Impact of Adaptation on Flex-Fuel Vehicle ... - ACS Publications

Feb 11, 2013 - emissions and fuel economy behavior of vehicles running on E40 showed results generally consistent with adaptation to the blend after...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/est

Impact of Adaptation on Flex-Fuel Vehicle Emissions When Fueled with E40 Janet Yanowitz,∥ Keith Knoll,‡,† James Kemper,§ Jon Luecke,‡ and Robert L. McCormick‡,* ‡

National Renewable Energy Laboratory, Golden, CO, United States Colorado Department of Public Health and Environment, Denver, CO, United States ∥ Ecoengineering, Inc., Boulder, CO, United States §

S Supporting Information *

ABSTRACT: Nine flex-fuel vehicles meeting Tier 1, light duty vehicle-low emission vehicle (LDV-LEV), light duty truck 2-LEV (LDT2-LEV), and Tier 2 emission standards were tested over hotstart and cold-start three-phase LA92 cycles for nonmethane organic gases, ethanol, acetaldehyde, formaldehyde, acetone, nitrous oxide, nitrogen oxides (NOx), carbon monoxide (CO), and carbon dioxide (CO2), as well as fuel economy. Emissions were measured immediately after refueling with E40. The vehicles had previously been adapted to either E10 or E76. An overall comparison of emissions and fuel economy behavior of vehicles running on E40 showed results generally consistent with adaptation to the blend after the length of the three-phase hot-start LA92 test procedure (1735 s, 11 miles). However, the single LDT2-LEV vehicle, a Dodge Caravan, continued to exhibit statistically significant differences in emissions for most pollutants when tested on E40 depending on whether the vehicle had been previously adapted to E10 or E76. The results were consistent with an overestimate of the amount of ethanol in the fuel when E40 was added immediately after the use of E76. Increasing ethanol concentration in fuel led to reductions in fuel economy, NOx, CO, CO2, and acetone emissions as well as increases in emissions of ethanol, acetaldehyde, and formaldehyde.

■

INTRODUCTION More than 95% of U.S. gasoline contains ethanol in an approximate 10% blend.1 Ethanol is also available in higher concentration blends marketed as E85 (also known as flexfuel), a blend containing 51% to 83% ethanol depending on geography and season.2 As of mid-2012 there were more than 2500 flex-fuel refueling stations nationwide.3 More recently, ethanol blender-pump dispensers have become available.4 Blender pumps draw fuel from two separate storage tanks (E10 and flex-fuel) and offer a choice of ethanol-blended gasoline products with concentrations between E15 and the seasonally adjusted flex-fuel ethanol content. Although blender pumps make consistent operation at midlevel ethanol concentrations straightforward, it has always been possible to use mixtures of varying levels of E10 and E85, either intentionally or not, by alternating between the two fuels at the filling station. Vehicles designed to use higher levels of ethanol in gasoline are known as flexible-fuel vehicles (FFVs). FFVs have a single fuel tank, fuel system, and engine designed to run on any combination of unleaded gasoline and ethanol up to a concentration of 85% ethanol. To optimize operation for varying fuel oxygen content, energy content, and octane, vehicles use feedback from sensors that can include the fuel gauge, exhaust oxygen sensor, a knock sensor in the engine, and © 2013 American Chemical Society

a separate fuel sensor in the fuel line. This input is used to optimize computerized engine control of air/fuel ratio and spark timing − with different manufacturers using different control and adaption strategies. As with standard vehicles, FFVs employ three-way catalyst systems to reduce exhaust emissions of carbon monoxide (CO), nitrogen oxides (NOx), and hydrocarbons, including aldehydes. FFVs are tested to ensure compliance with emissions regulations using certification fuel (0% ethanol) and an 85% ethanol blend with certification fuel.5 Operation and emissions testing on ethanol/gasoline blends between those levels is not part of the emissions certification process for new vehicles. Yanowitz and McCormick6 reviewed FFV certification testing results for certification E85 (85% ethanol) and E0 reported to the U.S. Environmental Protection Agency (EPA) for the 1999 to 2007 model years (primarily Tier 2 vehicles), and published results (Tier 1 cars), also primarily on true E85 (85% ethanol) versus E0 to estimate the impact of the use of ethanol fuels on tailpipe air emissions. They found wide differences between vehicles, with large 95% confidence Received: Revised: Accepted: Published: 2990

November 7, 2012 February 8, 2013 February 11, 2013 February 11, 2013 dx.doi.org/10.1021/es304552b | Environ. Sci. Technol. 2013, 47, 2990−2997

