested for emissions while the fuel sensor was reading incorrectly. After fuel exchanges and mileage accumulation on E85, with periodic checks to ensure the fuel sensor was reading the fuel composition correctly, the FFVs were again tested
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
for three repeats of the FTP with emissions characterization on consecutive days. 3. Results 3.1. Low blend ethanol study The complete set of measured emission rates is provided in the Supplementary material. ANOVA and regression analyses were used to identify and evaluate changes in emissions due to fuel ethanol content. For the regression analysis, the emission rate was plotted as a function of fuel ethanol content and the slope of the regression line and its 95% confidence limits were determined. If the confidence limits did not include zero, then the slope was considered statistically significantly different from zero. The Caravan ‘‘flex fuel’’ operation during this testing program was found to be unreliable as continuous monitoring of the on-board fuel ethanol sensor showed a fuel composition of zero ethanol for all test fuels. Therefore, it is possible that the engine did not realize any specially designed engine parameters for ethanol fuel operation. 3.1.1. Criteria emissions The usual trends of generally increased emissions on cold start and cold temperature operation and during aggressive driving were observed with these vehicles and will not be further discussed. The FTP composite and US06 emission rates for each vehicle and fuel are shown in Fig. 1. The slopes obtained from the regression analyses are presented in the Supplementary material. A negative slope was observed for CO, a positive slope for NOX, and a slope of approximately zero was observed for NMHC and NMOG; however, most of the observed trends were not statistically significantly different from zero. Referring to the per phase emission rates presented in the Supplementary material, ethanol blends tended to decrease the CO emissions and increase the NOX emissions primarily during engine cold start and aggressive driving conditions for the MPFI and GDI vehicles. There was minimal effect during hot engine start and stabilized driving. The reason for the elevated NOX emissions from the Sentra with E20 fuel at 10 1C for both the FTP and US06 is not known. Compared to the E10 tailor blend fuel, the splash blended E10 fuel resulted in 35–50% higher CO emissions during cold engine start at cold tempera-
4503
ture for the MPFI vehicles. At 20 1C, there was no statistical difference in CO, NOX, NMHC or NMOG emissions between the two fuels for any of the vehicles. As shown in Fig. 1, fuel ethanol content did not affect the specific reactivity or ozone forming potential of the exhaust from the MPFI vehicles. For the GDI vehicle, increasing fuel ethanol content resulted in decreasing specific reactivity and ozone forming potential of the exhaust. 3.1.2. Air toxic emissions The FTP composite and US06 emission rates for each vehicle and fuel are shown in Fig. 2. The slopes obtained from the regression analyses are presented in the Supplementary material. Formaldehyde and acetaldehyde emissions tended to increase, and 1,3butadiene and benzene emissions tended to decrease with increasing ethanol content. The changes for formaldehyde, 1,3-butadiene, and benzene were not statistically significant while the changes in acetaldehyde were nearly always statistically significant. The benzene response to change in ethanol content should be considered with caution as the total aromatics and benzene content of the fuels were different but did not change as a direct function of ethanol content as shown in Table 1. Referring to the Supplementary material, the presence of ethanol in the fuel increased the formaldehyde and acetaldehyde emissions primarily during cold engine start and under aggressive driving conditions, not usually for the hot start or stabilized driving. There were no statistically significant differences in these air toxic emissions between the E10 and E10-splash fuels. For all vehicles, ethanol emissions were highest for cold engine start. Once the catalyst was up to normal operating temperature, ethanol emissions were essentially not measurable. Operation at cold temperature resulted in higher ethanol emission rates as compared to standard temperature and mainly affected cold engine start emissions. Relatively low ethanol emissions were present during some of the tests with E0 fuel, likely due to hang up of ethanol in the vehicle fuel system. These findings indicate that the canister conditioning and vehicle preparation procedures minimized but did not completely eliminate fuel carry-over. 3.1.3. GHG emissions The FTP composite and US06 emission rates for each vehicle and fuel are shown in Fig. 3. The slopes
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
4504 4.5 4.0
10
E10-Spl
3.0
E20
2.5 2.0 1.5 1.0
US06 CO (g/km)
E10
3.5 FTP CO (g/km)
12
E0
0.0 Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
2.5
0.7
US06 NOX (g/km)
FTP NOX (g/km)
0.8 0.6 0.5 0.4 0.3 0.2
2.0 1.5 1.0 0.5
0.1 0.0
0.0 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
0.35 US06 NMOG (g/km)
0.30 FTP NMOG (g/km)
Escort (20C) 3.0
0.9
0.25 0.20 0.15 0.10 0.05 0.00 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0.00
Dion (20C) 6.0
E0
E10
E10-Spl
E20
US06 Specific Reactivity (g O3/g NMOG)
6.0 FTP Specific Reactivity (g O3/g NMOG)
4
0 Escort (20C)
4.0 3.0 2.0 1.0
5.0 4.0 3.0 2.0 1.0 0.0
0.0 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
1.2
0.35
1.0
0.30
US06 OFP (g O3/km)
FTP OFP (g O3/km)
6
2
0.5
5.0
8
0.8 0.6 0.4 0.2 0.0
0.25 0.20 0.15 0.10 0.05 0.00
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Fig. 1. Comparison of FTP composite and US06 criteria emissions, specific reactivity and ozone forming potential from the low blend study.
