- Email: [email protected]
Refueling emissions from cars in Japan: Compositions, temperature dependence and effect of vapor liquefied collection system
Atmospheric Environment 120 (2015) 455e462
Contents lists available at ScienceDirect
Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv
Refueling emissions from cars in Japan: Compositions, temperature dependence and effect of vapor liquefied collection system Hiroyuki Yamada a, *, Satoshi Inomata b, Hiroshi Tanimoto b a b
National Traffic Safety and Environment Laboratory, 7-42-27, Jindaiji-higashimachi, Chofu, Tokyo 182-0012, Japan National Institute for Environmental Studies, 16-2 Onogawa, Tsukuba, Ibaraki 305-8506, Japan
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Refueling emission factor in Japan was 1.02 ± 0.40 g/L and MIR was 3.49 ± 0.83. C4 alkene had the highest impact on the OFP of refueling emissions. Refueling emissions changed with temperature but MIR was almost constant. MIRrefueling z MIRbreakthrough z MIRfuel þ 0.5. The efficiency of a vapor liquefied collection system was similar to that of an ORVR.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 May 2015 Received in revised form 5 September 2015 Accepted 7 September 2015 Available online 9 September 2015
Refueling emissions from cars available on the Japanese market, which were not equipped with specific controlling devices, were investigated. For the composition analysis, a proton transfer reaction plus switchable reagent ion mass spectrometry (PTR þ SRI-MS), which is capable of real-time measurement, was used. In addition, the performance of a vapor liquefied collection system (VLCS), which is a recently developed controlling device, was evaluated and compared with an onboard refueling vapor recovery (ORVR) system. The refueling emission factor of uncontrolled vehicles at 20 C was 1.02 ± 0.40 g/L in the case dispensing 20 L of fuel. The results of composition analysis indicated that the maximum incremental reactivity (MIR) of refueling emissions in Japan was 3.49 ± 0.83. The emissions consist of 80% alkanes and 20% alkenes, and aromatics and di-enes were negligible. C4 alkene had the highest impact on the MIR of refueling emissions. The amounts of refueling emissions were well reproduced by a function developed by MOVE2010 in the temperature range of 5e35 C. The compositions of the refueling emissions varied in this temperature range, but the change in MIR was negligible. The trapping efficiency of VLCS was the same level as that of the ORVR (over 95%). The MIRs of refueling and evaporative emissions were strongly affected by that of the test fuel. This study and our previous study indicated that MIRbreakthrough z MIRrefueling z MIRfuel þ 0.5 and MIRpermeation z MIRfuel. The real-world estimated average MIRfuel in Japan was about 3.0. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Refueling emissions Vapor liquefied collection system ORVR PTRþSRI-MS Ozone formation potential
1. Introduction * Corresponding author. E-mail address: [email protected] (H. Yamada). http://dx.doi.org/10.1016/j.atmosenv.2015.09.026 1352-2310/© 2015 Elsevier Ltd. All rights reserved.
Efforts have been made in many countries to reduce emissions of volatile organic compounds (VOCs) to improve the ozone
456
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
situation. The emissions from automobiles are thought to be one of the major sources of VOCs. It is known that the VOCs from automobiles are emitted not only from the tailpipe during driving but also from the fuel tank of gasoline vehicles during parking and refueling. These are called evaporative emissions and refueling emissions, respectively. These emissions have been studied since the 1980's, mainly in the USA (Braddock et al., 1986; Braddock, 1987; Berglund and Petersson, 1990; Furey and Perry, 1991; Burns et al., 1992). For the evaporative emissions, recent studies have paid attention to estimating the amount in real-world conditions. Our previous study (Yamada, 2013) showed that the total evaporative emissions in Japan are higher than the tailpipe emissions, accounting for 4.6% of VOC emissions from stationary sources. In Europe, the evaporative emissions in real-world conditions often exceed the regulation limits (Martini et al., 2014). Dong et al. (2015) proposed an updated method for estimating the emissions from real parking activity data more accurately. On the other hand, there have been hardly any recent studies on refueling emissions. This is probably because prevention techniques have been already established and introduced in many countries. The strategy for preventing refueling emissions in the USA is to employ an onboard refueling vapor recovery (ORVR) system which traps the vapor displaced by the liquid fuel entering into the tank by using a carbon canister installed in the car. Unlike the USA, in Europe, the refueling emissions are vacuumed by a stage II system installed on a gasoline dispenser (European Parliament, 2009). These two are the major techniques in use, and usually countries which wish to control refueling emissions choose one of them after considering their merits and demerits. The ORVR system exhibits high trapping efficiency (95%). However, the benefits of ORVR will only be fully delivered after most vehicles on the road are replaced with those having ORVR, and this may take more than 10 years. In contrast, although the real world trapping efficiency of stage II systems is only 70%, they can be implemented more quickly (Fung and Maxwell, 2011). The theoretical efficiency of stage II system is similar to that of ORVR; in reality, however, they are often not wellmaintained, which results in lowered efficiency. Japan is one of the countries where refueling emissions have not been controlled, although its ozone situation is critical (Ministry of Environment, Japan, 2011). In Japan, gas stations emitted 107,082 ton/year out of the total VOC emissions of 736,612 ton/year (Ministry of Environment, Japan, 2014). Therefore, the introduction of prevention techniques has been demanded by local governments in Japan (Kanagawa Prefectural Government, 2014), and discussions will begin in the near future. However, there is no data on the effectiveness of control techniques for refueling emissions in Japan. As described above, there were many studies discussing the effectiveness of these techniques in the USA in the 1980's. However, the current situation in Japan is quite different from the situation in the USA in the 1980's, such as car specifications, fuel composition, and fuel dispenser specifications. The results of these evaluations will be useful not only to Japan but also to many other countries that have not yet introduced refueling emission controls. To discuss VOC emissions from various sources, it is important to consider the ozone formation potential (OFP) (Adam et al., 2011; Zhang et al., 2013; Gentner et al., 2013; Costagliola et al., 2014; Li et al., 2015), but there have been hardly any studies that discussed the OFP of refueling emissions in Japan. Even in the studies performed outside Japan, in most cases, the composition analyses were made using traditional analytical chemistry methods, such as GC-MS. Our previous study demonstrated that measurement of evaporative emissions using proton transfer reaction plus switchable reagent ion mass spectrometry (PTR þ SRI-MS) can be performed in real-time. Also, the results indicated the possibility of underestimating the concentrations of relatively high-molecular-
weight compounds due to adsorption on the walls of the variable-temperature sealed housing evaporative determination (VT-SHED) equipment (Yamada et al., 2015). Therefore, real-time measurements of VOCs are desirable in evaporative emission experiments. The Japanese gasoline dispenser maker Tatsuno has developed a new prevention device that can be deployed by replacing existing dispensers with new dispensers equipped with the device, even though there have been no regulations concerning refueling emissions in Japan. Their approach uses a vapor liquefied collection system (VLCS). Similar to the stage II system, the refueling emissions are vacuumed by a muffler-shaped inlet surrounding the nozzle. VLCS may be an effective tool; however, no evaluation of its emission preventing performance has been carried out. In this study, the refueling emissions from eight Japanese vehicles not equipped with refueling emission control systems were measured using a VT-SHED and a gasoline dispenser. In these experiments, the maximum incremental reactivity (MIR; Carter, 2010; Carter and Heo, 2013), which is one of the major indicators of OFP, was estimated from the results of real-time composition analysis with the PTR þ SRI-MS. Changes in the amounts of refueling emissions and their compositions were also observed in the temperature range from 5 to 35 C. In addition, the performance of the VLCS was compared with that of the ORVR. 2. Experimental methods 2.1. Refueling emissions measurements The refueling emissions from gasoline vehicles were measured by placing a test vehicle in a VT-SHED (VSH-9353; Hitachi Technology Engineering Inc., Tokyo, Japan) that met the requirements for the Japanese approval tests for evaporative emissions but that was not designed for refueling emission measurements. The VOC emissions were determined by measuring the increase in the total hydrocarbon (THC) concentration in the SHED using a gas analyzer (MEXA-1160TFL-L; Horiba Inc., Kyoto, Japan). A gasoline dispenser (MAB36621EVMBDTX0001A; Tatsuno Corporation, Tokyo, Japan) was set next to the SHED and the dispenser nozzle was brought into the SHED through a squareshaped hole with approximate dimensions of 150 mm 150 mm just before the test started. The hole was loosely sealed with a polytetrafluoroethylene sheet. Any leak from the hole was considered to be negligible because the hole was small enough compared with the SHED volume of 93.53 m3 and because of the relatively short test durations of approximately 5 min. The test was started upon removing the fuel tank cap and then inserting the nozzle of the dispenser into the tank as quickly as possible. Initially almost 10% of the fuel was left in the tank, and 20 L of gasoline was filled into the tank The gasoline flow rate was set to 35 L/min, which is the rate generally used in Japan. After refueling, the nozzle was left inserted, and the test ended when the indicated THC stabilized. The test fuel was specially blended for type approval tests, and its temperature was set equal to the SHED temperature. This procedure roughly agrees with that in the US. The differences are, in the US case, background measurement is performed with the tank cap opened and fuel is dispensed until automatic nozzle shut-off occurs (US Government Publishing Office, 1995). The shape of the nozzle was almost the same as the stage II nozzle which is surrounded with a muffler-shaped vacuum inlet. The vacuumed flow was sent to the VLCS (QE-1001; Tatsuno Corporation, Tokyo, Japan) at a flow rate of 40 L/min. The same nozzle was used even when the VLCS was off, for conducting measurements of the ORVR and the uncontrolled case. Alkanes, alkenes and di-ene species concentrations from carbon
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
numbers 3 to 11 were monitored in the SHED with PTR þ SRIeMS using NOþ as a reagent ion. Benzene, alkyl benzenes, naphthalene, and alkyl naphthalenes were also monitored. The details of the measurement technique with PTR þ SRI-MS can be found in our previous publication (Yamada et al., 2015). The detection sensitivities for some hydrocarbons were determined under ambient humid conditions: 3.9 ± 0.1 (benzene, m/z 78), 5.2 ± 0.3 (toluene, m/z 92), 4.8 ± 0.7 (p-xylene, m/z 106), 4.3 ± 0.9 (1,3,5-trimethylbenzene, m/z 120), 5.4 ± 0.1 (isoprene, m/z 68), 5.1 ± 0.5 (3-hexene, m/z 84), and 5.2 ± 0.6 ncps ppbv1 (3-heptene, m/z 99). Error limits represent 95% confidence levels determined by Student's t-test. The detection sensitivity of 4.3 ± 0.9 ncps/ppbv was used as the typical detection sensitivity for alkylbenzenes, naphthalene, and alkylnaphthalene, but not benzene, toluene, xylene, and trimethylbenzene. The detection sensitivity of 5.2 ± 0.7 ncps/ppbv was used as the typical detection sensitivity for alkenes and di-enes. For alkanes, the values reported in Yamada et al. (2015) were used. 2.2. Tested vehicles
457
study, the vacuum flow rate was set to 40 L/min, whereas the gasoline dispensing rate was 35 L/min. Another merit of the VLCS is that it is compatible with the ORVR system. In the case of the stage II system, vehicle fueling with ORVR results in a lower overall trapping efficiency (US Environmental Protection Agency, 2012b) because sometimes clean air is vacuumed into the underground tank by the stage II system and is mixed with fuel vapor. Then, it is released to the atmosphere without filtration to make room for newly vacuumed air. On the other hand, VLCS always exhausts the air after filtration of the fuel vapor. Therefore, this drop of overall efficiency does not occur in the case of VLCS in conjunction with the ORVR system. In addition, an uncontrolled dispenser can be equipped with a VLCS simply by replacing the dispenser with one equipped with a VLCS. This is easier than installing a stage II system, which requires providing a separate line to the underground tank.
