Click for next page ( 17

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 16
16 3.3 Truck Trip Generation, system (GIS) processing tools, it is possible to match the GPS Distribution, and records of stops and starts by latitude and longitude with Chaining Information these land use records to determine the land use of the truck trip end. Sequential stops can then be grouped using the Trip chaining of commercial trucks, including those mov- anonymous vehicle identifier and the sequence of stops; it may ing freight, requires specialized data not readily available to be possible to determine time between stops or stops per day. apply these methods in common usage. Information on the The information that could be available from providers of nature of truck tours, particularly the number of stops, the GPS data for truck operators is considerable, both in the average impedance between stops (e.g., time), and the nature number of firms involved and the geographies that are cov- of the land use at each stop on the tour can only be established ered. In order to prepare the data to be used in this topic, it though expensive surveys. was necessary to determine which vendors of GPS services to GPS data has been used in a number of passenger surveys truck operators could provide historical information with to collect data on passenger tours to assist in developing pas- sufficient detail to determine basic information concerning senger models.6,7,8 This information has included the deploy- trucking behavior, at a reasonable cost. In order to determine ment of GPS devices in passenger vehicles, by passengers, and if the information collected could be adapted to a number of the processing of that GPS data to determine information geographic settings, it was necessary to choose metropolitan concerning tours, including trip ends, the nature of the land areas with various geographic locations, representing a vari- use at trip ends, the time between trip ends or stops, and the ety of sizes and densities. organization of stops into tours. The use of GPS in these surveys improved the quality of the information collected, increased the response rate, eased the burden of data entry by Selection of GPS Vendors the passengers, and added additional information that could For the purpose of this research topic, the selection of a not otherwise be collected. GPS vendor had to satisfy several criteria. The GPS vendor Although it was necessary to deploy GPS units in these pas- must provide services nationally in order to have similar GPS senger studies, it is possible to collect much of the same infor- data for the variety of metropolitan areas that would be con- mation for truck activities without the need to deploy new sidered. Selection of different GPS vendors for each metropol- GPS units. Many truck fleet operators subscribe to GPS serv- itan area was not practical because of the differing reporting ices provided by vendors. These vendors currently collect and formats and standards that might be used by those vendors, electronically distribute GPS information provided by trucks as well as the administrative effort in acquiring data from var- equipped with units they sell. Although their business model ious sources. The GPS vendor must not only offer GPS track- is to provide GPS information to truck fleet operators, many ing services to truck operators, it also must store and be able vendors store the historical GPS information, and in all cases to provide this information from a historical database. The it would be easy to retrieve these data. It should be possible to truck operators that are the customers of the GPS vendor process the historic GPS information to obtain data to use in must include firms that primarily offer services within the better defining truck trip chaining. metropolitan area, in order to provide the desired informa- GPS data maintained by vendors typically will have an tion on trucking operations within metropolitan areas. The anonymous vehicle identifier, geographic latitude/longitude historical data provided by the GPS vendor must include, at and coordinates, and time. In order to successfully process GPS a minimum, information on GPS events identified by vehicle, information for this purpose, it will be necessary to identify which includes the date, time, and location. Ideally, the sta- stops and starts from truck GPS data, and to identify the land tus of the truck associated with the GPS event (e.g., moving use at these stops and starts (e.g., in the Calgary truck chaining or stopped) also will be available. Of the GPS data vendors model, land uses are classified by five categories: low-density, assessed for this study, it was determined that one best met residential, retail and commercial, industrial, and other high- the research parameters, for the following reasons: density employment). Most metropolitan areas maintain land use maps in the Meets the basics technical requirements (time stamped form of shapefiles. Using standard geographic information latitude/longitude data tracking truck trips and stop/start activities); 6 Bricka, S., and C. R. Bhat, A Comparative Analysis of GPS-Based and Travel Serves mostly metropolitan area short-haul trucks, which Survey-Based Data, Technical Paper, Department of Civil Engineering, Univer- would be those most likely to have frequent daily stops sity of Texas at Austin, July 2005. within the same metro area; 7 NuStats, Kansas City Regional Household Travel Survey: GPS Study Final Report. Provides substantial archived historical data for truck Mid-America Regional Council, Kansas City, 2004. 8 Lawson, T., et al., GPS Pilot Project: Phase VI Final Report, New York Metropol- movements within U.S. metropolitan regions; and itan Transportation Council, May 2009. Is available for a reasonable cost ($12,000 per month).

