Assessment of Gaseous and Particulate Emissions of a Euro 6d-Temp Diesel Vehicle Driven > 1300 km Including Six Diesel Particulate Filter Regenerations

: Diesel-fueled vehicles have classically had high particulate and NO x emissions. The introduction of Diesel Particulate Filters (DPFs) and Selective Catalytic Reduction for NO x (SCR) systems have decreased the Particle Number (PN) and NO x emissions, respectively, to very low levels. However, there are concerns regarding the emissions released during the periodic DPF regenerations, which are necessary to clean the ﬁlters. The absolute emission levels and the frequency of the regenerations determine the contribution of regenerations, but where they happen (city or highway) is also important due to di ﬀ erent contributions to human exposure. In this study, we measured regulated and non-regulated emissions of a Euro 6d-temp vehicle both in the laboratory and on the road. PN and NO x emissions were similar in the laboratory and on-the road, ranging around 10 10 p / km and 50 mg / km, respectively. Six regeneration events took place during the 1300 km driven, with an average distance between regeneration events of only 200 km. During regeneration events, the laboratory limits for PN and NO x , although not applicable, were exceeded in one of the two measured events. However, the on-road emissions were below the applicable not-to-exceed limits when regenerations occurred. The weighted PN and NO x emissions over the regeneration distance were approximately two times below the applicable limits. The N 2 O emissions were < 14 mg / km and NH 3 at instrument background level ( < 1 ppm), reaching 8 ppm only during regeneration. The results of this study indicate that due to the short interval between regenerations, studies of diesel vehicles should report the emissions during regeneration events.


Introduction
Air pollution is one of the main environmental threats worldwide as it impacts ecosystems [1], economic activities [2], and human health [3]. Short and long-term exposure to elevated concentrations of Particulate Matter (PM) and nitrogen dioxide (NO 2 ) in the air have been associated with respiratory [4] and cardiovascular [5] diseases, hypertension and diabetes [6], and carcinogenicity [7]. Recently, long-term exposure to elevated levels of fine PM (PM 2.5 ) and NO 2 have also been associated with an increase in the mortality rate from COVID-19 (coronavirus disease 2019) [8][9][10][11]. Air pollution has been identified as the most relevant public health risk in European cities with strong associations between PM and NO 2 concentrations in the air, population density, and road transport emissions [12,13]. The share of European population exposed to concentrations exceeding the WHO (World Health Organization) air quality guideline annual mean PM 2.5 (>10 µg/m 3 ) and NO 2 (>40 µg/m 3 ) in 2017 were 77% and 7%, respectively. Furthermore, densely populated and heavily trafficked cities such as London, Paris, Torino, and Munich, averaged an annual mean NO 2 concentration above 80 µg/m 3

Laboratory Tests
The laboratory emission tests were performed at JRC's (Joint Research Centre) vehicle emissions laboratory (VELA 2). The tests were conducted according to the Worldwide harmonized Light vehicles Test Procedure (WLTP) with cold engine start, i.e., the lubricant oil temperature was within 1 • C from the ambient temperature at the start of the test. The respective cycle was the Worldwide Light-duty Test Cycle (WLTC). The laboratory ambient temperature was set to 0 • C, 14 • C, 23 • C or 30 • C. The WLTP, in addition to the new test cycle, respected all provisions from Commission Regulation (EU) 2017/1151 including the preconditioning and soaking requirements for the 23 • C tests.
The whole exhaust gas was diluted in a dilution tunnel with Constant Volume Sampling (CVS) equipped with critical flow venturi to measure total air flow and a SAO (Smooth-Approach Orifice) to measure dilution air flow. Gas analyzers (MEXA 7400, Horiba, Kyoto, Japan) were measuring from the dilution tunnel in real time. A small part of the diluted gas was also collected in bags and was analyzed for the gaseous pollutants as required in the regulation ( Figure 1). The principle of operation of the analyzers was: non-dispersive infrared detection for CO and CO 2 , chemiluminescence for NO x , and hot (191 • C) flame ionization detection for total hydrocarbons (THC) and methane (CH 4 ).
The solid Particle Number (PN) emissions were measured from the full dilution tunnel with an AVL (Graz, Austria) Particle Counter (APC 489) [47]. The system had a hot dilution (150 • C), an evaporation tube (350 • C) and a secondary diluter at ambient temperature. Solid particles were counted with a Condensational Particle Counter (CPC) model 3790 from TSI Inc. (Shoreview, MN, USA), having a counting efficiency of 50% at 23 nm. In addition, a CPC with 50% counting efficiency at 10 nm (model 3010, TSI Inc.) was sampling from APC's secondary dilution exhaust outlet. No particle correction was applied for the sub-23 nm particles, as this will not be required in the future >10 nm regulation. A Fourier Transform Infrared Spectrometer (FTIR) (Sesam i60 from AVL, Austria, Graz) with a heated polytetrafluoroethylene sampling line at 191 °C was sampling from the tailpipe, to characterize non-regulated pollutants. The data acquisition frequency was 1 Hz. Compounds measured, among others, included NH3, N2O, formaldehyde (HCHO), acetaldehyde (CH3CHO) and isocyanic acid (HNCO).
The solid Particle Number (PN) emissions were measured from the full dilution tunnel with an AVL (Graz, Austria) Particle Counter (APC 489) [47]. The system had a hot dilution (150 °C), an evaporation tube (350 °C) and a secondary diluter at ambient temperature. Solid particles were counted with a Condensational Particle Counter (CPC) model 3790 from TSI Inc. (Shoreview, MN, USA), having a counting efficiency of 50% at 23 nm. In addition, a CPC with 50% counting efficiency at 10 nm (model 3010, TSI Inc.) was sampling from APC's secondary dilution exhaust outlet. No particle correction was applied for the sub-23 nm particles, as this will not be required in the future >10 nm regulation.
The PM mass was determined using a filter holder heated at 47 °C. A single filter was used for the whole measurement campaign in the laboratory in order to collect high enough mass on the filter and to minimize the influence of volatile species condensing on the filter. Even though this approach was not following the technical requirements described in the WLTP for particle mass (one filter is required per test), it should give a good estimation of the mean mass emissions.
Some engine and vehicle parameters (such as engine speed and coolant temperature) were logged through the OBD port at 1 Hz frequency.