Environmental Science & Technology

Article

intervals on the average emission values. However, certain general trends were statistically significant: an increase in ethanol content in the fuel reduced NOx, CO, benzene, and butadiene and increased ethanol, formaldehyde, and acetaldehyde emissions. The results for total organic emissions were mixed, with older Tier 1 vehicles showing a decrease in nonmethane hydrocarbons (NMHC) emissions and no effect for Tier 2 vehicles. Older literature studies had comparable results, with the exception that they showed an increase in CO emissions with E85, attributable to emissions from pre-1999 and European vehicles.7,8 A recent study by Haskew and Liberty9 showed that exhaust emissions [total hydrocarbons (THC), NMHC, nonmethane organic gases (NMOG), NOx, CO, acetaldehyde, and formaldehyde] using the LA92 test cycle, evaporative emissions (running loss, hot soak, and each day of the two-day diurnal) and fuel economy for intermediate-level fuel blends (E32 and E59) were consistent with those of E10 and E85 fuels at the 95% confidence level. However, they created the intermediate blends by first fueling with E6 allowing the vehicle to adapt to that concentration level and then refueling the partially full tank with E85 (or vice versa, starting with E85 and then refueling the partially full tank with E6). This study considers that the emission control and engine system, potentially dependent on complex algorithms that assume added fuel will either be E85 or E10, might not reliably adapt to fueling with intermediate blends that were not commonly available when older FFVs were built. Failure to rapidly adapt to a new fuel results in operation a nonoptimal air/fuel ratio and might have a significant impact on tailpipe emissions. We determined the fuel economy and tailpipe emissions impact of operation on E40 on nine in-use FFVs including both Tier 1 and Tier 2 vehicles. Testing was conducted after a fuel changeout to determine if the engine would be slow to adapt to the new ethanol concentration, resulting in higher emissions and/or changes in fuel economy.

Table 1. Analytical Results and Performance Properties for Fuels property

ASTM test method

density C

D4052 D5291

H

D5291

O

D5291

dry equivalent vapor pressure distillation: T10 distillation: T50 distillation: T90 research octane motor octane aromatic sulfur TV/L=20 lower heating value lower heating value ethanol

D5191

E10

E40

FlexFuel

g/cm3 mass fraction mass fraction mass fraction kPa

0.7377 0.82

0.7556 0.72

0.7801 0.59

0.14

0.14

0.13

0.040

0.1438

0.27

57.9

54.5

40.0

°C °C °C

56.9 73.5 138.9 102.0 88.4 15.5 80.0 58.1 36 500

68.4 77.9 79.0 101.6 90.0 7.1 28.9 66.9 30 240

D86 D86 D86 D2699 D2700 D1319 D5453 D5188 D240

wt % ppm °C kJ/kg

53.2 71.1 153.2 93.0 84.1 20.8 117.6 55.3 41 490

calculated

kJ/L

30 610

27 580

23 590

D5501

vol %

10.6

39.4

75.5

useful life per EPA regulations. The mileage level of the test vehicles was between 10 000 and 118 000 miles. The air and oil filters were changed, and the vehicle was driven at least 500 miles following the oil change. The catalyst was desulfated by running each vehicle on three consecutive US06 drive cycles on a dynamometer. This additional vehicle operation was conducted using nominally E10 retail fuel. EPA dynamometer target coefficients were used in conducting coast-downs to obtain set coefficients for the emission tests. Exhaust Analysis. Regulated emissions were measured via full-flow dilution per Title 40 Code of Federal Regulations Part 86 guidelines. The chassis dynamometer was a Horiba 48 in. (1.22 m) independent-axle, two-wheel drive motor/brake unit. Emissions were measured using Horiba series 200 emissions analyzers. Carbon dioxide (CO2) and CO were measured using a nondispersive infrared sensor. NOx was measured by chemiluminescence detector, and THC and methane were measured using a flame ionization detector with a methane cutter. Individual phase (bag) concentrations of CO, NOx, THC, methane, and CO2 were measured using a conventional Horiba FIA-220. Fuel economy was calculated by carbon balance. Exhaust gas ethanol was measured using the Innova Photacoustic Multigas Analyzer model 1312.12,13 Filter setup is included in Table S1 of the Supporting Information. Individual phase carbonyl emissions were measured using dinitrophenylhydrazine (DNPH) cartridges. Details are given in the Supporting Information. Vehicle Testing. Fuel change and adaptation procedures are shown graphically in the Supporting Information. For each vehicle, the fuel in the tank was drained and replaced with E10. The vehicle was adapted to E10 by running at idle, driving an LA4 cycle, a second idle, a hot-start LA4, and then driving an LA9214 followed by a second LA4. The vehicle was allowed to cool overnight, and emissions were tested and recorded over the LA92 test cycle the next day. Following this, the E10 fuel was drained and replaced with E40. A hot-start LA92 emissions

