The challenge to NO x emission control for heavy-duty diesel vehicles in China 9367

China’s new “Twelfth Five-Year Plan” set a target for total NOx emission reduction of 10 % for the period of 2011–2015. Heavy-duty diesel vehicles (HDDVs) have been considered a major contributor to NO x emissions in China. Beijing initiated a comprehensive vehicle test program in 2008. This program included a sub-task for measuring onroad emission profiles of hundreds of HDDVs using portable emission measurement systems (PEMS). The major finding is that neither the on-road distance-specific (g km −1) nor brake-specific (g kWh −1) NOx emission factors for diesel buses and heavy-duty diesel trucks improved in most cases as emission standards became more stringent. For example, the average NOx emission factors for Euro II, Euro III and Euro IV buses are 11.3 ± 3.3 g km−1, 12.5± 1.3 g km−1, and 11.8± 2.0 g km−1, respectively. No statistically significant difference in NOx emission factors was observed between Euro II and III buses. Even for Euro IV buses equipped with SCR systems, the NO x emission factors are similar to Euro III buses. The data regarding real-time engine performance of Euro IV buses suggest the engine certification cycles did not reflect their real-world operating conditions. These new on-road test results indicate that previous estimates of total NOx emissions for HDDV fleet may be significantly underestimated. The new estimate in total NO x emissions for the Beijing HDDV fleet in 2009 is 37.0 Gg, an increase of 45 % compared to the previous study. Further, we estimate that the total NO x emissions for the national HDDV fleet in 2009 are approximately 4.0 Tg, higher by 1.0 Tg (equivalent to 18 % of total NO x emissions for vehicle fleet in 2009) than that estimated in the official report. This would also result in 4 % increase in estimation of national anthropogenic NO x emissions. More effective control measures (such as promotion of CNG buses and a new in-use compliance testing program) are urged to secure the goal of total NOx mitigation for the HDDV fleet in the future.


Introduction
China's energy consumption is heavily dependent on coalbased energy (NBSC, 2011), resulting in severe air pollution during the 1990s and early 2000s.Consequently the Chinese government set a clear goal of a 10 % reduction of total SO 2 emissions (SO 2 is considered as a key indicator for coalbased pollution) between 2006 and 2010, the period of the "Eleventh Five-Year Plan" (11th FYP) (Zhao et al., 2009;Zhang et al., 2012).After an aggressive build up of flue gas desulfurization at coal-fired power plants, and phase-out of small, inefficient boilers in the coal-fired power sector, total SO 2 emissions decreased by 14 % during the 11th FYP (MEP, 2011a).
However, a surge of the vehicle population in China since the late 1990s, due in part to rapid economic growth and urbanization, triggers a concern about air pollution resulting Published by Copernicus Publications on behalf of the European Geosciences Union.2012-10-1 c NA d a Only for public fleets including public buses, postal and sanitation vehicles; b Euro IV emission standard for HDDVs was first planned to be implemented in 2010.However, it was postponed by the MEP due to several factors.For example, a sufficient supply of low sulfur fuel that complied with the Euro IV standard was unavailable.The most recent update is that the national Euro IV emission standard for HDDVs might be able to be implemented in 2013; c Euro V emission standard for public HDDV fleets in Beijing including public buses, postal and sanitation vehicles will probably be implemented, beginning from October 1 st , 2012.Prior to this, ultra low sulfur diesel that complied with Euro V standard will be fully delivered in the Beijing market from June 1 st , 2012; d not available.
from vehicle emissions.Vehicle sales jumped 32 % to 18.1 million in 2010 compared to that of 2009, topping all previous records and securing China's position as the world's largest vehicle market (CATARC and CAAM, 2011).Onroad vehicles are considered one of the most important sources of air pollution in China's major cities, changing the pollution pattern from coal-based air pollution to a mix of coal-and vehicle-based air pollution in large cities (Du et al., 2012;Hao et al., 2001;Li et al., 2011a;Wang et al., 2010;Yang et al., 2011).Among the criteria air pollutants, NO x is considered a key indicator for both vehicle and coal combustion sources.Over the past decade, observations from space and in situ indicated a worsening situation for NO x pollution in many regions of China (Richter et al., 2005).Chan and Yao (2008) revealed that annual average NO 2 concentrations had remained almost constant from around 2000 to 2005 and still exceeded the Grade-II standard in China's mega cities such as Beijing, Shanghai and Guangzhou.They also emphasized the necessity of vehicle emission control in the future due to major contribution from on-road vehicles to ground NO x emissions.In response to the NO x trend, the new 12th FYP establishes a target for total NO x emission reduction of 10 % for the period of 2011-2015(NPC, 2011)).However, average NO 2 concentration of China's 113 key environmental protection cities increased by 5.7 % during the first half of 2011 compared to same period of 2010 (MEP, 2011b), which indicated that NO x control would be much harder than SO 2 control.Furthermore, on February 29th, 2012, China's State Council approved the National Ambient Air Quality Standards (NAAQs) Amendment, which will be fully in effect in 2016.It adds annual and 24-h average PM 2.5 concentration limits, and 8-h average O 3 concentration limit.Meanwhile, the limits of PM 10 and NO 2 are tightened (MEP and AQSIQ, 2012).All the revisions of NAAQs call for more stringent controls of NO x emissions, especially in high-emitting sec-tors that include on-road vehicles (Wang and Hao, 2012;Zhang et al., 2012).Among the different vehicle categories, attention has been focused on heavy-duty diesel vehicles (HDDVs, including buses and trucks) due to their significantly higher NO x emission factors and higher travel mileage relative to gasoline cars (Imhof et al., 2006;Liu et al., 2009;Thornhill et al., 2010).The Ministry of Environmental Protection (MEP) reported the first-ever total national NO x emissions from on-road vehicles of 5.3 Tg in 2009, of which diesel vehicles (dominated by heavy-duty trucks and buses) contributed 60 % (MEP, 2010).Even in Beijing, with a large number of private gasoline cars, HDDVs emitted 25.0 Gg NO x , accounting for 30 % of total NO x emissions from vehicles based on our previous study (Wu et al., 2011).To mitigate emissions from HDDVs, MEP and local environmental protection bureaus (EPBs) have launched a series of control measures since 2000 (ICCT, 2010;Wu et al., 2011).For example, Table 1 illustrates new emission standards implemented by stage for HDDVs since 2000 in Beijing and at the national level.Accordingly, new advanced control technologies have penetrated into the HDDV fleet.For example, Selective Catalytic Reduction (SCR) technology has been required for new buses in Beijing since 2008 to comply with the Euro IV standard.
Previous studies of HDDV emissions in China relied heavily on results of engine tests on chassis dynamometers with typical driving cycles (e.g., US Environmental Protection Agency [EPA]'s Federal Test Procedure [FTP] cycle) (Wu et al., 2011;Zhou et al., 2010).However, such engine tests have often been criticized as an inaccurate representation of real-world on-road operation (Krishnamurthy et al., 2007;US DOJ and US EPA, 1998).In 2005, the US EPA released a final rule on a manufacturer-run, in-use testing program for HDDVs using Portable Emission Measurement Systems (PEMS) (US EPA, 2005).The in-use testing  (Krishnamurthy et al., 2007;US EPA, 2005a).Europe has also begun to set its own inuse compliance regulations to supplement certification testing for HDDVs.The in-use compliance regulations would implement the Averaging Window Method by using PEMS on-road (JRC, 2007).In order to improve the accuracy of emission factors and inventories, several on-road emission test programs for HDDVs using PEMS have been conducted in China recently (Liu et al., 2009;Liu et al., 2011).The Vehicle Emission Control Center of MEP has been collecting on-road emission test results of the HDDV fleet in several Chinese cities, attempting to develop a national mobile sources emission inventory for China (Tang et al., 2011 This study represents the real-world on-road NO x emission profiles of more than 130 HDDVs (including both buses and trucks) within the Beijing test program.We calculate and evaluate the NO x emission factors of HDDVs in detail for various emission standards and vehicle weight classifications under typical on-road driving cycles.We also explore detailed on-board diagnostic (OBD) data regarding real-time operating conditions of engines, discuss on-road brake-specific NO x emission factors, and evaluate the realworld SCR performance for the new Euro IV buses in Beijing.We then compare and discuss the new NO x emission factors of HDDVs with other previous estimates and suggest an update of the emission inventory for the HDDV fleet in Beijing and China.