On-Road Tests
The Portable Emissions Measurement System (PEMS) used was the MOVE (model 2016) from AVL, fulfilling the performance requirements from the real-driving emissions (RDE) Commission Regulation (EU) 2017/1151. It consisted of an Exhaust Flow Meter (EFM), exhaust gas analyzers and a solid particle counter and heated exhaust lines, a GPS (Global Positioning System) antenna, and a weather station measuring ambient temperature, pressure and relative humidity. The PEMS measured the exhaust gas concentrations of CO and CO2 with a non-dispersive infrared analyzer, and NO and NO2 with a non-dispersive ultra-violet analyzer (with independent chemical cells for NO and NO2). Solid PN >23 nm were measured by means of a diffusion charging-based detector downstream of a heated catalytic stripper at 300 °C [48]. The main PEMS unit was mounted inside the vehicle. The EFM used a Pitot tube (2.5 inches) to measure the exhaust flow rate. The average and The PM mass was determined using a filter holder heated at 47 • C. A single filter was used for the whole measurement campaign in the laboratory in order to collect high enough mass on the filter and to minimize the influence of volatile species condensing on the filter. Even though this approach was not following the technical requirements described in the WLTP for particle mass (one filter is required per test), it should give a good estimation of the mean mass emissions.
Some engine and vehicle parameters (such as engine speed and coolant temperature) were logged through the OBD port at 1 Hz frequency.

On-Road Tests
The Portable Emissions Measurement System (PEMS) used was the MOVE (model 2016) from AVL, fulfilling the performance requirements from the real-driving emissions (RDE) Commission Regulation (EU) 2017/1151. It consisted of an Exhaust Flow Meter (EFM), exhaust gas analyzers and a solid particle counter and heated exhaust lines, a GPS (Global Positioning System) antenna, and a weather station measuring ambient temperature, pressure and relative humidity. The PEMS measured the exhaust gas concentrations of CO and CO 2 with a non-dispersive infrared analyzer, and NO and NO 2 with a non-dispersive ultra-violet analyzer (with independent chemical cells for NO and NO 2 ). Solid PN >23 nm were measured by means of a diffusion charging-based detector downstream of a heated catalytic stripper at 300 • C [48]. The main PEMS unit was mounted inside the vehicle. The EFM used a Pitot tube (2.5 inches) to measure the exhaust flow rate. The average and maximum on-road measured exhaust flow rate was 133 kg/h and 700 kg/h, respectively. The maximum range of the flow meter (970 kg/h) was not exceeded. The mass of the PEMS was circa 135 kg and the maximum payload of the vehicle was not exceeded.
The tests on the road were performed in February 2020 on public paved roads in the proximity of the laboratories of JRC, over a set of routes within and outside the RDE boundary conditions. Table 2 summarizes the main characteristics of the five routes used. Figure A1 in the Appendix A displays the vehicle speed and altitude profiles of the routes used in the campaign. The vehicle was driven on two routes (RDE-1 and RDE-2) designed to meet all the criteria from the RDE regulation (trip duration, composition, temperature range, altitude range, cumulative positive elevation gain, etc.). Details on the routes can be found elsewhere [45]. The vehicle was additionally tested on both routes but driven with a more aggressive driving style (RDE-1-D and RDE-2-D), resulting in more dynamic trip indicators (above max. 95th percentile of v × a RDE limits, see Figure A2) in the Appendix. Even though the total distance between the normal and dynamic driving routes were the same, the urban, rural and motorway shares were slightly different due to the different speeds, but within typical experimental repeatability. The higher driving dynamics were achieved by performing more stop and go, and faster accelerations after vehicle stops at traffic lights, for example. All tests were performed respecting the Italian road safety code. Additionally, another three non-RDE-compliant routes that represent a combination of motorway-city-motorway drive (Motor-City), a prolonged motorway drive (Motorway) and a driving route simulating congested urban conditions with frequent stop and go situations (Traffic) were used. The traffic route consisted of five loops around the JRC site including several 30-40 s stops, two performed at an average speed (including stops) of 17 km/h, and three at an average speed of 22 km/h. One repetition per test was performed except for the Traffic route (2 repetitions) and RDE-1 (2 repetitions). In one of the two RDE-1 repetitions, a DPF regeneration took place. Regeneration also took place on the RDE-2-D, Motorway, and Motor-City tests, but these tests were not repeated.
The vehicle was soaked inside a facility at an ambient temperature of circa 18 • C. The ambient temperature during the test campaign ranged from 1 • C to 21 • C, with an average temperature of 13 • C. No driving occurred in extended temperature conditions (either < 0 • C or above 30 • C) or extended altitude conditions (>700 m). The start/stop system was used on all tests. Zero and span calibrations were performed systematically prior to and after the test. The zero and span drift were within the permissible drifts defined in the RDE regulation.

Calculations
The laboratory bag results (in g/km) and the second-by-second results from the dilution tunnel (in g/s) were provided by the chassis dynamometer automation software. The calculations were conducted according to the regulation.
Distance-specific emissions of non-regulated pollutants (in mg/km) were calculated from concentrations measured with the FTIR, exhaust mass flow was calculated as the difference between total air and dilution air flows, and distance was inferred from chassis dynamometer vehicle speed. The CO 2 and CH 4 values calculated from the FTIR were within ±3% of the bag results, while the NO x results were within ±20% (see Figure A3 in the Appendix A for a bag vs. FTIR intercomparison throughout the laboratory campaign).
The PEMS data was extracted into a Microsoft Excel file using AVL CONCERTO [49] Version 502. EMROAD version 6.03 [50] was used to calculate second-by-second emissions (in g/s) or Atmosphere 2020, 11, 645 7 of 25 distance-specific emissions (in g/km), using the GPS as the source for vehicle speed. A PEMS validation in the chassis dynamometer was performed immediately after the on-road campaign. The differences in gaseous and particle measurements of the PEMS against the laboratory bags were within the permissible tolerances of the RDE regulation: +10 mg/km for NO x (+20%), +10 g/km for CO 2 (+4%), <25 mg/km CO, −3 × 10 10 p/km for PN (−21%) [51,52].
The verification of the overall trip dynamics was done using the moving average window method as prescribed in EU regulation 2018/1832 [53], and using the declared distance-specific WLTP CO 2 value as reported in the CoC. The on-road emissions calculated and reported in this investigation are, however, the raw emissions (not corrected for extended conditions and without using the weighting function based on CO 2 emissions, as introduced in the fourth package of the RDE regulation EU 2018/1832). These raw emissions include all the emissions from test start to test end including cold start and idling periods. The emissions are compared against the not-to-exceed (NTE) limits for illustration purposes, as emissions compliance necessarily needs to be done against emissions calculated following Appendix 6 of the RDE regulation. For reference, only the two Traffic tests, and the Motorway test would require a 0.65 and 0.8 correction factor, respectively, due to the higher CO 2 on the road than in the laboratory, as described in Appendix 6. Note though that corrections and compliance to the limits are applicable only for the RDE-compliant tests.  Figure 2a) were almost one order of magnitude below the PN limit (6 × 10 11 p/km), at the cold WLTC at 23 • C and 30 • C (<7 × 10 10 p/km). They were three times higher at cold WLTC with a regeneration event (WLTC R 23 • C), but still half way from the limit (even though not applicable). They slightly exceeded the limit at the WLTC at 0 • C and they were four times higher at the WLTC with regeneration at 14 • C. The concentration of 10-23 nm particles was 11% to 51% of the >23 nm particles. The highest ratio was noted at 30 • C, where the absolute levels were the lowest. Quite a high percentage (28%) was also measured at the 14 • C WLTC with regeneration.  One filter was taken for the whole campaign and the PM mass emissions were 0.84 mg/km. The equivalent PN emissions were approximately 3-4 × 10 11 p/km (for >23 nm and >10 nm, respectively), giving a correlation factor of 3.8-4.8 × 10 11 p/mg.

