Case Studies of Truck Activity Data for Emissions Modeling (2019)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Guide to Truck Activity Data for Emissions Modeling F-1 Appendix F. Case Study #6: Truck Activity Analyses from Localized Fleet Telematics Data F.1 Emissions Model Inputs Supported • Starts per vehicle (by hour of day and weekday/weekend). • Start operating mode (distribution of starts by soak time, hour of day, and weekday/weekend). • Truck idling (distribution of idling by hour and duration). F.2 Level of Effort for Local Application This data source/method requires a high level of effort for local application. F.3 Overview This case study builds on Case Study #4, which investigates the potential for using truck fleet telematics data purchased from Vnomics to develop truck hotelling and starts inputs for the U.S. EPA’s MOVES model. Vnomics collects data from trucks of their customer fleets on an ongoing basis. The collected data include readings from onboard GPS units that can be used to identify the location and speed of the vehicle at any given time, and readings from the vehicle’s engine control unit which can be used to identify engine starts, stops, and idling events. In this case study, additional data were purchased from Vnomics and used to analyze a number of truck activities related to starts, soak, and idle. In contrast to the data purchased from Vnomics for Case Study #4 where we target 300 interstate long-haul trucks regardless of their domiciles, the data purchased for this case study are for any trucks operating in Denver County, CO, and Norfolk County, VA. These counties were used for the analysis of large-scale GPS data from StreetLight in Case Study #3. Therefore, comparisons of starts and soak time distributions derived in this case study can be made with those in Case Study #3. MOVES2014 provides a limited set of user input options for idle, start, and soak time parameters for trucks. Table F.1 summarizes the various idle-related, start, and soak time parameters that have been investigated in the case studies conducted to date, and how they relate to existing MOVES user inputs or data tables. In some cases, the project team is providing additional data that could be used to support future MOVES enhancements or to compare against underlying MOVES data even if they are not existing user input options. F.3.1 Engine Starts and Soak Time Inputs in MOVES MOVES uses data regarding the number of starts by time of day and soak time in the estimation of engine start emissions and evaporative fuel vapor losses. The base emission rate for engine starts is based on a 12-hour soak period. All engine soaks greater than 12 hours assume the same engine start emission rate as for 12 hours. However, for all engine soaks less than 12 hours, the base engine start emission rate is adjusted based on soak time bins (or start operation modes) shown in Table F.2. MOVES2014 allows users to specify the number of engine starts in each month, day type and hour of the day, as well as by source type and vehicle age. Therefore, it is of interest to investigate how MOVES users can create these data inputs for trucks in a specific area such as county.

Guide to Truck Activity Data for Emissions Modeling F-2 Table F.1 Summary of Start, Soak Time, and Hotelling Inputs in Case Studies MOVES Input/Data Table User Input or Data Table? StreetLight Case Study #3 Vnomics Case Study #4 Vnomics Case Study #6 Total daily hotelling hours User input (hotelling importer)  Hotelling operating mode—Fraction of long haul combination trucks (by model year) that are in extended idle mode, APU mode, or engine-off mode User input (hotelling importer) 1 Hotelling distribution—fraction of hotelling events by start hour and duration Data table  Off-network idling—fraction of idling events >5 minutes by duration Not currently an input  Start distributions—Total number of starts by hour of day and day of week, by source type User input 2  Start operating mode distributions— Fraction of starts by hour of day and day of week, by soak period User input 2  1 Idle versus engine-off only—no APU data. 2 Starts represented by vehicle movement, rather than engine on/off. Table F.2 Start Operating Modes in MOVES Nominal Soak Period (Mins) OpMode ID Definition 3 101 Soak Time < 6 minutes 18 102 6 minutes ≤ Soak Time < 30 minutes 45 103 30 minutes ≤ Soak Time < 60 minutes 75 104 60 minutes ≤ Soak Time < 90 minutes 105 105 90 minutes ≤ Soak Time < 120 minutes 240 106 120 minutes ≤ Soak Time < 360 minutes 540 107 360 minutes ≤ Soak Time < 720 minutes 720 108 720 minutes ≤ Soak Time Source: U.S. Environmental Protection Agency (2015). F.3.2 Truck Idling Activities Truck idling is a subject of interest for emission modelers and air quality regulators. Many states, counties, and municipalities across the U.S. have anti-idling regulation (U.S. EPA, 2006). However, truck idling activities are not well understood as trucks are diverse in terms of type and operation, which influences their idling activities. For example, an interstate long-haul truck may idle its engine overnight during the mandatory rest period (FMCSA, 2017). An urban delivery truck may idle for a short period throughout the day while delivering packages. A utility truck may idle for a long period while performing the work. The different truck idling activities will have different impacts on truck idling emissions and their control measures.

Guide to Truck Activity Data for Emissions Modeling F-3 In Case Study #4, the project team used data from 300 trucks to develop truck hotelling activity inputs for MOVES. MOVES defines hotelling as “any long period of time that drivers spend in their vehicles during mandated down times during long distance deliveries by tractor/trailer combination heavy-duty trucks.” (U.S. EPA, 2016). During the mandatory down time or rest period, drivers can stay in motels or other accommodations. On the other hand, most long-haul trucks have sleeping spaces built into the cab of the truck so drivers can also stay in their vehicles. While in their vehicles, the drivers may turn on the truck engine or use an auxiliary power unit to power air conditioning, heater, and other accessories. The emissions produced during this period are unaccounted for as part of other emission processes in MOVES. In MOVES2014, only long-haul combination trucks are assumed to have hotelling activity. However, other types of trucks, including single-unit trucks and short-haul trucks, may also engage in idling activities and produce emissions that are not currently accounted for in MOVES. For example, whereas emissions from idling activities while stopping at traffic signals or being stuck in traffic congestion are part of running exhaust emissions, emissions from idling while loading/unloading cargo, queuing at port terminal, being on restroom or meal break, and etc. are not captured by any emission processes in MOVES. Although these idling activities may not be as long as hotelling, they may occur more frequently and can add up to a significant amount of idling time. Therefore, in this case study the term off-network idling is introduced, which includes every idling activity that does not occur on a typical road network during a trip. More specifically, an off- network idling event is defined as a truck activity where the second-by-second vehicle speed values are lower than 5 mph for a period of at least 5 minutes with the engine running. The 5-mph speed threshold represents a typical speed limit in parking areas, distribution centers, etc. The 5-minute idling period threshold is a typical limit used in anti-idling regulations in many jurisdictions and is high enough to exclude on-network idling events such as stopping at traffic lights. Selection of a 5-minute period also made it possible to compare the Vnomics and StreetLight data, for which a similar threshold was set. F.4 Data Sources Additional truck fleet telematics data were purchased from Vnomics under the same nondisclosure agreement and contract provisions as in the previous purchase. The allocated budget and the pricing indicate that the project team would be able to purchase three county-months of data. In Case Study #3, the project team engaged StreetLight to develop starts and soak data inputs for three counties: 1) Denver, CO; 2) Fairfax, VA; and 3) Norfolk, VA. In this case study, instead of purchasing Vnomics data for one month for each of these three counties, the project team decided to purchase the data according to the matrix shown in Table F.3 in order to also explore seasonal differences. Table F.3 Vnomics Datasets Acquired Dataset Number County Month 1 Denver, CO January 2017 2 Denver, CO May 2017 3 Norfolk, VA January 2017 Two months of data were purchased for Denver as its weather swings the most between summer and winter among the three counties. The two months chosen are January 2017 (average air temperature around