Test vehicles and experimental procedures
The on-road emission tests were conducted in 2008-2010 in Beijing.A total of 135 HDDV test results were collected in this study, including 55 buses and 80 trucks.These tested vehicles are summarized in Table 2.They were mainly recruited from bus and freight transportation companies which represent the typical vehicle technology and maintenance conditions in Beijing.
According to the scrappage policy for the bus fleet in Beijing, public buses over 8 yr are to be phased out.Based on registration data from the bus company, Euro I and pre-Euro I diesel buses have been almost completely phased out.Therefore, our diesel bus samples are classified into 3 emission standard categories -Euro II, Euro III and Euro IV.Almost all of the tested buses have a vehicle length of 11-12 m and gross vehicle weight (GVW) of 15-18 tons.As for trucks, they are classified into 3 emission standard categories as Euro I, Euro II and Euro III.There have been delays in implementing Euro IV emission standards for heavy-duty diesel trucks (HDDTs) nationwide (see Table 1) and there is no formal scrappage policy for trucks.The weight of tested trucks varied considerably, which generally has significant influence on vehicle emissions.Thus, the tested trucks are further classified into 3 weight classifications (see Table 2).
Two Sensor Inc. SEMTECH-D PEMSs (Liu et al., 2009) and a Horiba OBS-2200 PEMS (Bougher et al., 2010) were used in the test program to measure the on-road emissions of gaseous pollutants.The SEMTECH-D employs a Y. Wu et al.: The challenge to NO x emission control for heavy-duty diesel vehicles in China non-dispersive ultraviolet analyzer to measure NO and NO 2 separately, a flame ionization detector to measure total hydrocarbons (THC), and a non-dispersive infrared analyzer to measure CO and CO 2 .The SEMTECH exhaust flow meter (SEMTECH-EFM) was used to measure the exhaust flow.The OBS-2200 PEMS employs a heated chemiluminescence detector to measure NO x (NO and NO 2 together), a heated flame ionization detector to measure THC, and a heated nondispersive infrared analyzer to measure CO and CO 2 .The flow measurement is conducted with a factory calibrated pitot tube exhaust flow meter in the a tailpipe attachment unit.The SEMTECH-D PEMS and OBS-2200 PEMS have both passed validation testing with laboratory systems for inuse compliance testing and were demonstrated to be both accurate and precise, meeting the Section 40 Code of Federal Regulations standards, Part 1065 (BMSTC and Beijing EPB, 2010;Durbin et al., 2007;US EPA, 2005b).Besides, a Dekati Mass Monitor (DMM) (Liu et al., 2009) and two Electric Low Pressure Impactors (ELPI) (Liu et al., 2011) were used in the test to measure the on-road particle emissions.They are both manufactured by Dekati Co. Ltd. and based on ELPI T M technology, which is combined the particle charging and electrical detection.They are able to provide real-time second-by-second PM mass and number emission profiles.Vehicle speed, altitude, latitude, and longitude were recorded second-by-second by GPS receivers.
The test cycles in Beijing included both urban and highway driving segments.Each test lasted more than 30 minutes so as to collect enough vehicle operation and emission profile data.A total of more than 323 000 second-by-second data for trucks and more than 163 000 second-by-second data for buses were collected.We have also recorded the driving condition data for more than 1 020 000 seconds with GPS receivers on 18 typical bus routes to develop a typical driving cycle for urban buses in Beijing.As heavy-duty trucks are not allowed within the Fourth-Ring Road (downtown in Beijing) during the daytime, the driving cycles for HDDTs are typically different than those for buses.For detailed descriptions of test routes and driving cycle development please refer to BMSTC and Beijing EPB (2010).