ARTICLE IN PRESS 1.4 1.2 1.0
E10
E10-Spl
E20
0.8 0.6 0.4 0.2 0.0 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
6.0 5.0 4.0 3.0 2.0 1.0 0.0 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
US06 Acrolein (mg/km)
FPT Acrolein (mg/km)
0.08 0.06 0.04 0.02
0.4 0.3 0.2 0.1 0.0 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
0.6 0.5 0.4 0.3 0.2 0.1 0.0
0.035 0.030 0.025 0.020 0.015 0.010 0.005 0.000
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
0.7
US06 1.3-Butadiene (mg/km)
FPT 1.3-Butadiene (mg/km)
0.5
0.040
0.10
0.00
0.6 0.5 0.4 0.3 0.2 0.1 0.0 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
0.20 0.15 0.10 0.05 0.00
Dion (20C)
6.0
7.0 US06 Benzene (mg/km)
FPT Benzene (mg/km)
0.6
Dion (20C)
0.12
4505
0.7
Dion (20C) US06 Acetaldehyde (mg/km)
FPT Acetaldehyde (mg/km)
E0
US06 Formaldehyde (mg/km)
FPT Formaldehyde (mg/km)
L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
5.0 4.0 3.0 2.0 1.0 0.0
6.0 5.0 4.0 3.0 2.0 1.0 0.0
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Fig. 2. Comparison of FTP composite and US06 toxic emissions from the low blend study.
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
4506
obtained from the regression analyses are summarized in the Supplementary material. For all vehicles and test temperatures, distancebased CO2 emission rates were essentially unchanged as ethanol content increased. The lower volumetric energy density of the ethanol blend fuels canceled out the lower carbon content. In general, increasing ethanol content did not result in any statistically significant changes to the CH4 emission rates. The exception was the Escort where CH4 emission rates decreased with increasing ethanol content at 20 1C only. N2O emissions in general tended to increase with ethanol content, though some changes were not statistically significant. The exception was the Sentra where there was
250
E0
E10
E10-Spl
E20
3.1.4. Evaporative emissions The evaporative emissions results are shown in Fig. 4. The differences in evaporative emissions standards were quite evident with these vehicles. The Sentra had the lowest diurnal and hot soak NMOG emissions (o0.04 g per test). The Caravan and the Escort had similar diurnal and hot soak NMOG emissions (o0.7 and o0.25 g per test, respectively). The Dion had the highest diurnal and hot soak NMOG emissions with diurnal emissions approaching 8 g per test and hot soak emissions around 1.5 g per test. The response of NMOG 300 250 US06 CO2 (g/km)
FTP Composite CO2 (g/km)
300
no change at 20 1C but a statistically significant decrease at 10 1C.
200 150 100 50
100
0 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
25
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
18 16
20
US06 N2O (mg/km)
FTP Composite N2O (mg/km)
150
50
0
15 10 5
14 12 10 8 6 4 2 0
0 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
30
30
25
25
US06 CH4 (mg/km)
FTP Composite CH4 (mg/km)
200
20 15 10 5
20 15 10 5
0
0 Escort (20C)
Escort (-10C)
Sentra (20C)
Sentra (-10C)
Caravan (20C)
Dion (20C)
Fig. 3. Comparison of FTP composite and US06 emissions for greenhouse gas emissions from the low blend study.