3. Results and discussion 3.1. Real-world refueling emissions in Japan
The list of vehicles is shown in Table 1. OR was the only vehicle equipped with an ORVR system complying with the refueling regulations in the USA. The others had no refueling emission control systems. The tested vehicles, except OR, are widely used in Japan. The ages and mileages of these vehicles varied from 0.7 to 11.9 year and 6046 to 186,596 km, respectively. UF was a so-called kei car (light automobile), a category of small car peculiar to Japan; light automobiles must have an engine displacement of 0.66 L or less and a size smaller than 3.4 m in length, 1.48 m in width, and 2 m in height. The ratio of light automobiles in Japan has increased recently, and 40% of all new cars sold in 2014 were in this category. 2.3. Vapor liquefied collection system The VLCS is a new type of device attached to fuel dispensers for controlling the refueling emissions. The operation of the VLCS is shown in Fig. 1. Similar to the stage II system, the refueling emissions are vacuumed by the muffler-shaped inlet surrounding the nozzle. The vacuumed emissions are not sent to an underground tank but to a condenser in the VLCS. At the condenser, the emissions are liquefied and mixed with the gasoline flowing from the underground tank to the dispenser. The exhaust from the condenser is sent to a gasoline trap, and unliquefied gasoline vapor is removed from the exhaust. The trapped gasoline vapor is sent back to the condenser. In the case of the conventional stage II system, a vacuum flow is generated by the airflow that replaces the decreasing volume of gasoline in the underground tank during refueling. Thus, the maximum vacuum flow rate cannot exceed the refueling flow rate. On the other hand, the vacuum flow rate of the VLCS is independent of the refueling flow rate and can be set higher than that of the stage II system. Theoretically, the increased vacuum flow rate results in a higher trapping efficiency than the stage II system. In this
We evaluated the refueling emissions from eight cars available on the Japanese market. The temperature in the SHED was set to 20 C. In this experiment, residual fuel in the tested vehicles' tanks, which affected the refueling emissions, was not replaced with the test fuel, because the aim of these experiments was to determine the emission factors that can be applied to real-world refueling emissions. Also, measuring the emissions with residual fuel is closer to the real-world conditions. Consequently, the results contain variations due to both differences in the vehicles and differences in the fuels. The results are shown in Table 2. The average value and 2s were 1.02 and 0.40 g/L, respectively. The detailed mass-based composition ratios of refueling emissions at 20 C are listed in Table 3, accompanied by mass ratio MIR. The MIR values shown were
Vapor
VLCS
Vapor
Dispenser
Gasoline
Exhaust
Pump Condenser
Gasoline Trap Gasoline Underground Tank
Release Valve
Liquefied Vapor
Fig. 1. Operation of VLCS.
Table 1 Specifications for the nine test vehicles used in this study. Name
OR
UA
UB
UC
UD
UE
UF
UG
UH
Category Maker Mileage (km) Age (year) Tank Volume (L) Displacement (L) Vehicle Weight (kg)
SUV Ford 3820 0.2 70 2.0 2020
Compact Mitsubishi 11,543 4.0 45 1.5 1090
Compact Toyota 46,528 7.2 40 1.5 1050
mini-van Toyota 54,352 7.8 55 2.0 1580
Van Toyota 186,596 9.8 70 2.0 2140
mid-size Toyota 39,046 6.9 50 1.8 1260
Kei Honda 6046 0.7 35 0.66 970
Wagon Toyota 25,881 11.9 70 2.4 1770
Wagon Honda 16,361 7.9 70 2.4 1870
458
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
refueling emissions (MIRrefueling), and permeation emissions (MIRpermeation) are related by:
taken from a report by the California Environmental Protection Agency (2009). For the alkanes, the values of straight and branched alkanes were averaged according to the ratio of test fuel. The averaged total MIR and 2s of refueling emissions were estimated to be 3.49 and 0.83, respectively. The effect of experimental errors was typically ±8%, as shown later. C4 alkene had the highest impact on the MIR of refueling emissions. The mass ratio and mass ratio MIR of the emissions from each vehicle are shown in Fig. 2. In the mass ratio results, there was no large difference among the vehicles, and the main components were alkanes, which occupied up to 80%, and the rest were alkenes. Aromatics and di-enes were negligible. On the other hand, in the mass ratio MIR results, alkenes showed the highest impact, at almost 70%.
where MIRfuel is the MIR of the test fuel. The average MIRrefueling with the tank filled with real-world fuel shown in Table 1 is 3.49 ± 0.83. Using this value and eq. (1), the estimated MIRfuel of real-world fuels is about 3.0 on average, suggesting that the test fuels used in this and previous studies were likely to have higher MIRs than the average MIR of real-world fuels.