Operating mode binning methodology
Vehicle specific power (VSP), defined as the instantaneous power demand by the engine per unit vehicle mass, was first introduced by Jimenez (1999).VSP can represent driving conditions in an integrated way and show better correlation with pollutant emission rates than other operating parameters.As a result, VSP is now widely used in developing new emission factor models, such as the International Vehicle Emissions Model (IVE) developed by the University of California, Riverside (CE-CERT et al., 2008) and the Motor Vehicle Emission Simulator (MOVES) developed by US EPA (US EPA, 2009).In this study, we applied the equation from the MOVES model (Koupal et al., 2005) for VSP cal-  kW s 3 m −3 t −1 0.000542 0.000356 0.000331 0.000279 Note: With reference to the MOVE model, those coefficients are estimated according to the typical vehicle weight (Koupal et al., 2005).culation of all HDDV samples, which is shown below.
where m is vehicle weight, tons; v is instantaneous vehicle speed, m s −1 ; a is instantaneous vehicle acceleration, m s −2 ; θ is road grade, radians; A is the rolling resistance coefficient, kW s m −1 ; B is the rotational resistance coefficient, kW s 2 m −2 ; C is the aerodynamic drag coefficient, kW s 3 m −3 .These road-load coefficients (i.e., A, B, and C) by each major HDDV category used in this study are summarized in Table 3.We further developed operating mode bins defined by VSP and vehicle speed with reference to the operating mode binning methodology applied in MOVES (US EPA, 2009).These bins were slightly modified due to the availability of test data in each bin, specifically in the high-speed segments.Finally, we established 22 operating modes as described in Table 4, including a deceleration or braking bin, an idling bin, and 20 bins to represent cruise or acceleration driving modes.

Emission rates and emission factors
The second-by-second emission rates (in g s −1 ) for each individual HDDV were collected.Equation 2 summaries the way we calculated the average emission rate for each group by emission standard, GVW and operating mode bin (US EPA, 2009).
where T k is the number of second-by-second data for each vehicle in operating mode bin k; N i is the total number of vehicles in vehicle category i; ER j is the instantaneous emission rate of pollutant j , g s −1 ; and ER i,j,k is the average emission rate of pollutant j for vehicle category i and operating mode bin k, g s −1 .The distance-specific emission factor in g km −1 was further developed based on the average emission rates and time distribution of operating mode bins within a specific driving cycle, as shown in Eq. (3).Bin 18 Bin 28 Bin 38 * Bin 35 indicates the operating modes with speed higher than 80 km h −1 and VSP lower than 4 kW t −1 .Time percentages of high-speed mode bins are significantly lower than low-speed and middle-speed mode bins.According to the operating mode bins developed in MOVES 2004 (Koupal et al., 2005), those original five mode bins with VSP lower than 4 kW t −1 were ultimately merged into the Bin 35.
where EF i,j is the emission factor for vehicle category i of pollutant j , g km −1 ; ER i,j,k is the average emission rate of pollutant j for vehicle category i and operating mode bin k, g s −1 ; P k is the time percentage of operating mode bin k to the total driving cycle; T is the total time of the driving cycle, s; and L is the total distance of the cycle, km.
For comparison with the emission limits (in g kW h −1 ), we convert the distance-specific emission factors into brakespecific emission factors using on-road CO 2 and NO x emission data.First, we estimated the fuel mass based NO x emission factors (g kg fuel −1 ) with Eq. (4) (Liu et al., 2009).
where W Cdiesel is the carbon mass in diesel fuel, 870 g carbon per kg diesel.
We then applied the brake-specific fuel consumption (BSFC) value in converting fuel consumption-based emission factors (g kg fuel −1 ) into brake-specific emission factors (g kWh −1 ) with Eq. ( 5).The BSFC values for buses and trucks by weight are estimated based on the method applied by the MOBILE6 model (US EPA, 2002).
3 Results and discussion