■

EXPERIMENTAL METHODS Fuels. The two fuels procured for this study were a retail E10 meeting ASTM D4814 Class A-2 standards and a flex fuel meeting ASTM D5798 Class 1 standards. Fuels were purchased in a single batch to minimize test-to-test variation. Performance properties and ethanol content of these fuels and the E40 blend gravimetrically prepared from them are listed in Table 1. The flex fuel contained 76 vol% ethanol comparable to a typical D5798 Class 1 fuel. A recent survey of flex-fuel quality found an average ethanol content for Class 1 of 80 vol%.10 The dry equivalent vapor pressure met the D5798 requirements, and the fuel exhibited a research octane number (RON) of 102. The E10 meets the volatility requirements of Class A-2 gasoline, and it exhibits a RON of 93. The elevated sulfur content of 117.6 ppm indicates that this gasoline was likely produced under the small refiner standard in place in 2010 when the fuel was acquired.11 Vehicles. Nine flex-fuel vehicles ranging from less than 1 year old at the time of testing to 10 years old were obtained from either the State of Colorado fleet or a rental car agency. The vehicles are described in Table 2. Prior to beginning test work, all vehicles were inspected to ensure they were in good working order and had no existing or pending malfunctions that would trigger the malfunction indicator light. Mileage was checked to verify that each vehicle had accumulated at least 4000 miles but was still within its full 2991

dx.doi.org/10.1021/es304552b | Environ. Sci. Technol. 2013, 47, 2990−2997

Environmental Science & Technology

Article

cycle, incorporating a 10 min engine-off hot-soak period following phase 2. Phase 3 was then run as a hot-start replicate of the cold-start phase 1. Composite emissions were calculated using the same weighting factors as specified for the FTP.

Table 2. Description of Vehicles Tested tier (full useful life) Tier 2/Bin 4 (120k miles)

Tier 2/Bin 5: (50k miles)

Tier 2/Bin 8: (50k miles)

LDV-LEV: (50k − 120k miles)

LDT2a-LEV (120k miles)

Tier 1, LDT4 (120k miles)

vehicles (and mileage)

regulated emission limits

■

2011 GMC Terrain, 3.0L DI (10k miles)

NMOG = 0.07 g/mile CO = 2.1 g/mile 2010 Chrysler T&C, 3.3L (28k NOx = 0.04 g/ miles) mile HCHO = 0.011 g/mile 2010 Toyota Tundra, 5.7L (17k NMOG = 0.075 miles) g/mile 2009 Nissan Titan, 5.6L (21k CO = 3.4 g/mile miles) 2011 Ford Fusion, 3.0L (11k NOx = 0.05 g/ miles) mile HCHO = 0.015 g/mile 2007 Chevy Silverado P/U, 5.3L NMOG = 0.125 (10k miles) g/mile CO = 3.4 g/mile NOx = 0.14 g/ mile HCHO = 0.015 g/mile 2002 Ford Taurus, 3.3L (115k NMOG = 0.075 miles) g/mile CO = 3.4 g/mile NOx = 0.2 g/mile HCHO = 0.015 g/mile 2002 Dodge Caravan, 3.3L NMOG = 0.2 g/ (110k miles) mile CO = 5.5 g/mile NOx = 0.5 g/mile HCHO = 0.023 g/mile 2002 Chevy Tahoe (118k miles) NMHC = 0.56 g/ mile CO = 7.3 g/mile NOx = 1.53 g/ mile HCHO = N/A

RESULTS No automated alerts or discernible changes in operation were noted when E40 was added to the fuel tanks of FFVs previously adapted to either E10 or E76. Emission results are provided in the Supporting Information. Composite values are compared in Tables 3−6 using Student’s paired t test, two-tailed. Hot-Start Emissions. Vehicles were tested once (no replicates) using the hot-start LA92 test immediately after E40 was added to the tank to represent the scenario of an already warmed-up car pulling out from a fueling station. To test adaptation, the hot-start E40 emissions tests that were done immediately following adaptation to E76 were compared to those done immediately following adaptation to E10. Composite results from all vehicles are shown in Table 3, and results for individual vehicles are included in the Supporting Information. On average, significantly more CO and acetaldehyde were emitted from the E40 via E76 adapted vehicles − as much as double the CO and more than four times the acetaldehyde. Eight of nine vehicles exhibited an increase in CO (all except the 2002 Tahoe), and seven of the nine showed an increase in acetaldehyde (all except the 2002 Tahoe and the 2011 Fusion). These differences in emissions on E40 after adaption to either E10 or E76 indicate that adaptation is not immediate in most of these vehicles, and, that upon startup after changing fuels, vehicles previously adapted to E76 will continue to operate as though still running on higher ethanol content (i.e., running too rich or with advanced timing), or conversely, after adapting to E10 will continue to operate as though still running on lower ethanol content. In support of this hypothesis, formaldehyde, ethanol, and NMOG emissions also increase on average, and fuel economy decreases, although at levels that are not statistically fully persuasive. NOx emissions are virtually unchanged. Cold-Start Emissions. Replicate cold-start tests were performed 2 to 5 times on each vehicle/fuel combination. The two sets of average E40 emissions measurements (made in the vehicle that was either previously adapted to E10 or previously adapted to E76) were compared for each pollutant and each vehicle individually and then the percent change for each pollutant averaged for all the vehicles together. The results of these calculations are in Table 4 and in Figures 1−3 for the criteria pollutants (NMOG, NOx, and CO) and in a table and figures in the Supporting Information for several other pollutants. Overall, vehicles adapted on E76 showed higher NMOG and CO emissions when operating on E40 than did vehicles adapted on E10; however, if there is an overall lack of adaptation, the effect is significantly more muted than that shown during the hot-start tests immediately after refueling