OCR for page 16
17 Selection of Metropolitan Areas vides its subscribers with a number of product lines based on how often the data is transmitted. In addition to regular trans- The research team selected four pilot cities for GPS data col- mittals of data, the subscriber may query (i.e., "ping") the GPS lection and processing. Although there were no technological unit, which will generate and transmit information. All GPS limitations influencing the number of metropolitan areas that and engine information received from these products is cen- could be investigated, constrained project resources required a trally stored and available for a historical period of 5 years. limited sampling. Metropolitan areas with active truck and/or The information is provided in both XML and CSV file for- freight studies were considered desirable because the data may mats. The primary difference between the file formats is in the have immediate application. To ensure that the data developed manner in which the data are stored and accessed. Some attrib- could be broadly applied, it was desirable to select metropoli- utes are meaningless when the status of the unit is "parked" or tan areas with various sizes, densities, and geographic loca- "stopped" (e.g., heading or speed would have no meaning for tions. Similarly, it was desirable to select metropolitan areas a stopped unit). Similarly, other information is meaningless that have existing data for comparison against the GPS find- when the vehicle is moving (e.g., stop duration is meaningless ings. Finally, the metropolitan area must have a shapefile of for a moving unit). The XML format includes only the data land uses, with attributes of land use coded in a conventional items appropriate for the specified status, defined as format, which could be used in processing the GPS data. Based on these criteria, GPS data were obtained for the following Stop-Not Moving/Engine On; metropolitan areas: Park-Not Moving/Engine Off; Moving-Vehicle in Motion; and Los Angeles, which currently is processing GPS data from Status-0 for moving, 1 for short stop, 2 for medium stop, October 2008 whose results could be compared to this 3 for long stop. study. This is a large metropolitan area, with low-density land use development. It has an active freight study that For determining information about truck stops, records may be able to utilize any data developed. It has a land use with a status of "Moving" or "Status-0" need not be processed. shapefile available that could be used to process the data. For this study, GPS records were acquired for the month of Chicago, which has active freight and truck studies. This is September 2009. Latitude and longitude boxes were defined a large metropolitan area with high-density land use devel- to encompass the areas covered by the metropolitan area land opment. It has a land use shapefile available that could be use shapefiles. GPS records falling outside of this area were used to process the data. dropped. All remaining records were processed and sorted to Phoenix, which has active freight and truck studies. It provide the required information by metropolitan area (and has recently completed a commercial vehicle survey that land use). GPS records for Saturdays, Sundays, and the Labor included the development of a land-use-to-land-use inter- Day holiday (September 7, 2009) were excluded in the calcu- change matrix. It is a mid-sized metropolitan area with lation of average weekday truck information. The remaining low-density land use development. It has a land use shape- records were processed to produce the following information: file available that could be used to process the data. Baltimore, which has active freight and truck studies. It is Number of trucks--Number of unique GPS IDs. a mid-sized metropolitan area with low-density land use Number of GPS events--Transmittals that trigger a GPS development. It has a land use shapefile available that could event. be used to process the data. Number of stops--Number of GPS event records exclud- ing moving and maintenance records. In a chain of trips by Although these four cities were selected, for preliminary the same vehicle, a stop is both the destination of one trip research, duplication of the data collection and processing and the origin