Laboratory Tests
The NOx emissions were around 30 mg/km at the cold WLTC at 23 °C and 30 °C, and around 50 One filter was taken for the whole campaign and the PM mass emissions were 0.84 mg/km. The equivalent PN emissions were approximately 3-4 × 10 11 p/km (for >23 nm and >10 nm, respectively), giving a correlation factor of 3.8-4.8 × 10 11 p/mg.
The NO x emissions were around 30 mg/km at the cold WLTC at 23 • C and 30 • C, and around 50 mg/km at the WLTC at 0 • C and the regenerating WLTC at 23 • C, below the limit of 80 mg/km. They exceeded two times the limit (not applicable) at the regenerating WLTC at 14 • C. The NO/NO x ratio varied between 70% and 91%, without any particular trend between regenerating or non-regenerating cycles. Figure 3 gives examples of PN and NO x concentrations for the cold WLTC at 23 • C with and without regeneration and at 14 • C with regeneration. High PN emissions were measured at the cold start of the cycle and during the regeneration (Figure 3a). Similarly, for NO x , high concentrations were measured during cold start and regeneration. One filter was taken for the whole campaign and the PM mass emissions were 0.84 mg/km. The equivalent PN emissions were approximately 3-4 × 10 11 p/km (for >23 nm and >10 nm, respectively), giving a correlation factor of 3.8-4.8 × 10 11 p/mg.
The NOx emissions were around 30 mg/km at the cold WLTC at 23 °C and 30 °C, and around 50 mg/km at the WLTC at 0 °C and the regenerating WLTC at 23 °C, below the limit of 80 mg/km. They exceeded two times the limit (not applicable) at the regenerating WLTC at 14 °C. The NO/NOx ratio varied between 70% and 91%, without any particular trend between regenerating or nonregenerating cycles. Figure 3 gives examples of PN and NOx concentrations for the cold WLTC at 23 °C with and without regeneration and at 14 °C with regeneration. High PN emissions were measured at the cold start of the cycle and during the regeneration (Figure 3a). Similarly, for NOx, high concentrations were measured during cold start and regeneration.   Figure 4 gives examples of NH 3 and N 2 O for the WLTC at 23 • C and 14 • C with and without regeneration, as measured with the FTIR from the tailpipe. Regarding NH 3 (Figure 4a), the emissions were at the instrument background levels (<1 ppm), except during regeneration, where up to 8 ppm were detected. This would translate to 1 mg/km for the regenerating cycle. The N 2 O emissions ( Figure 4b) were variable over the cycle and the regeneration did not result in a significant increase in N 2 O. The average concentration for the WLTC at 23 • C was 11 mg/km (2.8 CO 2 -equivalent) and reached 12.5-13.5 mg/km at the regenerating cycles. Other non-regulated pollutants such as HCHO (0.3 mg/km) and CH 3 CHO (19 mg/km) were low over the WLTC at 23 • C. HNCO averaged 0.35 mg/km over the WLTC at 23 • C, but ten times higher at 0 • C. Emissions of CH 4 , which were not particularly affected by ambient temperature (6.5 mg/km at 0 • C, 5.4 mg/km at 23 • C, 5.7 mg/km at 30 • C), were twice as high during the cycles where a regeneration occurred (12.8 mg/km at 14 • C and 12.4 mg/km at 23 • C). CH 4 emissions at 23 • C corresponded to <0.2% CO 2 -equivalent. N 2 O and CH 4 accounted for circa 1% of the CO 2 -equivalent of the vehicle on the WLTP 23 • C test. Table A1 in the Appendix A provides the emissions for regulated and non-regulated pollutants as measured with FTIR from the tailpipe over the WLTC at different ambient temperatures. affected by ambient temperature (6.5 mg/km at 0 °C, 5.4 mg/km at 23 °C, 5.7 mg/km at 30 °C), were twice as high during the cycles where a regeneration occurred (12.8 mg/km at 14 °C and 12.4 mg/km at 23 °C). CH4 emissions at 23 °C corresponded to <0.2% CO2-equivalent. N2O and CH4 accounted for circa 1% of the CO2-equivalent of the vehicle on the WLTP 23 °C test. Table A1 in the Appendix provides the emissions for regulated and non-regulated pollutants as measured with FTIR from the tailpipe over the WLTC at different ambient temperatures.   Figure 5 summarizes the on-road PN and NOx results. The PN emissions ( Figure 5a) were very low 2 × 10 10 p/km at normal RDE-compliant or traffic conditions, 3 times higher at RDE dynamic driving (RDE D). They reached on average 2 × 10 11 p/km during RDE trips with regenerations (RDE* R). The maximum PN emissions during a regenerating RDE were 4.5 × 10 11 p/km, half of the on-road PN limit with a Conformity Factor (CF) 1.5. The scatter at the (complete) dynamic RDEs was due to the different PN levels during the regeneration events. In particular, for the urban part, the scatter was even higher due to one regeneration event taking place in the urban part.

On-Road Tests
The RDE NOx emissions ( Figure 5b) were below the applicable NTE limit for the specific vehicle (with CF = 2.1), even the future Euro 6d limit with CF = 1.43. They were on average below 100 mg/km at the traffic cycle or even at RDE tests with regenerations, although they could reach 120 mg/km in some routes. The dynamic driving resulted in emissions of around 200 mg/km. The levels of the complete and urban part were similar for the different trips, indicating a good NOx management even during urban driving. The NO/NOx ratio varied between 45% (RDE) and 85% (dynamic driving).  The maximum PN emissions during a regenerating RDE were 4.5 × 10 11 p/km, half of the on-road PN limit with a Conformity Factor (CF) 1.5. The scatter at the (complete) dynamic RDEs was due to the different PN levels during the regeneration events. In particular, for the urban part, the scatter was even higher due to one regeneration event taking place in the urban part.   The RDE NO x emissions ( Figure 5b) were below the applicable NTE limit for the specific vehicle (with CF = 2.1), even the future Euro 6d limit with CF = 1.43. They were on average below 100 mg/km at the traffic cycle or even at RDE tests with regenerations, although they could reach 120 mg/km in some routes. The dynamic driving resulted in emissions of around 200 mg/km. The levels of the complete and urban part were similar for the different trips, indicating a good NO x management even during urban driving. The NO/NO x ratio varied between 45% (RDE) and 85% (dynamic driving). Figure 6 gives examples of PN and NO x emissions during a DPF regeneration (time 4800 s until 5300 s). The PN emissions ( Figure 6a) increased 2-3 orders of magnitude from baseline levels. The regeneration was initiated at the motorway part of the RDE test. During the regeneration, there was a stop (toll in the highway) where the vehicle was idling. Nevertheless, the regeneration continued and finished the regeneration at the second part of the motorway phase. Then, the emissions remained relatively elevated until the end of the test as the DPF was empty and the filtration efficiency low.