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516 100
0.6 E0
E10
E10-Spl
E20
10 Evap NMOG (g/test)
Evap Ethanol (mg/test)
0.5 0.4 0.3 0.2 0.1
1 0.1 0.01 0.001
0 Diurnal
Soak
Escort
Diurnal
Soak
Sentra
Diurnal
Soak
Caravan
Diurnal
Soak
Diurnal
Dion
Soak
Escort
4.0
Diurnal
Soak
Sentra
Diurnal
Soak
Caravan
Diurnal
Soak
Dion
100
3.5 Evap OFP (g O3/test)
Evap Specific Reactivity (g O3/g NMOG)
4507
3.0 2.5 2.0 1.5 1.0
10 1 0.1 0.01
0.5 0.0
0.001 Diurnal
Soak
Escort
Diurnal
Soak
Sentra
Diurnal
Soak
Caravan
Diurnal
Soak
Dion
Diurnal
Soak
Escort
Diurnal
Soak
Sentra
Diurnal
Soak
Caravan
Diurnal
Soak
Dion
Fig. 4. Comparison of evaporative emissions from the low blend study.
emissions to ethanol content was different for each vehicle as shown in Fig. 4. The results of the regression analysis are summarized in the Supplementary material and show that each vehicle responds differently, and that in general, the responses were not statistically significant. The differences observed between the E10 and E10splash fuels were also not statistically significant. 3.2. E85 study The complete set of measured emission rates is provided in the Supplementary material. FTP composite emission rates are compared in Figs. 5–7. 3.2.1. Criteria emissions The first step of the study was to determine if the paired vehicles differed in emissions when operated on certification fuel. Since the conventional Sebring was certified to a different emission standard than the FFV Sebring, it was excluded from the study. As seen in Figs. 5 and 6, the two Caravans showed no statistically significant differences in emissions, except for NOX for which the FFV showed 23%
lower emissions than the conventional vehicle during Phase 1 of the FTP. When compared to the conventional Caravan on certification 2 fuel, the FFV Caravan on E85 showed statistically significant decreases in CO, NOX, and NMHC emissions by 72%, 48%, and 55%, respectively. There was no statistically significant difference in NMOG emissions. For the FFV Caravan, operation on E85 resulted in statistically significant decreases in CO, NOX and NMHC emissions but no change in NMOG emissions, as compared to operation on certification fuel. The decreases in CO and NMHC emissions occurred in Phases 1 and 3 while the decrease in NOX emissions occurred only in Phase 1. For the FFV Sebring, operation on E85 also resulted in statistically significant decreases in CO, NOX, and NMHC emissions, as compared to operation on certification fuel. The decrease in CO occurred across all three phases. The decreases in NOX and NMHC emissions occurred in Phases 1 and 3. While there was no statistically significant change in FTP composite NMOG emissions, Phase 1 showed a statistically significant increase and Phase 3 showed a statistically significant decrease.
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
4508 0.9
0.20
Cal. RFG 2
0.8
0.18
E85
0.16 FTP NOX (g/km)
FTP CO (g/km)
0.7 0.6 0.5 0.4 0.3
0.10 0.08 0.06 0.04
0.1
0.02 0.00 Conv Caravan
FFV Caravan
FFV Sebring
0.07
0.07
0.06
0.06 FTP NMOG (g/km)
FTP NMHC (g/km)
0.12
0.2
0.0
0.05 0.04 0.03 0.02 0.01
Conv Caravan
FFV Caravan
FFV Sebring
Conv Caravan
FFV Caravan
FFV Sebring
Conv Caravan
FFV Caravan
FFV Sebring
0.05 0.04 0.03 0.02 0.01
0.00
0.00 Conv Caravan
FFV Caravan
FFV Sebring
4.5
0.18
4.0
0.16
3.5
0.14
FTP OFP (g O3/km)
FTP SR (g O3/g NMOG)
0.14
3.0 2.5 2.0 1.5 1.0 0.5
0.12 0.10 0.08 0.06 0.04 0.02
0.0
0.00 Conv Caravan
FFV Caravan
FFV Sebring
Fig. 5. Criteria emissions for conventional and flexible fuel vehicles operating on Cal. RFG 2 fuel and E85.