3.2. Effects of fuel variety on MIR values of refueling, breakthrough and permeation emissions
3.3. Temperature dependence of the refueling emissions and compositions
To observe the features of refueling emissions, the composition was compared with that of the test fuel. The fuel was analyzed by GC-MS using the method outlined in the Japanese Industrial Standard JIS K2536e2:2003 (Japanese Standards Association, 2003), and the refueling emissions from UB were measured by PTR þ SRIMS. To avoid the effects of different residual fuel in the tank, the fuel was replaced with the test fuel before the experiment. The SHED temperature and the fuel temperature were set to 20 C. The test was conducted only once, and similar experiments were conducted at different temperatures to obtain the temperature dependence, as shown later. Compared with these results, the data obtained at 20 C seems to have acceptable accuracy. The results are shown in Fig. 3. The error limits in this figure were calculated by propagating the errors of the ion signals and the detection sensitivities for each compound, but the error caused by the fact that typical detection sensitivities were used for alkenes, di-enes, and some aromatics was not included. The detailed composition of the refueling emissions was quite different from that of the fuel. The ratio of aromatic compounds in the refueling emissions was less than 10% and the main components were C4eC6 alkanes and alkenes. These features were similar to the breakthrough emissions of evaporative emissions observed in our previous study (Yamada et al., 2015). The estimated MIR of the refueling emissions was 4.21 ± 0.35 higher than that of the test fuel (3.78). The error was calculated by summing the errors for each compound. The error was smaller than the variations of test results with the eight cars shown in Table 3. Fig. 3 also shows the results for the same vehicle (UB) with the tank filled with the real-world fuel shown in Fig. 2. The estimated MIR in this case was 3.83 ± 0.31, which is smaller than the case where the tank was filled with the test fuel (4.21 ± 0.35). This smaller MIR was caused by a smaller C4 and C5 alkene ratio, whose MIRs are quite high. It seems that this difference was due to the different fuel composition. In our previous study (Yamada et al., 2015), we obtained MIRs of breakthrough and permeation emissions of 3.89 and 3.31, respectively using a fuel whose MIR was 3.33. The difference in MIR between the breakthrough emissions and the test fuel was almost the same as that between the refueling emissions and the test fuel in this study. The MIRs of breakthrough emissions (MIRbreakthrough),
The amounts of uncontrolled refueling emissions and their compositions in the temperature range from 5 to 35 C were measured using the vehicle UB. Similar to the experiment in Fig. 3, residual fuel in the tank was replaced with the test fuel to avoid fluctuations caused by different residual fuel compositions. The THC results are shown in Fig. 4, together with a couple of calculated results. One of the functions used for the estimation was taken from the results of MOVE 2010 in the USA (US Environmental Protection Agency, 2012a). The refueling emissions can be expressed as:
Table 2 The refueling emissions from eight cars which were not equipped with emission control devices. The test temperature was 20 C. The error limits of the average are represented by 2s (experimental uncertainties are not included). Vehicle
UA
UB
UC
UD
UE
UF
UG
UH
Average
Emissions (g/L) 0.95 1.03 1.06 1.41 0.75 1.14 0.98 0.86 1.02 ± 0.40
MIRbreakthrough zMIRrefueling zMIRfuel þ 0:5
(1)
MIRpermeation zMIRfuel
(2)
HCuncontrol ¼ 5:909 0:0949 TDFDIF þ 0:0884 DFTEMP þ 0:485 RVP
(3)
where HCuncontrol is the refueling emissions (g/L), DFTEMP is the dispensed gasoline temperature ( F), RVP is the raid vapor pressure of the test fuel (psi), and TDFDIF is the tank gasoline temperature ( F), defined as:
TDFDIF ¼ 0:418 DFTEMP 16:6
(4)
Another function has been used for estimating the VOC inventory in Japan (Ministry of Environment, Japan, 2014), namely:
. HCuncontrol ¼ 0:97 Tref þ 11:12 21
(5)
where Tref is the ambient temperature ( C). The estimated results obtained with (eq. (3)) well represented the experimental results, whereas (eq. (5)) overestimated the results by a factor of almost 1.5. The emissions of total aromatics, alkanes, alkenes and di-enes from UB as a function of temperature are shown in Fig. 5 as the ratio of each emission at 20 C. Aromatics showed a gradual increase with increasing temperature. The other emissions seemed to be constant, having no dependence on the temperature. However, the detailed emissions shown in Fig. 6 indicated that the compositions changed as the temperature changed, even in alkanes, alkenes and di-enes. These three groups exhibited common features. At low temperature, the emissions of higher mass species were lower than those at 20 C, and the emissions of smaller mass species were also higher. As a result, the total emissions of the groups were similar to those at 20 C. The emissions at temperatures higher than 20 C were similar to those at 20 C. In the results for the aromatics shown in Fig. 6 (a), the emission of benzene was constant in this temperature range. The other species showed increasing emissions as the temperature increased. This was more apparent in species with higher mass. Fig. 7 shows the MIR of refueling emissions calculated using the results shown in Fig. 6 as a function of the temperature. There was no significant temperature
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
459
Table 3 Average species composition ratio (mass ratio) and the mass ratio MIR of the refueling emissions for eight Japanese cars available on the market, obtained by PTR þ SRI-MS. SHED temperature and fuel temperature were 20 C. The error limits of the average are represented by 2s (experimental uncertainties are not included). Aromatics MIRb Mass ratio Ratio MIR Alkanes MIR Mass ratio Ratio MIR Alkenes MIR Mass ratio Ratio MIR Di-ene MIR Mass ratio Ratio MIR
a M78 (benzene)
M92 (Toluene)
M106 (C8-benzene)
M120 (C9-benzene)
M134 (C10-benzene)
M128 (naphthalene)
Aromatic compounds total
0.72 0.001 ± 0.001 0.00
4 0.009 ± 0.009 0.04
6.57 0.002 ± 0.001 0.01
6.23 0.001 ± 0.000 0.00
7.49 0.000 ± 0.000 0.00
3.34 0.000 ± 0.000 0.00
0.013 ± 0.009 0.05 ± 0.04
M58 (C4)
M72 (C5)
M86 (C6)
M100 (C7)
M114 (C8)
Alkanes total
1.17 0.425 ± 0.214 0.50
1.40 0.220 ± 0.148 0.31
1.30 0.048 ± 0.022 0.06
1.45 0.007 ± 0.002 0.01
1.44 0.003 ± 0.003 0.00
0.766 ± 0.262 0.91 ± 0.33
M56 (C4)
M70 (C5)
M84 (C6)
M98 (C7)
M112 (C8)
Alkenes total
11.66 0.140 ± 0.050 1.59
12.22 0.056 ± 0.039 0.67
8.84 0.014 ± 0.009 0.12
6.98 0.002 ± 0.000 0.01
5.37 0.001 ± 0.000 0.00
0.214 ± 0.065 2.47 ± 0.76
M54 (C4)
M68 (C5)
M82 (C6)
M96 (C7)
M110 (C8)
Di-ene total
Total
12.61 0.000 ± 0.000 0.00
11.028 0.003 ± 0.004 0.03
8.68 0.002 ± 0.001 0.02
7.29 0.001 ± 0.000 0.00
4.89 0.000 ± 0.000 0.00
0.006 ± 0.004 0.06 ± 0.04
3.49 ± 0.83
Bold indicates sum of horizontal cells. a M ¼ molecular weight; bMIR ¼ maximum incremental reactivity.
dependence of the MIR of the refueling emissions (Slope: 0.005 ± 0.021), although temperature dependence of the
composition was observed as shown in Fig. 6.
3.4. Evaluation of VLCS
Fig. 2. Mass ratio (a) and mass ratio MIR (b) of the refueling emissions obtained by PTR þ SRI-MS from eight cars available in the Japanese market. SHED temperature and fuel temperature were 20 C. Error bars indicates 2s.
The performances of the VLCS and ORVR were evaluated in the temperature range from 8 to 33 C. For the tested vehicles, OR, which was equipped with the ORVR complying with the regulations in the USA, and UA were used. The regulation limit in the USA is 0.05 g/L (0.2 g/gal) at 26.7 C (80 F). In the experiments with UA, the VLCS was used to accumulate the refueling emissions, and it was switched off with OR. The results are listed in Table 4. The emissions with the VLCS and ORVR relative to uncontrolled emissions are shown in Fig. 8. The emissions with the ORVR were almost 5% and were independent of the temperature. This result agrees with a former observation (US Environmental Protection Agency, 2012b) and is below the regulation limit in the USA, except for the results at 32.9 C. The emissions with the VLCS were mostly lower than those with the ORVR. Also, the trapping efficiency of the VLCS showed a temperature dependence. Fig. 9 show the emissions from the ORVR and the uncontrolled case when the fuel cap was opened but the nozzle of the dispenser was not inserted in the tank, at 20 C. OR was equipped with a system that restrict these emissions, and the emission rate was small enough compared with the refueling emissions listed in Table 4. On the other hand, the emission in the uncontrolled case was comparable with the emission of the VLCS. Vehicle UA, which was used for evaluating the VLCS, did not have a system for restricting these emissions, and it is suggested that the emissions before inserting the nozzle are a major source of refueling emissions with the VLCS. Also, the observed temperature dependence of the VLCS trapping efficiency seems to be caused by the temperature dependence of these emissions. As a result, the VLCS had a similar level of performance in preventing refueling emissions to the ORVR, suggesting that the VLCS has a higher trapping efficiency than the stage II system. Thus, the VLCS will become one of the choices for controlling refueling emissions. The cost associated with introducing one VLCS system is 365 million yen. There are 119,000 fuel dispensers in Japan (Yugyohouchisimbun, Co, Ltd, 2015). To replace all of them, the
460
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
0.12
0.50
(a) Aromatics
UB(Tested fuel) UB(Real world) Fuel
0.10
Mass ratio
Mass ratio
0.04
0.30 0.20 0.10
0.02 0.00
0.00 M78
0.20
UB(Tested fuel) UB(Real world) Fuel
0.40
0.08 0.06
(b) Alkanes
M92 M106 M120 M134 M128 M148
C4 0.90
(c) Alkenes
C7
C8
C9
(d) Comparisons of groups UB(Tested fuel) UB(Real world) Fuel
0.70 0.60
Mass ratio
Mass ratio
0.10
C6
0.80
UB(Tested fuel) UB(Real world) Fuel
0.15
C5
0.50 0.40 0.30
0.05
0.20 0.10
0.00 C4
C5
C6
C7
C8
C9
0.00
Aromatics total Alkanes total
Alkenes total
Fig. 3. Ratio of refueling emissions at 20 C from UB filled with the test fuel and the real-world fuel: (a) aromatic compound, (b) alkanes, (c) alkenes, and (d) summary of the groups. The compositions of the test fuel are also shown. Estimated MIRs from these results were 3.78 for the test fuel, 4.21 ± 0.35 for the refueling emissions with the test fuel and 3.83 ± 0.31 for the refueling emissions with real-world fuel for residual. The error bars were calculated from the errors of ion signals and detection sensitivities of each compound.
total cost would be almost 434 billion yen. On the other hand, the cost of ORVR is 6e8 dollars per vehicle (Fung and Maxwell, 2011), and the total cost of introducing the ORVR systems would be almost 58.5 billion yen if replacing the carbon canisters of all gasoline vehicles in Japan (66.8milion; Automobile Inspection & Registration Information Association, 2013). By assuming the efficiencies of ORVR and VLCS are similar according to the results shown above, the ORVR is more cost-effective than the VLCS. But this rough analysis excluded effect of the differences in spreading
Fig. 4. Experimental and estimated results of uncontrolled refueling emissions (Ministry of Environment, Japan, 2014; US Environmental Protection Agency, 2012a).
period and maintenance cost. And a more comprehensive comparative analysis of ORVR and VLCS, taking the factor into account, is needed. 4. Conclusions The refueling emissions were measured from eight gasoline cars available on the Japanese market and a car equipped with an ORVR,
Fig. 5. Uncontrolled refueling emissions of total aromatics, alkanes, alkenes and dienes from UB as a function of temperature. The error bars were calculated by summing the errors for each compound.