On-Road Tests
The NO x (Figure 6b) had a similar behavior.
Examples of normal and dynamic driving at the rural and motorway parts are given in Figure 7. During normal driving (Figure 7a), the NO x emissions during rural driving were very low and only at one part of the motorway, where the exhaust gas temperature exceeds 550 • C, the NO x emissions increased to 10 mg/s. During dynamic driving, NO x spikes appeared during accelerations (Figure 7b). They also depended on the exhaust gas temperature and the exhaust gas flow. They were up to 15 mg/s at the rural part, but >40 mg/s at the motorway part. They reached 90 mg/s when the exhaust gas temperature was >600 • C.

Figure 5.
Emissions during on-road tests for the urban part or the complete trip: (a) Particle Number (PN) >23 nm; (b) NOx as sum of NO and NO2. Percentage gives the ratio NO/NOx. "RDE" refers to RDE-1 and RDE-2 (compliant routes) without any regeneration events. "D" stands for dynamic driving. "RDE* R" includes RDE-1, Motor-City and Motorway routes, all with regeneration events. "Traffic" is the urban route with many stop and go. Error bars show min-max values of two or three routes. The red dotted line gives the Real-Driving Emissions (RDE) limit with the Conformity Factor (CF). Figure 6 gives examples of PN and NOx emissions during a DPF regeneration (time 4800 s until 5300 s). The PN emissions ( Figure 6a) increased 2-3 orders of magnitude from baseline levels. The regeneration was initiated at the motorway part of the RDE test. During the regeneration, there was a stop (toll in the highway) where the vehicle was idling. Nevertheless, the regeneration continued and finished the regeneration at the second part of the motorway phase. Then, the emissions remained relatively elevated until the end of the test as the DPF was empty and the filtration efficiency low. The NOx (Figure 6b) had a similar behavior.
Examples of normal and dynamic driving at the rural and motorway parts are given in Figure  7. During normal driving (Figure 7a), the NOx emissions during rural driving were very low and only at one part of the motorway, where the exhaust gas temperature exceeds 550 °C, the NOx emissions increased to 10 mg/s. During dynamic driving, NOx spikes appeared during accelerations (Figure 7b). They also depended on the exhaust gas temperature and the exhaust gas flow. They were up to 15 mg/s at the rural part, but >40 mg/s at the motorway part. They reached 90 mg/s when the exhaust gas temperature was >600 °C.

Discussion
In this study, we measured the emissions of regulated and non-regulated pollutants of a diesel fueled vehicle both in the laboratory and on the road. For all regulated testing conditions, all currently regulated pollutants were below the applicable limits, in agreement with other studies of Euro 6dtemp diesel vehicles [45,46]. Particularly for NOx, which was the main attention topic for diesel vehicles until very recently, the emissions were below the applicable (and future) limits on the road

Discussion
In this study, we measured the emissions of regulated and non-regulated pollutants of a diesel fueled vehicle both in the laboratory and on the road. For all regulated testing conditions, all currently regulated pollutants were below the applicable limits, in agreement with other studies of Euro 6d-temp diesel vehicles [45,46]. Particularly for NO x , which was the main attention topic for diesel vehicles until very recently, the emissions were below the applicable (and future) limits on the road even when DPF regeneration events occurred, and much lower compared to the pre-RDE Euro 6 diesel vehicles that were typically emitting on the road >4 times the limit [23,29,[54][55][56][57]. Only dynamic driving brought the NO x levels (200 mg/km) slightly higher than the limit (168 mg/km, using the applicable 2.1 conformity factor). Others have also reported that dynamic driving significantly increases the emissions [36,46]. The NO/NO x ratio was on average 78% in the laboratory tests and 65% in the on-road tests. Partly, this has to do with the different instruments used: FTIR at the laboratory, non-dispersive ultra-violet (NDUV) analyzer on the road. Differences of 5-15% between NO x measurement instruments are common [58][59][60]. In our case, the FTIR NO x were on average 20% lower compared to the regulated bags methodology (Figure A3), and the PEMS 20% higher. Possible drift of the PEMS analyzers during the long duration on-road tests could also have affected the results. For example, 1 ppm drift in one of the NDUV analyzers (which is much lower than the 5 ppm allowed by the regulation) would change the percentages by 5%. The other reason is a true difference. A closer look at the data revealed that the high NO laboratory emissions are due to the high contribution of cold start, where the catalyst is inactive in converting NO to NO 2 . Furthermore, based on the tailpipe exhaust gas measurements, the laboratory cycles had lower exhaust gas temperatures, which could mean lower engine out NO and lower NO to NO 2 conversion at the oxidation catalyst [61]. On-road reported NO 2 /NO x percentages for Euro 6 vehicles with SCR around 45% (±12%) to 54% (±23%) [55,62], similar if only RDE-compliant cycles are considered: 56% (±3%); but higher if all cycles are considered: 35% (±23). The reason is that regenerations and dynamic driving had high NO/NO x ratios.

Cold Start
For most pollutants, the majority of the emissions were released during the first 3-4 min of the vehicle operation, as also reported by others [23,67]. During cold start, defined here as the first 300 s of a test, the engine and aftertreatment systems of a vehicle are cold and do not function in optimal conditions [68]. As most vehicles start operating in urban environments where population density is high, cold start emissions are relevant as they make an important contribution to the increase in population exposure to air pollutants. Figure 8 plots the cumulative NO x over the WLTCs (Figure 8a) and the two on-road Traffic routes (Figure 8b) for various ambient temperatures.
Most of the NO x were emitted during the first 300 s following the first ignition of the engine. For the tests without a DPF regeneration, the share of NO x emissions released in the first five minutes (17% of the test time) ranged from 85% to circa 100%, proving that cold start still represents a major contribution to NO x emissions for current state-of-the-art diesel vehicles. In the laboratory tests, after 300 s, the coolant temperature reached 60 • C when the ambient temperature was 23 • C, 55 • C at 14 • C and 48 • C at 0 • C. It took an additional 120 s to reach 60 • at an ambient temperature of 14 • C, and 220 s at 0 • C. The emissions during the first 300 s of the WLTCs were 220-290 mg/km for the 23 • C tests and 30 • C, 380 mg/km for the 14 • C test and 550 mg/km for the 0 • C test, proving that cold start at lower ambient temperatures results in increased NO x emissions. Different combustion and abatement strategies could explain this different NO x emission rates occurring at different ambient temperatures. For all tests, the curves of the cumulative NO x emission flattened after circa 250 s (300 s for the 0 • C test) when the SCR catalyst started effectively reducing NO x emissions. A detailed presentation of regulated and non-regulated pollutant emission during the cold start as compared to the complete WLTC is given in the Appendix A ( Figure A4).
Atmosphere 2020, 11, x FOR PEER REVIEW 12 of 25 by others [46,66]. Two points need more discussion: emissions during cold start at different ambient temperatures, and emissions during regeneration events.