of the next trip. techniques described below could be completed by other Number of stops per truck per day--Number of stops metropolitan areas relatively easily with minimal resources. divided by the number of trucks, adjusted by the operating days of the trucks. Airline distance to next stop--From records in time GPS File sequence sorted by ID and by date. Records include fields The GPS vendor provides data feeds of its subscribers for a indicating event time. Travel time is the difference between user-specified period of time (typically in one-month incre- event times for a given stop and the next stop in sequence ments, within user-specified bounding "boxes" of latitude and for that same GPS unit (truck). For information by land longitude). These GPS products use wireless technologies to use stop, the next stop need not be of the same land use. transmit GPS location and engine condition information to its Airline distance to next stop--From records in time central locations for transmittal to its subscribers, and pro- sequence sorted by ID and by date. Records include fields

OCR for page 16
18 indicating event time. Airline distance is the difference, distance between stops is the difference between cumula- expressed in miles as the great circle distance between the tive mileage for a given stop and the next stop in sequence latitude and longitude of a given stop and the latitude and for that same GPS unit (truck). For information by land longitude of the next stop in the sequence for that vehicle on use stop, the next stop need not be of the same land use. that day for the GPS unit (truck). For information by land use stop, the next stop need not be of the same land use. For the four pilot cities, processed data for GPS events and Mileage to next stop--From records in time sequence stops are included in Table 3.1, and for distances in Table 3.2 sorted by ID and by date. Records include fields indicating (number of event and number of stops). Although not pre- cumulative vehicle mileage (odometer reading). Highway sented, these data could be processed to determine additional Table 3.1. GPS-derived truck characteristics. Percent Number Number Number of Number of Origins of Origins per Metro Area Land Use of Trucks GPS Events of Origins by Land Use Truck per day Los Angeles Total 6,901 3,926,611 853,049 100% 9.05 Industrial 5,702 640,084 202,187 24% 3.32 Low density 4,631 230,703 44,876 5% 2.10 Other high-density 5,830 1,420,470 164,858 19% 3.73 employment Residential 4,919 773,228 176,728 21% 3.56 Retail and 6,083 862,126 264,400 31% 3.92 commercial Chicago Total 3,290 1,955,033 432,311 100% 10.59 Industrial 2,730 357,130 116,749 27% 4.38 Low density 2,441 241,271 39,584 9% 2.28 Other high-density 2,650 554,915 77,209 18% 4.14 employment Residential 2,298 348,463 76,076 18% 3.29 Retail and 2,888 453,254 122,693 28% 3.97 commercial Baltimore Total 2,797 1,044,132 258,578 100% 8.61 Industrial 1,894 186,544 52,747 20% 3.24 Low density 2,058 273,476 36,396 14% 2.53 Other high-density 1,310 59,669 18,996 7% 2.75 employment Residential 1,917 287,941 78,102 30% 3.84 Retail and 2,343 236,502 78,937 28% 4.25 commercial Phoenix Total 2,851 1,491,659 436,758 100% 10.80 Industrial 2,446 179,345 54,718 13% 0.90 Low density 2,554 409,615 92,673 21% 4.33 Other high-density 2,163 146,677 43,062 10% 0.97 employment Residential 2,258 418,321 135,281 31% 2.72 Retail and 2,693 337,701 111,024 25% 3.49 commercial

OCR for page 16
19 Table 3.2. GPS-derived average trip characteristics. Airline Travel Time Distance Mileage to Next Stop to Next Stop to Next Stop Circuity Metro Area Land Use (Minutes) (Miles) (Miles) Ratio Los Angeles Total 43.61 7.44 11.27 1.51 Industrial 54.59 8.71 13.19 1.51 Low density 43.36 8.26 12.62 1.53 Other high-density employment 41.11 8.56 12.75 1.49 Residential 37.96 4.88 7.47 1.53 Retail and commercial 41.27 7.58 11.48 1.51 Chicago Total 40.90 5.73 10.82 1.89 Industrial 44.81 5.97 13.37 1.88 Low density 43.28 7.40 13.92 2.06 Other high-density employment 36.77 5.38 11.11 1.52 Residential 43.07 4.73 7.18 1.68 Retail and commercial 37.80 5.85 9.86 2.24 Baltimore Total 37.92 4.64 9.53 2.05 Industrial 44.16 4.97 10.27 2.07 Low density 37.84 5.13 9.22 1.80 Other high-density employment 36.43 3.87 6.57 1.70 Residential 37.43 4.14 11.03 