Evaluation of On-Board Sensor-Based NOx Emissions from the Heavy-Duty Vehicles in an Inspection and Maintenance Program

Heavy-duty diesel vehicles (HDDVs) are important sources of urban nitrogen oxides (NOx) in an actual application for environmental compliance. The remote on-board sensing (OBS) is a cost-effective emission reduction approach for HDDVs. NOx emissions from fifty-four sediment hauling trucks were evaluated in a construction area, and this information was utilized to make changes to the hauling trucks being used on the project. Approximately 1/3 of the trucks had emissions that were comparable to or below 0.2 g/bhp-hr level. The fleet average NOx emissions were 0.38 g/bhp-hr for non-credit engines and 0.77 g/bhp-hr for credit engines, with 2013 and newer credit engines averaging 0.29 g/bhp-hr. The overall NOx conversion efficiencies of the selective catalytic reduction (SCR) system on many vehicles were in 80% to 90%, and SCR efficiencies for some high emitter vehicles were down to approximately 60%. This is due to high engine operation fractions at the lowest SCR inlet temperature zone (> 200℃). This study also found that the NOx emissions for non-2010–2012 Family Emission Limit (FEL) trucks were well below the California Clean Idle certification of 30 g/hour. Recommendation was provided to efficiently maintain the emission levels in the reservoir.


Introduction
On-road heavy-duty diesel vehicles are a major source of NOx emissions in the United State (U.S.). Although heavyduty vehicles (HDDVs) represent only 2% of the total population of vehicles on the road, they presented an estimated 30% of the total NOx emission inventory in California in 2020 [1]. This is considerably higher than the U.S. average, where trucks are estimated to represent 15% of the emission inventory [1,2]. To reduce NOx emissions from HDDVs, the U.S. Environmental Protection Agency (EPA) and California Air Recourses Board (CARB) adopted new on-highway heavy-duty diesel engine certification standards that required model year (MY) 2010 and newer heavy-duty diesel engines to meet a NOx emission standard of 0.20 g/bhp-hr or 0.35 g/ bhp-hr for certain credit engines over the transient Federal Test Procedure (FTP) and Supplemental Emissions Test (SET) on an engine dynamometer [3]. These engines are also subject to in-use testing standards of 0.45 g/bhp-hr, or 0.52 g/bhp-hr for credit engines, for at least 90% of operation under not-to-exceed (NTE) testing conditions where engine speed and load conditions are above ~ 30%, among other conditions, for 30 consecutive seconds or longer.
To meet the NOx emission standard, heavy-duty diesel engines in the U.S. market have utilized selective catalytic reduction (SCR) together with exhaust gas recirculation (EGR) and other in-cylinder NOx control methods [4][5][6]. For the SCR system, NOx is converted into nitrogen and water by reaction with ammonia over a special catalyst. Although SCR provides the potential to reduce NOx emissions by more than 90% reduction of NOx from HDDVs still represents one of the most important challenges to improving urban air quality [7]. Furthermore, NOx emissions from on-road HDDVs can differ significantly from in the laboratory certification levels. Conditions, such as urban driving, stop-and-go traffic, excessive idling and low load/ low speed operation, can all lead to reduced SCR conversion efficiencies and increased tailpipe NOx [8][9][10][11]. This implies that controlling NOx emissions from HDDVs remains a significant challenge, especially during real-world driving conditions.
To better monitor in-use emissions from HDDVs, Portable Emission Measurement Systems (PEMS) have been developed for both in-use compliance and research purposes [5,[8][9][10][11][12][13][14][15][16]. Several studies have shown that NOx emissions measured from the HDDVs with PEMS under in-use conditions are generally higher compared to the U.S. laboratory certification standards and can also exceed U.S. in-use testing limits [8][9][10][11]. NOx emissions for diesel goods movement and delivery vehicles equipped with engines certified to the 0.2 g/bhp-hr standard were on the order of 1 g/bhp-hr for normal in-use operation [11]. Heavy-Duty In-Use Testing (HDIUT) averages have shown average emissions of 0.42 g/ bhp-hr [17]. Although European standards are generally higher than those in the U.S., with the Euro VI standards for heavy-duty diesel engines being 0.46 g/kWh (0.34 g/ bhp) with a 1.5 conformity factor for in-use testing, some studies have shown lower emissions for European HDDVs compared to U.S. HDDVs NOx [18]. In a study of five Euro VI diesel trucks, Grigoratos et al. found most of these trucks to be below the certification/in-use limits, with the exception of one truck that was slightly above the conformity limits and under low-speed conditions [9]. Although PEMS trends have been used to identify conditions for higher off-cycle NOx emissions, these instruments are expensive and their setup and operation is time-consuming. So, the number of vehicles that can be tested with PEMS is still small.