NO x emission profiles for buses
Figure 1 presents the average emission rates of diesel buses for each operating mode bin from Euro II to Euro IV.Generally, the emission rates increase with VSP, similar to the trend in other studies (Liu et al., 2009;US EPA, 2009).However, we did not observe a clear decrease in emission rates from Euro II to Euro IV, although NO x emission limits were tightened.Using one-way ANOVA method, there are no statistically significant differences in NO x emission rates between Euro II, Euro III and Euro IV buses for operating mode bins 0 to 18.For buses in urban driving conditions, operating mode bins 0 to 18 usually cover most of the driving time (see more detailed discussion later).Furthermore, the average NO x emission rates for Euro III buses for those operating modes in the median-speed segment with VSP higher than 0 kW ton −1 (e.g.operating mode bins 24-28) are all statistically significantly higher than those for Euro II buses for the corresponding operating modes.On average, such increments in NO x emission rates for operating mode bins 24 to 28 are 42-67 %.It indicates Euro III buses would have higher on-road NO x emission levels on-average than Euro II buses under relatively high speed and high power demand conditions.
To eliminate the distinction of operating conditions between each individual test, all the emission factors were corrected to a representative driving cycle.This typical driving  cycle was developed based on more than 1 million secondby-second driving condition data on 18 typical bus routes in Beijing.Figure 2 illustrates the time allocation of each operating mode bin to total time of this typical driving cycle.Due to frequent stops and congestion for the bus fleet in urban areas, the idling time share is as high as 33 % and low-speed zones with operating mode bins 11 to 18 contribute more than 50 % of total time in such a driving cycle.There are no data in the high speed segment, because the maximum speed is lower than 80 km h −1 .Not surprisingly, the average speed of this typical bus driving cycle is as low as 18 km h −1 .
Figure 3 presents the distance-specific emission factors of NO x as well as other three air pollutants (i.e., CO, THC and PM 2.5 ) under this typical bus driving cycle for each emission standard category.For comparison, the diesel bus emission factors from MOBILE-China model (Wu et al., 2011) and COPERT4 v9.0 model (Gkatzoflias et al., 2011) are also provided in the figure.The NO x emission limits of certification testing for heavy-duty diesel engines have been significantly tightened.As shown in Table 1, the NO x emission limits for Euro III and Euro IV HDDVs are reduced by 29 % and 50 %, respectively, relative to Euro II HDDVs.However, the on-road NO x emission factors for diesel buses in this study did not improve as expected.The average NO x emission factors (with 95 % confidence interval, CI) for Euro II, Euro III and Euro IV urban buses are 11.3 ± 3.3 g km −1 , 12.5 ± 1.3 g km −1 , and 11.8 ± 2.0 g km −1 , respectively.With one-way ANOVA method, no statistically significant difference in NO x emission factors has been observed between Euro II and III buses.Even Euro IV buses equipped with an SCR system specifically to control NO x did not have significantly lower NO x emission factors than Euro III buses.The results are clearly different from the estimates calculated by the MOBILE-China model that were used in our previ- ous study (see Fig. 3c) (Wu et al., 2011).For example, with MOBILE-China we estimate a significant decrease in NO x emission factors (54 %) from Euro II to Euro III buses, and a significant decrease of 50 % from Euro III to Euro IV buses.Such estimates using a continuous decrease in NO x emission factors as emission standards tightened for HDDV have been used widely in previous policy evaluations (Wu et al., 2011;Zhou et al., 2010).Thus, NO x emissions for China's diesel bus fleet may be significantly underestimated based on these new NO x results.
The trend for other gaseous pollutants, such as CO and THC, is different.As shown in Fig. 3a and b, both THC and CO emission factors of diesel buses decrease as emission standards tightened, especially for Euro IV buses.On-road emission factors of CO and THC for Euro IV buses decreased by 54 % and 70 %, respectively, compared to Euro III buses.In addition, the results of PM 2.5 measurements in this test program indicate that PM 2.5 emissions of buses improved as well, as shown in Fig. 3d.For buses, average emission factors of PM 2.5 for Euro IV are reduced by 59 %, compared to Euro III buses.These trends are likely a result of engine modifications.More discussion on this issue is in the next sub-section.
To further evaluate the SCR performance, two more Euro IV diesel buses with an OBD system installed were selected.Using the OBD systems, we obtained detailed data regarding real-time engine profiles, shown in Table 5.Their exhaust temperature was measured as the SCR system would not activate when exhaust temperature is too low (Liu et al., 2011).The detailed information and test results of these two buses are listed in Table 5.For Bus 1, its exhaust temperature was lower than 170 • C during 29 % of its operating time, and Bus 2 operated at low temperatures as much as 68 % of operating time.This is a major reason why the SCR of these two  buses did not perform well.As a result, their NO x emission factors are both high (15.0 and 12.0 g km −1 ).The low exhaust temperature might be a frequent symptom, especially when the average speed and load mass of a bus both are low.Therefore, a challenge remains for manufacturers to ensure satisfactory performance of SCR systems for bus fleets under real-world conditions, especially when speed and load mass are low.Similar results were found in Europe.Several on-road PEMS and remote sensing studies for the European fleet also indicated unsatisfactory performance of SCR systems for Euro IV and even Euro V HDDVs in low-speed urban driving circumstances (Carslaw et al., 2011;Verbeek et al., 2010;Velders et al., 2011).As the Euro IV emission standard has not been implemented in the HDDT fleet in China, on-road emission data of Euro IV trucks equipped with SCR systems are unavailable.It might be reasonable to believe SCR systems for trucks could perform better than those for buses because the average load and driving speed of trucks are usually significantly higher than buses (Carslaw et al., 2011;Verbeek et al., 2010;Velders et al., 2011).