2.67 Retail and commercial 34.73 4.93 8.32 1.69 Phoenix Total 40.61 4.72 7.73 1.64 Industrial 54.46 6.09 10.95 1.80 Low density 40.08 5.26 8.45 1.61 Other high-density employment 44.80 4.66 7.01 1.50 Residential 34.14 3.24 4.99 1.54 Retail and commercial 40.10 5.30 8.92 1.68 information, such as median, standard deviation, and distri- of truck activities in that metropolitan area, but may indicate bution around the median. Similarly, in addition to calculat- that the subscribers of GPS services in these metropolis areas ing information for average stops, the same information can represent a different mix of fleets with different truck activity. be calculated by stop sequence (1st, 2nd, 3rd, etc.). The abil- Although the number of events by land use do add to the ity to develop this information may assist in developing metropolitan totals, the truck rates by land use are not additive. chaining and/or distribution models. This is due to the fact that some truck trips may begin in one As can be seen in Table 3.1, the GPS data provide a tremen- land use type and end in another land use type. The land use dous number of records on events. The number of records activity reported by the GPS data varies by metropolitan area. processed ranged from more than 1 million records in Balti- Stops by truck in industrial land uses are 24 percent of the total more to more than 3.9 million in Los Angeles. A large number in Los Angeles, 27 percent in Chicago, 20 percent in Baltimore, of these records were GPS events in which the vehicle was mov- and 13 percent in Phoenix. Stops by trucks in retail and com- ing. For this research, only records in which the vehicle was mercial land uses are 31 percent of the total in Los Angeles, stopped are of interest. The number of records that are stop 28 percent in Chicago, 28 percent in Baltimore, and 25 percent events, where a trip originates, ranges from more than 258,000 in Phoenix. Stops by trucks in residential land uses are 21 per- in Baltimore to more than 853,000 in Los Angeles. The num- cent of the total in Los Angeles, 18 percent in Chicago, 30 per- ber of origins per truck per day of operation ranged from a low cent in Baltimore, and 31 percent in Phoenix. This could of 8.6 origins per day in Baltimore to a high of 10.8 origins per represent different patterns of truck usage in these areas, but day in Phoenix. The differences could indicate different levels more likely reflects the bias of different fleets of trucks that

OCR for page 16
20 make up the GPS vendor's customer base in these metropoli- which the GPS stop event was located, to the activity serving tan areas. the next stop for the same vehicle was determined by examin- ing and processing the GPS records. The results of the inter- changes of individual truck trips, based on the land use activity Land Use Interchange Matrix at the originating stop and the terminating stop, are shown in Trip chaining recognizes that the probability of making a Table 3.3. truck trip in a tour depends both on the type of activity the The percentage of truck trips by the land use in interchange truck is serving at its current stop, as well as the type of activ- at the origin and the destination of trips in Table 3.3 total to ity at the next stop. For example, a truck that is currently 100 percent within each metropolitan area. Even on this stopped at a manufacturing facility might be expected to make basis, the tables show similar patterns of interchanges. If the its next stop at another manufacturing facility. This informa- cells are weighted based on the total trips to or from that land tion could be used to weight the attractiveness of truck trip use pattern, the data appear even more consistent. In each distribution for individual trips, and to organize these trips case, as would be expected, the activity within a land use (e.g., into chains (tours). The data on the probability of making a trips with a manufacturing land use as the origin and a man- truck trip from one activity, as determined by the land use in ufacturing land use as the destination) is the highest value Table 3.3. GPS-derived land use interchange matrix. Los Angeles Destination Other Retail High-Density and Industrial Employment Commercial Residential Low Density 14.80% 2.20% 4.00% 1.70% 0.50% Industrial 