On-board sensing (OBS) is another method for monitoring vehicle emissions. This method can utilize NOx data from sensors that are already equipped on HDDVs, in particular those that are equipped with SCR NOx control [5,[13][14][15][16]. NOx sensors are usually installed upstream and downstream of the SCR to measure NOx conversion efficiency to nitrogen gas. This method not only provides NOx data in a labor-and cost-effective manner but also can be used to estimate instantaneous NOx emissions over a wide range of real-world operation. Given the potential of monitoring NOx sensors to ensure lower NOx emissions can be maintained in-use, CARB has moved forward with implementing the Real Emissions Assessment Logging (REAL) which is designed to track NOx, greenhouse gas, and smogrelated emissions. This new program requires recording emissions data in diesel trucks starting in 2022 [19,20]. However, few studies to date have been conducted where OBS has been used to monitor NOx emissions of larger numbers HDDVs; however, a previous study estimated NOx emissions using OBS readings from 72 HDDVs and found that there were large differences between in-use and certification NOx emissions, with 12 HDDVs emitting more than 3 times the standard in real-world operation [13]. In another study [5], SCR temperature and engine load profiles were created from monitoring 92 HDDVs for a variety of vocations by using OBS readings. It was found that the vehicles in this study operated with SCR temperatures lower than 200 °C for 11-70% of the time depending on their vocation type. The data available from the engine control module (ECM) can also provide information about the condition of the vehicle from a maintenance perspective and identity issues that need repair. Jiang et al. [15] monitored the NOx emissions from 45 HDDVs on a chassis dynamometer before and after repair for OBD identified diagnostic trouble codes (DTCs) using OBS readings. The results showed large reductions of NOx emissions compared to pre-repair levels.
While monitoring of sensor-based NOx emissions offers the potential for enhanced compliance for in-use NOx emissions, there are few studies [10,13,21] where the methodology has been used in an actual application for environmental compliance. The objective of this study was to characterize NOx emissions of sediment hauling trucks in a construction area located in the densely populated Los Angeles basin and more specifically at the Devil's Gate (DG) Reservoir near La Canada, CA, in the South Coast Air Basin (SCAB). This work addressed concerns from local communities as to the possible impact of the increased truck traffic related to the sediment removal project on local air quality. Field monitoring was conducted on 56 trucks to monitor their OBS system for NOx and to data log the truck activity and emissions over a typical route between the sediment removal site and the sediment placement site. The results of NOx emissions will inform policymakers on the importance of real-world OBS system monitoring for NOx emissions from heavy-duty diesel vehicles when operated in urban settings.

Data Collection
The data collection for this project focused on monitoring in-use NOx emissions using the OBS that is equipped on the vehicle itself. The data loggers utilized for this study were J1939 Mini (HEM Data Corporation) data loggers capable of collecting a full range of information from the ECM. The data loggers are configured to collect upwards of 200 ECM parameters at a frequency of 1 Hz. A subset of the type of data that are collected is provided in Table S1. The data loggers communicate with the engine's ECM/ OBD through industry standard communication protocols. The data loggers are also equipped to collect Global Positioning System (GPS) data on a second-by-second basis. GPS is capable of measuring the truck's location (latitude 1 3 and longitude) and altitude, from which speed can also be derived. The HEM data loggers are a small unit that can be attached quickly to the truck's J1939 CAN (Controller Area Network) port in the cab on the driver's side (see Figure S1). The HEM data loggers are self-triggering to start automatically when a test truck engine is started and to stop automatically when the truck engine is stopped and can store data for up to 6 months.

Test Trucks
A total of 56 Class-8 trucks were monitored as part of this field monitoring study effort. Information about each test truck is shown in Table S2, including the manufacturer and model year of the truck's engine (27 2010-2012 MY and 13 2013 + engines), and its odometer reading at the time the data logger was installed. Information on the engines applicable emissions standards including whether the engine family was certified using credits [22], and its latest PSIP (Periodic Smoke Inspection Program) results, was also obtained.