Guide to Truck Activity Data for Emissions Modeling F-4 30°F10; more likely to have hotelling activities with engine on in order to keep the cabin warm) and May 2017 (milder air temperature with an average of 50 – 60°F; less likely to need heater). Also, Vnomics has customer terminals in the Denver area. For a county in Virginia, the project team chose Norfolk over Fairfax because: 1) based on the truck stop databases used in Case Study #4, there are a few truck stops in Norfolk while there are none in Fairfax; and 2) Vnomics has customer terminals in the Norfolk area but not in the Fairfax area. For Norfolk, January 2017 was chosen in order to be consistent with Denver. It had an average air temperature of around 40°F.11 For each county, the project team provided Vnomics with a bounding box that defines the boundary of data to be included, as shown in Figure F.1 and Figure F.2 for Denver and Norfolk, respectively. This bounding box is larger than the county boundary itself and covers most of the populated area surrounding the county. Vnomics provided data for all the trips with any data point inside the bounding box. Defining the bounding box for data purchase in this way instead of based on the actual county boundary reduced the data processing time for Vnomics, which resulted in a lower price. The project team then performed a spatial analysis ourselves to identify which data points fall inside the county boundary. Thus, in this report we refer to county and area differently. For example, Denver area is the area inside the bounding box for the Denver data purchase while Denver County is the area inside the Denver county boundary. The delivered datasets include 43,830 data files from 443 trucks in the Denver area and 3,379 data files from 52 trucks in the Norfolk area. All the trucks are Class 7 (26,000 – 33,000 lbs.) or Class 8 (> 33,000 lbs.), which are equivalent to the “heavy-duty” category in Case Study #3.The majority of these trucks are less than six years old as shown in Figure F.3. These trucks are from fleets that operate a less-than-truckload model with home domiciles in the Denver or Norfolk areas. Most of their operation are pickups and deliveries within the area, which can be considered as short haul in the context of MOVES. 10 City-Data.com. “Denver, Colorado.” http://www.city-data.com/city/Denver-Colorado.html, accessed August 2017. 11 City-Data.com. “Norfolk Virginia.” http://www.city-data.com/city/Norfolk-Virginia.html, accessed August 2017.

Guide to Truck Activity Data for Emissions Modeling F-5 Figure F.1 Data Bounding Box and County Boundary of Denver, Colorado Map source: Google Maps. Figure F.2 Data Bounding Box and County Boundary of Norfolk, Virginia Map source: Google Maps.

Guide to Truck Activity Data for Emissions Modeling F-6 Figure F.3 Number of Trucks in Vnomics Datasets by Vehicle Model Year F.5 Data Processing and Analysis The raw data obtained included a number of data files containing second-by-second time stamp, latitude, longitude, vehicle speed, accelerator pedal position, fuel rate, engine RPM, total distance (vehicle odometer), and reference torque, as well as the make, model, and year of the vehicle. The project team inspected the completeness of the data files and removed the data files where one or more of the following data fields is missing—time stamp, latitude, longitude, vehicle speed, and RPM. These five data fields are necessary for the analysis of truck activity in this case study. Note that Vnomics’ data logger starts recording data after the key is switched on (key-on event) and it stops recording data after the key is switched off (key-off event). This entire period makes up a data file. On the other hand, a trip in the context of this case study is the period from when the engine is turned on (engine-on event) to when the engine is turned off (engine-off event) as shown in Figure F.4. Thus, one data file may consist of more than one trip if the engine is turned off and then on without a key-off event. On the other hand, a data file may not consist of any trip if the key is switched on and then off without an engine-on event. Since the recorded data include engine speed, engine-on and engine-off events can be identified. Due to possible noise in the engine speed data, an engine speed of 300 RPM is used as the threshold for determining whether the engine is running. This threshold is consistent with the one used in a previous research (Boriboonsomsin et al., 2017). Figure F.4 Illustration of Events in Truck Activity 1 0 1 18 7 38 12 4 4 10 59 29 50 45 60 60 45 1 3 0 0 3 3 4 2 1 2 8 4 7 3 4 7 0 0 10 20 30 40 50 60 70 2002 2004 2006 2008 2010 2012 2014 2016 2018 Count of Trucks Vehicle Model Year Denver (total = 443) Norfolk (total = 52)

Guide to Truck Activity Data for Emissions Modeling F-7 The project team applied the following data analysis steps: 1. Removed data files where one or more of the following data fields is missing— time stamp, latitude, longitude, vehicle speed, and RPM. 2. Identified trip starts and ends based on engine speed threshold of 300 RPM. Compiled all the trips into a trip table containing information about vehicle ID, trip ID, trip duration, trip distance, trip average speed, start and end times, start and end locations, and day of week. 3. Using the trip table, compiled a Soak table containing information about vehicle ID, trip ID, soak start and end times, soak location, and day of week. 4. Identified on-network and off-network idling events. An off-network idling event (meal/restroom breaks, loading/unloading, etc.) is defined as a truck activity where the second-by-second vehicle speed values are lower than 5 mph for a period of at least 5 minutes with the engine running. On-network idling (traffic lights, congestion, etc.) is defined as other events when the instantaneous vehicle speed is less than 1 mph. (On-network idling is not an input parameter for MOVES, but was evaluated for research purposes.) 5. Compiled all the off-network idling events into an idling table containing information about vehicle ID, trip ID, idling event ID, start and end times, start and end locations, maximum and average speeds, traveled distance, duration, and day of week. 6. Counted the number of starts occurring in each hour of the day, for each month and county. 7. Counted the total number of vehicles with activity in each county by month and hour of the day, defining the vehicle population. 8. Analyzed the distributions of number of starts per vehicle by hour of day (total starts in month and hour divided by total vehicles active in month and hour) and compared them to those in Case Study #3. 9. Analyzed the distributions of start operating mode (i.e., number of starts by soak period) by hour of day. 10. Analyzed the distributions of off-network idling events by idling duration. Evaluated the fraction of off-network idling duration in the total operating time. Additionally, in order to evaluate the impacts of having engine speed data on the calculated truck activity statistics, the project team also analyzed the data as if no engine speed data were available. That is, each data file was regarded as a trip, and an off-network idling event was defined as truck activity where the second-by-second speed values were lower than 5 mph for a period of at least 5 minutes without consideration of engine speed. This made it possible to compare the Vnomics and StreetLight data.