3.2.2. Air toxic emissions Fig. 6 shows FTP composite emission rates for selected air toxics. As summarized in the Supplementary material, there were no statistically significant differences in air toxic emissions between the conventional Caravan and FFV Caravan operating on certification fuel. Operation of the FFV on E85 resulted in no statistically significant change in formaldehyde emissions, a statistically significant increase in acetaldehyde emissions, and statistically significant decreases in benzene and BTEX emissions, as compared to the conventional Caravan. Comparing the FFV Caravan operation on E85 to its operation on the certification fuel, similar trends to the conventional Caravan were observed, except
now, a statistically significant increase in formaldehyde emissions was observed. Comparing the FFV Sebring operation on E85 to its operation on the certification fuel, similar trends to the FFV Caravan were observed. The 1,3-butadiene emissions from all vehicles on the certification fuel were below detection limits so changes could not be reliably evaluated. 3.2.3. GHG emissions Fig. 7 shows FTP composite emission rates for the greenhouse gases. As summarized in the Supplementary material, there were no statistically significant differences in GHG emissions between the conventional Caravan and the FFV Caravan
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516 1.0 Cal. RFG 2 E85
0.10 0.08 0.06 0.04 0.02
0.9 FTP Acetaldehyde (mg/km)
FTP Formaldehyde (mg/km)
0.12
FFV Caravan
0.6 0.5 0.4 0.3 0.2
FFV Sebring
0.012
Conv Caravan
FFV Caravan
FFV Sebring
Conv Caravan
FFV Caravan
FFV Sebring
Conv Caravan
FFV Caravan
FFV Sebring
Conv Caravan
FFV Caravan
FFV Sebring
FTP Total Carbonyl (mg/km)
1.8
0.010 FTP Acrolein (mg/km)
0.7
0.0 Conv Caravan
0.008 0.006 0.004 0.002 0.000
1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0
Conv Caravan
FFV Caravan
FFV Sebring
0.35
2.0 1.8
0.30 FTP Benzene (mg/km)
FTP 1.3-butadiene (mg/km)
0.8
0.1 0.00
0.25 0.20 0.15 0.10 0.05
1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2
0.00
0.0 Conv Caravan
FFV Caravan
FFV Sebring
12
30
10
25 FTP Ethanol (mg/km)
FTP BTEX (mg/km)
4509
8 6 4 2 0
20 15 10 5 0
Conv Caravan
FFV Caravan
FFV Sebring
Fig. 6. Air toxic emissions for conventional and flexible fuel vehicles operating on Cal. RFG 2 fuel and E85.
operating on certification fuel. The FFV Caravan operating on E85 showed a statistically significant decrease in N2O emissions but no change in CH4
emissions, as compared to the conventional Caravan operating on Certification fuel. The CH4 change was almost significant at the 0.05 level (p-value ¼ 0.058).
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
4510
350 300
E85
10 FTP CH4 (mg/km)
FTP CO2 (g/km)
12
Cal. RFG 2
250 200 150 100
8 6 4 2
50 0
0 Conv Caravan
FFV Caravan
FFV Sebring
Conv Caravan
FFV Caravan
FFV Sebring
Conv Caravan
FFV Caravan
FFV Sebring
FTP N2O (mg/km)
25 20 15 10 5 0
Fig. 7. Greenhouse gas emissions for conventional and flexible fuel vehicles operating on Cal. RFG 2 fuel and E85.
Sebring
Caravan Fuel Sesnor (%Ethanol)
Fuel Sesnor (%Ethanol)
70 60 50 40 30 20 10 0 0
100
200
300
400
500
Accumulation (km)
90 80 70 60 50 40 30 20 10 0 0
100
200
300
400
Accumulation (km)
Fig. 8. Fuel sensor change by mileage accumulation. Solid symbols indicate fuel exchange or top-up prior to mileage accumulation to reach that point.
The FFV Caravan, when operating on E85, showed a statistically significant increase in CH4 emissions and no statistically significant change in N2O emissions, as compared to operation on Certification fuel. The N2O change was almost significant at the 0.05 level (p-value ¼ 0.084). The FFV Sebring also showed a statistically significant increase in CH4 emissions and a statistically significant decrease in N2O emissions, as compared to operation on Certification fuel.
3.2.4. Fuel sensor behavior Fig. 8 shows how the FFVs’ sensors changed as mileage accumulated. The FFV Caravan’s sensor, over 356 km, changed from 0% to 64%. The FFV Sebring’s sensor, over 273 km, changed from 0% to 83%. Fig. 1a and b in the Supplementary material shows linear trend lines fit to the measured emission rates as a function of fuel sensor reading. All tests shown in these figures were conducted on E85 and the sensor reading was recorded at the beginning of
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
the test. For the FFV Caravan, regression analysis showed that CO, NOX, and NMHC emissions decreased and NMOG emissions increased during Phase 1 of the FTP as the sensor adjusted to the fuel composition. The Phases 2 and 3 trend lines showed no statistically significant change in emissions with fuel sensor change. For the Sebring, CO and NOX emissions decreased as the sensor reading increased during all three phases. These changes were not statistically significant for Phases 2 and 3, but were statistically significant for Phase 1. For NMHC and NMOG, emissions increased during Phase 1 but no change was observed during Phase 2 or 3. 4. Discussion We have reported the results of two studies that have examined the effect on emissions due to the use of ethanol-blended gasolines. These two studies, like other studies reported in the literature, are limited (mainly for financial reasons) to a fleet of a few test vehicles. The question of how representative any one of these test vehicles is of the population from which it is taken is always asked. As can be seen from the literature reviews discussed below, results from individual studies often show contradicting results, and one reason for this is the small fleet size of any one study. In order to clear away some of these contradictions, we take the results of the current studies and combine them with the results reported in the literature to increase the fleet size. It is expected that, as a result, a clearer picture of the effects on emissions will be obtained.