Test Routes
NOx monitoring was conducted over several sessions in August and October of 2020. Each truck was data logged for one or two trips from the DG reservoir site to the sediment placement site and then back to the DG site. Initially in August, trucks were monitored for a single trip, but it was determined that approximately 200-800 s of NOx data was lost during the single trips because engine and exhaust cooled off during the data logger installation, and NOx sensors do not provide readings when the exhaust temperature is below approximately 200 °C to prevent humidity damage to the ceramic sensing element [23]. For the subsequent sessions in October, the data loggers were left on the trucks for two full trips, such that NOx sensor data could be obtained for the full second trip, as the engine was fully warm for the second trip.
Two different locations were used for sediment disposal in August and October, as shown in Fig. 1. In August, the sediment was placed at the Sheldon sediment placement site (SPS) in Sun Valley, CA, to the west of the DG site and in October the sediment was placed at Manning Pit SPS in Irwindale, CA, to the east of the DG site. The test route for the Sheldon SPS is provided in Fig. 1a. This test route was on average about 38 miles in length and took on average 1 h 19 min for the full round trip. The test route for the Manning Pit SPS is provided in Fig. 1b. This test route was on average about 41 miles in length and took on average 1 h 31 min for the full round trip.

OBD Results
In addition to the NOx monitoring, an OBD scan was performed on each of the trucks as part of the testing protocol to identify any active DTCs being stored in the OBD system. The OBD data collection was done with an OBD scanner provided by an outside contractor and a HEM data logger. For the OBD scanning, the HEM data logger was configured with a configuration file specifically designed to request OBD information related to fault codes. Of the 56 trucks scanned as part of this project, 6 trucks were found to have active emission-related DTCs that triggered the MIL light on, and one vehicle was found to have an active emission-related DTCs that had not illuminated the MIL. These vehicles were all identified during the initial August testing. A summary of the DTCs found for each truck is provided in Table 1. The DTC codes included those related to an SCR conversion fault for two trucks, an oxygen sensor for three trucks, a NOx sensor for two trucks, an EGR problem for one truck, and diesel emission fluid (DEF) tank quality for one truck. Note that this includes some trucks with multiple fault codes. Unfortunately, NOx data was only available for one of the trucks with an active emission-related DTC and the MIL on and one truck with an active emission-related DTC and the MIL off. It should be noted that the trucks found to have an active DTC were not allowed to return to the site on subsequent days until the issue causing the DTC was fixed.

Data Analysis
The real-time NOx data from the NOx sensor was processed along with the actual engine and nominal friction percent torque (%), reference engine torque (Nm), and engine revolutions per minute (rpm) data to provide NOx emissions on a g/bhp-hr basis. For these calculations, the exhaust flow rate was obtained from the broadcast fuel flow rate and intake air mass flow rate, which was in turn used in conjunction with the NOx concentrations to calculate NOx emissions. Additional analyses were also conducted to evaluate NOx emissions during operation specifically in the DG reservoir, with a particular emphasis on idle emissions. It should be noted that NOx data was available for only 43 of the 56 trucks monitored, since the broadcasting of NOx data was not a full requirement for all of the model year engines that were tested. It should also be noted that for periods where the NOx data was not obtained due to the exhaust temperature being below 200 °C, as discussed in Sect. 2.3, the corresponding miles traveled and bhp-hr work were also excluded in determining the g/bhp-hr and g/mi emission rates.

NOx Emissions over the Test Route
An important element in understanding the potential impact of the NOx emissions for the test trucks is to understand the distribution of NOx emissions over the test route. Figure 2 shows the NOx emissions for a typical truck over a full circuit of the test route to the Sheldon sediment placement site (SPS) in August 2020 (see Fig. 2a) and the Manning SPS in October 2020 (see Fig. 2b). Note that the figure for the Manning Pit SPS represents two round trips from DG reservoir to the sediment placement site and back. Profiles of NOx emissions over the route from the GPS data are provided in Figure S2a for the Sheldon SPS and in Figure S2b for the Manning SPS. Additional information showing comparisons between NOx emissions over the two trips for the same vehicle are shown in Fig. 3a for a lower emitter and in Fig. 3b for a high emitter. Figure 2(a) shows NOx emissions in gram basis over the monitoring time, along with the SCR inlet temperature, with arrows marking the freeway sections. For the trip to the Sheldon SPS, the data show that the highest NOx emissions were found for the first part of the freeway trip to the SPS and on the trip back from the SPS to DG reservoir. Overall, these sections represent high power operation on uphill parts of the freeway for both the outbound and return routes. In contrast, NOx emissions appear to be relatively modest over the remaining portions of the trip route. It should be noted that the freeway segment on the return trip back from the Sheldon SPS to the DG reservoir shows lower NOx emissions, which could be due to the fact that the truck is unloaded at this point, after placing the sediment at the disposal site. It has been previously reported that NOx emissions tend to decrease with decreasing vehicle weight [24]. As can be seen in the figures, the NOx sensor does not begin to reach its activation temperature and turn on to provide readings until after approximately the first 500-600 s of operation after installation. This can be attributed to the fact that the temperature of the exhaust was too cold to activate the NOx sensor since the trucks were shut off for the installation of the data loggers.