Atmos
In addition, directly emitted NO 2 (i.e., primary NO 2 ) is of more concern due to its greater impacts on local air quality and public health.In particular, several studies have suggested that oxidation catalysts such as the diesel oxidation www.atmos-chem-phys.net/12/9365/2012/catalyst (DOC) and the diesel particulate filter (DPF) might increase primary NO 2 emissions significantly (Grice et al., 2009;Hu et al., 2012;Kousoulidous et al., 2008).In China, most heavy-duty diesel trucks and buses that will comply with the Euro IV or V standard in the near future will apply SCR but not DPF since DPF is more sensitive to sulfur content in diesel fuel.In this study, we collected on-road NO and NO 2 emission test profiles by using SEMTECH-D PEMS for four Euro IV diesel buses with SCR and eight Euro III diesel buses without SCR for comparison.The average fraction of primary NO 2 to total NO x (by volume, with 95 % CI) for those eight Euro III buses is 3.2 ± 1.5 %.For comparison, such a ratio for those four tested Euro IV buses equipped with SCR system is as low as 1.0 ± 1.3 %.The results suggest that those HDDVs equipped with SCR might have very low primary NO 2 emission factors under the real-world conditions, which is consistent with the results from Europe (Kousoulidous et al., 2008;Ntziachristos et al., 2009;Velders et al., 2011).They reported that HDDV equipped with SCR would reduce the fraction of NO 2 to total NO x effectively compared to other technologies (such as HDDV with DPF).
Beijing EPB planned to release more stringent Euro V emission standards for the public-bus fleet in late 2012 (Beijing EPB, 2012).However, the technology option would still heavily rely on SCR.We envision that those SCR devices installed to comply with Euro V would not operate well for diesel buses during the low-speed stop-and-go driving cycles.Other solution options are necessary for the bus fleet in Beijing as well as other cities in China.Alternativefuels and advanced vehicle technologies have been introduced in the Beijing bus fleet since 2000.Compressed natural gas (CNG) buses represent such a success story.By 2009, the CNG buses reached over 4,200, accounting for nearly 20 % of bus stock in Beijing.Since 2008, new CNG or liquefied natural gas (LNG) buses in Beijing could even meet the Euro EEV (Enhanced Environmentally-friendly Vehicle) emission standards (Wu et al., 2011).We employed SEMTECH-D PEMS to measure emission factors of gaseous pollutants for 6 CNG EEV buses for comparison.The detailed information of each CNG bus and their emission factors are summarized in Table 6.The average NO x emission factor (with 95 % CI) is 5.8 ± 2.6 g km −1 .A significant decrease of ∼50 % is achieved as compared to Euro IV diesel buses.This indicates that the CNG bus could be a competitive alternative to NO x control for the bus fleet especially under the low-speed urban driving conditions.Since 2009, hybrid diesel buses and pure electric buses have also been introduced in Beijing.Many other cities in China also plan to substantially expand the bus stock for these so-called "Energy-Saving and New-Energy Vehicles" during the next decade (Wu et al., 2012).However, for NO x control the results from different studies are quite mixed.NO x emission factors for tested hybrid diesel buses are approximately within ± 20 % of those for diesel buses (León, 2011;Li et al., 2009Li et al., , 2011b)).At this stage, we cannot confirm a considerable improvement -34 - in NO x emissions for hybrid buses.For pure electric buses, no doubt they have zero NO x emissions at vehicle operation stage; however, Huo et al. (2010) pointed out a significant increase in that high upstream NO x emissions might offset the NO x reduction benefit from the tailpipe using a life-cycle perspective due to the fact that coal power plants dominate in China.It should also be noted that zero emissions at vehicle operation stage have larger positive impact on local air quality than increased same amount of emissions at power plants since the majority of vehicle emissions are concentrated in the urban area.Therefore, the environmental impact of electric vehicle to local/regional air quality needs to be carefully evaluated in the future.