4511
significance could be done for the changes resulting from individual vehicle/fuel pairs. For each vehicle/ fuel pair, the relative change in emission rate ((E10-Ref)/Ref) was calculated to minimize the effect of different emission control technologies. The relative change data was examined to determine if it was normally or symmetrically distributed. The Kolmogorov–Smirnov goodness of fit test and Q–Q plots were used to compare each dataset to the normal distribution. The skewness parameter was also calculated. Descriptive statistics are summarized in Table 3 and the relative changes for each vehicle/fuel pair are provided in the Supplementary material. In all cases, the data were not normally distributed and almost always moderately to highly skewed; therefore, the median value instead of the mean was used for quantifying the change due to fuel. From the results of the entire dataset, all changes in emissions were statistically significant except for NOX, formaldehyde and the GHGs. For the 31 or 32 tests for which individual tests for significance could be done, only a few of the individual vehicles showed a statistically significant change for all pollutants except acetaldehyde. Because of the magnitude of the change in acetaldehyde emissions, 11 of the 31 vehicles showed statistically significant changes. These statistically significant changes were in the same direction as the entire dataset, and at least twice the magnitude of the median value of the entire dataset. The box plot shown in Fig. 9 illustrates these results. 4.2. Literature review E85
4.1. Literature review E10 Direct comparisons between reference fuels and E10 blends for FTP composite emission rates were possible for two published studies (Knapp et al., 1998; Durbin et al., 2006) in addition to the results from this study. Changes in criteria emissions, air toxic and GHG emissions (CO, NOX, NMHC, NMOG, formaldehyde, acetaldehyde, 1,3-butadiene, benzene, CO2, N2O, CH4) as a result of fuel change were analyzed. Knapp et al. did not report NMHC, NMOG, or GHG emission rates. Durbin et al. did not report N2O or CH4 emission rates. A total of 43 vehicle/fuel pairs were available from these studies. For two of the three studies, individual test results were available for each vehicle, so tests for
Direct comparisons between reference fuels and E85 blends for FTP composite emission rates were possible for four published studies (Benson et al., 1995; Kelly et al., 1996; Winebrake and Deaton, 1999; Black et al., 1998) in addition to the results reported in this study. Winebrake and Deaton reported only toxic emissions. A total of 11 vehicle/fuel pairs were included in the dataset. The relative changes for each vehicle/fuel pair are provided in the Supplementary material. The results from the literature for E85 were generally for older vehicles than those in the E10 comparison and the present study. The analysis described above for E10 was conducted and the results are summarized in Table 3. In all cases, the datasets were not normally distributed
4512
Table 3 Summary of descriptive statistics and results of test for significance for relative changes in emissions due to the use of ethanol blended fuels
Wilcoxon signed-rank test (p-value) Significance of change (a ¼ 0.05)
(b) E85 fuel Minimum (%) 1st quartile (%) Median (%) 3rd quartile (%) Maximum (%) Total N Missing data Mean (%) S.D. (%) Skewness Shape of distribution Kolmogorov–Smirnov test (p-value) Normality Wilcoxon signed-rank test (p-value) Significance of change (a ¼ 0.05)
NMHC
NMOG
Formaldehyde
Acetaldehyde
1,3Butadiene
Benzene
CO2
N2O
CH4
82 31 16 0 70 43 0 15 28 0.086 Fairly symmetric 0.502
58 15 3 18 77 43 1 4 26 0.506 Moderately skewed 0.462
17 0 9 22 50 43 11 11 18 0.490 Fairly symmetric 0.776
14 5 14 21 66 43 12 15 19 1.004 Highly skewed 0.458
83 8 5 30 137 43 1 13 40 1.166 Highly skewed
26 68 108 160 486 43 1 129 107 1.859 Highly skewed
0.337
0.267
83 3 16 44 267 43 1 28 63 2.035 Highly skewed 0.111
71 5 15 29 85 43 1 12 31 0.127 Fairly symmetric 0.782
2.4 0.0 0.6 1.4 3.3 43 11 0.6 1.2 0.209 Fairly symmetric 1.782
30 7 2 5 16 43 35 3 16 1.00 Highly skewed 2.782
45 21 9 1 19 43 35 12 22 0.48 Fairly symmetric 3.782
Not normal
Not normal
Not normal
0.5035
Not normal 0.0002
Not normal
0.0004
Not normal 0.0035
0.0689
0.0000
Not normal 0.0043
Not normal 0.0075
Not normal 1.0075
Not normal 2.0075