a

Greater than 3750 lbs loaded vehicle weight and less than 6000 lb gross vehicle weight.

test was performed immediately. This was followed by an overnight soak and a cold-start LA92 emissions test. To determine if the adaptation process would be different if the initial fuel were E76, the fuel change-out procedure outlined above was followed using E76 instead of E10. Test cycles were run to adapt the vehicle to the fuel, and then duplicate LA92 emissions tests were performed on E76. For a final time, the fuel tank was drained, refilled with E40, and the vehicle was tested, first with a hot start and then, after an overnight soak, with a cold start. Both the hot-start and cold-start LA92 were executed as a three-phase test, similar to the federal test procedure (FTP)

Table 3. Comparison of E40 Emissions Results for Hot-Start LA92 Tests after Prior Adaptation to E10 or E76a all vehicles E40 via E76 hot vs E40 via E10 hot

NMOG (g/ mi)

NOx (g/ mi)

CO (g/ mi)

FE (mpg)

EtOH (g/ mi)

acetaldehyde (mg/ mi)

formaldehyde (mg/ mile)

average percentage change p-value

88% 0.23

3% 0.57

101% 0.02

−2% 0.39

382% 0.52

340% 0.04

29% 0.66

a

Those tests for which the p-value is ≤0.05, suggesting the two data sets are different at the 95% confidence interval, are highlighted in bold italic. 2992

dx.doi.org/10.1021/es304552b | Environ. Sci. Technol. 2013, 47, 2990−2997

Environmental Science & Technology

Article

Table 4. Comparison of E40 Emissions Results after Vehicle Is First Adapted to E76 versus after Vehicle Is First Adapted to E10. Cold tests onlya average percentage change, p-value, paired t test of log of emissions value 02Caravan 02Taurus 07Silverado 10T&C 11Fusion 02Tahoe 10Tundra 09Titan 11Terrain all vehicles, log of emissions measurements, paired t test a

NMOG (g/ mi)

NOx (g/ mi)

CO (g/ mi)

FE (mpg)

EtOH (g/ mi)

acetaldehyde (mg/mi)

formaldehyde (mg/mi)

39% 0.02 16% 0.23 −13% 0.08 47% 0.54 49% 0.55 5% 0.73 −2% 0.83 −3% 0.47 30% 0.05 19% 0.05

−26% 0.00 −20% 0.08 −2% 0.85 23% 0.09 −15% 0.19 2% 0.48 −24% 0.29 15% 0.60 25% 0.07 −2% 0.55

139% 0.01 38% 0.45 2% 0.65 6% 0.47 44% 0.01 −5% 0.32 3% 0.33 3% 0.70 56% 0.09 32% 0.05

−2% 0.01 1% 0.19 0% 0.75 0% 0.95 0% 0.90 1% 0.21 0% 0.60 0% 0.85 1% 0.07 0% 0.72

34% 0.18 45% 0.01 −31% 0.17 100% 0.58 64% 0.69 44% 0.56 −24% 0.47 −28% 0.01 134% 0.13 38% 0.17

40% 0.00 27% 0.09 1% 0.94 23% 0.20 91% 0.01 3% 0.44 −31% 0.38 26% 0.10 35% 0.11 24% 0.09

−1% 0.84 24% 0.13 −10% 0.32 −16% 0.18 −37% 0.18 −8% 0.12 5% 0.84 17% 0.05 28% 0.34 0% 0.82

Those tests for which the p-value is ≤0.05, suggesting the two data sets are different at the 95% confidence interval, are highlighted in bold italic.