Guide to Truck Activity Data for Emissions Modeling F-8 F.6 Findings from Sample Data F.6.1 Summary Statistics Table F.4 summarizes the number of vehicles, data files, and trips in the datasets. Observations from this table include: • There are many more trucks in the Denver datasets than in the Norfolk dataset. Note that this does not represent the characteristics of the truck population in these two areas but rather of the truck samples in Vnomics database. • In total, about 20 percent of the received data files are invalid where one or more of the following data fields is missing—time stamp, latitude, longitude, vehicle speed, and RPM. Most of these invalid data files are empty files, which may be created when the key is switched on and then off without an engine- on event. • For both areas, the majority of starts is outside of the county boundary as shown in Figure F.5 and Figure F.6 for Denver and Norfolk, respectively. Only 17 percent and 20 percent of trips in Denver and Norfolk, respectively, were started inside the county boundary. This suggests that even trucks operating within the county boundaries spend most of their time operating outside of the county. Also, it could be that the home bases of the truck fleets are located outside the county boundary, which could be due to local land use ordinances and cheaper land costs compared to the central cities of each region. Note that the subsequent analysis of start, soak, and idling activities was focused on the area inside the county boundary only. • When comparing the number of files with starts inside the county boundary and the number of trip starts inside county boundary, it was found that there were about 20 percent more trips than data files. This indicates that truck activity data without engine operation data (such as those collected by GPS) could result in significant underestimation of the number of trip starts. Table F.4 Summary of Number of Vehicles, Data Files, and Trips Number of Denver Jan 2017 Denver May 2017 Denver Combined Norfolk Jan 2017 Received data files 19,867 23,962 43,829 3,379 Vehicles in received data files 317 298 443 52 Valid data files 15,873 19,819 35,692 2,280 Vehicles in valid data files 292 312 418 40 Files that start inside county boundary 2,904 3,452 6,356 445 Files that start outside county boundary 12,969 16,367 29,336 1,835 Trips 20,872 24,852 45,724 2,597 Trips that start inside county boundary 3,467 4,084 7,551 524 Trips that start outside county boundary 17,405 20,768 38,173 2,073

Guide to Truck Activity Data for Emissions Modeling F-9 Figure F.5 Start Locations for Denver Datasets Background image: Google Earth. Figure F.6 Start Locations for Norfolk Dataset Background image: Google Earth.

Guide to Truck Activity Data for Emissions Modeling F-10 F.6.2 Starts per Vehicle To compute starts per truck per hour, the vehicle population needed to be defined. As discussed in Case Study #3, it is not entirely clear what is meant by “vehicle population” for bounded modeling domains when vehicles migrate in and out of the domain over the course of the day. In this analysis, the vehicle population was defined as the total number of vehicles in the provided dataset that showed activity (key on) within the county in the given hour of the day, for the given day type (weekday or weekend), over the one-month period of observation. The denominator (vehicle population) therefore depends upon the period of observation, since not every truck will be active on every day in the observation period. This is problematic and raises the question of whether starts per vehicle-mile traveled in the modeling domain would be a better metric for measuring start activity. Using data of trip starts inside the county boundary, the project team generated plots of trip starts per vehicle by hour of day and day type. Figure F.7 shows the plots for Denver County in January and May 2017. The plots for weekday (wd) look similar to each other for the most part except during late night and early morning hour where the plot for May shows significantly higher numbers of starts per vehicle. When inspecting the data, it was found that these unusually high numbers of starts per vehicle were caused by a few vehicles making several starts during those hours. The plots for weekend (we) also show a similar trend where there were barely any trip starts on a weekend. Figure F.7 Starts per Truck (Heavy-Duty) for Denver County January and May 2017 Figure F.8 shows the plots of trip starts per vehicle by hour of day for weekday, comparing between Denver and Norfolk. The plot for Denver was generated using data for both months combined (with the total vehicle population equal to the sum of the populations in each month). It is observed that the two plots have different characteristics where the trucks in Norfolk did not usually start a trip between 12 and 8 a.m. while the trucks in Denver started trips as early as 4 a.m. In this figure, the plots generated using data from Vnomics are also compared to those generated using data from StreetLight as well as the MOVES2014 defaults for -0.05 0.05 0.15 0.25 0.35 0.45 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Starts per Truck per Hour Vnomics/Denver_Jan/Wd Vnomics/Denver_Jan/We Vnomics/Denver_May/Wd Vnomics/Denver_May/We

Guide to Truck Activity Data for Emissions Modeling F-11 combination short-haul trucks. It is obvious that the numbers of starts per vehicle derived from Vnomics data are much lower. For Denver, it is 3.6 starts per vehicle per weekday versus 16.0 based on StreetLight data versus 7.4 in MOVES2014. For Norfolk, it is 2.0 starts per vehicle per weekday versus 20.5 based on StreetLight data versus 7.5 in MOVES2014. These findings indicate that the trucks in the Vnomics datasets had much lower levels of vehicle utilization than the trucks in StreetLight data (based on a much larger sample size) and those used to develop the default values in MOVES (based on a sample of 120 trucks collected in 1997 – 1998). Upon inspection of the Vnomics datasets, it was found that the trucks were not necessarily used every day. And on some days that they were used, they did not make any trip start inside the county boundary. These may be characteristics specific to the truck samples in the Vnomics datasets and do not necessarily reflect the characteristics of the entire truck population in these two counties. Figure F.9 show similar plots to those in Figure F.8 but for weekend. As pointed out earlier, there were barely any trip starts on weekend in the Vnomics datasets for both Denver and Norfolk. Again, this characteristic may be specific to the truck samples in the Vnomics datasets and is not necessarily true for other trucks in these two counties as evident by the plots derived from StreetLight data. Figure F.8 Starts per Truck (Heavy-Duty) by Hour of Day during Weekday 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Vnomics/Denver/Wd StreetLight/Denver/Wd MOVES_61/Wd Vnomics/Norfolk/Wd StreetLight/Norfolk/Wd Starts per Truck per Hour