For the trip to the Manning Pit SPS, as shown in Fig. 2b, the highest NOx emissions are then along the initial segment of the freeway, which could be due in part of the aftertreatment still heating up and the fact that the truck is fully loaded on the outgoing trip to the SPS. Focusing on the second-round trip in Fig. 2b, the data show some NOx peaks while the truck is in DG reservoir, from about 4200 to 5200 s, and on the road to the freeway. The truck does show some NOx peaks in the surface street area around and in the SPS. Similar to the results presented above, the truck shows very low emissions for the freeway driving on the return trip from the placement site back to DG reservoir, which could be because it is unloaded at that point.

NOx Emissions
The distribution of average NOx emissions for individual trucks on a g/bhp-hr basis over the test route is provided in Fig. 3. Figure 3 shows the vehicles from the lowest to highest emissions from left to right on the graph, with the sorting based on the initial August test run data. A total of 43 vehicles provided valid NOx sensor data that was used for this distribution. The vehicles are identified using a letter to designate the manufacturer (D = Detroit Diesel, P = Paccar, C = Cummins, and N = Navistar), a number to identify the engine model year, and an "F" to indicate engines that were certified using banked emission credits. Note that the engines certified using banked emission credits can be certified at higher emissions levels that are offset by credits generated from the production of engines that have been certified to lower emission levels, such that the aggregate total emissions of all combined engine family certifications remain below the primary certification standard [25]. A separate bar is also included for vehicles that were monitored for a second test run in October, which are denoted by "_Two" at the end of the name.
The results on a g/bhp-hr can be compared to the certification standard of 0.2 g/bhp-hr or 0.35 g/bhp-hr for credit engines or the NTE testing limits for in-use testing of 0.45 g/ bhp-hr or 0.52 g/bhp-hr for credit engines. It should be noted that the certification standards and the NTE limits are designed for laboratory testing on an engine dynamometer over the FTP cycle and NTE conditions where engine speed and load are above ~ 30%, among other conditions, for 30 consecutive seconds or longer, respectively, so the numbers are not necessarily directly comparable. But this still provides a rough estimate of the types of emissions levels that might be expected based on the certification standards and in-use compliance limits.
On a g/bhp-hr basis, NOx emissions varied from under 0.2 g/bhp-hr to 1.73 g/bhp-hr. The fleet average emissions were 0.45 g/bhp-hr. For these averages, the values for the second test run for the October tests on a given truck were utilized where data was available for the full test route, as the NOx sensor was fully warm and NOx sensor data was available for a full trip. The fleet average value is very comparable to values obtained from analysis of HDIUT from 160 HDDVs, which showed full trip emission rates of 0.42 g/ bhp-hr [18]. The fleet average values are lower, however, than those found by McCaffery et al. for a fleet of heavyduty diesel goods movement vehicles tested with PEMS, whose average in-use NOx emissions were closer to 1 g/ bhp-hr [11]. The higher emissions for the goods movement vehicles in the McCaffery et al. study could be attributed to the fact that they have a greater number of stops along their route, with significant periods of idling [11]. Tan et al. also found significantly higher NOx emission rates for HDDVs from a wide range of vocations, with the exception of instate and out-of-state line haul trucks, food distribution, and utility repair, which average NOx emissions were less than 0.42 g/bhp-hr [13]. Approximately 1/3 of the vehicles had emissions that were comparable to or below the 0.2 g/bhp-hr NOx level. Approximately 1/3 of the vehicles had emissions that were comparable to or below the NTE limits of 0.45 to 0.52 of NOx g/bhp-hr. The remaining approximately 1/3 of the vehicles had average emissions higher than the in-use NTE limits.