Table 5. Comparison of Emissions at E10 and E76 Emissions Results Using LA92, Cold Results Only; Those Tests for Which the p-Value is ≤0.05, Suggesting the Two Data Sets Are Different at the 95% Confidence Interval, Are Highlighted in Bold Italic average percentage change, p-value, paired t test of log of emissions value all vehicles E76 vs E10 pre-Tier 2 vehicles Only E76 vs E10 Tier 2 Vehicles Only E76 vs E10 2009 review of literature E85 vs E06 EPA certification data E85 vs E06 (FTP cycle)

NMOG (g/ mi)

NOx (g/ mi)

CO (g/ mi)

FE (mpg)

EtOHa (g/ mi)

acetaldehyde (mg/mi)

formaldehyde (mg/mi)

11% 0.62 8% 0.65 12% 0.75 +12% 0.43 +26% 0.00

−25% 0.03 −19% 0.40 −27% 0.07 −18% 0.00 −14% 0.51 −13% 0.05

−19% 0.05 −38% 0.09 −10% 0.32 −20% 0.30 −19% 0.00 +16% 0.75

−22% 0.00 −22% 0.00 −22% 0.00

10 713% 0.00 580% 0.00 15 779% 0.07

582% 0.00 567% 0.00 589% 0.00 1786% 0.00

83% 0.00 77% 0.04 86% 0.02 63% 0.00 59% 0.00

average of 8 vehicles tested on 3 different cycles9 a

Two of the vehicles tested had emissions of ethanol below background levels in two of three repeats (2011 Fusion E10) or one of two repeats (2010 Tundra E10). Those values were rounded up to zero and averaged with the other repeat(s).

ratio and suggests that the vehicle control system for the Caravan overestimates the amount of ethanol in the fuel when E40 is used after the vehicle has first been adapted to E76. How significant is the effect of less efficient adaptation at E40? As shown in Figures 1 and 3, the impact of poor adaption on the 2002 Caravan is only slightly higher NMOG, about 0.01 g/mile, but as much as 1 g/mile of additional CO over E10 or more than 2 g/mile in comparison to E76. NOx emission values appear to fall between those of the car fully adapted to E10 and to E76. It appears that the Caravan was also ineffective at adapting to E40 via prior adaptation to E10, but the effects were more benign. Organic concentrations were reduced below levels expected when extrapolating between E10 and E76, whereas NOx emissions were unaffected and fuel economy was somewhat improved from what would be expected based on the ethanol concentration.

with E40 with an average increase in NMOG of only 19% and an increase in CO of only 32%. The CO increase between E40 via E76 vs E40 via E10 can be attributed primarily to four vehicles: the 2002 Caravan, the 2002 Taurus, the 2011 Terrain, and the 2011 Fusion. CO is a clear indicator of lean or rich engine operation. Thus, whereas a case could be made that any of these vehicles with appreciably different level of CO at E40 via E76 versus E40 via E10 is not fully adapted, of the four, only the 2002 Caravan and the 2002 Taurus produced higher CO emissions at E40 than when using either E10 or E76 (detailed emission results in the Supporting Information). Additionally, the 2002 Caravan also produced higher NMOG emissions at E40 than when using either E10 or E76 (the Taurus did not) and showed a statistically stronger trend of higher organic emissions and lower fuel economy and NOx emissions than any of the other vehicles (Table 4). This trend is consistent with incomplete adaptation resulting in an excessively low air to fuel 2993

dx.doi.org/10.1021/es304552b | Environ. Sci. Technol. 2013, 47, 2990−2997

Environmental Science & Technology

Article

Table 6. Comparison of Emissions at E40 with E10 and E76 Emissions Results Using LA92. Cold start results only; Those Tests for Which the p-Value Is ≤0.05, Suggesting the Two Data Sets Are Different at the 95% Confidence Interval, Are Highlighted in Bold Italic average percentage change, p-value, paired t test of log of emissions value all vehicles E40 via E10 vs E10 all vehicles E40 via E76 vs E10 all vehicles E76 vs E40 via E10 all vehicles E76 vs E40 via E76 average of 8 vehicles tested on 3 different cycles E32 vs E69

NMOG (g/ mi)

NOx (g/ mi)

CO (g/ mi)

FE (mpg)

EtOHa (g/ mi)

acetaldehyde (mg/mi)

formaldehyde (mg/mi)

−5% 0.36 14% 0.39 16% 0.18 −2% 0.55

−8% 0.27 −13% 0.08 −18% 0.02 −14% 0.06 +9% 0.89 +4% 0.71

−10% 0.17 14% 0.34 −6% 0.39 −25% 0.04 +15% 0.31 +29% 0.02

−9% 0.00 −9% 0.00 −14% 0.00 −14% 0.00

8255% 0.03 13 505% 0.01 140% 0.02 72% 0.02

188% 0.00 245% 0.00 136% 0.00 101% 0.00

12% 0.39 9% 0.43 67% 0.00 69% 0.00

average of 8 vehicles tested on 3 different cycles E59 vs E69 a

Two of the vehicles tested had emissions of ethanol below background levels in two of three repeats (2011 Fusion E10) or one of two repeats (2010 Tundra E10). Those values were rounded up to zero and averaged with the other repeat(s).

