While evaluating the magnitude and distribution of commercial vehicles and identifying methods, parameters and data sources that can be used to forecast these vehicles in urban transportation planning models, the project team has uncovered a number of areas for future data development. These areas are determined primarily by the gaps in the data and the need for data sources to support the development of advanced commercial vehicle models. Once the data development has been undertaken, research and evaluation of these new data sources can provide a more robust evaluation of the methods for estimating commercial vehicle travel for urban transportation planning models.
The team's recommendations for future data development are designed to support the three methods of estimating commercial vehicle travel in urban area planning models: Aggregate Demand, Network-based Quick Response, and Model Estimation techniques. Future research should focus on collecting data for specific commercial vehicle categories, based on a level of priority. The priority levels are determined based on the overall impact on VMT of a particular vehicle category and the current availability of data (or lack thereof) for that vehicle category. The recommendations are designed to support the development of traditional four-step transportation planning models and/or tour-based transportation planning models. Tour-based modeling techniques should be considered when evaluating Model Estimation techniques, but are considered to be beyond the scope of this initial work effort.
The remainder of this section outlines recommendations on the identification of vehicle by type, a discussion of establishment surveys for each vehicle category, limitations for forecasting and priorities for data to support individual vehicle categories. All of the recommendations are designed to support the evaluation of commercial vehicle travel from a national perspective and would support the development of national default model parameters for commercial vehicle models. The discussion of issues related to establishment surveys and forecasting also apply to the evaluation of commercial vehicle travel from a local perspective and would support the development of local model parameters for commercial vehicle models.
Based on the definition of commercial vehicles established for this study (all vehicles being used for commercial purposes), it is very difficult to clearly identify personal and commercial vehicle travel using existing datasets. Vehicle registration data, when augmented by significant levels of additional processing, provides the best available identification of personal and commercial vehicle fleet sizes, based on whether the vehicle is registered for personal or commercial use. However, in the Business and Personal Services category, a personal vehicle may be used for commercial purposes (for example, real estate agents and door-to-door salesmen) and a commercial vehicle may be used for personal use (for example, a construction worker using the company truck after hours). The VIUS shows that 88 percent of business and personal services vans, pickup trucks, and sport utility vehicles are being used for personal use.
Household and establishment surveys do not typically classify vehicles as personal or commercial vehicles. It is assumed that all vehicles in a household survey are personal and that all vehicles in an establishment survey are commercial. It may be desirable to modify establishment surveys to specifically determine whether the vehicle is being used for personal or commercial purposes.
Contrary to popular belief, many commercial vehicles are automobiles and buses rather than trucks. The vehicle registration databases provide a distribution of vehicles by type for each commercial vehicle category in this study, but these databases are not currently available in most states. In order to compare commercial vehicle models with ground counts, it will be necessary to classify all commercial vehicles by type (autos, buses and trucks) and preferably to separate trucks by weight (light, medium and heavy duty). Most commercial vehicle models to date are aimed at quantifying trucks and therefore do not include commercial vehicles moving people and public sector commercial vehicles providing services. These vehicles are primarily autos, vans, and buses.
Further research is needed to design and collect surveys to support the estimation of commercial vehicles by category using advanced modeling methods. Seven types of surveys are identified, in order of priority for supporting the model development modeling methods (based on the percent of total VMT contributions from each vehicle type). All are establishment surveys and are intended to include the commercial vehicles used by both public and private establishments.
The surveys would collect data on commercial vehicle movements, including the type and size of vehicle, number of trips, purpose of the trip, products carried, service provided, and origin and destination characteristics. The following industries would require surveys and these are presented in order of priority:
The surveys could be standardized or adapted for unique types of establishments, although the project team recommends standardizing these surveys as much as possible to improve the usefulness for model development. All of the surveys should include a complete day's travel diary information for a sample of vehicles in the establishment. In the transportation industry, it will be important specifically to identify shuttle services, taxis, paratransit, and rental car companies to ensure that enough samples in these categories are represented. In all industries, it will be important to design the survey to capture all 12 categories of commercial vehicles.
The current proposed methods for forecasting commercial vehicle travel are necessarily limited by the expected exogenous forecast data that would be available to a metropolitan planning organization. These methods could be expanded to provide more accurate assessment of future commercial vehicle travel as these future data sources become available. For example, school bus travel could be a function of the geographic coverage of the school district and enrollment, but school districts change over time to accommodate growth; these changes are not currently part of the regional forecasting activities.
The current proposed methods for forecasting commercial vehicle travel also are unable to estimate micro-level transportation impacts. These impacts can be estimated for specific categories of commercial vehicle travel from the network-based approaches identified in the Network-based Quick Response Method or model development modeling techniques, but significant effort will be needed if these techniques were developed only for a micro-level analysis (and not for a macro- or meso-level analysis). Specific techniques to estimate micro-level transportation impacts would be more appropriate, but should be developed after the regional-level impacts are better understood.