For 2010 and newer engines, it is useful to further differentiate emissions between engines that are certified to the primary 0.2 g/bhp-hr NOx standard and those that are certified on a credit basis to a 0.35 g/bhp-hr NOx standard. It should be noted that for the truck fleet, some of the higher emitting vehicles were equipped with credit engines. For the truck fleet, the fleet average NOx was 0.38 g/bhp-hr for non-credit engines and 0.77 g/bhp-hr for credit engines. It is also worth noting that the 2013 and newer credit engines performed considerably better than the 2010-2012 credit engines, averaging only 0.29 NOx g/bhp-hr. It should be noted that these values are comparable to those seen in other studies [9,15]. NOx emissions from HDIUT program have been found to average 0.37 g/bhp-hr for non-credit engines and 0.70 g/bhp-hr for "credit" engines [26]. Based on these results, the sediment haul trucks for the project were revised to exclude 2010-2012 credit engines from the final season of work at the DG reservoir. Figure 4 shows NOx emissions on a g/mile basis for each vehicle vocation. NOx emissions varied from under 0.3 g/mile to 6.2 g/mile. The fleet average emissions were 1.81 g/mile, and the average NOx emissions were 1.53 g/mile for non-credit engines and 2.97 g/mile for credit engines. It is also worth noting that the 2013 and newer credit engines performed considerably better than the 2010-2012 credit engines, averaging only 1.13 NOx g/mile. This average value is very comparable to values obtained from analysis of HDIUT program, which showed NOx emissions from HDIUT program have been found to average 1.11 g/bhp-hr for non-credit engines and 2.42 g/ bhp-hr for "credit" engines [26].
To better understand NOx emissions as a function of different driving conditions, the NOx emissions for a higher emitting truck, with an average NOx emission rate of 0.82 g/ bhp-hr, tested in August and October, are plotted in Fig. 5 as function of engine load and vehicle speed bins. For comparison purposes, these speed and load bins correspond with the bins being used for the CARB Real Emissions Assessment Logging (REAL) specifications [19,20], as described in Table 2. The results in Fig. 5(b) show that the largest fraction of NOx occurred under high load conditions (64.2%), bins 11-14, even though only 26.6% of the operation was under these conditions (see Fig. 5a). The other load/idle conditions all represented much smaller fractions of the total NOx. Idle emissions under low, medium, and high loads under bin 2 represented 15.6% of operation, but only 2.2%  of the total NOx. Low load operation, bins 3-6, represented 38.3% of operation, but only 20.3% of total NOx. Medium load operations, bins 7-10, represented 19.6% of operation and 15.0% of total NOx. On a g/bhp-hr basis, NOx emissions were higher under low load conditions, as the work as a function of time is much less than under the medium or high load conditions. Under medium and high load conditions, the emissions were below 0.5 on a g/bhp-hr basis for all of the bins. NOx emission profiles for August and October testing were relatively similar over the range of different load and speed bins. The results are consistent with those of McCaffery et al., who showed higher NOx fractions for high load conditions, compared to the low and medium speed bins [11]. Tan et al., on the other hand, showed that the NOx emission rates for more vocational trucks were highest in the lowest load category and also represented the highest fraction of activity [13].
It is also useful to evaluate trucks with active DTCs to see if they have excessively high NOx emissions. Among those trucks with active DTCs and the MIL on, as described in Sect. 2.4 and in Table 1, NOx data was only available for one of the trucks. This truck (D10) showed NOx emissions of 0.56 g/bhp-hr. Another vehicle (D15) with an active emissions-related DTC without the MIL that had NOx data had an emission rate of 0.35 g/bhp-hr. It is interesting to note that emissions for some of the vehicles that showed active DTC were not significantly different from the average NOx emissions levels. This is not necessarily surprising, however, as the OBD system is designed to identify issues with various emissions components as they are starting to deteriorate to a point where they could impact emissions. As such, the OBD system serves not only to identify vehicles that already have higher emissions but also vehicles that require maintenance to prevent them from become higher emitting vehicles.