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
461
Fig. 6. Detailed uncontrolled refueling emissions of (a) aromatics, (b) alkanes, (c) alkenes, and (d) di-enes from UB as a function of temperature.
Fig. 7. MIR of uncontrolled refueling emissions as a function of temperature. The error bars were calculated by summing the errors for each compound. Fig. 8. Temperature dependences of the refueling emissions with the VLCS and ORVR relative to the uncontrolled emissions. Table 4 The refueling emissions of the uncontrolled case, the VLCS, and the ORVR at various temperatures. Temperature ( C)
Uncontrolled
VLCS
ORVR
8.2 12.3 20.0 24.5 32.9
0.64 0.69 0.95 0.93 1.24
0.000 0.014 0.011 0.024 0.066
0.037 0.029 0.037 0.039 0.059 Unit: g/L
using VT-SHED and a gasoline dispenser. The average emission from the Japanese cars at 20 C was 1.02 ± 0.40 g/L.
PTR þ SRI-MS was adopted for the first time to observe the refueling emission compositions. The main components in the refueling emissions were alkanes, which occupied almost 80% of the mass. The rest was mostly alkenes, but C4 alkene showed the highest impact on the MIR of refueling emissions. Aromatics and dienes were negligible. The average refueling MIR of the eight cars, whose tanks were filled with Japanese real-world fuel, was 3.49 ± 0.83. When using a test fuel whose MIR was 3.78, the MIR of refueling emissions was 4.21 ± 0.35. The MIR of the emissions was higher than that of the fuel because the decreased ratio of aromatics resulted in a higher ratio of small alkenes, whose MIRs are high.
462
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
Fig. 9. Emissions of uncontrolled case and the ORVR while the fuel cap was opened and the dispensing nozzle was not inserted, as a function of duration.
This feature was similar to breakthrough emissions observed in a previous study. The MIRs of refueling and evaporative emissions were strongly affected by that of test fuel. They can be calculated with the equations MIRbreakthrough z MIRrefueling z MIRfuel þ 0.5 and MIRpermeation z MIRfuel. The average MIRfuel of Japanese real-world gasoline was 3.0. The refueling emissions in the experiments were well reproduced by a function developed in MOVE2010. The function conventionally used for estimating the VOC inventory in Japan overestimated the emission by a factor of almost 1.5. The compositions of the emissions changed with temperature. The emissions of aromatics increased as the temperature increased. Total alkane, alkene and di-ene was almost constant in this temperature range. Detailed composition analyses of alkanes, alkenes, and di-enes indicated that the emissions of higher-mass species were low and those of lower-mass species were high, at temperatures below 20 C. The MIR of the refueling emissions decreased slightly with increasing temperature. The performance of VLCS was evaluated for the first time. The refueling emissions with VLCS were almost 2% of the uncontrolled case, and the results were below the regulation limit of ORVR in the USA. A rough cost analysis indicated that the ORVR is more cost effective, but the VLCS can be one of the choices for controlling refueling emissions. The emissions with the VLCS seem to be due to the emissions occurring when opening the tank cap to insert the dispenser nozzle into the tank. Acknowledgment The authors are grateful to Tatsuno Corporation and Ford Japan Ltd. for providing the experimental instruments. References Automobile Inspection & Registration Information Association, 2013. Monthly Report of Vehicle Numbers, June 2013. Japanese Automobile Inspection & Registration Information Association, Tokyo (in Japanese). Adam, T.W., Astorga, C., Clairotte, M., Duane, M., Elsasser, M., Kransenbrink, A., Larsen, B.R., Manfredi, U., Martini, G., Montero, L., Sklorz, M., Zimmermann, R., Peruji, R., 2011. Chemical analysis and ozone formation potential of exhaust from dual-fuel (liquefied petroleum gas/gasoline) light duty vehicles. Atmos.
Environ. 45, 2842e2848. Berglund, P.M., Pertersson, G., 1990. Hazardous petrol hydrocarbons from refueling with and without vapour recovery. Sci. Total Environ. 91, 49e57. Braddock, J.N., 1987. PB87e212775. Factors Influencing the Composition and Quantity of Passenger Car Refueling Emissions e Part 2. Atmospheric Sciences Research Laboratory, U. S. Environment Protection Agency. Research Triangle Park NC 27711. Braddock, J.N., Gabele, P.A., Lemmons, T.J., 1986. Factors Influencing the Composition and Quantity of Passenger Car Refueling Emissions ePart 1. SAE Technical Paper 861558. Burns, V.R., Gorse, R.A., Reuter, R.M., Hochhauser, A.M., Benson, J.D., Koehl, W.J., Painter, L.J., 1992. Effects of Gasoline Composition on Evaporative and Running Loss Emissions e Auto/oil Air Quality Improvement Research Program. SAE Technical Paper 920323. California Environmental Protection Agency, 2009. Tables of Maximum Incremental Reactivity (MIR) Values. http://www.arb.ca.gov/regact/2009/mir2009/mir2009. htm (accessed 19.04.15.). Carter, W.P.L., 2010. Development of the SAPRC-07 chemical mechanism. Atmos. Environ. 44, 5324e5335. Carter, W.P.L., Heo, G., 2013. Development of revised SAPRC aromatics mechanisms. Atmos. Environ. 77, 404e414. Costagliola, M.A., Murena, F., Prati, M.V., 2014. Exhaust emissions of volatile organic compounds of powered two-wheelers: effect of cold start and vehicle speed. Contribution to greenhouse effect and tropospheric ozone formation. Sci. Total Environ. 468e469, 1043e1049. Dong, X., Tschantz, M., Fu, J.S., 2015. Estimating evaporative vapor generation from automobiles based on parking activities. Environ. Pollut. 202, 104e111. European Parliament, 2009. Council Directive 2009/126/EC of the European Parliament and of the Council of 21 October 2009 on Stage II Petrol Vapour Recovery during Refuelling of Motor Vehicles at Service Stations. Fung, F., Maxwell, B., 2011. Onboard Refueling Vapor Recovery: Evaluation of the ORVR Program in the United States. The International Council on Clean Transportation Working Paper 2011-12, available at: http://www.theicct.org/ onboard-refueling-vapor-recovery-evaluation-orvr-program-united-states. Furey, R.L., Perry, K.L., 1991. Composition and Reactivity of Fuel Vapor Emissions from Gasoline e Oxygenated Blends. SAE Technical Paper 912429. Gentner, D.R., Worton, D.R., Isaacman, G., Davis, L.C., Dallmann, T.R., Wood, E.C., Herndon, S.C., Goldstein, A.H., Harley, R.A., 2013. Chemical composition of gasphase organic carbon emissions from motor vehicles and implications for ozone production. Environ. Sci. Technol. 47, 11837e11848. Japanese Standards Association, 2003. Liquid Petroleum Products e Testing Method of Components Part 2: Determination of Total Components by Gas Chromatography. Japanese Standards Association, Tokyo. JIS K2536-2:2003. Kanagawa Prefectural Government, 2014. Activities for Spreading the ORVR (in Japanese). Available at: http://www.pref.kanagawa.jp/cnt/f531508/p852505. html. Li, L., Ge, Y., Yang, M., Li, J., Peng, Z., Song, Y., Zhang, L., 2015. Effect of gasoline/ methanol blends on motorcycle emissions: exhaust and evaporative emissions. Atmos. Environ. 102, 79e85. Martini, G., Paffumi, E., De Gennaro, M., Mellios, G., 2014. European type-approval test procedure for evaporative emissions from passenger cars against realworld mobility data from two Italian provinces. Sci. Total Environ. 487, 506e520. Ministry of the Environment, Japan, 2011. Air Quality Monitoring Report. Ministry of the Environment Japan, Tokyo (in Japanese). Ministry of the Environment, Japan, 2014. Report on the VOC Emission Inventory in the 2013 Fiscal Year. Ministry of the Environment Japan, Tokyo (in Japanese). US Environmental Protection Agency, 2012a. Development of Evaporative Emissions Calculations for the Motor Vehicle Emissions Simulator MOVES2010. EPA420-R-12-027, available at: http://www.epa.gov/oms/models/moves/ documents/420r12027.pdf. US Environmental Protection Agency, 2012b. Rulemaking. Air Quality: Widespread Use for Onboard Refueling Vapor Recovery and Stage II Waiver. EPA-HQ-OAR2010-1076, available at: https://www.federalregister.gov/articles/2012/05/16/ 2012-11846/air-quality-widespread-use-for-onboard-refueling-vapor-recoveryand-stage-ii-waiver. US Government Publishing Office, 1995. Measurement Procedure; Refueling Tests. Title 40-(Chapter 1)-Subchapter C - Part 86. 154e98 of the US Code of Federal Regulations (CFR), available at: http://www.ecfr.gov/cgi-bin/text-idx? SID¼eb43af3600813c84aba33e893b216400&mc¼true&node¼se40.19.86_ 1154_698&rgn¼div8. Yamada, H., 2013. Contribution of evaporative emissions from gasoline vehicles toward total VOC emissions in Japan. Sci. Total Environ. 449, 143e149. Yamada, H., Inomata, S., Tanimoto, H., 2015. Evaporative emissions in three-day diurnal breathing loss tests on passenger cars for the Japanese market. Atmos. Environ. 107, 166e173. Yugyohouchishimbun, Co, Ltd, 2015. Gas Filing Station Japan. spring. Zhang, Y., Wang, X., Zhang, Z., Lu, S., Shao, M., Lee, F.S.C., Yu, J., 2013. Species profiles and normalized reactivity of volatile organic compounds from gasoline evaporation in China. Atmos. Environ. 79, 110e118.