Cold Start
For most pollutants, the majority of the emissions were released during the first 3-4 min of the vehicle operation, as also reported by others [23,67]. During cold start, defined here as the first 300 s of a test, the engine and aftertreatment systems of a vehicle are cold and do not function in optimal conditions [68]. As most vehicles start operating in urban environments where population density is high, cold start emissions are relevant as they make an important contribution to the increase in population exposure to air pollutants. Figure 8 plots the cumulative NOx over the WLTCs (Figure 8a) and the two on-road Traffic routes (Figure 8b) for various ambient temperatures. Most of the NOx were emitted during the first 300 s following the first ignition of the engine. For the tests without a DPF regeneration, the share of NOx emissions released in the first five minutes (17% of the test time) ranged from 85% to circa 100%, proving that cold start still represents a major contribution to NOx emissions for current state-of-the-art diesel vehicles. In the laboratory tests, after 300 s, the coolant temperature reached 60 °C when the ambient temperature was 23 °C, 55 °C at 14 °C and 48 °C at 0 °C. It took an additional 120 s to reach 60° at an ambient temperature of 14 °C, and 220 s at 0 °C. The emissions during the first 300 s of the WLTCs were 220-290 mg/km for the 23 °C tests and 30 °C, 380 mg/km for the 14 °C test and 550 mg/km for the 0 °C test, proving that cold start at lower ambient temperatures results in increased NOx emissions. Different combustion and abatement strategies could explain this different NOx emission rates occurring at different ambient temperatures. For all tests, the curves of the cumulative NOx emission flattened after circa 250 s (300 s for the 0 °C test) when the SCR catalyst started effectively reducing NOx emissions. A detailed presentation of regulated and non-regulated pollutant emission during the cold start as compared to the complete WLTC is given in the Appendix ( Figure A4).
Despite the natural variability of on-road testing, the two Traffic tests were performed with similar settings (same driver, same route, soak ambient temperature circa 18 °C) and emissions at cold start are compared for illustration purposes, as the first test took place in the morning at an ambient temperature that was circa 10 °C lower than the second test performed in the evening. While the emissions at the beginning of the routes were similar, at around 150 s they started to deviate. The different ambient temperatures sensed by the vehicle probably changed the EGR operation and Despite the natural variability of on-road testing, the two Traffic tests were performed with similar settings (same driver, same route, soak ambient temperature circa 18 • C) and emissions at cold start are compared for illustration purposes, as the first test took place in the morning at an ambient temperature that was circa 10 • C lower than the second test performed in the evening. While the emissions at the beginning of the routes were similar, at around 150 s they started to deviate. The different ambient temperatures sensed by the vehicle probably changed the EGR operation and consequently the emissions, as reported by others [23]. The results are in good agreement with the laboratory tests with similar behavior. The cold start contribution ranged between 35% and 39% of the total NO x , which is sensibly lower than for the laboratory tests without regeneration. This can be explained by the higher dynamicity in the Traffic tests as compared to the WLTC ( Figure A2) and the lower share of cold start with respect to the duration of the test (<10%), as the Traffic test lasted 53 min (Table 1). Figure 9 displays the cumulative PN emissions over the WLTCs (Figure 9a) and the two on-road Traffic routes (Figure 9b) for various ambient temperatures. Almost all of the particles are emitted in the first 100-150 s of the cold start, both in the laboratory tests and on the Traffic on-road tests. In particular, most of the PN is released with the ignition of the engine and in the first acceleration of the vehicle (first 50 s from ignition) [69]. As for the case of NO x , during cold start, PN emissions are higher in tests performed at lower ambient temperatures. The cold start spikes for DPF-equipped vehicles are well known. On the one hand, the engine out emissions are higher, and, on the other hand, the DPF has reduced efficiency due to defects in the mat employed to mount the brick in the canister [70]. As the DPF heats up, the defects close. The effect is more pronounced at lower temperatures which require a longer time, especially for the mounting material being in direct contact with the canister. The cold start contribution to the total PN is higher in the laboratory tests than on the road for the same reasons as for NO x (lower share of cold start with respect to the total duration of the Traffic tests, and higher dynamicity).
the vehicle (first 50 s from ignition) [69]. As for the case of NOx, during cold start, PN emissions are higher in tests performed at lower ambient temperatures. The cold start spikes for DPF-equipped vehicles are well known. On the one hand, the engine out emissions are higher, and, on the other hand, the DPF has reduced efficiency due to defects in the mat employed to mount the brick in the canister [70]. As the DPF heats up, the defects close. The effect is more pronounced at lower temperatures which require a longer time, especially for the mounting material being in direct contact with the canister. The cold start contribution to the total PN is higher in the laboratory tests than on the road for the same reasons as for NOx (lower share of cold start with respect to the total duration of the Traffic tests, and higher dynamicity).

Regeneration
One unique characteristic of this study is that the car was driven approximately 1300 km (900 km on the road and 400 km in the laboratory) in order to capture many regenerations and have a more complete picture of the emissions, i.e., including the regenerations. Figure 10 summarizes the regeneration frequency as identified in this study and values reported in the literature [38,40,[70][71][72][73][74][75][76][77]. For Euro 5 vehicles, the mean distance was around 495 km (±215 km), while for Euro 6, around 415 km (±155 km). In this study, the distance between regenerations varied from 60 km to 360 km, with a mean of 196 km (±110 km). The 60 km distance could be explained by the incomplete previous regeneration, which did not completely empty the DPF. This is the first study, according to our knowledge, in which such short distances have been reported. It is also

Regeneration
One unique characteristic of this study is that the car was driven approximately 1300 km (900 km on the road and 400 km in the laboratory) in order to capture many regenerations and have a more complete picture of the emissions, i.e., including the regenerations. Figure 10 summarizes the regeneration frequency as identified in this study and values reported in the literature [38,40,[70][71][72][73][74][75][76][77]. For Euro 5 vehicles, the mean distance was around 495 km (±215 km), while for Euro 6, around 415 km (±155 km). In this study, the distance between regenerations varied from 60 km to 360 km, with a mean of 196 km (±110 km). The 60 km distance could be explained by the incomplete previous regeneration, which did not completely empty the DPF. This is the first study, according to our knowledge, in which such short distances have been reported. It is also important to note that the scatter around the mean was high. Such high scatter has also been reported in one more study [75].