Guide to Truck Activity Data for Emissions Modeling F-12 Figure F.9 Starts per Truck (Heavy-Duty) by Hour of Day during Weekend F.6.3 Start Operating Mode The start operating mode input for MOVES is in the form of a two-dimensional distribution of the number of starts by hour of day and soak period. There are eight soak period bins as defined in Figure F.10. Figure F.10 shows the start operating mode distributions for Denver during weekday, comparing between January and May 2017. According to the figure, the two distributions look very similar to each other. They have a peak around 12 – 2 p.m. and almost one-half of the starts were preceded by a soak period between 6 and 30 minutes. This indicates that the trucks might be doing the same type of operation (revenue service) in both months. Another observation is that there were very few overnight soak events (i.e., those longer than 12 hours). Most of the long soak events were between 2 and 6 hours as there were starts late into the night hours except between 12 and 3 a.m. This implies that these truck samples rarely had cold starts during a weekday. -0.10 0.10 0.30 0.50 0.70 0.90 1.10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Vnomics/Denver/We StreetLight/Denver/We MOVES_61/We Vnomics/Norfolk/We StreetLight/Norfolk/We Starts per Truck per Hour

Guide to Truck Activity Data for Emissions Modeling F-13 Figure F.10 Start Operating Mode Distributions for Denver during Weekday for January and May 2017 Figure F.11 shows the comparison of start operating mode distributions during weekdays between Denver and Norfolk. The distribution for Denver is based on the data for both months combined, and thus, has a very similar pattern to those shown in Figure F.10. The distribution for Norfolk shows a slightly different pattern. Although the typical soak period is between 6 and 30 minutes as in the case of Denver, the starts were most frequent between 10 a.m. and 12 p.m. instead of 12 – 2 p.m. Also, unlike the truck samples in Denver which did not start between 12 and 3 a.m., the truck samples in Norfolk did not have any starts between 11 p.m. and 6 a.m. Figure F.12 shows the comparison of start operating mode distributions during weekends between Denver and Norfolk. There were a few rare starts made by the truck samples in Denver. The truck samples in Norfolk had no starts at all on weekends. Soak Period 3 18 45 75 105 240 540 720 Sum 19.9 48.2 13.0 5.7 3.2 7.7 1.2 1.2 100 Hour 1 0.0 0 0 0 0 0 0 0 0.0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0.1 0 0.1 0.1 0 0 0 0 0 5 0.2 0.1 0 0 0.0 0.0 0.0 0 0 6 0.9 0.2 0.2 0.0 0.1 0.0 0.3 0 0 7 1.2 0.2 0.5 0.2 0.1 0 0.2 0.1 0 8 2.2 0.5 1.0 0.3 0.1 0.1 0.1 0.1 0.1 9 5.4 0.7 2.4 0.8 0.4 0.2 0.7 0.3 0 10 8.9 1.4 4.0 0.9 0.5 0.4 1.2 0.4 0.1 11 9.8 1.8 4.8 1.0 0.5 0.3 1.2 0.3 0 12 11.2 2.3 5.3 1.3 0.5 0.2 1.5 0.0 0.0 13 12.2 2.0 6.6 1.6 0.6 0.4 0.8 0.1 0.1 14 11.8 1.9 6.6 1.5 0.7 0.5 0.5 0 0 15 11.8 2.5 6.0 1.6 0.4 0.4 0.7 0 0.1 16 10.3 2.4 4.8 1.4 0.8 0.5 0.3 0 0.1 17 6.7 1.4 2.9 1.5 0.6 0.1 0.1 0 0.1 18 3.4 1.0 1.6 0.4 0.2 0 0 0 0.3 19 1.7 0.5 0.6 0.3 0.0 0 0 0 0.3 20 1.0 0.4 0.4 0.1 0.1 0 0 0 0.1 21 0.5 0.2 0.3 0 0.0 0 0 0 0 22 0.3 0.1 0.1 0 0.1 0.0 0 0 0 23 0.2 0.0 0.1 0.1 0 0 0 0 0 24 0.1 0.1 0 0 0 0 0 0 0 Sum 100 100 Denver/Jan 2017/Weekday Soak Period 3 18 45 75 105 240 540 720 Sum 20.8 47.5 13.3 5.8 3.3 7.4 1.1 0.9 100 Hour 1 0.0 0 0 0 0 0 0 0 0.0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0.5 0.1 0.2 0.0 0.0 0 0.0 0.0 0.0 5 0.4 0.1 0 0.1 0.0 0.1 0.1 0 0 6 0.7 0.1 0.1 0.1 0.1 0.1 0.1 0 0 7 1.6 0.3 0.7 0.3 0.0 0.1 0.1 0.0 0 8 2.6 0.5 1.2 0.4 0.3 0.1 0.1 0.0 0.0 9 6.0 0.7 3.1 0.9 0.4 0.2 0.5 0.1 0.0 10 8.3 1.1 3.6 1.3 0.6 0.3 1.0 0.3 0.0 11 11.0 1.9 5.3 1.1 0.7 0.3 1.3 0.4 0 12 11.8 2.4 5.6 1.3 0.6 0.5 1.3 0.1 0.0 13 11.5 2.6 5.9 1.3 0.3 0.3 1.0 0.0 0 14 11.4 2.5 6.0 1.2 0.5 0.4 0.8 0 0.0 15 10.8 2.6 5.2 1.5 0.7 0.3 0.5 0 0.0 16 10.1 2.3 4.6 1.6 0.7 0.5 0.3 0.0 0.1 17 6.7 1.8 3.2 1.1 0.3 0.1 0.1 0 0.1 18 3.2 0.8 1.6 0.7 0.1 0 0.0 0 0.0 19 1.2 0.3 0.5 0.2 0.1 0 0 0 0.1 20 0.4 0.1 0.1 0.0 0 0 0 0 0.1 21 0.3 0.2 0.0 0 0 0 0 0 0.0 22 0.6 0.2 0.2 0.1 0 0 0 0 0.0 23 0.6 0.3 0.2 0.1 0.0 0 0 0 0 24 0.2 0.0 0.1 0 0 0 0 0 0.0 Sum 100 100 Denver/May 2017/Weekday