NO x emission profiles for trucks
Similar to buses, the emission factors for trucks were normalized to a typical driving cycle to eliminate the distinction of operating conditions between each individual test.Not surprisingly, the driving cycle for trucks is far different than that for buses.Figure 4 presents the allocation of time of each operating mode bin to total time of this typical driving cycle for trucks.The idling time share for trucks is 6 %, much lower than the 33 % for buses.Consequently, the time share of medium-speed zones with operating mode bins 21 to 28 increases significantly to ∼55 %, versus only ∼10 % for buses in the same zones.The average speed of the truck driving cycle is 40 km h −1 .It should be noted that we were unable to collect data for high-speed zones with speed more than 80 km h −1 due to the Beijing city speed limit.
The GVWs for test trucks vary widely from each other, from ∼3.5 tons to as heavy as more than 30 tons.As shown in Fig. 5  usually equipped with higher powered engines.To minimize the impact of vehicle weight when comparing different emission standards, the test trucks were grouped into three weight classifications as HDDT1, HDDT2 and HDDT3.Within each weight group, the average GVW results of trucks complied with each of the three different emission standards.
Figure 6 shows the on-road distance-specific emission factors of THC, CO, NO x and PM 2.5 for the three emission standards and three vehicle weight classifications, together with emission factors estimated by MOBILE-China and COPERT4 v9.0 models.The NO x emission factors (with 95 % CI) for Euro I, Euro II and Euro III HDDT1 are 4.8 ± 1.5 g km −1 , 4.6 ± 1.8 g km −1 , 2.0 ± 0.7 g km −1 , respectively.For HDDT2 the NO x emission factors for Euro I, Euro II and Euro III are 7.8 ± 14.0 g km −1 , 4.3 ± 2.0 g km −1 and 6.0 ± 2.9 g km −1 , respectively.For HDDT3, the NO x emission factors for Euro I, Euro II and Euro III are 9.2 ± 2.8 g km −1 , 6.7 ± 1.7 g km −1 , 7.2 ± 1.2 g km −1 , respectively.
For HDDT2 and HDDT3, the NO x emission factor trend from Euro I to Euro III is similar.The average NO x emission factors decrease 45 % (HDDT2) and 27 % (HDDT3) from Euro I to Euro II, then rebound from Euro II to Euro III at an increment of either 39 % for HDDT2 or 6 % for HDDT3.With one-way ANOVA method, there is no statistically significant difference between the Euro II and Euro III HDDT2 and HDDT3 fleet.Again, we did not observe any improvement in NO x emissions from Euro II to Euro III, similar to the findings for the bus fleet.A similar trend has also been identified by a recent on-road chasing study in Beijing and Chongqing (Wang et al., 2012).However, we did see a clear decrease in THC, CO, and PM 2.5 for most transitions from Euro II to Euro III (see Fig. 6a, b and d).These test results for HDDT2/HDDT3 fleets indicate that on-road engine performance has been modified significantly.Under real-world driving conditions, the internal combustion performance of these newer engines might be significantly improved.These improvements help reduce PM 2.5 , CO, and THC, but increase NO x emissions.To comply with the Euro III emission standards, the HDDV fleet in China is usually either equipped with High-pressure Common Rail (HPCR) systems or Exhaust Gas Recirculation (EGR) systems.However, it seems the new control technologies cannot offset the increment of NO x emissions due to the engine improvement, at least in real-world driving conditions.
It should be noted that HDDT1 is an exception.The difference in NO x emission factors for Euro II and Euro III is statistically significant.The average NO x emission factor for Euro III HDDT1 is 57 % lower than for Euro II HDDT1 in this study.A probable reason is the engine improvement in combustion performance for this relatively lighter truck category is not as significant as that for the heavier truck category (e.g., HDDT2 or HDDT3).However, due to limited test data available for Euro III HDDT1 in this study (only four  samples), more on-road test data are necessary to support this judgment.
As Euro III trucks have penetrated the market of mega cities (e.g., Beijing) since 2006 and nationwide since mid 2007 and Euro IV standards for HDDT were postponed to no sooner than 2013, Euro III HDDT has become the dominant category in the HDDT fleet, especially in big cities.Compared to previous estimates for Euro III HDDT by the MOBILE-China model (Wang et al., 2010;Wu et al., 2011), the new on-road NO x emission factors for HDDT2 and HDDT3 are usually ∼50-80 % higher.Therefore, similar to the bus fleet, an update in total NO x emissions for the HDDT fleet would be expected.

Comparison between emission factors under real-world condition and standard limits
As discussed above, the real-world operating conditions of HDDVs vary significantly.However, our certification standard test cycle is fixed.With OBD data from the two Euro IV diesel buses, we constructed a graph of real-time engine speed and engine power, as shown in Fig. 7.The average speeds of these two Euro IV buses were as low as 15 and 12 km h −1 , respectively.The engine speed was below 1400 rpm 85 % of the time based on real-time their OBD data regarding engine speed and engine power, which indicates for a majority of the real-world operation the engines worked beyond the emission control zone under the European Steady Cycle (ESC) (i.e., the area within 13 steady modes) (Krahl et al., 2009).The NO x brake-specific emission factors for the two tested buses are 9.4 g kWh −1 and 6.6 g kWh −1 , respectively.Not surprisingly, their on-road NO x brake-specific emission factors are much higher than the emission standard limits of 3.5 g kW h −1 for Euro IV engines.Some other studies have similarly found that the emission test results of heavy-duty engines under the certification standard test cycle do not reflect real-world emissions (Liu et al., 2011;Ligterink et al., 2009).To develop a specific compliance in-use testing program for buses is clearly necessary in the future since their operating conditions are quite unique with a typical low-speed profile.Otherwise, newer engines with advanced controls that make it possible to pass the certificate test might not approach standards in real-world conditions when their controls fail or operate suboptimally.Figure 8 presents the estimated brake-specific emission factors for test buses from Euro II to Euro IV and for test trucks from Euro I to Euro III.With one-way ANOVA method, there is no statistically significant difference in brake-specific emission factors of NO x between the different vehicle categories and weight classifications.Clearly, the estimated brake-specific NO x emission factors do not improve under real-world operating conditions with more stringent emission standards, although the emission limits for certification tests are tighter.As a result, the gap between the on-road brake-specific emission factors and emission limits for newer engines becomes larger and larger.For example, NO x brake-specific emission factors for Euro II HDDVs are typically 20-70 % higher than the Euro II limit.However, for Euro III HDDVs they are 60-130 % higher and Euro IV diesel buses average 170 % higher.Even for modern Euro V trucks in Europe, recent on-road PEMS test results also indicate significantly higher emission factors relatively to emission limits and estimates based on laboratory tests (Ligterink et al., 2009).Such a situation demonstrates the clear need for China's policy-makers to develop in-use compliance testing programs and relevant regulations for HDDVs to complement the current engine certificate test program, and to secure a real on-road emission reduction benefit of NO x for the HDDV fleet in China in the future.