Regeneration Frequency
Active/forced regenerations are not only triggered when a specific distance (or time) has been accumulated since the last regeneration event, but other parameters such as engine back pressure, and/or estimated accumulated soot play an important role [78]. Thus, the driving style and the ambient temperature are important. Furthermore, sometimes regenerations are not triggered for cleaning the filter, but for resetting the system to a known state or desulfation of catalysts. Thus, the high scatter is understandable due to the big variety of driving conditions and/or different strategies of the different vehicle manufacturers. As there is no information regarding the DPFs, it is not possible to explain why the distance between regenerations has decreased. Nevertheless, one reason could be that DPFs have been increasingly replaced by SCR-coated DPFs, which have reduced passive regeneration due to ammonia dosing and the competition for NO 2 between soot and SCR reactions [79], and consequently, more active regenerations are needed.
Atmosphere 2020, 11, x FOR PEER REVIEW 14 of 25 important to note that the scatter around the mean was high. Such high scatter has also been reported in one more study [75]. Active/forced regenerations are not only triggered when a specific distance (or time) has been accumulated since the last regeneration event, but other parameters such as engine back pressure, and/or estimated accumulated soot play an important role [78]. Thus, the driving style and the ambient temperature are important. Furthermore, sometimes regenerations are not triggered for cleaning the filter, but for resetting the system to a known state or desulfation of catalysts. Thus, the high scatter is understandable due to the big variety of driving conditions and/or different strategies of the different vehicle manufacturers. As there is no information regarding the DPFs, it is not possible to explain why the distance between regenerations has decreased. Nevertheless, one reason could be that DPFs have been increasingly replaced by SCR-coated DPFs, which have reduced passive regeneration due to ammonia dosing and the competition for NO2 between soot and SCR reactions [79], and consequently, more active regenerations are needed.
There are a few implications of the frequent regenerations: 1. Do the emissions exceed the limits and, if yes, how much? 2. In which driving conditions do the regeneration events take place? 3. Do the weighted emissions, including regeneration events, still respect the limits?
Regarding the first point, it was shown that the laboratory limits can be exceeded in some cases. The PN emissions exceeded the limit by 300% at the regenerating WLTC cycle at 14 °C, but were only 40% of the emission limit at the regenerating WLTC cycle at 23 °C. The PN on-road tests were below the NTE limit even when regenerations took place (four cases). For NOx, the emissions were exceeded only at the regenerating WLTC cycle at 14 °C, by a factor of 2. At all other tests in the lab or on the road, the emissions were below the applicable limits. Since there are cases where the limits are exceeded, one question is where do the regeneration events take place (i.e., which driving conditions are compatible with a regeneration event). In the laboratory, both regeneration events took place at the high speed part of the cycle, which corresponds to a typical motorway drive. On the road, from the four regeneration events, two took place after the motorway part, entering the city or the rural section, while the other two took place at the high speed part of the routes. The rural/urban regenerations (1.5-3.5 km) emitted 0.4-1.5 × 10 13 particles, while the motorway (12-24 km), 1.3-2.5 × 10 13 particles. In the laboratory, the regenerations (5.8-8.8 km) emitted 0.2-5.0 × 10 13 particles. Such levels have also been reported by others [38,70,73]. DPF regenerations tend to occur at high speed, although in one out of the six registered events, the regenerations took place in the city while the There are a few implications of the frequent regenerations:

1.
Do the emissions exceed the limits and, if yes, how much? 2.
In which driving conditions do the regeneration events take place? 3.
Do the weighted emissions, including regeneration events, still respect the limits?
Regarding the first point, it was shown that the laboratory limits can be exceeded in some cases. The PN emissions exceeded the limit by 300% at the regenerating WLTC cycle at 14 • C, but were only 40% of the emission limit at the regenerating WLTC cycle at 23 • C. The PN on-road tests were below the NTE limit even when regenerations took place (four cases). For NO x , the emissions were exceeded only at the regenerating WLTC cycle at 14 • C, by a factor of 2. At all other tests in the lab or on the road, the emissions were below the applicable limits. Since there are cases where the limits are exceeded, one question is where do the regeneration events take place (i.e., which driving conditions are compatible with a regeneration event). In the laboratory, both regeneration events took place at the high speed part of the cycle, which corresponds to a typical motorway drive. On the road, from the four regeneration events, two took place after the motorway part, entering the city or the rural section, while the other two took place at the high speed part of the routes. The rural/urban regenerations (1.5-3.5 km) emitted 0.4-1.5 × 10 13 particles, while the motorway (12-24 km), 1.3-2.5 × 10 13 particles. In the laboratory, the regenerations (5.8-8.8 km) emitted 0.2-5.0 × 10 13 particles. Such levels have also been reported by others [38,70,73]. DPF regenerations tend to occur at high speed, although in one out of the six registered events, the regenerations took place in the city while the vehicle was being driven at slow speeds in stop and go traffic conditions, which is a relevant issue as human exposure may be large.
In order to assess the contribution of regeneration events to the emission levels, the emitted particles of the regenerating cycles were added to the sum of particles of the non-regenerating cycles ( Figure 11). The sum of particles was divided by the total driven distance. This procedure was done for all tests in the laboratory (Overall lab in Figure 11a), for all tests on the road (Overall RDE*) and for all tests both in the lab and on the road (Overall). The (weighted) PN emissions were below the applicable limits in the lab or on the road by a factor of 2 and 9, respectively. The inclusion of the sub-23 nm particles did not change the picture, as the emissions remained below the PN limit.
particles of the regenerating cycles were added to the sum of particles of the non-regenerating cycles ( Figure 11). The sum of particles was divided by the total driven distance. This procedure was done for all tests in the laboratory (Overall lab in Figure 11a), for all tests on the road (Overall RDE*) and for all tests both in the lab and on the road (Overall). The (weighted) PN emissions were below the applicable limits in the lab or on the road by a factor of 2 and 9, respectively. The inclusion of the sub-23 nm particles did not change the picture, as the emissions remained below the PN limit. The same procedure was followed for NOx ( Figure 11b). The (weighted) laboratory emissions were 50 mg/km, below the 80 mg/km limit. The on-road NOx emissions were 105 mg/km, also below the 168 mg/km (CF = 2.1) applicable limit for this car or the future limit of 114 mg/km (CF = 1.43).
It should be emphasized that the inclusion of the regeneration emissions is applicable for NOx in the lab (not for PN). Our tests showed that the same approach could be applied to PN as well. For RDE tests, a test with regeneration can be repeated, but if a regeneration occurs at the second test, then the limit has to be respected for both NOx and PN. Figure 12 plots the PN emissions during the WLTC at 14 °C with regeneration. In the first two minutes of the cycle there is a spike of particles, and the 10-23 nm particles are 28% higher than the 23 nm. The higher sub-23 nm fraction during the first two minutes has also been reported by others [41,80,81]. During regeneration (i.e., time after 1350 s) there is an increase in particles at the levels of cold start, but there is no difference between the >23 and >10 nm concentrations, indicating that they consist mainly of soot particles (mean diameter >50 nm) [82]. There is, however, a high difference at the end of the regeneration (time 1650 s). This has sometimes been seen by others as well [70,83]. This difference has been attributed many times to volatile re-nucleation downstream of the evaporation tube (volatile artefact) [84]. Recently, it was also suggested that these sub-23 nm particles are nonvolatile in nature as their concentration did not change with different dilutions or using catalytic stripper [85]. Elastomer connectors, which was not the case at our tests, can release non-volatile particles at elevated temperatures [86]. Another possibility is that these particles consist of compounds with boiling points around 350 °C, which is the temperature of the evaporation tube of the PN system. It is suspected that a source of such compounds is the transfer line between the vehicle and the dilution tunnel [87]. The contribution of volatile material is indirectly supported by the mass measurements. Even though we used one filter for the whole campaign to minimize volatile artifacts The same procedure was followed for NO x (Figure 11b). The (weighted) laboratory emissions were 50 mg/km, below the 80 mg/km limit. The on-road NO x emissions were 105 mg/km, also below the 168 mg/km (CF = 2.1) applicable limit for this car or the future limit of 114 mg/km (CF = 1.43).