Guide to Truck Activity Data for Emissions Modeling F-14 Figure F.11 Start Operating Mode Distributions during Weekday Figure F.12 Start Operating Mode Distributions during Weekend Soak Period 3 18 45 75 105 240 540 720 Sum 20.4 47.9 13.1 5.7 3.2 7.6 1.1 1.0 100 Hour 1 0.0 0 0 0 0 0 0 0 0.0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0.3 0.0 0.1 0.0 0.0 0 0.0 0.0 0.0 5 0.3 0.1 0 0.1 0.0 0.1 0.1 0 0 6 0.8 0.2 0.2 0.1 0.1 0.1 0.2 0 0 7 1.4 0.3 0.6 0.2 0.1 0.1 0.1 0.1 0 8 2.4 0.5 1.1 0.4 0.2 0.1 0.1 0.1 0.0 9 5.7 0.7 2.8 0.8 0.4 0.2 0.6 0.2 0.0 10 8.5 1.3 3.8 1.1 0.6 0.3 1.1 0.4 0.0 11 10.5 1.8 5.1 1.0 0.6 0.3 1.3 0.3 0 12 11.5 2.4 5.5 1.3 0.5 0.4 1.4 0.1 0.0 13 11.8 2.3 6.2 1.5 0.5 0.3 0.9 0.1 0.0 14 11.6 2.2 6.3 1.4 0.6 0.4 0.7 0 0.0 15 11.2 2.5 5.6 1.6 0.6 0.3 0.6 0 0.1 16 10.2 2.3 4.7 1.5 0.8 0.5 0.3 0.0 0.1 17 6.7 1.6 3.1 1.2 0.5 0.1 0.1 0 0.1 18 3.3 0.9 1.6 0.5 0.1 0 0.0 0 0.1 19 1.4 0.4 0.5 0.2 0.1 0 0 0 0.2 20 0.7 0.3 0.3 0.1 0.0 0 0 0 0.1 21 0.4 0.2 0.1 0 0.0 0 0 0 0.0 22 0.4 0.1 0.2 0.1 0.0 0.0 0 0 0.0 23 0.4 0.2 0.1 0.1 0.0 0 0 0 0 24 0.1 0.0 0.1 0 0 0 0 0 0.0 Sum 100 100 Denver/Weekday Soak Period 3 18 45 75 105 240 540 720 Sum 19.7 43.3 14.0 5.7 4.4 9.2 0.4 3.3 100 Hour 1 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 6 0 0 0 0 0 0 0 0 0 7 0.2 0 0 0 0 0 0 0.2 0 8 0.4 0 0 0 0 0.2 0.2 0 0 9 2.9 0 2.1 0.4 0.2 0 0.2 0 0 10 10.9 2.3 5.4 0.6 0.4 0.6 1.5 0.2 0 11 14.6 3.6 6.7 1.0 0.8 0 2.5 0 0 12 13.0 2.3 6.1 1.7 0.8 0.8 1.1 0 0.2 13 11.7 2.5 5.2 1.7 0.6 0.8 1.0 0 0 14 10.9 1.3 5.7 1.9 0.2 0.8 1.0 0 0 15 10.2 2.1 4.0 1.7 0.8 0.4 0.6 0 0.6 16 7.5 1.9 3.4 1.3 0.2 0.2 0 0 0.4 17 5.4 1.3 1.9 0.8 0.2 0 0.2 0 1.0 18 1.9 0 0.8 0.4 0 0.2 0 0 0.6 19 4.0 0.4 1.0 1.0 0.4 0.2 0.6 0 0.6 20 1.9 0 0.4 0.8 0.2 0.4 0.2 0 0 21 1.5 0.6 0 0.4 0.6 0 0 0 0 22 2.1 0.8 0.2 0.4 0.6 0 0.2 0 0 23 1.0 0.6 0.4 0 0 0 0 0 0 24 0 0 0 0 0 0 0 0 0 Sum 100 100 Norfolk/Weekday Soak Period 3 18 45 75 105 240 540 720 Sum 25.0 25.0 0 0 0 0 25.0 25.0 100 Hour 1 25.0 25.0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 5 25.0 0 25.0 0 0 0 0 0 0 6 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 0 0 0 0 8 0 0 0 0 0 0 0 0 0 9 0 0 0 0 0 0 0 0 0 10 25.0 0 0 0 0 0 0 0 25.0 11 0 0 0 0 0 0 0 0 0 12 0 0 0 0 0 0 0 0 0 13 0 0 0 0 0 0 0 0 0 14 0 0 0 0 0 0 0 0 0 15 0 0 0 0 0 0 0 0 0 16 0 0 0 0 0 0 0 0 0 17 0 0 0 0 0 0 0 0 0 18 0 0 0 0 0 0 0 0 0 19 0 0 0 0 0 0 0 0 0 20 0 0 0 0 0 0 0 0 0 21 0 0 0 0 0 0 0 0 0 22 0 0 0 0 0 0 0 0 0 23 25.0 0 0 0 0 0 0 25.0 0 24 0 0 0 0 0 0 0 0 0 Sum 100 100 Denver/Weekend Soak Period 3 18 45 75 105 240 540 720 Sum 0 0 0 0 0 0 0 0 0 Hour 1 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 6 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 0 0 0 0 8 0 0 0 0 0 0 0 0 0 9 0 0 0 0 0 0 0 0 0 10 0 0 0 0 0 0 0 0 0 11 0 0 0 0 0 0 0 0 0 12 0 0 0 0 0 0 0 0 0 13 0 0 0 0 0 0 0 0 0 14 0 0 0 0 0 0 0 0 0 15 0 0 0 0 0 0 0 0 0 16 0 0 0 0 0 0 0 0 0 17 0 0 0 0 0 0 0 0 0 18 0 0 0 0 0 0 0 0 0 19 0 0 0 0 0 0 0 0 0 20 0 0 0 0 0 0 0 0 0 21 0 0 0 0 0 0 0 0 0 22 0 0 0 0 0 0 0 0 0 23 0 0 0 0 0 0 0 0 0 24 0 0 0 0 0 0 0 0 0 Sum 0 0 Norfolk/Weekend