Contents lists available at ScienceDirect
Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv
Refueling emissions from cars in Japan: Compositions, temperature dependence and effect of vapor liquefied collection system Hiroyuki Yamada a, *, Satoshi Inomata b, Hiroshi Tanimoto b a b
National Traffic Safety and Environment Laboratory, 7-42-27, Jindaiji-higashimachi, Chofu, Tokyo 182-0012, Japan National Institute for Environmental Studies, 16-2 Onogawa, Tsukuba, Ibaraki 305-8506, Japan
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Refueling emission factor in Japan was 1.02 ± 0.40 g/L and MIR was 3.49 ± 0.83. C4 alkene had the highest impact on the OFP of refueling emissions. Refueling emissions changed with temperature but MIR was almost constant. MIRrefueling z MIRbreakthrough z MIRfuel þ 0.5. The efficiency of a vapor liquefied collection system was similar to that of an ORVR.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 May 2015 Received in revised form 5 September 2015 Accepted 7 September 2015 Available online 9 September 2015
Refueling emissions from cars available on the Japanese market, which were not equipped with specific controlling devices, were investigated. For the composition analysis, a proton transfer reaction plus switchable reagent ion mass spectrometry (PTR þ SRI-MS), which is capable of real-time measurement, was used. In addition, the performance of a vapor liquefied collection system (VLCS), which is a recently developed controlling device, was evaluated and compared with an onboard refueling vapor recovery (ORVR) system. The refueling emission factor of uncontrolled vehicles at 20 C was 1.02 ± 0.40 g/L in the case dispensing 20 L of fuel. The results of composition analysis indicated that the maximum incremental reactivity (MIR) of refueling emissions in Japan was 3.49 ± 0.83. The emissions consist of 80% alkanes and 20% alkenes, and aromatics and di-enes were negligible. C4 alkene had the highest impact on the MIR of refueling emissions. The amounts of refueling emissions were well reproduced by a function developed by MOVE2010 in the temperature range of 5e35 C. The compositions of the refueling emissions varied in this temperature range, but the change in MIR was negligible. The trapping efficiency of VLCS was the same level as that of the ORVR (over 95%). The MIRs of refueling and evaporative emissions were strongly affected by that of the test fuel. This study and our previous study indicated that MIRbreakthrough z MIRrefueling z MIRfuel þ 0.5 and MIRpermeation z MIRfuel. The real-world estimated average MIRfuel in Japan was about 3.0. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Refueling emissions Vapor liquefied collection system ORVR PTRþSRI-MS Ozone formation potential
1. Introduction * Corresponding author. E-mail address: [email protected] (H. Yamada). http://dx.doi.org/10.1016/j.atmosenv.2015.09.026 1352-2310/© 2015 Elsevier Ltd. All rights reserved.
Efforts have been made in many countries to reduce emissions of volatile organic compounds (VOCs) to improve the ozone
456
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
situation. The emissions from automobiles are thought to be one of the major sources of VOCs. It is known that the VOCs from automobiles are emitted not only from the tailpipe during driving but also from the fuel tank of gasoline vehicles during parking and refueling. These are called evaporative emissions and refueling emissions, respectively. These emissions have been studied since the 1980's, mainly in the USA (Braddock et al., 1986; Braddock, 1987; Berglund and Petersson, 1990; Furey and Perry, 1991; Burns et al., 1992). For the evaporative emissions, recent studies have paid attention to estimating the amount in real-world conditions. Our previous study (Yamada, 2013) showed that the total evaporative emissions in Japan are higher than the tailpipe emissions, accounting for 4.6% of VOC emissions from stationary sources. In Europe, the evaporative emissions in real-world conditions often exceed the regulation limits (Martini et al., 2014). Dong et al. (2015) proposed an updated method for estimating the emissions from real parking activity data more accurately. On the other hand, there have been hardly any recent studies on refueling emissions. This is probably because prevention techniques have been already established and introduced in many countries. The strategy for preventing refueling emissions in the USA is to employ an onboard refueling vapor recovery (ORVR) system which traps the vapor displaced by the liquid fuel entering into the tank by using a carbon canister installed in the car. Unlike the USA, in Europe, the refueling emissions are vacuumed by a stage II system installed on a gasoline dispenser (European Parliament, 2009). These two are the major techniques in use, and usually countries which wish to control refueling emissions choose one of them after considering their merits and demerits. The ORVR system exhibits high trapping efficiency (95%). However, the benefits of ORVR will only be fully delivered after most vehicles on the road are replaced with those having ORVR, and this may take more than 10 years. In contrast, although the real world trapping efficiency of stage II systems is only 70%, they can be implemented more quickly (Fung and Maxwell, 2011). The theoretical efficiency of stage II system is similar to that of ORVR; in reality, however, they are often not wellmaintained, which results in lowered efficiency. Japan is one of the countries where refueling emissions have not been controlled, although its ozone situation is critical (Ministry of Environment, Japan, 2011). In Japan, gas stations emitted 107,082 ton/year out of the total VOC emissions of 736,612 ton/year (Ministry of Environment, Japan, 2014). Therefore, the introduction of prevention techniques has been demanded by local governments in Japan (Kanagawa Prefectural Government, 2014), and discussions will begin in the near future. However, there is no data on the effectiveness of control techniques for refueling emissions in Japan. As described above, there were many studies discussing the effectiveness of these techniques in the USA in the 1980's. However, the current situation in Japan is quite different from the situation in the USA in the 1980's, such as car specifications, fuel composition, and fuel dispenser specifications. The results of these evaluations will be useful not only to Japan but also to many other countries that have not yet introduced refueling emission controls. To discuss VOC emissions from various sources, it is important to consider the ozone formation potential (OFP) (Adam et al., 2011; Zhang et al., 2013; Gentner et al., 2013; Costagliola et al., 2014; Li et al., 2015), but there have been hardly any studies that discussed the OFP of refueling emissions in Japan. Even in the studies performed outside Japan, in most cases, the composition analyses were made using traditional analytical chemistry methods, such as GC-MS. Our previous study demonstrated that measurement of evaporative emissions using proton transfer reaction plus switchable reagent ion mass spectrometry (PTR þ SRI-MS) can be performed in real-time. Also, the results indicated the possibility of underestimating the concentrations of relatively high-molecular-
weight compounds due to adsorption on the walls of the variable-temperature sealed housing evaporative determination (VT-SHED) equipment (Yamada et al., 2015). Therefore, real-time measurements of VOCs are desirable in evaporative emission experiments. The Japanese gasoline dispenser maker Tatsuno has developed a new prevention device that can be deployed by replacing existing dispensers with new dispensers equipped with the device, even though there have been no regulations concerning refueling emissions in Japan. Their approach uses a vapor liquefied collection system (VLCS). Similar to the stage II system, the refueling emissions are vacuumed by a muffler-shaped inlet surrounding the nozzle. VLCS may be an effective tool; however, no evaluation of its emission preventing performance has been carried out. In this study, the refueling emissions from eight Japanese vehicles not equipped with refueling emission control systems were measured using a VT-SHED and a gasoline dispenser. In these experiments, the maximum incremental reactivity (MIR; Carter, 2010; Carter and Heo, 2013), which is one of the major indicators of OFP, was estimated from the results of real-time composition analysis with the PTR þ SRI-MS. Changes in the amounts of refueling emissions and their compositions were also observed in the temperature range from 5 to 35 C. In addition, the performance of the VLCS was compared with that of the ORVR. 2. Experimental methods 2.1. Refueling emissions measurements The refueling emissions from gasoline vehicles were measured by placing a test vehicle in a VT-SHED (VSH-9353; Hitachi Technology Engineering Inc., Tokyo, Japan) that met the requirements for the Japanese approval tests for evaporative emissions but that was not designed for refueling emission measurements. The VOC emissions were determined by measuring the increase in the total hydrocarbon (THC) concentration in the SHED using a gas analyzer (MEXA-1160TFL-L; Horiba Inc., Kyoto, Japan). A gasoline dispenser (MAB36621EVMBDTX0001A; Tatsuno Corporation, Tokyo, Japan) was set next to the SHED and the dispenser nozzle was brought into the SHED through a squareshaped hole with approximate dimensions of 150 mm 150 mm just before the test started. The hole was loosely sealed with a polytetrafluoroethylene sheet. Any leak from the hole was considered to be negligible because the hole was small enough compared with the SHED volume of 93.53 m3 and because of the relatively short test durations of approximately 5 min. The test was started upon removing the fuel tank cap and then inserting the nozzle of the dispenser into the tank as quickly as possible. Initially almost 10% of the fuel was left in the tank, and 20 L of gasoline was filled into the tank The gasoline flow rate was set to 35 L/min, which is the rate generally used in Japan. After refueling, the nozzle was left inserted, and the test ended when the indicated THC stabilized. The test fuel was specially blended for type approval tests, and its temperature was set equal to the SHED temperature. This procedure roughly agrees with that in the US. The differences are, in the US case, background measurement is performed with the tank cap opened and fuel is dispensed until automatic nozzle shut-off occurs (US Government Publishing Office, 1995). The shape of the nozzle was almost the same as the stage II nozzle which is surrounded with a muffler-shaped vacuum inlet. The vacuumed flow was sent to the VLCS (QE-1001; Tatsuno Corporation, Tokyo, Japan) at a flow rate of 40 L/min. The same nozzle was used even when the VLCS was off, for conducting measurements of the ORVR and the uncontrolled case. Alkanes, alkenes and di-ene species concentrations from carbon
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
numbers 3 to 11 were monitored in the SHED with PTR þ SRIeMS using NOþ as a reagent ion. Benzene, alkyl benzenes, naphthalene, and alkyl naphthalenes were also monitored. The details of the measurement technique with PTR þ SRI-MS can be found in our previous publication (Yamada et al., 2015). The detection sensitivities for some hydrocarbons were determined under ambient humid conditions: 3.9 ± 0.1 (benzene, m/z 78), 5.2 ± 0.3 (toluene, m/z 92), 4.8 ± 0.7 (p-xylene, m/z 106), 4.3 ± 0.9 (1,3,5-trimethylbenzene, m/z 120), 5.4 ± 0.1 (isoprene, m/z 68), 5.1 ± 0.5 (3-hexene, m/z 84), and 5.2 ± 0.6 ncps ppbv1 (3-heptene, m/z 99). Error limits represent 95% confidence levels determined by Student's t-test. The detection sensitivity of 4.3 ± 0.9 ncps/ppbv was used as the typical detection sensitivity for alkylbenzenes, naphthalene, and alkylnaphthalene, but not benzene, toluene, xylene, and trimethylbenzene. The detection sensitivity of 5.2 ± 0.7 ncps/ppbv was used as the typical detection sensitivity for alkenes and di-enes. For alkanes, the values reported in Yamada et al. (2015) were used. 2.2. Tested vehicles
457
study, the vacuum flow rate was set to 40 L/min, whereas the gasoline dispensing rate was 35 L/min. Another merit of the VLCS is that it is compatible with the ORVR system. In the case of the stage II system, vehicle fueling with ORVR results in a lower overall trapping efficiency (US Environmental Protection Agency, 2012b) because sometimes clean air is vacuumed into the underground tank by the stage II system and is mixed with fuel vapor. Then, it is released to the atmosphere without filtration to make room for newly vacuumed air. On the other hand, VLCS always exhausts the air after filtration of the fuel vapor. Therefore, this drop of overall efficiency does not occur in the case of VLCS in conjunction with the ORVR system. In addition, an uncontrolled dispenser can be equipped with a VLCS simply by replacing the dispenser with one equipped with a VLCS. This is easier than installing a stage II system, which requires providing a separate line to the underground tank.