To better understand the potential reasons for the higher NOx emissions for some vehicles, additional comparisons were made comparing the engine out and tailpipe emissions for the subset of vehicles where engine out NOx emissions were available. Figure 6 shows the SCR efficiency values based on the ratio of the full trip tailpipe emissions to the full trip engine out emissions. The results show SCR efficiencies in the 80% to 90% + range for the more typical emitters, which had emissions rates ranging from 0.14 to 0.42 g/ bhp-hr. For some, but not necessarily all, of the higher emitters, which had emissions rates ranging from 0.77 to 1.73 g/ bhp-hr, the SCR efficiencies were lower, however, ranging down to approximately 60% in some cases. Several factors can affect SCR performance, such as SCR operating temperature, residence time, diesel exhaust fluid (DEF) dosing quantity, engine-out NOx concentrations, and catalyst reactivity [13]. Tan et al. found SCR efficiencies of 59.0 ± 18.6% for SCR temperatures < 200℃, of 73.7 ± 14.3% for SCR temperatures 200℃ < x < 250 °C, and of 81.1 ± 14.0% for SCR temperatures above 250 °C [13]. Figure 7 shows average NOx emission rates versus MY of the trucks along with the emission rates from California's EMission FACtors (EMFAC) mobile source emission inventory (EMFAC2021). For this comparison, the EMFAC NOx emission rates were based an average speed of 30 mph, the average speed of the trucks operating at the Devil's Gate reservoir, for class tractors. The EMFAC emission rates for each MY were obtained for the statewide 2020 calendar year from the CARB EMFAC emission inventory [27]. The EMFAC2021 emission factors for MY 2010-2015 SCR diesel vehicles in EMFAC ranged from 1.84 to 7.98 g/mile for diesel Class 8 tractors. The DG emission rate for MY2010 to MY2015 SCR diesel vehicles, on the other hand, ranged from 0.92 to 2.87 g/ mile and was significantly lower than that predicted by

SCR Temperature Results
The average SCR temperatures for the trucks shown in Fig. 3 suggest that on average the SCR temperature is in a range where the SCR should be operating relatively effectively, with average temperatures of approximately 250 °C. Interestingly, the emissions for the credit engines showed generally lower SCR temperatures, with average SCR temperatures of only 230 °C. As shown in Fig. 3, the SCR temperature also varies throughout the trip. The SCR temperature is the lowest immediately after the installation of the data loggers in the DG reservoir after the truck engine has been shut off for the installation of the data logger. The temperature also shows a drop at the very end of the trip, which is when the truck reenters the DG reservoir and operates at a low load up until the time that the data logger is removed, and at the SPS after the truck exits the freeway. In general, the SCR temperature stays steadily above 250 °C during the higher load operation on the freeways. Figure 6 also shows average engine operation time of the trucks in three different SCR inlet temperature zones in comparison with the SCR efficiency. The SCR temperature zone (< 250℃) had the highest fraction of engine operation time (42.9 ± 8.2%), with good NOx conversion efficiencies. The results show that the lowest SCR inlet temperature zone (> 200℃) had the second highest engine operation fraction (40.1 ± 9.0%) and the lowest NOx conversion efficiencies. In agreement with previous studies, NOx conversion efficiencies were significantly affected by the catalyst reactivity, showing the low conversion efficiency for the SCR when it is below its light-off temperature (< 200℃) [11,13,29]. McCaffery et al. [11] found a much wider range in terms of the fraction of time the SCR was operating < 200℃. McCaffery et al. [11] found a subset of goods movement  vehicles with 20 to 40% of the operating time < 200℃, while another group of goods movement vehicles showed between 50 and 70% < 200℃, with the latter group showing much higher emission rates; Tan et al. [13], on the hand, found that operation with the SCR temperature < 200℃ represented only 11.7 ± 9.5% for a fleet including a wide range of HDDVs from different vocations.

Analysis of Idle and Low Speed Operational NOx Emission in the Reservoir
Additional analyses were also conducted to evaluate NOx emissions during operation specifically in the DG reservoir.
A particular emphasis of this analysis was on idle emissions, where idle was defined as when truck speed was determined to be zero via either GPS or the vehicle speed information from the ECM, with identified idle periods in all cases being at least 30 s or longer. The data for this section was separated into two groups representing the two different monitoring campaigns. For the vehicles monitored in October 2020, NOx emission data was available for the full operation within the reservoir site. For the vehicles monitored in August 2020, NOx emission data was available for only the operation within the reservoir site up to the point where the vehicles exited the weight scales and the data loggers were removed. Hence, the data for these trucks as they exited the reservoir was not available. Typical paths through the DG reservoir during the October and August campaigns are presented in Figure S3a and Figure S3b, respectively. A total of 13 vehicles were characterized during the October campaign, and a total of 33 vehicles were characterized during the August campaign. Summary average results for the data logging in the reservoir are presented in Table 3, including separate results for the October campaign trucks, the August campaign trucks, and for the averages for all trucks. These averages represent a single trip in the reservoir. This includes the average time, miles traveled, speed, and idle time in the reservoir, as well as the amount of grams of NOx emitted while the trucks were in the reservoir, and the portion of NOx emitted during idle periods and while the truck was moving in the reservoir.