Guide to Truck Activity Data for Emissions Modeling F-15 F.6.4 Off-Network Idling This subsection presents the results of off-network idling activities inside the county boundary. Figure F.13 shows a comparison of the percent distributions of idling duration in Denver County between January 2017 and May 2017. According to the figure, off-network idling activities in the two months were similar in terms of both frequency and duration. This finding adds to the previous indication from start and soak activity patterns that the trucks might be doing the same type of operation (revenue service) in both months. Figure F.13 Off-Network Idling Events in Denver County in January and May 2017 Figure F.14 shows a comparison of the percent distributions of idling duration during weekday between Denver County and Norfolk County. The distribution for Denver County is based on the data for both months combined, and thus, is very similar to those shown in Figure F.13. When compared to the trucks in Norfolk County, the trucks in Denver County had 14 percent fewer idling events with duration of 5 to 10 minutes, but had 13 percent more idling events with duration of 10 – 15 minutes. F.6.5 Impacts of Having Engine Speed Data This subsection presents the impacts of having engine speed data on the calculated truck activity statistics. When there is no engine speed data from ECU (such as data collected from GPS alone), each data file would represent a trip and the first data point in the data file would be regarded as the trip start irrespective of whether the engine is actually on or not. We call this data point as file start in this analysis. Figure F.15 compares starts per truck during weekdays. It shows that the number of trip starts was consistently higher than the number of file starts. For Denver, there were 3.6 trip starts but only 3.1 file starts per vehicle per weekday. For Norfolk, there were 2.0 trip starts but only 1.8 file starts per vehicle per weekday. As a reference, the starts per vehicle default numbers for combination short-haul trucks in MOVES2014 are 7.4 for Denver and 7.5 for Norfolk. These default numbers would be equivalent to file starts in this analysis. Trip starts numbers are higher than file starts numbers because many data files consist of more than one trip. A new trip in the middle of a data file occurs when the engine is turned off and then on without a key-off event. The period of such sequence of events is usually short. In which case, having engine speed data would 65 15 15 3 1 1 0 0 0 73 15 8 2 1 0 0 0 0 0 10 20 30 40 50 60 70 80 5-10 10-15 15-30 30-45 45-60 60-90 90-120 120-180 180+ % of Events Idling Duration (min) Jan 2017 (total = 993) May 2017 (total = 1,081)

Guide to Truck Activity Data for Emissions Modeling F-16 result in having a higher fraction of short soak events as evident in Figure F.16 for Denver County and Figure F.17for Norfolk County. Figure F.14 Off-Network Idling Events during Weekday Figure F.15 Trip Starts versus File Starts per Truck by Hour of Day during Weekday 69 15 12 3 1 0 0 0 0 55 28 12 5 0 1 0 1 0 0 10 20 30 40 50 60 70 80 5-10 10-15 15-30 30-45 45-60 60-90 90-120 120-180 180+ % of Events Idling Duration (min) Denver (total = 2,074) Norfolk (total = 152) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Denver/Trip Starts Norfolk/Trip Starts Denver/File Starts Norfolk/File Starts Starts per Truck per Hour