PN Instrumentation
It should be emphasized that the inclusion of the regeneration emissions is applicable for NO x in the lab (not for PN). Our tests showed that the same approach could be applied to PN as well. For RDE tests, a test with regeneration can be repeated, but if a regeneration occurs at the second test, then the limit has to be respected for both NO x and PN. Figure 12 plots the PN emissions during the WLTC at 14 • C with regeneration. In the first two minutes of the cycle there is a spike of particles, and the 10-23 nm particles are 28% higher than the 23 nm. The higher sub-23 nm fraction during the first two minutes has also been reported by others [41,80,81]. During regeneration (i.e., time after 1350 s) there is an increase in particles at the levels of cold start, but there is no difference between the >23 and >10 nm concentrations, indicating that they consist mainly of soot particles (mean diameter >50 nm) [82]. There is, however, a high difference at the end of the regeneration (time 1650 s). This has sometimes been seen by others as well [70,83]. This difference has been attributed many times to volatile re-nucleation downstream of the evaporation tube (volatile artefact) [84]. Recently, it was also suggested that these sub-23 nm particles are non-volatile in nature as their concentration did not change with different dilutions or using catalytic stripper [85]. Elastomer connectors, which was not the case at our tests, can release non-volatile particles at elevated temperatures [86]. Another possibility is that these particles consist of compounds with boiling points around 350 • C, which is the temperature of the evaporation tube of the PN system. It is suspected that a source of such compounds is the transfer line between the vehicle and the dilution tunnel [87]. The contribution of volatile material is indirectly supported by the mass measurements. Even though we used one filter for the whole campaign to minimize volatile artifacts on the filter, the PN/PM ratio was 3.8-4.8 × 10 11 p/mg, lower than the typically expected 1-2 × 10 12 p/mg, indicating that the contribution of volatiles was significant. We suspect that the origin is released material from the transfer tube to the dilution tunnel during the high exhaust gas temperature regeneration events [88].

PN Instrumentation
Atmosphere 2020, 11, x FOR PEER REVIEW 16 of 25 on the filter, the PN/PM ratio was 3.8-4.8 × 10 11 p/mg, lower than the typically expected 1-2 × 10 12 p/mg, indicating that the contribution of volatiles was significant. We suspect that the origin is released material from the transfer tube to the dilution tunnel during the high exhaust gas temperature regeneration events [88].

Fuel Penalty
Regarding the fuel penalty due to regeneration, comparison of the WLTCs at 23 °C with and without regeneration gave a 4% difference. For comparison, the test at 0 °C had 11% higher CO2 than at 23 °C. For the on-road trips, the regeneration effect was <1%, which is much less than the effect of dynamic driving, which was between 6.7% and 9.1%.

Conclusions
In this study, we measured the emissions of a Euro 6d-temp vehicle driven more than 400 km in the laboratory and 900 km on the road, at ambient temperatures between 0 °C and 30 °C. During the testing, six DPF regenerations took place (two in the laboratory). The vehicle respected all applicable emission limits, both in the laboratory and on the road (WLTP 23 °C and RDE-compliant tests). The PN limit was exceeded only during a cold start test at 0 °C (+40%), and during a regeneration at 14 °C, by a factor of 4. The NOx limit of 80 mg/km was exceeded only during one of the two laboratory regeneration events by a factor of two, and during dynamic driving on the road (reaching up to 200 mg/km). The NO to NOx ratio was, in most cases, around 70% to 90%, but lower values (45% to 85%) were measured during the on-road tests. The difference was attributed to the high contribution of cold start in the laboratory tests, where the NOx emissions consisted mainly of NO. As compared to the complete WLTC, cold start emissions (first 300 s of the WLTC) increased by a factor of 10 for PN, and 7-8 for NOx and CH4. Over WLTC, practically, the majority (>85%) of NOx and PN were emitted during the first 300 s, whereas on the road, where the tests last longer, the contribution of the cold start was less relevant.

Fuel Penalty
Regarding the fuel penalty due to regeneration, comparison of the WLTCs at 23 • C with and without regeneration gave a 4% difference. For comparison, the test at 0 • C had 11% higher CO 2 than at 23 • C. For the on-road trips, the regeneration effect was <1%, which is much less than the effect of dynamic driving, which was between 6.7% and 9.1%.