Figure 1. Effect of ethanol content and adaption on cold-start LA92 composite NMOG emissions.

Supporting Information) are slightly but consistently reduced with higher ethanol content fuel, an average of 2.7% difference between E76 and E10. Because all energy from the fuel is derived from the combustion of carbon and hydrogen, the somewhat higher ratio of hydrogen to carbon in E76 versus E10 results in the slightly lower emissions of CO2 for the same amount of energy expended. There is a clear trend of increasing ethanol and acetaldehyde emissions with ethanol concentrations in the fuel (Table 5), in agreement with previous studies.6,15,16 Formaldehyde emissions also increased significantly for the higher ethanol blend (Table 5). Two of the Tier 2 vehicles tested had emissions of ethanol below background levels in two of three repeats (2011 Fusion E10) or one of two repeats (2010 Tundra E10). Those values were rounded up to zero and averaged with the other repeat(s). Whereas tailpipe concentrations cannot be less than zero, it is possible for measured background to exceed the tailpipe measurements either through normal experimental error or because ethanol present in the ambient air is combusted in the

Generally, the relative emissions from the different vehicles were consistent with the certification values for each of the vehicles. The highest emissions in each figure were, in most cases, from two of the oldest vehicles, the 2002 Caravan and 2002 Tahoe. The 2002 Taurus, certified as a LDV/LEV vehicle, showed considerably lower emissions, as would be expected. Table 5 shows the average percent change in emissions and fuel economy between E10 and E76 both for these vehicles and for similar testing reported in the literature. Considering all vehicles as a group, increasing ethanol content from E10 to E76 in fully adapted vehicles results in significant reductions in NOx (25%) and CO (19%), but no significant change in NMOG. These emission changes are reasonably consistent with the observations of previous studies.6,9,15,16 It should be noted that only the 2002 Caravan was not required to meet an NMOG standard as it was regulated under the older NMHC standard. Fuel economy was reduced by 22%, approximately equal to the 23% reduction in energy density of E76 versus E10 (see Table 1 and Figure 4). CO2 emissions (shown in the 2994

dx.doi.org/10.1021/es304552b | Environ. Sci. Technol. 2013, 47, 2990−2997

Environmental Science & Technology

Article

Figure 2. Effect of ethanol content and adaption on cold-start LA92 composite CO emissions.

Figure 3. Effect of ethanol content and adaption on cold-start LA92 composite NOx emissions.

engine. In either case, the upward rounding may bias the emission results for E10 slightly upward in comparison to the other measurements done on E40 or E76. Even with this slight bias, the results clearly show increasing ethanol emissions with increasing ethanol in the fuel. However, it should be noted that small changes in the value of emissions very close to zero dramatically affect the percentage increase in ethanol emissions. The difference between Tier 2 vehicle emissions (affected by the below-zero measurements) and pre-Tier 2 vehicle emissions (not affected by below-zero measurements) seem especially large for ethanol in comparison to the other compounds evaluated. Thus, care should be used in applying these values. There was a significant discontinuity in ethanol emissions for the 2002 Taurus and the 2009 Titan when tested on E40 following fully adapted E10 versus when following fully adapted E76 (Table 4). However, these vehicles did not exhibit this type of discontinuity for any other pollutants suggesting

that perhaps this was due only to the wide range of emissions values measured for ethanol. Table 6 summarizes the changes in emissions for E40 compared to E10 and E76. To calculate p-values, parameters were normalized between vehicles by taking the log of the average for each vehicle/pollutant/fuel combination. The only significant changes in regulated pollutants were a decrease in NOx for E76 vs E40 for vehicles adapted on E10 (and very nearly significant for vehicles adapted on E76), and a decrease in CO for E76 versus E40 for vehicles adapted on E76. Acetone emissions were all below about 0.7 mg/mile and decreased on average 42% as ethanol content in the fuel increased from E10 to E76. Nitrous oxide emissions were nearly flat for all vehicles tested, showing no impact of ethanol content and were relatively unaffected by hot or cold operation. Other organics measured included acrolein, methacrolein, propionaldehyde, benzaldehyde m-, p-, and o-tolualdehyde. 2995

dx.doi.org/10.1021/es304552b | Environ. Sci. Technol. 2013, 47, 2990−2997

Environmental Science & Technology

Article

Figure 4. Effect of ethanol content and adaption on vehicle fuel economy, cold-start LA92 composite.