64.00% 9.50% 17.20% 7.20% 2.20% 63.30% 12.40% 13.00% 7.10% 10.10% Other 2.40% 10.80% 2.90% 1.30% 0.50% High-Density 13.20% 60.60% 16.20% 7.10% 2.90% Employment 10.10% 61.30% 5.50% 5.50% 10.20% 4.10% 1.00% 18.40% 5.10% 1.00% Origin Retail 13.10% 3.00% 58.40% 16.20% 3.00% and Commercial 17.60% 19.00% 60.20% 21.80% 19.00% 1.50% 1.20% 4.40% 14.50% 0.70% 6.90% 5.20% 19.60% 65.10% 3.30% Residential 6.60% 6.60% 14.30% 62.20% 14.80% 0.60% 0.50% 0.90% 0.80% 2.30% Low-Density 11.30% 10.20% 17.70% 15.60% 45.20% 2.50% 2.90% 2.90% 3.40% 45.90% Chicago Destination Other Retail High-Density and Industrial Employment Commercial Residential Low Density 15.30% 2.10% 3.50% 1.50% 1.20% Industrial 64.80% 8.90% 15.00% 6.40% 4.90% 64.30% 13.60% 12.60% 6.30% 13.70% Other 2.20% 9.10% 2.40% 1.10% 0.90% High-Density 13.90% 57.70% 15.40% 7.20% 5.90% Employment 9.20% 59.00% 4.70% 4.70% 10.90% 3.70% 2.40% 15.60% 5.40% 1.70% Origin Retail 12.80% 8.40% 54.10% 18.70% 5.90% and Commercial 15.60% 15.70% 55.20% 22.40% 20.10% 1.50% 1.00% 4.90% 14.70% 1.20% 6.90% 4.40% 21.00% 63.50% 5.30% Residential 6.60% 6.60% 17.20% 61.00% 14.50% 1.30% 0.80% 1.80% 1.30% 3.50% Low-Density 14.80% 9.30% 20.60% 15.50% 39.80% 5.40% 5.30% 6.40% 5.60% 40.80%

OCR for page 16
21 Table 3.3. (Continued). Baltimore Destination Other Retail High-Density and Industrial Employment Commercial Residential Low Density 11.90% 0.80% 3.20% 2.30% 1.60% Industrial 60.40% 3.80% 16.10% 11.40% 8.20% 60.60% 11.00% 11.90% 6.80% 12.30% Other 0.80% 2.90% 1.40% 1.20% 0.60% High-Density 10.80% 42.10% 20.30% 17.70% 9.00% Employment 3.80% 42.70% 3.70% 3.70% 4.80% 3.20% 1.40% 14.30% 6.50% 2.60% Origin Retail and 11.30% 5.10% 51.00% 23.20% 9.30% Commercial 16.10% 20.80% 53.00% 19.50% 19.80% 1.50% 1.10% 5.60% 20.50% 2.70% 6.90% 3.50% 17.50% 64.20% 8.40% Residential 6.60% 16.30% 20.70% 61.40% 20.30% 1.80% 0.60% 2.50% 2.90% 5.70% Low Density 13.10% 4.70% 18.50% 21.60% 42.10% 9.00% 9.20% 9.20% 8.70% 42.80% Phoenix Destination Other Retail High-Density and Industrial Employment Commercial Residential Low Density 6.30% 0.60% 2.70% 1.10% 1.40% Industrial 52.20% 5.30% 22.00% 9.40% 11.10% 52.60% 6.60% 10.80% 3.50% 6.60% Other 0.60% 4.00% 2.00% 1.70% 1.10% High-Density 6.70% 42.50% 21.50% 17.60% 11.60% Employment 5.20% 41.50% 5.10% 5.10% 5.40% 2.50% 2.10% 12.40% 5.40% 2.90% Origin Retail and 10.10% 8.20% 48.80% 21.40% 11.50% Commercial 21.10% 21.40% 49.60% 16.50% 14.20% 1.50% 1.80% 5.10% 21.20% 3.20% 6.90% 5.50% 15.60% 65.50% 10.00% Residential 6.60% 18.40% 20.30% 64.50% 15.90% 1.40% 1.20% 2.80% 3.50% 11.80% Low Density 6.90% 5.70% 13.40% 16.70% 57.20% 11.90% 12.10% 11.20% 10.50% 57.90% Note: For each cell in the table: the first value, shown in bold, is the percent of the total table, the second value, shown in italic, is the percent of the origin, and the third value, shown in regular type, is the percent of the destination. within origins and destinations, and in all but a few cases characteristics of trips between these stops. In addition to (low-density interchanges in Los Angeles, Chicago, and Bal- being able to identify the land use at the destination for trips timore, and other high-density employment in Baltimore and from a given land use origin, the GPS information can be Phoenix) it is the majority of trips to or from that land use. used to estimate the travel time and distances between stops Even when those intra land use exchanges are not the major- in the chain. The averages of these times and distances in total ity of truck trips, they are still the highest percentage. also can be used to develop friction factors for truck trip dis- tribution models. This same information by land use can be used to develop friction factors between specific types of land Trip Characteristics uses that might be used in trip chaining. The distance for a As discussed previously, trip chaining recognizes that the trip was calculated both as the airline distance between the probability of making a truck trip in a tour depends both on latitudes and longitudes reported for the GPS records, as well the type of activity the truck is serving at its current stop and as the difference in odometer readings reported by these GPS the type of activity at the next stop. It also depends on the records. It is worth noting that the GPS odometer reading