Not normal 3.0075
Significant
Not significant
Significant
Significant
Not significant
Significant
Significant
Significant
Not significant
Not significant
Not significant
CO
NOX
NMHC
NMOG
Formaldehyde
Acetaldehyde
1,3-Butadiene
Benzene
69 43 8 31 62 11 3 5 51 0.161 Fairly symmetric
87 55 48 39 7 11 1 45 24 0.662 Moderately skewed 0.790
41 13 5 35 74 11 3 12 39 0.407 Fairly symmetric 0.878
5 55 73 102 164 11 0 80 44 0.400 Fairly symmetric 0.871
1275 2005 2540 3821 4967 11 0 2855 1189 0.512 Moderately skewed 0.719
89 82 77 59 0 11 3 64 31 1.590 Highly skewed
0.863
60 50 45 34 13 11 3 42 15 0.843 Moderately skewed 0.966
0.380
94 82 76 60 50 11 3 72 15 0.177 Fairly symmetric 0.910
Not normal 0.7422
Not normal 0.0078
Not normal 0.0039
Not normal 0.5469
Not normal 0.0010
Not normal 0.0010
Not normal 0.0170
Not normal 0.0078
Not significant
Significant
Significant
Not significant
Significant
Significant
Significant
Significant
ARTICLE IN PRESS
Kolmogorov–Smirnov test (p-value) Normality
NOX
L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
(a) E10 fuel Minimum (%) 1st quartile (%) Median (%) 3rd quartile (%) Maximun (%) Total N Missing data Mean (%) S.D. (%) Skewness Shape of distribution
CO
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
4513
Relative Change (E10-Ref)/Ref
6
4
2
0
-2 Acetald Butadiene Benzene CH4
CO
Formald CO2
NMHC N2O
NOX NMOG
Fig. 9. Box plot showing relative change in emissions with fuel change from reference fuel to E10 for combined data from published studies and this study.
and also generally not symmetrically distributed. Statistically significant changes in emissions were found for all pollutants except CO and NMOG. Individual test data were available for between three and seven vehicles for each pollutant, allowing individual tests for significance to be done (see Supplementary material). Since the changes in emissions were generally much larger than seen for E10, the sign and magnitude of the individual tests for significance agree quite well with what the entire dataset suggests. The exception to this trend is CO, which showed a much wider variation in effect. Upon closer examination, it appears that the older vehicles (Tier 0 and older) have different CO and NMOG responses than newer vehicles (Tier 1 and newer). Four of the five older vehicles showed large increases in CO emissions, while the newer vehicles showed a decrease in CO emissions of similar magnitude. Three of these five older vehicles also showed a greater increase in NMOG emissions than the decrease shown by the newer vehicles. The box plot in Fig. 10 illustrates these results. These changes were generally consistent with those used by Jacobson in a recent assessment of potential health effects of widespread E85 use in the USA (Jacobson, 2007).
5. Conclusions The results of two recent studies on the effects on tailpipe and evaporative emissions of low blend ethanol gasolines and E85 were presented. The results from these two studies were combined with published literature results to better assess the effect of ethanol on vehicle emissions. The results of the low blend ethanol gasoline study suggests that up to 20% ethanol can lead to statistically significant reductions in FTP composite emissions of CO, statistically significant increases in NOX and acetaldehyde emissions, and no change in NMHC, NMOG, formaldehyde, 1,3-butadiene, Benzene, or GHG emissions. The magnitude of the change appears to depend on vehicle technology for the four vehicles studied. Much smaller changes were observed for US06 emissions. Operation at cold temperature tended to increase the magnitude of the emissions and the magnitude of the changes but not, in general, the direction of the changes. Low blend ethanol gasolines showed no statistically significant effect on evaporative NMOG emissions. While changes in the detailed composition of the tailpipe and evaporative emissions were observed (increases in acetaldehyde and ethanol, decreases in gasoline hydrocarbons), there appeared to be
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
4514 60
2
Relative Change (E85-Ref)/Ref
1 40
0
-1 20
-2 Benzene Butadiene
CO
Formald
NMHC
NMOG
NOX
0
-20 Acetald
Benzene Butadiene
CO
Formald
NMHC
NMOG
NOX
Fig. 10. Box plot showing relative change in emissions with fuel change from reference fuel to E85 for combined data from published studies and this study.