standards over just this past decade. Thus, care must be taken to account for the effects of fleet turnover as well as changes to existing and projected emission standards (which will apply equally to FFVs operating on E85 and standard vehicles) in projecting air quality impacts based on changes in fuel usage. It is also important to consider the uncertainty associated with the specific estimates of emissions changes between fuel types. For example, for the nine vehicles tested here, the NOx reduction between E76 and E10 averaged −25%, with a range of −54% to +10%. Jacobson modeled the effect of complete conversion of all spark ignition vehicles to E85 in the Los Angeles basin, and found that predicted ozone-related fatalities were very sensitive to the amount of NOx reduction: a 30% reduction in NOx would result in about 120 more ozone-related deaths in 2020 than would occur in the base gasoline case, but only 24 additional ozone-related deaths would occur with an assumed 15% reduction in NOx.8 Extrapolating from these modeling results suggests that for an estimated 10% reduction in NOx with E85, it is possible that the model would predict no change or even reduced ozone deaths. A reduction of 10% in overall NOx with E85 is well within the 95% confidence interval of values measured in this study and historically.6 Moreover, total VOC emissions from all vehicles have been reduced by a factor of 4 over the past 40 years, and VOC emissions per mile by as much as 2 orders of magnitude, a trend which is predicted to continue into the next decade.18 Jacobson points out that the ratio between VOC and NOx is critical in estimating ozone, underlining the need for a careful consideration of the uncertainty of the combined value of the two pollutants in future ozone modeling of fuel changes. Specific toxics are also affected by the same considerations of variations in engine tuning and fuelling algorithms as well as the implementation of new emissions standards that will impact all organic emissions. Nonetheless, it seems apparent that ethanol emissions (which are an atmospheric precursor to aldehydes) and aldehydes will increase with increased use of E85; and benzene and 1,3 butadiene will decrease. Millet and co-workers have found that even with ethanol emission increases consistent with the large increase in ethanol use required by the Energy Independence and Security Act,19 the atmospheric impacts will

Measured tailpipe values below background occurred in numerous tests and were included as zeros prior to averaging with other values as required by the California NMOG test procedure.17 No consistent increasing or decreasing trends for these compounds with changes in fuel ethanol content could be detected. Measured values for these compounds as well as acetone and nitrous oxide are included in the Supporting Information. Implications. There was evidence of incomplete adaptation during the initial LA92 hot test immediately after refueling with E40. However, following the initial LA92 hot test the FFVs, on average, adapted successfully to fueling with midrange blends such as those available at blender pump fueling stations with average emissions falling between those of E10 and E76 in the second test, a cold-start LA92. However, the 2002 Dodge Caravan did not appear to adapt successfully, resulting in higher CO and NMOG emissions than when using either E10 or E76. Further testing of older vehicles will be needed to determine whether this issue is limited to certain makes and models or is widespread among older FFVs. In addition to this apparent lack of adaption in one vehicle, there is also a wide variability in the impact of E76 versus E40 and E10 on emissions between vehicles. Vehicle-to-vehicle variability is caused by different FFV manufacturers having developed different engine control strategies to adapt to changes in fuel energy and oxygen content while balancing good fuel economy and the need to meet all emission standards for both E85 and E10. A full discussion of how fuel impacts on emissions affect air quality is well beyond the scope of this study, but a few general conclusions can be drawn. Poor adaption primarily affected CO and partial combustion product emissions, and it should be noted that even in these cases (for example, CO emissions in the 2002 Caravan using E40) the increase in emissions was about a factor of 2; equal to the change in CO emission limits between the Tier 2 Bin 4 (2.1 g/mile) and Tier 2 Bin 5 emission standards (4.2 g/mile). The point being that the difference in criteria pollutant emissions between E10, E40, and E76 in these tests, whether fully adapted or not, is very small compared to the measured differences between the different vehicle models, and in comparison to the changes in emissions 2996