3. Results and discussion 3.1. Real-world refueling emissions in Japan
The list of vehicles is shown in Table 1. OR was the only vehicle equipped with an ORVR system complying with the refueling regulations in the USA. The others had no refueling emission control systems. The tested vehicles, except OR, are widely used in Japan. The ages and mileages of these vehicles varied from 0.7 to 11.9 year and 6046 to 186,596 km, respectively. UF was a so-called kei car (light automobile), a category of small car peculiar to Japan; light automobiles must have an engine displacement of 0.66 L or less and a size smaller than 3.4 m in length, 1.48 m in width, and 2 m in height. The ratio of light automobiles in Japan has increased recently, and 40% of all new cars sold in 2014 were in this category. 2.3. Vapor liquefied collection system The VLCS is a new type of device attached to fuel dispensers for controlling the refueling emissions. The operation of the VLCS is shown in Fig. 1. Similar to the stage II system, the refueling emissions are vacuumed by the muffler-shaped inlet surrounding the nozzle. The vacuumed emissions are not sent to an underground tank but to a condenser in the VLCS. At the condenser, the emissions are liquefied and mixed with the gasoline flowing from the underground tank to the dispenser. The exhaust from the condenser is sent to a gasoline trap, and unliquefied gasoline vapor is removed from the exhaust. The trapped gasoline vapor is sent back to the condenser. In the case of the conventional stage II system, a vacuum flow is generated by the airflow that replaces the decreasing volume of gasoline in the underground tank during refueling. Thus, the maximum vacuum flow rate cannot exceed the refueling flow rate. On the other hand, the vacuum flow rate of the VLCS is independent of the refueling flow rate and can be set higher than that of the stage II system. Theoretically, the increased vacuum flow rate results in a higher trapping efficiency than the stage II system. In this
We evaluated the refueling emissions from eight cars available on the Japanese market. The temperature in the SHED was set to 20 C. In this experiment, residual fuel in the tested vehicles' tanks, which affected the refueling emissions, was not replaced with the test fuel, because the aim of these experiments was to determine the emission factors that can be applied to real-world refueling emissions. Also, measuring the emissions with residual fuel is closer to the real-world conditions. Consequently, the results contain variations due to both differences in the vehicles and differences in the fuels. The results are shown in Table 2. The average value and 2s were 1.02 and 0.40 g/L, respectively. The detailed mass-based composition ratios of refueling emissions at 20 C are listed in Table 3, accompanied by mass ratio MIR. The MIR values shown were
Vapor
VLCS
Vapor
Dispenser
Gasoline
Exhaust
Pump Condenser
Gasoline Trap Gasoline Underground Tank
Release Valve
Liquefied Vapor
Fig. 1. Operation of VLCS.
Table 1 Specifications for the nine test vehicles used in this study. Name
OR
UA
UB
UC
UD
UE
UF
UG
UH
Category Maker Mileage (km) Age (year) Tank Volume (L) Displacement (L) Vehicle Weight (kg)
SUV Ford 3820 0.2 70 2.0 2020
Compact Mitsubishi 11,543 4.0 45 1.5 1090
Compact Toyota 46,528 7.2 40 1.5 1050
mini-van Toyota 54,352 7.8 55 2.0 1580
Van Toyota 186,596 9.8 70 2.0 2140
mid-size Toyota 39,046 6.9 50 1.8 1260
Kei Honda 6046 0.7 35 0.66 970
Wagon Toyota 25,881 11.9 70 2.4 1770
Wagon Honda 16,361 7.9 70 2.4 1870
458
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
refueling emissions (MIRrefueling), and permeation emissions (MIRpermeation) are related by:
taken from a report by the California Environmental Protection Agency (2009). For the alkanes, the values of straight and branched alkanes were averaged according to the ratio of test fuel. The averaged total MIR and 2s of refueling emissions were estimated to be 3.49 and 0.83, respectively. The effect of experimental errors was typically ±8%, as shown later. C4 alkene had the highest impact on the MIR of refueling emissions. The mass ratio and mass ratio MIR of the emissions from each vehicle are shown in Fig. 2. In the mass ratio results, there was no large difference among the vehicles, and the main components were alkanes, which occupied up to 80%, and the rest were alkenes. Aromatics and di-enes were negligible. On the other hand, in the mass ratio MIR results, alkenes showed the highest impact, at almost 70%.
where MIRfuel is the MIR of the test fuel. The average MIRrefueling with the tank filled with real-world fuel shown in Table 1 is 3.49 ± 0.83. Using this value and eq. (1), the estimated MIRfuel of real-world fuels is about 3.0 on average, suggesting that the test fuels used in this and previous studies were likely to have higher MIRs than the average MIR of real-world fuels.