no net effect on the specific reactivity of the emissions. There were also no statistically significant differences in emissions from the E10 and E10-splash fuels, except during cold temperature cold start where an increase in CO emissions was observed. The results of the E85 study indicate that the only statistically significant difference in emissions between the conventional Caravan and the FFV Caravan certified to the same emission standard and operating on certification fuel was a 23% decrease in NOX emissions for the FFV Caravan. Operation of the FFV Caravan and the FFV Sebring on E85 resulted in statistically significant increases in formaldehyde (86–117%), acetaldehyde (1300–5000%) and methane emissions (37–49%); statistically significant decreases in CO (37–60%), NOX (32–47%), NMHC (3–45%), benzene (58–72%) emissions and no statistically significant changes in NMOG, 1,3-butadiene, CO2, or N2O emissions. Results also suggest that it takes some time for the fuel oxygen sensor to respond to the step change in change fuel composition that occurred on fuel exchange from certification fuel to E85. It appeared that for both vehicles, the observed changes in fuel composition during 300+ km mileage accumulation were associated with fuel exchange or top-up followed by mileage accumulation, not just mileage accumulation. Limited emissions testing was conducted while the fuel sensor
was adjusting to the fuel composition and these results show that CO and NOX emissions can be higher while NMHC and NMOG emissions can be lower while the fuel sensor is adjusting to the correct fuel composition. The differences are largest during cold start but in some cases are observed during all phases of the test. For low blend ethanol fuels and E85, unburned ethanol was found almost exclusively during cold and hot engine start portions of the test, when the catalytic converter was cold, not during stabilized operation or during aggressive driving, when the catalytic converter was hot. Ethanol was found in the evaporative emissions during the low blend ethanol study and increased with increasing ethanol content. The difference in RVP of the E10 and E10splash blends did not appear to increase ethanol or NMOG evaporative emissions. When reviewing the conclusions of each study individually, contradictory results are often found. But, when all available data for a particular ethanol blend is considered as a single dataset, as done in this paper, the following general conclusions can be drawn. As compared to reference fuels with no ethanol, operating on E10 blends generally result in:
statistically significant decreases in CO emissions (16%),
ARTICLE IN PRESS L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
statistically significant increases in NMHC (9%), NMOG (14%), acetaldehyde (108%), 1,3-butadiene (16%), and benzene (15%) emissions, and no statistically significant change in NOX, formaldehyde, CO2, CH4, or N2O emissions.
4515
managers of Natural Resources Canada and the Royal Canadian Mounted Police. The authors would also like to thank Dr. Thomas Durbin for providing the individual test results for the CRC E-67 study.
The increase in benzene emissions may seem curious as one may expect a dilution effect proportional to the ethanol content. This may result from differences in benzene and/or total aromatic content of the reference and E10 fuels as not all fuels were splash blends of ethanol with the reference fuel. As compared to reference fuels with no ethanol, operating on E85 generally results in:
Appendix A. Supplementary materials
Baas, C., Graham, L.A., 2006a. Emissions from 4 different light duty vehicle technologies operating on low blend ethanol gasoline—tailpipe greenhouse gas emissions. Environment Canada ERMD Report 04-27A, 2004. Baas, C., Graham, L.A., 2006b. Emissions from 4 different light duty vehicle technologies operating on low blend ethanol gasoline—tailpipe regulated and unregulated gaseous emissions (CO, NOX, THC, NMHC, NMOG, ethanol, carbonyls, VOC). Environment Canada ERMD Report 04-27B. Baas, C., Graham, L.A., 2006c. Emissions from 4 different light duty vehicle technologies operating on low blend ethanol gasoline—diurnal and hot soak evaporative emissions (ethanol, NMOG and speciated NMHC). Environment Canada ERMD Report 04-27D. Belisle, S., Graham, L.A., 2006. Comparison of emissions of conventional and flexible-fuel vehicles operating on gasoline and E85 fuels. Environment Canada, ERMD Report #05-39. Benson, J.D., et al., 1995. Emissions with E85 and gasolines in flexible/variable fuel vehicles—the auto/oil air quality improvement research program. SAE Technical Paper Series, No. 952508. Black, F., et al., 1998. Alternative fuel motor vehicle tailpipe and evaporative emissions composition and ozone potential. Journal of the Air and Waste Management Association 48, 578–591. Durbin, T.D., Miller, J.W., Pisano, J.T., Younglove, T., Sauer, C.G., Rhee, S.H., Huai, T., 2003. The effect of fuel on NH3 and other emissions from 2000–2001 model year vehicles. Final Report CRC Project E-60, May 2003. Durbin, T.D., et al., 2006. Effects of ethanol and volatility parameters on exhaust emissions. CRC Project No. E-67. Coordinating Research Council, Inc., Alpharetta, GA. Environment Canada, 2007a. Fossil fuels. /http://www.ec.gc. ca/cleanair-airpur/Fuels-WS0E66B313-1_En.htmS (accessed 4.09.07). Environment Canada, 2007b. Renewable fuels. /http://www. ec.gc.ca/cleanair-airpur/Bio_Fuels-WSCD7D0FD7-1_En.htmS (accessed 4.09.07). Jacobson, M.Z., 2007. Effects of ethanol (E85) versus gasoline vehicles on cancer and mortality in the United States. Environmental Science and Technology 41, 4150–4157.