Revisiting NO x emissions of the HDDV fleet in Beijing and China
Using the on-road test data in this study, we updated total NO x emissions for Beijing's HDDV fleet.In our previous study, the emission inventory for the vehicle fleet in Beijing in 2009 was based on emission factors simulated by the most recent version of MOBILE-China (Wu et al., 2011).
The NO x emission factors for HDDVs from this model are also presented in Figs. 3 and 6.We assume the vehicle registration data and activity data for buses and trucks remain the same as in the previous study.It should be noted that these new test data could not be directly used for national NO x emission inventory development because there are other factors that need to be considered (e.g., the impact of the relatively higher sulfur content in diesel fuel on NO x emissions in regions other than Beijing).However, we made a rough estimate of total NO x emissions for the national HDDV fleet using these new onroad test data.Using statistical data and published reports and papers (CACART and CAAM, 2011, MEP et al., 2010, NBSC, 2010), we gathered HDDV fleet registration data by major vehicle category and vehicle activity data.We estimate that the total NO x emissions for the national HDDV fleet in 2009 are approximately 3.9 Tg.This is higher than previous estimates.For example, MEP reported that national total NO x emissions from vehicles were 5.3 Tg in 2009, of which diesel vehicles contributed 60 %, close to 3.2 Tg (MEP, 2010).With the data split between the HDDV and light-duty diesel vehicle (LDDV) fleets, we estimate that HDDV contribute roughly 95 % of total diesel vehicle NO x emissions.Thus, the reported NO x emission data for the HDDV fleet should be around 3.0 Tg.Our new estimate indicates a total increment of ∼1.0 Tg of NO x for the national HDDV fleet.This is 18 % of total NO x emissions from vehicles in China in 2009.It would also result in a 4 % increase in estimation of national anthropogenic NO x emissions (State Council, 2010).Such a significant underestimate in vehicular Y. Wu et al.: The challenge to NO x emission control for heavy-duty diesel vehicles in China NO x emissions triggers an urgent need to re-evaluate and update the national emission inventory as well as other regional/international emission inventories (e.g., GAINS-Asia, INTEX-B, etc.) (GAINS-Asia, 2008;Saikawa, et al., 2011;Xing et al., 2011;Zhang et al., 2009).This effort will also provide a better understanding of the future modeling for ambient NO x concentration profiles as well as other relevant secondary pollutant profiles in the air (e.g., O 3 and nitrate particles).
Furthermore, the revised estimate of HDDV NO x would also challenge the national control efforts for NO x .As mentioned earlier, the new 12th FYP establishes a control target for NO x emission reductions of 10 % for the period of 2011-2015.Previously, we expected the decrease in NO x emissions from Euro II to Euro III HDDV could contribute to a considerable reduction in NO x emissions for the vehicle fleet.However, the new on-road test results do not support this judgment.Therefore, to achieve the goal of NO x mitigation from the HDDV fleet, policy-makers may have to rely on other effective control measures such as promotion of CNG/LNG buses, the development of an effective inuse compliance testing program and relevant regulations for HDDVs, and the implementation of Euro IV for heavy-duty diesel trucks nationwide.

Conclusions
More than 130 HDDVs in Beijing are tested using PEMS, and the real-world on-road emission profiles for NO x as well as other pollutants (i.e., THC, CO and PM 2.5 ) are explored.An operating mode binning methodology is applied to obtain instantaneous emission characteristics and relate them to real-time operating conditions.We evaluate the NO x emission factors for both buses and trucks, in detail by various emission standards and vehicle weight classifications under typical on-road driving cycles.
The average NO x emission factors for Euro II, Euro III and Euro IV urban buses are 11.3 ± 3.3 g km −1 , 12.5 ± 1.3 g km −1 , and 11.8 ± 2.0 g km −1 , respectively, under a typical urban driving cycle.Different from a clearly decreasing trends in emission factors of THC, CO and PM 2.5 , no statistically significant improvement in NO x emission factors has been observed as emission standards get tightened from Euro II to Euro IV.With the detailed OBD data regarding real-time operating conditions of engines, the unsatisfactory SCR performance for Euro IV buses under low-speed driving cycles is identified.Therefore, to develop a specific compliance in-use testing program for buses is clearly necessary since their operating conditions are quite unique with a typical low-speed profile.Furthermore, the CNG bus could be a competitive alternative to NO x control for the urban bus fleet.
Similar to diesel buses, we did not observe statistically significant improvement in NO x emissions from Euro I to Euro III trucks for most cases.Under a typical urban driving cycle, the NO x emission factors for Euro I, Euro II and Euro III HDDT1 are 4.8 ± 1.5 g km −1 , 4.6 ± 1.8 g km −1 , 2.0 ± 0.7 g km −1 , respectively.For HDDT2, the NO x emission factors for Euro I, Euro II and Euro III are 7.8 ± 14.0 g km −1 , 4.3 ± 2.0 g km −1 and 6.0 ± 2.9 g km −1 , respectively.For HDDT3, the NO x emission factors for Euro I, Euro II and Euro III are 9.2 ± 2.8 g km −1 , 6.7 ± 1.7 g km −1 , 7.2 ± 1.2 g km −1 , respectively.They are lower than the NO x emission factors for diesel buses because the average speed of the typical driving cycle for trucks is much higher.
The brake-specific NO x emission factors did not improve either under real-world operating conditions with more stringent emission standards, although the emission limits for certification tests are tighter.Consequently, the gap between the on-road brake-specific emission factors and emission limits for newer engines becomes larger and larger.For example, NO x brake-specific emission factors for Euro II HDDVs are typically 20-70 % higher than the Euro II limit.However, for Euro III HDDVs they are 60-130 % higher and Euro IV diesel buses average 170 % higher.Such a situation demonstrates again the clear need for China's policy-makers to develop in-use compliance testing programs and relevant regulations for HDDVs to complement the current engine certificate test program, and to secure a real on-road emission reduction benefit of NO x for the HDDV fleet in China in the future.
Updated on-road NO x emission factors for both buses and trucks would result in a significant underestimate in NO x emissions for HDDV fleet from other previous studies.Our new estimate in total NO x emissions for the Beijing HDDV fleet in 2009 is 37.0 ± 3.1 Gg, an increase of 45 % compared to the previous study.Furthermore, we estimate that the total NO x emissions for the national HDDV fleet in 2009 are approximately 4.0 Tg, ∼ 1.0 Tg higher than that estimated in the official report.The increment is equivalent to 18 % of total NO x emissions from vehicles in China in 2009.This would also result in 4 % increase in estimation of national anthropogenic NO x emissions.Such an underestimate in vehicular NO x emissions triggers an urgent need to re-evaluate and update the national emission inventory as well as other regional/international emission inventories.