3.2. Effects of fuel variety on MIR values of refueling, breakthrough and permeation emissions
3.3. Temperature dependence of the refueling emissions and compositions
To observe the features of refueling emissions, the composition was compared with that of the test fuel. The fuel was analyzed by GC-MS using the method outlined in the Japanese Industrial Standard JIS K2536e2:2003 (Japanese Standards Association, 2003), and the refueling emissions from UB were measured by PTR þ SRIMS. To avoid the effects of different residual fuel in the tank, the fuel was replaced with the test fuel before the experiment. The SHED temperature and the fuel temperature were set to 20 C. The test was conducted only once, and similar experiments were conducted at different temperatures to obtain the temperature dependence, as shown later. Compared with these results, the data obtained at 20 C seems to have acceptable accuracy. The results are shown in Fig. 3. The error limits in this figure were calculated by propagating the errors of the ion signals and the detection sensitivities for each compound, but the error caused by the fact that typical detection sensitivities were used for alkenes, di-enes, and some aromatics was not included. The detailed composition of the refueling emissions was quite different from that of the fuel. The ratio of aromatic compounds in the refueling emissions was less than 10% and the main components were C4eC6 alkanes and alkenes. These features were similar to the breakthrough emissions of evaporative emissions observed in our previous study (Yamada et al., 2015). The estimated MIR of the refueling emissions was 4.21 ± 0.35 higher than that of the test fuel (3.78). The error was calculated by summing the errors for each compound. The error was smaller than the variations of test results with the eight cars shown in Table 3. Fig. 3 also shows the results for the same vehicle (UB) with the tank filled with the real-world fuel shown in Fig. 2. The estimated MIR in this case was 3.83 ± 0.31, which is smaller than the case where the tank was filled with the test fuel (4.21 ± 0.35). This smaller MIR was caused by a smaller C4 and C5 alkene ratio, whose MIRs are quite high. It seems that this difference was due to the different fuel composition. In our previous study (Yamada et al., 2015), we obtained MIRs of breakthrough and permeation emissions of 3.89 and 3.31, respectively using a fuel whose MIR was 3.33. The difference in MIR between the breakthrough emissions and the test fuel was almost the same as that between the refueling emissions and the test fuel in this study. The MIRs of breakthrough emissions (MIRbreakthrough),
The amounts of uncontrolled refueling emissions and their compositions in the temperature range from 5 to 35 C were measured using the vehicle UB. Similar to the experiment in Fig. 3, residual fuel in the tank was replaced with the test fuel to avoid fluctuations caused by different residual fuel compositions. The THC results are shown in Fig. 4, together with a couple of calculated results. One of the functions used for the estimation was taken from the results of MOVE 2010 in the USA (US Environmental Protection Agency, 2012a). The refueling emissions can be expressed as:
Table 2 The refueling emissions from eight cars which were not equipped with emission control devices. The test temperature was 20 C. The error limits of the average are represented by 2s (experimental uncertainties are not included). Vehicle
UA
UB
UC
UD
UE
UF
UG
UH
Average
Emissions (g/L) 0.95 1.03 1.06 1.41 0.75 1.14 0.98 0.86 1.02 ± 0.40
MIRbreakthrough zMIRrefueling zMIRfuel þ 0:5
(1)
MIRpermeation zMIRfuel
(2)
HCuncontrol ¼ 5:909 0:0949 TDFDIF þ 0:0884 DFTEMP þ 0:485 RVP
(3)
where HCuncontrol is the refueling emissions (g/L), DFTEMP is the dispensed gasoline temperature ( F), RVP is the raid vapor pressure of the test fuel (psi), and TDFDIF is the tank gasoline temperature ( F), defined as:
TDFDIF ¼ 0:418 DFTEMP 16:6
(4)
Another function has been used for estimating the VOC inventory in Japan (Ministry of Environment, Japan, 2014), namely:
. HCuncontrol ¼ 0:97 Tref þ 11:12 21
(5)
where Tref is the ambient temperature ( C). The estimated results obtained with (eq. (3)) well represented the experimental results, whereas (eq. (5)) overestimated the results by a factor of almost 1.5. The emissions of total aromatics, alkanes, alkenes and di-enes from UB as a function of temperature are shown in Fig. 5 as the ratio of each emission at 20 C. Aromatics showed a gradual increase with increasing temperature. The other emissions seemed to be constant, having no dependence on the temperature. However, the detailed emissions shown in Fig. 6 indicated that the compositions changed as the temperature changed, even in alkanes, alkenes and di-enes. These three groups exhibited common features. At low temperature, the emissions of higher mass species were lower than those at 20 C, and the emissions of smaller mass species were also higher. As a result, the total emissions of the groups were similar to those at 20 C. The emissions at temperatures higher than 20 C were similar to those at 20 C. In the results for the aromatics shown in Fig. 6 (a), the emission of benzene was constant in this temperature range. The other species showed increasing emissions as the temperature increased. This was more apparent in species with higher mass. Fig. 7 shows the MIR of refueling emissions calculated using the results shown in Fig. 6 as a function of the temperature. There was no significant temperature
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
459
Table 3 Average species composition ratio (mass ratio) and the mass ratio MIR of the refueling emissions for eight Japanese cars available on the market, obtained by PTR þ SRI-MS. SHED temperature and fuel temperature were 20 C. The error limits of the average are represented by 2s (experimental uncertainties are not included). Aromatics MIRb Mass ratio Ratio MIR Alkanes MIR Mass ratio Ratio MIR Alkenes MIR Mass ratio Ratio MIR Di-ene MIR Mass ratio Ratio MIR
a M78 (benzene)
M92 (Toluene)
M106 (C8-benzene)
M120 (C9-benzene)
M134 (C10-benzene)
M128 (naphthalene)
Aromatic compounds total
0.72 0.001 ± 0.001 0.00
4 0.009 ± 0.009 0.04
6.57 0.002 ± 0.001 0.01
6.23 0.001 ± 0.000 0.00
7.49 0.000 ± 0.000 0.00
3.34 0.000 ± 0.000 0.00
0.013 ± 0.009 0.05 ± 0.04
M58 (C4)
M72 (C5)
M86 (C6)
M100 (C7)
M114 (C8)
Alkanes total
1.17 0.425 ± 0.214 0.50
1.40 0.220 ± 0.148 0.31
1.30 0.048 ± 0.022 0.06
1.45 0.007 ± 0.002 0.01
1.44 0.003 ± 0.003 0.00
0.766 ± 0.262 0.91 ± 0.33
M56 (C4)
M70 (C5)
M84 (C6)
M98 (C7)
M112 (C8)
Alkenes total
11.66 0.140 ± 0.050 1.59
12.22 0.056 ± 0.039 0.67
8.84 0.014 ± 0.009 0.12
6.98 0.002 ± 0.000 0.01
5.37 0.001 ± 0.000 0.00
0.214 ± 0.065 2.47 ± 0.76
M54 (C4)
M68 (C5)
M82 (C6)
M96 (C7)
M110 (C8)
Di-ene total
Total
12.61 0.000 ± 0.000 0.00
11.028 0.003 ± 0.004 0.03
8.68 0.002 ± 0.001 0.02
7.29 0.001 ± 0.000 0.00
4.89 0.000 ± 0.000 0.00
0.006 ± 0.004 0.06 ± 0.04
3.49 ± 0.83
Bold indicates sum of horizontal cells. a M ¼ molecular weight; bMIR ¼ maximum incremental reactivity.
dependence of the MIR of the refueling emissions (Slope: 0.005 ± 0.021), although temperature dependence of the
composition was observed as shown in Fig. 6.
3.4. Evaluation of VLCS
Fig. 2. Mass ratio (a) and mass ratio MIR (b) of the refueling emissions obtained by PTR þ SRI-MS from eight cars available in the Japanese market. SHED temperature and fuel temperature were 20 C. Error bars indicates 2s.
The performances of the VLCS and ORVR were evaluated in the temperature range from 8 to 33 C. For the tested vehicles, OR, which was equipped with the ORVR complying with the regulations in the USA, and UA were used. The regulation limit in the USA is 0.05 g/L (0.2 g/gal) at 26.7 C (80 F). In the experiments with UA, the VLCS was used to accumulate the refueling emissions, and it was switched off with OR. The results are listed in Table 4. The emissions with the VLCS and ORVR relative to uncontrolled emissions are shown in Fig. 8. The emissions with the ORVR were almost 5% and were independent of the temperature. This result agrees with a former observation (US Environmental Protection Agency, 2012b) and is below the regulation limit in the USA, except for the results at 32.9 C. The emissions with the VLCS were mostly lower than those with the ORVR. Also, the trapping efficiency of the VLCS showed a temperature dependence. Fig. 9 show the emissions from the ORVR and the uncontrolled case when the fuel cap was opened but the nozzle of the dispenser was not inserted in the tank, at 20 C. OR was equipped with a system that restrict these emissions, and the emission rate was small enough compared with the refueling emissions listed in Table 4. On the other hand, the emission in the uncontrolled case was comparable with the emission of the VLCS. Vehicle UA, which was used for evaluating the VLCS, did not have a system for restricting these emissions, and it is suggested that the emissions before inserting the nozzle are a major source of refueling emissions with the VLCS. Also, the observed temperature dependence of the VLCS trapping efficiency seems to be caused by the temperature dependence of these emissions. As a result, the VLCS had a similar level of performance in preventing refueling emissions to the ORVR, suggesting that the VLCS has a higher trapping efficiency than the stage II system. Thus, the VLCS will become one of the choices for controlling refueling emissions. The cost associated with introducing one VLCS system is 365 million yen. There are 119,000 fuel dispensers in Japan (Yugyohouchisimbun, Co, Ltd, 2015). To replace all of them, the
460
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
0.12
0.50
(a) Aromatics
UB(Tested fuel) UB(Real world) Fuel
0.10
Mass ratio
Mass ratio
0.04
0.30 0.20 0.10
0.02 0.00
0.00 M78
0.20
UB(Tested fuel) UB(Real world) Fuel
0.40
0.08 0.06
(b) Alkanes
M92 M106 M120 M134 M128 M148
C4 0.90
(c) Alkenes
C7
C8
C9
(d) Comparisons of groups UB(Tested fuel) UB(Real world) Fuel
0.70 0.60
Mass ratio
Mass ratio
0.10
C6
0.80
UB(Tested fuel) UB(Real world) Fuel
0.15
C5
0.50 0.40 0.30
0.05
0.20 0.10
0.00 C4
C5
C6
C7
C8
C9
0.00
Aromatics total Alkanes total
Alkenes total
Fig. 3. Ratio of refueling emissions at 20 C from UB filled with the test fuel and the real-world fuel: (a) aromatic compound, (b) alkanes, (c) alkenes, and (d) summary of the groups. The compositions of the test fuel are also shown. Estimated MIRs from these results were 3.78 for the test fuel, 4.21 ± 0.35 for the refueling emissions with the test fuel and 3.83 ± 0.31 for the refueling emissions with real-world fuel for residual. The error bars were calculated from the errors of ion signals and detection sensitivities of each compound.
total cost would be almost 434 billion yen. On the other hand, the cost of ORVR is 6e8 dollars per vehicle (Fung and Maxwell, 2011), and the total cost of introducing the ORVR systems would be almost 58.5 billion yen if replacing the carbon canisters of all gasoline vehicles in Japan (66.8milion; Automobile Inspection & Registration Information Association, 2013). By assuming the efficiencies of ORVR and VLCS are similar according to the results shown above, the ORVR is more cost-effective than the VLCS. But this rough analysis excluded effect of the differences in spreading
Fig. 4. Experimental and estimated results of uncontrolled refueling emissions (Ministry of Environment, Japan, 2014; US Environmental Protection Agency, 2012a).
period and maintenance cost. And a more comprehensive comparative analysis of ORVR and VLCS, taking the factor into account, is needed. 4. Conclusions The refueling emissions were measured from eight gasoline cars available on the Japanese market and a car equipped with an ORVR,
Fig. 5. Uncontrolled refueling emissions of total aromatics, alkanes, alkenes and dienes from UB as a function of temperature. The error bars were calculated by summing the errors for each compound.
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
461
Fig. 6. Detailed uncontrolled refueling emissions of (a) aromatics, (b) alkanes, (c) alkenes, and (d) di-enes from UB as a function of temperature.