statistically significant decreases in NOX (45%), NMHC (48%), 1,3-butadiene (77%), and benzene (76%) emissions, statistically significant increases in formaldehyde (73%) and acetaldehyde (2540%) emissions, and no statistically significant change in CO, CO2, or NMOG emissions.
While each of the individual studies generally followed the trends suggested by the analysis of the aggregated datasets, some differences in conclusions were found. This finding reinforces the importance of having a large enough vehicle fleet to minimize influences of a single vehicle on the conclusions drawn and to increase the ability to detect smaller changes. Acknowledgements The authors would like to acknowledge the vehicle testing and chemistry laboratory staff of Environment Canada’s Emissions Research and Measurement Division for their efforts in conducting the emissions testing and analyses reported in this paper. Funding for the low blend ethanol study was provided by Natural Resources Canada’s Program for Energy Research and Development and Environment Canada. Vehicles for this study were provided by Transport Canada’s Advanced Technology Vehicle program (Nissan Sentra and Mitsubishi Dion) and by Natural Resources Canada (FFV Caravan). Funding for the E85 study was provided by Natural Resources Canada and the federal fleet managers’ Federal Vehicles Initiative (FVI). Vehicles for this study were provided by the federal fleet
Supplementary data associated with this article can be found in the online version at doi:10.1016/ j.atmosenv.2008.01.061.
References
ARTICLE IN PRESS 4516
L.A. Graham et al. / Atmospheric Environment 42 (2008) 4498–4516
Kelly, K.J., et al., 1996. Federal test procedure emissions test results from ethanol variable-fuel vehicle Chevrolet Luminas. SAE Technical Paper Series, No. 961092. Knapp, K.T., et al., 1998. The effect of ethanol fuel on the emissions of vehicles over a wide range of temperatures. Journal of the Air and Waste Management Association 48, 646–653. Loo, J.F., Parker, D.T., 2000. Evaluation of a photoacoustic gas analyzer for ethanol vehicle emissions measurement. Society of Automotive Engineers Technical Paper No. 2000-01-0794. McCarthy, J.E., Tiemann, M., 2001. CRS Report for Congress 98-290: MTBE in Gasoline: Clean Air and Drinking Water Issues, Updated May 15, 2001. /http://www.ncseonline.org/ NLE/CRSreports/air/air-26.cfm#Back12S (accessed 1.08.07). US EPA, 2007a. MTBE in fuels. /http://www.epa.gov/mtbe/ gas.htmS (accessed 4.09.07). US EPA, 2007b. Reformulated gas—basic information. /http:// www.epa.gov/otaq/rfg/information.htmS (accessed 4.09.07).
US EPA, 2007c. State actions banning MTBE (statewide), EPA420-B-07-013, August 2007. /http://www.epa.gov/mtbe/ 420b07013.pdfS (accessed 4.09.07). US EPA, 2007d. Regulatory announcement: removal of reformulated gasoline oxygen content requirement and revision of commingling prohibition to address non-oxygenated reformulated gasoline. /http://www.epa.gov/otaq/regs/fuels/rfg/ 420f06020.htmS (accessed 4.09.07). US EPA, 2007e. E85 and flex fuel vehicles. /http://www.epa.gov/ smartway/growandgo/documents/factsheet-e85.htmS (accessed 4.09.07). US EPA, 2007f. Regulatory impact analysis: renewable fuel standard program. EPA420-R-07-004, April 2007. /http://www.epa.gov/ otaq/renewablefuels/420r07004.pdfS (accessed 9.01.08). Winebrake, J.J., Deaton, M.L., 1999. Hazardous air pollution from mobile sources: a comparison of alternative fuel and reformulated gasoline vehicles. Journal of the Air and Waste Management Association 49, 576–581.