Guide to Truck Activity Data for Emissions Modeling F-17 Figure F.16 Trip Starts (Left) versus File Starts (Right) Operating Modes in Denver County Figure F.17 Trip Starts (Left) versus File Starts (Right) Operating Modes in Norfolk County Soak Period 3 18 45 75 105 240 540 720 Sum 20.4 47.9 13.1 5.7 3.2 7.6 1.1 1.0 100 Hour 1 0.0 0 0 0 0 0 0 0 0.0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0.3 0.0 0.1 0.0 0.0 0 0.0 0.0 0.0 5 0.3 0.1 0 0.1 0.0 0.1 0.1 0 0 6 0.8 0.2 0.2 0.1 0.1 0.1 0.2 0 0 7 1.4 0.3 0.6 0.2 0.1 0.1 0.1 0.1 0 8 2.4 0.5 1.1 0.4 0.2 0.1 0.1 0.1 0.0 9 5.7 0.7 2.8 0.8 0.4 0.2 0.6 0.2 0.0 10 8.5 1.3 3.8 1.1 0.6 0.3 1.1 0.4 0.0 11 10.5 1.8 5.1 1.0 0.6 0.3 1.3 0.3 0 12 11.5 2.4 5.5 1.3 0.5 0.4 1.4 0.1 0.0 13 11.8 2.3 6.2 1.5 0.5 0.3 0.9 0.1 0.0 14 11.6 2.2 6.3 1.4 0.6 0.4 0.7 0 0.0 15 11.2 2.5 5.6 1.6 0.6 0.3 0.6 0 0.1 16 10.2 2.3 4.7 1.5 0.8 0.5 0.3 0.0 0.1 17 6.7 1.6 3.1 1.2 0.5 0.1 0.1 0 0.1 18 3.3 0.9 1.6 0.5 0.1 0 0.0 0 0.1 19 1.4 0.4 0.5 0.2 0.1 0 0 0 0.2 20 0.7 0.3 0.3 0.1 0.0 0 0 0 0.1 21 0.4 0.2 0.1 0 0.0 0 0 0 0.0 22 0.4 0.1 0.2 0.1 0.0 0.0 0 0 0.0 23 0.4 0.2 0.1 0.1 0.0 0 0 0 0 24 0.1 0.0 0.1 0 0 0 0 0 0.0 Sum 100 100 Denver/Weekday Soak Period 3 18 45 75 105 240 540 720 Sum 12.3 50.8 15.0 6.6 3.8 8.9 1.3 1.2 100 Hour 1 0.0 0 0 0 0 0 0 0 0.0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0.3 0 0.2 0.0 0.0 0 0.0 0.0 0.0 5 0.3 0 0 0.1 0.0 0.1 0.1 0 0 6 0.7 0.1 0.1 0.1 0.1 0.1 0.2 0 0 7 1.5 0.2 0.6 0.3 0.1 0.1 0.2 0.1 0 8 2.3 0.3 1.1 0.4 0.2 0.1 0.1 0.1 0.0 9 6.1 0.5 3.0 1.0 0.5 0.2 0.7 0.2 0.0 10 9.0 0.8 4.0 1.3 0.7 0.4 1.3 0.4 0.0 11 10.9 1.2 5.6 1.1 0.7 0.4 1.5 0.4 0 12 11.7 1.7 5.7 1.5 0.6 0.5 1.7 0.1 0.0 13 11.9 1.4 6.8 1.7 0.5 0.4 1.1 0.1 0.0 14 11.9 1.5 6.7 1.6 0.7 0.5 0.8 0 0.0 15 11.2 1.5 6.0 1.8 0.7 0.4 0.7 0 0.1 16 10.1 1.4 5.0 1.7 0.9 0.6 0.4 0.0 0.1 17 6.5 0.9 3.2 1.4 0.5 0.2 0.1 0 0.1 18 2.9 0.4 1.6 0.5 0.2 0 0.0 0 0.2 19 1.1 0.1 0.4 0.3 0.1 0 0 0 0.2 20 0.5 0.1 0.2 0.0 0.0 0 0 0 0.1 21 0.3 0.1 0.1 0 0.0 0 0 0 0.0 22 0.5 0.1 0.2 0.1 0.0 0.0 0 0 0.0 23 0.4 0.1 0.2 0.1 0.0 0 0 0 0 24 0.1 0.0 0.0 0 0 0 0 0 0.0 Sum 100 100 Denver/Weekday Soak Period 3 18 45 75 105 240 540 720 Sum 19.7 43.3 14.0 5.7 4.4 9.2 0.4 3.3 100 Hour 1 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 6 0 0 0 0 0 0 0 0 0 7 0.2 0 0 0 0 0 0 0.2 0 8 0.4 0 0 0 0 0.2 0.2 0 0 9 2.9 0 2.1 0.4 0.2 0 0.2 0 0 10 10.9 2.3 5.4 0.6 0.4 0.6 1.5 0.2 0 11 14.6 3.6 6.7 1.0 0.8 0 2.5 0 0 12 13.0 2.3 6.1 1.7 0.8 0.8 1.1 0 0.2 13 11.7 2.5 5.2 1.7 0.6 0.8 1.0 0 0 14 10.9 1.3 5.7 1.9 0.2 0.8 1.0 0 0 15 10.2 2.1 4.0 1.7 0.8 0.4 0.6 0 0.6 16 7.5 1.9 3.4 1.3 0.2 0.2 0 0 0.4 17 5.4 1.3 1.9 0.8 0.2 0 0.2 0 1.0 18 1.9 0 0.8 0.4 0 0.2 0 0 0.6 19 4.0 0.4 1.0 1.0 0.4 0.2 0.6 0 0.6 20 1.9 0 0.4 0.8 0.2 0.4 0.2 0 0 21 1.5 0.6 0 0.4 0.6 0 0 0 0 22 2.1 0.8 0.2 0.4 0.6 0 0.2 0 0 23 1.0 0.6 0.4 0 0 0 0 0 0 24 0 0 0 0 0 0 0 0 0 Sum 100 100 Norfolk/Weekday Soak Period 3 18 45 75 105 240 540 720 Sum 12.0 46.7 14.7 6.8 5.2 10.4 0.5 3.8 100 Hour 1 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 6 0 0 0 0 0 0 0 0 0 7 0.2 0 0 0 0 0 0 0.2 0 8 0.5 0 0 0 0 0.2 0.2 0 0 9 2.9 0.2 1.8 0.5 0.2 0 0.2 0 0 10 11.7 1.8 6.1 0.7 0.5 0.7 1.8 0.2 0 11 14.4 2.5 7.0 1.1 0.9 0 2.9 0 0 12 13.5 1.4 7.0 1.8 0.9 0.9 1.4 0 0.2 13 9.9 0.9 4.3 2.0 0.9 0.7 1.1 0 0 14 11.7 0.9 6.8 2.0 0.2 0.9 0.9 0 0 15 10.2 0.7 5.0 2.0 0.9 0.5 0.5 0 0.7 16 7.9 1.1 4.3 1.4 0.2 0.5 0 0 0.5 17 5.4 0.7 2.0 1.1 0 0.2 0.2 0 1.1 18 1.8 0 0.5 0.5 0 0.2 0 0 0.7 19 3.6 0.5 0.7 0.5 0.5 0.2 0.7 0 0.7 20 1.6 0 0.2 0.7 0.2 0.2 0.2 0 0 21 1.1 0 0.5 0 0.7 0 0 0 0 22 2.3 0.7 0.2 0.5 0.7 0 0.2 0 0 23 1.1 0.7 0.5 0 0 0 0 0 0 24 0 0 0 0 0 0 0 0 0 Sum 100 100 Norfolk/Weekday

Guide to Truck Activity Data for Emissions Modeling F-18 In terms of off-network activities, vehicle speed data (such as from GPS) can be used to identify off-network parking events, which are defined in this case study as truck activity where the second-by-second speed values are lower than 5 mph for a period of at least 5 minutes. However, engine speed data is needed to determine the time fraction of the parking event where the engine is on (i.e., off-network idling). Table F.5 summarizes off-network idling and parking activities in the Vnomics datasets. The results show that for these truck samples, 71 – 78 percent of the number of parking events were actually idling events with the engine on. And the total idling duration accounted for only 62 – 65 percent of the total parking duration. The truck samples in Denver idled for an average of 10.3 minutes per event while the truck samples in Norfolk idled for an average of 12.5 minutes per event. Table F.5 Summary of Off-Network Idling and Parking Activities Denver Jan 2017 Denver May 2017 Denver Combined Norfolk Jan 2017 Total idling duration (minutes) 10,967 10,318 21,285 1,893 Total parking duration (minutes) 16,971 16,227 33,198 3,070 Ratio of idling to parking duration 0.65 0.64 0.64 0.62 Number of idling events 993 1,081 2,074 152 Number of parking events 1,365 1,394 2,759 215 Ratio of idling to parking events 0.73 0.78 0.75 0.71 Idling duration per event (minutes) 11.0 9.5 10.3 12.5 Parking duration per event (minutes) 12.4 11.6 12.0 14.3 Figure F.18 shows the distributions of off-network idling and parking events in Denver County while Figure F.19 shows the distributions of off-network idling and parking events in Norfolk County. Both figures indicate that not having engine speed data, and then assuming every parking event to be an idling event, would result in an overestimation of the number of idling events across all duration bins. The overestimation tends to be more severe for long-duration bins than shorter ones.