Fig. 7. MIR of uncontrolled refueling emissions as a function of temperature. The error bars were calculated by summing the errors for each compound. Fig. 8. Temperature dependences of the refueling emissions with the VLCS and ORVR relative to the uncontrolled emissions. Table 4 The refueling emissions of the uncontrolled case, the VLCS, and the ORVR at various temperatures. Temperature ( C)
Uncontrolled
VLCS
ORVR
8.2 12.3 20.0 24.5 32.9
0.64 0.69 0.95 0.93 1.24
0.000 0.014 0.011 0.024 0.066
0.037 0.029 0.037 0.039 0.059 Unit: g/L
using VT-SHED and a gasoline dispenser. The average emission from the Japanese cars at 20 C was 1.02 ± 0.40 g/L.
PTR þ SRI-MS was adopted for the first time to observe the refueling emission compositions. The main components in the refueling emissions were alkanes, which occupied almost 80% of the mass. The rest was mostly alkenes, but C4 alkene showed the highest impact on the MIR of refueling emissions. Aromatics and dienes were negligible. The average refueling MIR of the eight cars, whose tanks were filled with Japanese real-world fuel, was 3.49 ± 0.83. When using a test fuel whose MIR was 3.78, the MIR of refueling emissions was 4.21 ± 0.35. The MIR of the emissions was higher than that of the fuel because the decreased ratio of aromatics resulted in a higher ratio of small alkenes, whose MIRs are high.
462
H. Yamada et al. / Atmospheric Environment 120 (2015) 455e462
Fig. 9. Emissions of uncontrolled case and the ORVR while the fuel cap was opened and the dispensing nozzle was not inserted, as a function of duration.
This feature was similar to breakthrough emissions observed in a previous study. The MIRs of refueling and evaporative emissions were strongly affected by that of test fuel. They can be calculated with the equations MIRbreakthrough z MIRrefueling z MIRfuel þ 0.5 and MIRpermeation z MIRfuel. The average MIRfuel of Japanese real-world gasoline was 3.0. The refueling emissions in the experiments were well reproduced by a function developed in MOVE2010. The function conventionally used for estimating the VOC inventory in Japan overestimated the emission by a factor of almost 1.5. The compositions of the emissions changed with temperature. The emissions of aromatics increased as the temperature increased. Total alkane, alkene and di-ene was almost constant in this temperature range. Detailed composition analyses of alkanes, alkenes, and di-enes indicated that the emissions of higher-mass species were low and those of lower-mass species were high, at temperatures below 20 C. The MIR of the refueling emissions decreased slightly with increasing temperature. The performance of VLCS was evaluated for the first time. The refueling emissions with VLCS were almost 2% of the uncontrolled case, and the results were below the regulation limit of ORVR in the USA. A rough cost analysis indicated that the ORVR is more cost effective, but the VLCS can be one of the choices for controlling refueling emissions. The emissions with the VLCS seem to be due to the emissions occurring when opening the tank cap to insert the dispenser nozzle into the tank. Acknowledgment The authors are grateful to Tatsuno Corporation and Ford Japan Ltd. for providing the experimental instruments. References Automobile Inspection & Registration Information Association, 2013. Monthly Report of Vehicle Numbers, June 2013. Japanese Automobile Inspection & Registration Information Association, Tokyo (in Japanese). Adam, T.W., Astorga, C., Clairotte, M., Duane, M., Elsasser, M., Kransenbrink, A., Larsen, B.R., Manfredi, U., Martini, G., Montero, L., Sklorz, M., Zimmermann, R., Peruji, R., 2011. Chemical analysis and ozone formation potential of exhaust from dual-fuel (liquefied petroleum gas/gasoline) light duty vehicles. Atmos.
Environ. 45, 2842e2848. Berglund, P.M., Pertersson, G., 1990. Hazardous petrol hydrocarbons from refueling with and without vapour recovery. Sci. Total Environ. 91, 49e57. Braddock, J.N., 1987. PB87e212775. Factors Influencing the Composition and Quantity of Passenger Car Refueling Emissions e Part 2. Atmospheric Sciences Research Laboratory, U. S. Environment Protection Agency. Research Triangle Park NC 27711. Braddock, J.N., Gabele, P.A., Lemmons, T.J., 1986. Factors Influencing the Composition and Quantity of Passenger Car Refueling Emissions ePart 1. SAE Technical Paper 861558. Burns, V.R., Gorse, R.A., Reuter, R.M., Hochhauser, A.M., Benson, J.D., Koehl, W.J., Painter, L.J., 1992. Effects of Gasoline Composition on Evaporative and Running Loss Emissions e Auto/oil Air Quality Improvement Research Program. SAE Technical Paper 920323. California Environmental Protection Agency, 2009. Tables of Maximum Incremental Reactivity (MIR) Values. http://www.arb.ca.gov/regact/2009/mir2009/mir2009. htm (accessed 19.04.15.). Carter, W.P.L., 2010. Development of the SAPRC-07 chemical mechanism. Atmos. Environ. 44, 5324e5335. Carter, W.P.L., Heo, G., 2013. Development of revised SAPRC aromatics mechanisms. Atmos. Environ. 77, 404e414. Costagliola, M.A., Murena, F., Prati, M.V., 2014. Exhaust emissions of volatile organic compounds of powered two-wheelers: effect of cold start and vehicle speed. Contribution to greenhouse effect and tropospheric ozone formation. Sci. Total Environ. 468e469, 1043e1049. Dong, X., Tschantz, M., Fu, J.S., 2015. Estimating evaporative vapor generation from automobiles based on parking activities. Environ. Pollut. 202, 104e111. European Parliament, 2009. Council Directive 2009/126/EC of the European Parliament and of the Council of 21 October 2009 on Stage II Petrol Vapour Recovery during Refuelling of Motor Vehicles at Service Stations. Fung, F., Maxwell, B., 2011. Onboard Refueling Vapor Recovery: Evaluation of the ORVR Program in the United States. The International Council on Clean Transportation Working Paper 2011-12, available at: http://www.theicct.org/ onboard-refueling-vapor-recovery-evaluation-orvr-program-united-states. Furey, R.L., Perry, K.L., 1991. Composition and Reactivity of Fuel Vapor Emissions from Gasoline e Oxygenated Blends. SAE Technical Paper 912429. Gentner, D.R., Worton, D.R., Isaacman, G., Davis, L.C., Dallmann, T.R., Wood, E.C., Herndon, S.C., Goldstein, A.H., Harley, R.A., 2013. Chemical composition of gasphase organic carbon emissions from motor vehicles and implications for ozone production. Environ. Sci. Technol. 47, 11837e11848. Japanese Standards Association, 2003. Liquid Petroleum Products e Testing Method of Components Part 2: Determination of Total Components by Gas Chromatography. Japanese Standards Association, Tokyo. JIS K2536-2:2003. Kanagawa Prefectural Government, 2014. Activities for Spreading the ORVR (in Japanese). Available at: http://www.pref.kanagawa.jp/cnt/f531508/p852505. html. Li, L., Ge, Y., Yang, M., Li, J., Peng, Z., Song, Y., Zhang, L., 2015. Effect of gasoline/ methanol blends on motorcycle emissions: exhaust and evaporative emissions. Atmos. Environ. 102, 79e85. Martini, G., Paffumi, E., De Gennaro, M., Mellios, G., 2014. European type-approval test procedure for evaporative emissions from passenger cars against realworld mobility data from two Italian provinces. Sci. Total Environ. 487, 506e520. Ministry of the Environment, Japan, 2011. Air Quality Monitoring Report. Ministry of the Environment Japan, Tokyo (in Japanese). Ministry of the Environment, Japan, 2014. Report on the VOC Emission Inventory in the 2013 Fiscal Year. Ministry of the Environment Japan, Tokyo (in Japanese). US Environmental Protection Agency, 2012a. Development of Evaporative Emissions Calculations for the Motor Vehicle Emissions Simulator MOVES2010. EPA420-R-12-027, available at: http://www.epa.gov/oms/models/moves/ documents/420r12027.pdf. US Environmental Protection Agency, 2012b. Rulemaking. Air Quality: Widespread Use for Onboard Refueling Vapor Recovery and Stage II Waiver. EPA-HQ-OAR2010-1076, available at: https://www.federalregister.gov/articles/2012/05/16/ 2012-11846/air-quality-widespread-use-for-onboard-refueling-vapor-recoveryand-stage-ii-waiver. US Government Publishing Office, 1995. Measurement Procedure; Refueling Tests. Title 40-(Chapter 1)-Subchapter C - Part 86. 154e98 of the US Code of Federal Regulations (CFR), available at: http://www.ecfr.gov/cgi-bin/text-idx? SID¼eb43af3600813c84aba33e893b216400&mc¼true&node¼se40.19.86_ 1154_698&rgn¼div8. Yamada, H., 2013. Contribution of evaporative emissions from gasoline vehicles toward total VOC emissions in Japan. Sci. Total Environ. 449, 143e149. Yamada, H., Inomata, S., Tanimoto, H., 2015. Evaporative emissions in three-day diurnal breathing loss tests on passenger cars for the Japanese market. Atmos. Environ. 107, 166e173. Yugyohouchishimbun, Co, Ltd, 2015. Gas Filing Station Japan. spring. Zhang, Y., Wang, X., Zhang, Z., Lu, S., Shao, M., Lee, F.S.C., Yu, J., 2013. Species profiles and normalized reactivity of volatile organic compounds from gasoline evaporation in China. Atmos. Environ. 79, 110e118.