For the October data, which includes all operation within the reservoir, it was found that the trucks on average spent approximately 14.5 min in the reservoir, traveling 1.4 miles at an average speed of 5.8 miles per hour (mph), and with an idle time of 5.0 min. An average total of 12.7 and 8.0 g of NOx were emitted in the reservoir for all October trucks and for all October trucks excluding the 2010-2012 FELs, respectively. For the non-2010-2012 FEL trucks, this included 0.4 g of NOx during idling and 7.6 g of NOx while the truck was moving. For all trucks, this included 1.8 g of NOx during idling and 10.9 g of NOx while the truck was moving. The idle emissions rates were found to be 19.5 g/ hour for all trucks and 4.6 g/hour for the non-2010-2012 FEL trucks. These idle emission rates are below to well below the 30 g/hr levels utilized in EMFAC2021 [30] which are based on engine certification limits for meeting the California Certified Clean Idle rule.
The averages for the trucks monitored in August, which only included operation up to the scales, showed a slightly shorter average time in the reservoir (13.3 min) and less miles traveled (0.6 miles), but a similar average speed (4.9 mph) and actually a higher average idle time (6.4 min). These differences can be attributed, in part, to the fact that the August truck monitoring data did not include operation past the scales, but it also reflects differences in the path in the reservoir itself, as debris was being removed from different locations inside of the reservoir and the specific routes that the trucks followed were adjusted accordingly. The total emissions in the reservoir for the August trucks were considerably lower than those for the October trucks, at 2.6 and 1.7 g of NOx, respectively, for all trucks and for the non-2010-2012 FEL trucks. This was mostly attributed to NOx emissions while the trucks were moving (1.9 and 1.4 g of NOx for all and non-2010-2012 FEL trucks), with lower levels of emissions during idle (0.7 and 0.3 g of NOx for all and non-2010-2012 FEL trucks). For the August testing, the idle emissions rates were found to be 5.5 g/hour for all trucks and 2.7 g/hour for the non-2010-2012 FEL trucks.
Analysis of the real-time NOx emissions in the reservoir suggests that by keeping the truck turned on while it travels through the reservoir, including during idle periods, the temperatures of the SCR aftertreatment system can be maintained at levels where there is not a significant loss in the catalytic activity of the SCR. In contrast, it was observed that when the trucks were turned off for the installation of the data loggers, the exhaust temperature and temperature of the SCR fell to levels below those need for effective NOx reduction for periods of up to 500 s after the vehicle was turned on again. Based on the results of this analysis, including the relatively low levels of idle emissions found, and the fact that the SCR aftertreatment temperature can be maintained at relatively effective levels if the trucks are not turned off, it was recommended that the most effective strategy for achieving lower emission levels in the DG reservoir would be to keep the truck running during idle periods in the reservoir.

Conclusions
In this study, NOx emissions from sediment hauling trucks were evaluated in a construction area and this information was utilized to modify the haul truck fleet for the project. The results showed the average NOx emissions were 0.45 g/bhphr, and approximately 1/3 of the trucks had emissions that were comparable to or below the 0.2 g/bhp-hr NOx level. The fleet average NOx emissions were 0.38 g/bhp-hr for non-credit engines and 0.77 g/bhp-hr for credit engines, with 2013 and newer credit engines averaging 0.29 g/bhp-hr. Based on this information, it was decided to remove the 2010-2012 credit engine trucks for the following season. SCR efficiencies were in the 80% to 90% plus range for the more typical emitters, with SCR efficiencies ranging down to approximately 60% for some high emitters. SCR temperatures averaged approximately 250℃ over the routes, with temperatures consistently above 250℃ for freeway driving and lower SCR temperatures for the surface street and DG reservoir SPS operation. This study found that the NOx emissions for non-2010-2012 FEL trucks were well below the California Clean Idle certification of 30 g/hour and the estimates used in EMFAC2021. Also, this study recommended that the most effective strategy for maintaining the emission levels feasible will be to keep the truck running during idle periods in the reservoir.