Guide to Truck Activity Data for Emissions Modeling F-19 Figure F.18 Off-Network Idling and Parking Events inside Denver County Figure F.19 Off-Network Idling and Parking Events inside Norfolk County 1441 319 240 55 13 5 0 1 0 1641 521 431 127 26 11 1 1 0 0 400 800 1200 1600 2000 5-10 10-15 15-30 30-45 45-60 60-90 90-120 120-180 180+ Count Duration (min) Idling (total = 2,074) Parking (total = 2,759) 83 42 18 7 0 1 0 1 0 106 58 34 11 1 2 2 1 0 0 20 40 60 80 100 120 5-10 10-15 15-30 30-45 45-60 60-90 90-120 120-180 180+ Count Duration (min) Idling (total = 152) Parking (total = 215)

Guide to Truck Activity Data for Emissions Modeling F-20 F.6.6 On-Network Idling On-network idling was investigated to help inform future MOVES model development. The entire Denver/Norfolk Vnomics dataset was used in this analysis. This means that some portions of the data are outside the boundaries of Denver and Norfolk counties to reduce processing effort. Idling rates were also compared with the 300-truck national dataset. Table F.6 compares the fraction of time spent idling across geographies, time periods, and datasets. The table shows that the idle fraction is fairly consistent for regional fleets such as those in the Denver/Norfolk dataset where time spent idling ranges from 26 to 30 percent. The idle fraction for long-haul fleets in the national dataset is somewhat lower at 21 percent. Although it is not vastly different from the idle fraction for regional fleets, the type of idling activities that the long-haul trucks engaged in is quite different. They tend to be for longer periods as shown in Case Study #4. Table F.6 Comparison of Idle Fractions Denver Jan 2017 Denver May 2017 Denver Combined Norfolk Jan 2017 National Nov 2016 No. of vehicles 292 312 4181 40 300 Total idling hours 2,717 2,805 5,522 263 14,943 Total engine-on hours 9,162 10,911 20,073 906 71,183 Idle fraction 30% 26% 28% 29% 21% 1 Some vehicles appear in both months of data. F.6.7 Starts per Mile Table F.7 shows the results of (engine) starts per mile for both the Denver/Norfolk and national datasets. It also shows the results in terms of miles per start, which is perhaps easier to understand. The results are vastly different between the regional fleets in the Denver/Norfolk dataset and the long-haul fleets in the national dataset. This implies that, as in the case of starts per vehicle, calculating total starts based on starts per mile could also be problematic if the trucks are not differentiated by vocation or the type of operation. Table F.7 Comparison of Idle Fractions Denver Jan 2017 Denver May 2017 Denver Combined Norfolk Jan 2017 National Nov 2016 No. of vehicles 292 312 4181 40 300 Total engine starts 20,944 24,957 45,901 2,600 51,218 Total vehicle-miles 262,345 337,791 600,136 21,970 3,127,475 Engine starts/mile 0.080 0.074 0.076 0.118 0.016 Miles/engine start 12.5 13.5 13.1 8.5 61.1 1 Some vehicles appear in both months of data.

Guide to Truck Activity Data for Emissions Modeling F-21 F.6.8 Summary of Findings This case study builds on Case Study #4 by analyzing additional fleet telematics data from Vnomics. The geographic areas and time periods of the new datasets were strategically chosen to allow for comparison with the results of StreetLight data analysis in Case Study #3 and for examination of seasonal differences in truck activity patterns. Some general observations about the datasets used in this case study include: • The Vnomics data are unique among the datasets investigated for this project in that they include ECU data. ECU data are critical to identifying actual engine operation, rather than simply whether a vehicle is moving or stopped, and therefore to developing more accurate estimates of truck activities. • The dataset was useful in comparing start activity observed using engine on/off data with estimates of starts based on vehicle activity data only (key on/off, as might be recorded with a GPS device). It was also useful for looking at hourly distributions of start activity as well as the frequency of idling events by duration. • The definition of truck “population” is based on the number of vehicles observed operating within the analysis domain during a given time period. This definition does not directly correspond to the definition of vehicle population in MOVES. Furthermore, the population estimate is sensitive to the time period being analyzed. The ability to use telematics data such as obtained for this case study to directly estimate starts per vehicle per hour in a bounded local area and consistent with the current MOVES definition of starts per vehicle, is therefore called into question. Further investigation should consider whether starts per truck VMT might be a better metric for local MOVES inputs given the available data sources; or even whether another approach altogether, such as the use of an employment-based truck trip generation model, should be considered. • Although the datasets were obtained for specific geographic areas (Denver and Norfolk counties), the data only included trucks in the fleets that are customers of Vnomics, and the vocations of the sampled vehicles are not known. The observed fleet therefore may or may not be representative of the entire fleet of heavy-duty vehicles in the geographic areas. Observations from the analysis of the datasets in this case study include: • The truck samples in Denver do not exhibit significant seasonal differences in start, soak, and idling activity patterns between the two months studied (January 2017 and May 2017). This implies that these trucks might be doing the same type of operation (revenue service) in both months. • The truck samples in both Denver and Norfolk rarely had trip starts on weekends. Almost one-half of their starts were preceded by a soak period of 6 – 30 minutes. They did not engage in any idling events longer than 3 hours. These characteristics indicate that these trucks perform regional haul type of service. • The truck samples in this case study had much fewer starts per vehicle as compared to those estimated from StreetLight data in Case Study #3. This may be due to the two data sources drawing data from different sets of truck samples in the same areas. Some of the difference may also relate to the time period over which truck population was observed and defined. • Having engine speed data from ECU allows for a determination of whether the truck engine is on, and thus a more accurate estimation of truck activities. The impacts of having engine speed data in the datasets used in this case study include roughly 20 percent more starts and 8 percent more very short soak events (less than 6 minutes).

Guide to Truck Activity Data for Emissions Modeling F-22 • The ratio of engine-on to key-on duration during off-network parking is 0.62 – 0.65 for the three datasets. This factor could be used to estimate the duration of engine-on and the amount of off-network idling based on the key-on duration as calculated from other truck activity datasets with only GPS data. F.7 Transferability MOVES defaults are estimated from truck GPS data collected in 1997 – 1998. The sample data evaluated in this case study as well as Case Study #4 are real-world and more up-to-date. However, all the datasets used are from a relatively small sample size with unknown vocations and may or may not be representative of the truck population in the study areas. Furthermore, the datasets are not large enough to evaluate whether there might be significant differences in activity patterns across different states or regions of the country. Finally, resident vehicle population is difficult to define for the purposes of measuring starts per vehicle.