Fig. 1 .
Figures 1 2 Fig.2.Allocation of time of each operating mode bin to total time of the typical driving cycle for urban buses in Beijing, with an average speed of 18 km h -1 .

Fig. 2 .
Fig. 2. Allocation of time of each operating mode bin to total time of the typical driving cycle for urban buses in Beijing, with an average speed of 18 km h −1 .

Fig. 3 .
Fig. 3. Emission factors of (a) THC, (b) CO, (c) NO x and (d) PM 2.5 for Euro II, Euro III and Euro IV diesel buses under a typical urban driving cycle with an average speed of 18 km h −1 .

6 Fig. 4 .
Fig. 4. Allocation of time of each operating mode bin to total time of the typical driving cycle for trucks in Beijing, with an average speed of 40 km h −1 .
Fig. 5.The scatter graph of the relationship between vehicle weight and emission factor under the typical driving cycle for trucks with an average speed of 40 km h -1 , from Euro I to Euro III.

Fig. 5 .
Fig. 5.The scatter graph of the relationship between vehicle weight and emission factor under the typical driving cycle for trucks with an average speed of 40 km h −1 , from Euro I to Euro III.

Y.
Fig. 6.Emission factors of (a) THC, (b) CO, (c) NO X and (d) PM 2.5 for Euro I, Euro II and Euro III HDDTs under a typical driving cycle with an average speed of 40 km h -1 .

Fig. 6 .Fig. 7 .Fig. 8 .
Fig. 6.Emission factors of (a) THC, (b) CO, (c) NO x and (d) PM 2.5 for Euro I, Euro II and Euro III HDDTs under a typical driving cycle with an average speed of 40 km h −1 .

Table 1 .
Summary of emission standards for new HDDVs adopted in Beijing and China.
requires that engines meet the Not-to-Exceed limits by using PEMS during real-world operation to compose a complementary test suite with the FTP cycle and steady-state Supplemental Emission Test

Table 3 .
Summary of road-load coefficient values for calculating VSP of each major HDDV category.

Table 4 .
Definition of operating mode bins by using VSP and vehicle speed (v).

Table 5 .
Summary of information of two tested Euro IV buses with data regarding with real-time engine conditions recorded by OBD system.The load mass was estimated based on the mass of the PEMS and the number of people on board during the test.Such load mass data are equivalent to 13 to 16 passengers.It could represent a typical passenger load during non-rush hours in Beijing. *

Table 6 .
Summary of on-road PEMS test results for CNG EEV buses.
a Those distance-specific emission factors in this table are normalized to the typical driving cycle (see Fig.2); b Not available due to malfunction of the flame ionization detector for SEMTECH-D PEMS.
(Wu et al., 2011)for pre-Euro II non-public buses and pre-Euro I trucks are provided by the MOBILE-China model(Wu et al., 2011).The new results show that NO x emissions (with 95 % CI) for the public diesel bus fleet, other diesel bus fleets, and the HDDT fleet in Beijing in 2009 are 9.1 ± 1.4 Gg, 4.8 ± 0.5 Gg, and 23.0 ± 2.7 Gg, respectively.These three values increase by 119 ± 15 %, 71 ± 11 % and 24 ± 12 % compared to our previous estimates.The most significant increase in NO x emissions was found for the public bus fleet, which is attributed to Euro III and Euro IV diesel buses becoming the majority of the public bus fleet in Beijing.Such poor on-road performance in NO x control for Euro III and Euro IV diesel buses causes a significant underestimate in NO x emissions in previous studies.Overall, total NO x emissions of the Beijing HDDV fleet in 2009 reached 37.0 ± 3.1 Gg, approximately an increase of 45 % compared to the previous study (25.0 Gg).