Conclusions
In this study, we measured the emissions of a Euro 6d-temp vehicle driven more than 400 km in the laboratory and 900 km on the road, at ambient temperatures between 0 • C and 30 • C. During the testing, six DPF regenerations took place (two in the laboratory). The vehicle respected all applicable emission limits, both in the laboratory and on the road (WLTP 23 • C and RDE-compliant tests). The PN limit was exceeded only during a cold start test at 0 • C (+40%), and during a regeneration at 14 • C, by a factor of 4. The NO x limit of 80 mg/km was exceeded only during one of the two laboratory regeneration events by a factor of two, and during dynamic driving on the road (reaching up to 200 mg/km). The NO to NO x ratio was, in most cases, around 70% to 90%, but lower values (45% to 85%) were measured during the on-road tests. The difference was attributed to the high contribution of cold start in the laboratory tests, where the NO x emissions consisted mainly of NO. As compared to the complete WLTC, cold start emissions (first 300 s of the WLTC) increased by a factor of 10 for PN, and 7-8 for NO x and CH 4 . Over WLTC, practically, the majority (>85%) of NO x and PN were emitted during the first 300 s, whereas on the road, where the tests last longer, the contribution of the cold start was less relevant.
Regarding non regulated pollutants, ammonia was always at the instrument (FTIR) background levels (<1 ppm), except during regenerations in which up to 8 ppm were measured (1 mg/km). N 2 O was around 11 mg/km, and reached 13.5 mg/km during a cycle with regeneration. CH 4 was around 6 mg/km. N 2 O and CH 4 accounted for 1% of the CO 2 -equivalent emissions of this vehicle. Isocyanic acid (HNCO) emissions reached 3.5 mg/km in the 0 • C test. The rest of the pollutants, such as formaldehyde and acetaldehyde, were at background levels, even during regeneration events. The fraction of particles below 23 nm, the lowest size limit defined in the current regulation, was 11-51%, with the highest percentage at the lowest absolute levels.
One important finding of study was the high frequency of regeneration events: on average, it was every 196 km. This frequency is much lower compared to the mean of Euro 5 vehicles (495 km) and Euro 6 vehicles (415 km) found searching the literature. Such frequent regenerations need to be considered when emissions from diesel vehicles are reported. Nevertheless, the emissions, weighted also for the regenerations, were below the applicable limits; the absolute levels were two times (for NO x ) to four times (for PN) lower than the type approval values. In one case, the regeneration took place in the urban part, which is relevant as human exposure is increased due to population density. Our tests showed that the methodology to include the regeneration emissions in the weighted result, applicable to NO x , could also be applied to PN >23 nm. An interesting finding, though, was that during one regeneration event, the sub-23 nm fraction of solid particles increased (23%). These seem to be artefact particles, probably due to the high exhaust gas temperature of the regeneration event. Thus, more research will be needed in that direction when the future >10 nm regulation is in place. The reported emissions correspond to a vehicle with low mileage (3000 km at the start of the campaign), and further investigation needs to confirm that similar environmental performance is maintained during the vehicle's lifetime.  Figure A1 presents the vehicle speed and altitude profiles of the routes used in the test campaign. For the routes that were driven more than once, a test-to-test variability in the speed profile can expected due to the stochastic nature of the on-road testing. The tests driven dynamically over routes RDE-1 and RDE-2 share the altitude profile of those routes.
The driving dynamics of the on-road tests, expressed as velocity times positive acceleration per phase (Urban/Rural/Motorway) and for the WLTC phases (Low/Medium/High/Extra high) are plotted in Figure A2. Figure A3 presents a comparison of emissions as measured in the bag and with the FTIR for the six laboratory tests presented in Section 2.2, and three additional preconditioning cycles performed prior to the 23 • C tests.
The results are presented for the complete WLTC tests as well as for the WLTC phases (Low, Medium, High, and Extra-high) for CO 2 , NO x and CH 4 . The correlation between the two measurement methods is good over the measured emission ranges, with R 2 being 0.90 to 0.99. The FTIR accuracy for CO 2 and CH 4 is good as compared to the bag measurement, whereas a 20% underestimation of NO x is observed.
Atmosphere 2020, 11, x FOR PEER REVIEW 18 of 25 FTIR accuracy for CO2 and CH4 is good as compared to the bag measurement, whereas a 20% underestimation of NOx is observed.   Figure A4 presents an overview of cold start emissions of a selection of air pollutants and greenhouse gases as measured with the 23 nm particle counter (PN) and FTIR (gases) in the laboratory at different ambient temperatures. On the WLTC, 300 s corresponds to a distance of 2 km driven at an average speed of 24.3 km/h, including a stop of 30 s. The WLTC Low phase, which characterizes urban driving, lasts 589 s over a distance of 3.1 km and an average speed of 18.9 km/h, including stops. In general, distance-specific emissions are higher at cold start and WLTC Low than they are over the complete WLTC on all tested ambient temperatures. For example, cold start PN distance-specific emissions are circa ten times higher than they are over the complete cycle on all tested temperatures. Cold start NO x is 7-8 times higher than it is over the complete WLTC. The highest PN and NO x emissions occur at the lowest ambient temperature (0 • C) reaching 9 × 10 12 p/km and 487 mg/km, respectively, at cold start. NH 3 emissions are equivalent at cold start, WLTC Low, and complete WLTC, and emissions are higher at 30 • C than at other ambient temperatures. The distance-specific emissions of CO 2 and N 2 O are 30% higher at cold start than over the complete WLTC at 23 • C, whereas the increase for CH 4 is more significant (8.5 times higher).  Figure A4 presents an overview of cold start emissions of a selection of air pollutants and greenhouse gases as measured with the 23 nm particle counter (PN) and FTIR (gases) in the laboratory at different ambient temperatures. On the WLTC, 300 s corresponds to a distance of 2 km driven at an average speed of 24.3 km/h, including a stop of 30 s. The WLTC Low phase, which characterizes urban driving, lasts 589 s over a distance of 3.1 km and an average speed of 18.9 km/h, including stops. In general, distance-specific emissions are higher at cold start and WLTC Low than they are over the complete WLTC on all tested ambient temperatures. For example, cold start PN distance-specific emissions are circa ten times higher than they are over the complete cycle on all tested temperatures. Cold start NOx is 7-8 times higher than it is over the complete WLTC. The highest PN and NOx emissions occur at the lowest ambient temperature (0 °C) reaching 9 × 10 12 p/km and 487 mg/km, respectively, at cold start. NH3 emissions are equivalent at cold start, WLTC Low, Figure A2. Overview of the driving dynamics, in terms of the 95th percentile of v × a for the laboratory and RDE tests. There are three markers per RDE test corresponding to the respective Urban/Rural/Motorway phase. Note that the Traffic tests are Urban-only tests. Tests labelled with an R at the end correspond to those where a DPF regeneration occurred. The red dotted line indicates the 95th percentile of the v × a limit, as defined for M1 vehicles in Appendix 7 of the RDE regulation. The WLTC laboratory cycle is divided into its respective phases. RDE = Real-Driving Emissions; WLTC = Worldwide harmonized Light vehicles Test Cycle.   Figure A4 presents an overview of cold start emissions of a selection of air pollutants and greenhouse gases as measured with the 23 nm particle counter (PN) and FTIR (gases) in the laboratory at different ambient temperatures. On the WLTC, 300 s corresponds to a distance of 2 km driven at an average speed of 24.3 km/h, including a stop of 30 s. The WLTC Low phase, which characterizes urban driving, lasts 589 s over a distance of 3.1 km and an average speed of 18.9 km/h, including stops. In general, distance-specific emissions are higher at cold start and WLTC Low than they are over the complete WLTC on all tested ambient temperatures. For example, cold start PN distance-specific emissions are circa ten times higher than they are over the complete cycle on all tested temperatures. Cold start NOx is 7-8 times higher than it is over the complete WLTC. The highest PN and NOx emissions occur at the lowest ambient temperature (0 °C) reaching 9 × 10 12 p/km and 487 mg/km, respectively, at cold start. NH3 emissions are equivalent at cold start, WLTC Low, Figure A3. Scatterplot of Bag vs. FTIR distance-specific emissions for CO 2 , NO x and CH 4 for the complete laboratory WLTC tests and their respective sub-phases. Number of tests considered: 9. The brown dotted line represents the fitted line to the scatter data of each pollutant, considering all cycles and sub-phases.
Atmosphere 2020, 11, x FOR PEER REVIEW 20 of 25 and complete WLTC, and emissions are higher at 30 °C than at other ambient temperatures. The distance-specific emissions of CO2 and N2O are 30% higher at cold start than over the complete WLTC at 23 °C, whereas the increase for CH4 is more significant (8.5 times higher).