dx.doi.org/10.1021/es304552b | Environ. Sci. Technol. 2013, 47, 2990−2997

Environmental Science & Technology

Article

(13) Loo, J.; Parker, D. “Evaluation of a Photoacoustic Gas Analyzer for Ethanol Vehicle Emissions Measurement,” SAE Technical Paper 2000−01−0794, 2000, doi: 10.4271/2000-01-0794. (14) http://www.dieselnet.com/standards/cycles/uc.php, accessed January 31, 2013. (15) Yasinne, M. K.; La Pan, M. “Impact of Ethanol Fuels on Regulated Tailpipe Emissions” SAE Technical Paper 2012−01−0872, 2012, doi:10.4271/2012-01-0872. (16) Karavalakis, G.; Durbin, T. D.; Shrivastava, M.; Zheng, Z.; Villela, M.; Jung, H. Impacts of Ethanol Fuel Level on Emissions of Regulated and Unregulated Pollutants from a Fleet of Gasoline LightDuty Vehicles. Fuel 2012, 93, 549−558. (17) California Environmental Protection Agency, Air Resources Board, “California Non-Methane Organic Gas Test Procedure,” July 2002. (18) Wallington, T. J.; Anderson, J. E.; Winkler, S. L. Comment on “Natural and Anthropogenic Ethanol Sources in North America and Potential Atmospheric Impacts of Ethanol Fuel Use. Environ. Sci. Technol. 2012. (19) Public Law 110−140 “Energy Independence and Security Act of 2007, December 19, 2007. (20) Millet, D. B.; Apel, E.; Henze, D. K.; Hill, J.; Marshall, J. D.; Singh, H. B.; Tessum, C. W. Natural and Anthropogenic Ethanol Sources in North America and Potential Atmospheric Impacts of Ethanol Fuel Use. Environ. Sci. Technol. 2012, 46, 8484−8492. (21) Kelly, K.; Eudy, L.; Coburn, T. Light-Duty Alternative Fuel Vehicles: Federal Test Procedure Emissions Results; Report No. NREL/ TP-54−25818, National Renewable Energy Laboratory: Golden, CO, September 1999.

be minimal, because significant sources of atmospheric acetaldehyde already exist.20 Because of the large amount of acetaldehyde emitted with E85, total toxic compound mass emissions are likely to go up, but because acetaldehyde is far less toxic, the potency-weighted toxicity will be reduced.21 Air modeling showed no clear trend in predicted cancer deaths with the use of E85.8

■

ASSOCIATED CONTENT

S Supporting Information *

Additional experimental details as well as emission testing data are provided. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address †

SGS Environmental Testing Corporation, Aurora, CO

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS

■

REFERENCES

This work was supported by the U.S. Department of Energy under Contract No. DE-AC36-99GO10337 with the National Renewable Energy Laboratory. The authors wish to thank Dennis Smith, Co-director of the Clean Cities Program.

(1) AAM Alliance of Automobile Manufacturers North American Fuel Survey, 2011. (2) ASTM D5798−11 “Standard Specification for Ethanol Fuel Blends for Flexible-Fuel Automotive Spark-Ignition Engines” Volume 05.02, 2011. (3) http://www.afdc.energy.gov/afdc/fuels/stations_counts.html . Accessed August 23, 2012. (4) http://www.afdc.energy.gov/technology_bulletin_0210.html . Accessed August 23, 2012. (5) Control of Emissions from New and In-Use Highway Vehicles and Engines. Title 40 Code of Federal Regulations, Part 86, July 1, 2007. (6) Yanowitz, J.; McCormick, R. L. Effect of E85 on Tailpipe Emissions from Light-Duty Vehicles. J. Air & Waste Manage. Assoc. 2009, 59, 172−182. (7) Graham, L. A.; Belisle, S. L.; Baas, C.-L. Emissions from Light Duty Gasoline Vehicles Operating on Low Blend Ethanol Gasoline and E85. Atmos. Environ. 2008, 42, 4498−4516. (8) Jacobson, M. Z. Effects of Ethanol (E85) versus Gasoline Vehicles on Cancer and Mortality in the United States. Environ. Sci. Technol. 2007, 41, 4150−4157. (9) Haskew, H. M.; Liberty, T. F. Exhaust and Evaporative Emissions Testing of Flexible-Fuel Vehicles, CRC Report No. E-80, August, 2011. (10) Alleman, T. L. National 2010−2011 Survey of E85: CRC Project E-85−2, National Renewable Energy Laboratory: Golden, CO. NREL Technical Report NREL/TP-5400−52905, December 2011. (11) Regulation of Fuels and Fuel Additives. What are the small refiner gasoline sulfur standards? Title 40 Code of Federal Regulations, Part 80 (80.240), July 1, 2005. (12) Yassine, M., Nayfeh, K.; Michell, R.; Schubert, E. “Integration of Photoacoustic Innova Analyzer Within Bag Bench for Direct Measurement of Ethanol in Vehicle Emissions,” SAE Technical Paper 2009−01−1518, 2009, doi: 10.4271/2009-01-1518. 2997

dx.doi.org/10.1021/es304552b | Environ. Sci. Technol. 2013, 47, 2990−2997