All of the vehicle categories have specific future data needs that were identified and prioritized during the course of the evaluation. These data need priorities are described in the following sections. Priorities are based on the overall impact that a particular vehicle category has on VMT (higher VMT categories receive a high priority) and on the availability of data for the category (categories will little or no data receive a higher priority). The priorities by category are summarized below:
Interestingly, the vehicle categories in the greatest need of better data sources (business and personal services, rental cars, public service, safety and utility vehicles) are also those that rank high- or medium-priority based on their overall impact on VMT.
All of the recommended methods for estimating commercial vehicles in urban transportation planning models (Aggregate Demand analysis, Network-based Quick Response Method, and Model Estimation Methods) use existing model forms and are not expected to require any future research to support these efforts. Other methods, such as activity-based or tour-based models, would advance the methods proposed in this project, but are beyond the scope of this initial effort, which was primarily aimed at developing Network-based Quick Response Methods that could be adapted or transferred by metropolitan planning agencies. Tour-based models should be considered during the development of any locally specific models identified as Model Estimation Methods in this report; these could be developed using the same data recommended here to support the Model Estimation Methods.
Retail and service surveys of establishments are needed to obtain information regarding fleet size, number of trips, origins and destinations, departure times, and mileages for business and personal services. It also would be useful to collect data on how vehicles are registered, to allow for a better interpretation of vehicle registration data for vehicles used for business and personal services (i.e., those that are registered as commercial but used for personal activities and those that are registered as personal but used for commercial activities).
Urban freight is the movement of freight within the urban area, but does not include trips to, from, and traveling through the area. Current commercial vehicle surveys collect comprehensive data on this category of commercial vehicle trips. Urban freight trips are more likely to be short-haul trips made using medium- and light-duty vehicles. Methods for modeling urban freight vehicles in urban transportation planning models are currently based on the Quick Response Freight Manual. However, the trip rates developed in the Manual represent all goods movement and service-related commercial trucks in an urban area and these trucks have very different behavior. In addition, trucks include all freight, of which urban freight is only part of the total. What is referred to as freight in the Manual includes long-haul trips that are primarily heavy vehicles, which are often modeled using national and international commodity flow data.
Comprehensive business surveys should be conducted to determine the fleet size and composition of vehicles maintained at an establishment. Businesses that are likely to be urban freight generators or destinations such as warehouses, distribution centers, and manufacturing employers need to be identified and contacted for follow-up surveys. Follow-up surveys should determine vehicle fleet size and composition and include trip activity diaries with numbers of trips, products carried, origins and destinations, mileages, time of travel, and costs.
Among commercial vehicles, rental cars have the largest impact on VMT, accounting for two percent of total VMT nationwide. However, very limited data are available on rental cars. The project team could find no specific studies related to rental car demand, fleet size, or VMT analysis.
The only comprehensive source of data for rental cars is the augmented California DMV database. The California Energy Commission processed California DMV datasets and identified the rental cars from the master list of rental companies. Processing DMV data and extracting information about rental cars is very expensive and time-consuming, and there is no indication that other states have done this kind of processing.
Rental car trips can be divided into three types: 1) business-related trips, 2) recreational car trips, and 3) home-based trips. Most business trips start at the airport and go to business districts or employment centers. Business-related rental car trips can be estimated from airport surveys. Recreational rental car trips usually start from airports or car rental centers and go to tourist or recreational areas. In order to more accurately predict the impact of rental cars on the urban transportation network, data on rental cars also is needed. These data can be collected by surveying rental car companies, which collect information on miles traveled, number of vehicles, and cost, and by asking rental car drivers to provide information on number of trips, trip purpose, origins and destinations, and time of travel.
The primary data source for public service vehicles is the augmented Department of Motor Vehicles registration database. These data are only useful to identify fleet size and do not contain any data on miles traveled, origins and destinations, time of travel, and cost. Some states maintain miles traveled as part of their registration database. Data on number of trips, origins and destinations, services provided, departure times, and costs should be collected using an establishment survey of public service agencies. These data are necessary to support both the Network-based Quick Response Method and Model Estimation Methods of estimating commercial vehicle models.
Trip data for construction transport vehicles are adequately collected by existing commercial vehicle surveys. Vehicle fleet data are available through the commercial vehicle surveys and for select cities in California from the California DMV. Other than these data sources, data are limited for construction transport vehicles.
Similar to urban freight vehicles, construction transport vehicle research should begin with a comprehensive business survey. From this survey, firms that produce or maintain construction materials and/or equipment or have fleets of transport vehicles can be identified. Travel diaries sent to these firms can be used to determine the characteristics of the trips.
A business survey also will identify construction companies. These companies, like suppliers of construction material and equipment, can provide fleet information and trip characteristics, as well as data regarding construction sites. Details from a construction site survey could include the type of project, the length of project, the number of workers on site, and the number of shipments received at site.
A construction site classification should be developed. Construction project types are as numerous as they can be different from each other. For example, trip rates associated with housing construction, office building construction, and highway reconstruction all would be expected to vary. Composite trip rates could be developed from the different classifications for all construction site types.
The primary data source for safety vehicles is the augmented Department of Motor Vehicles registration database. These data are only useful to identify fleet size and do not contain any data on miles traveled, origins and destinations, time of travel, and cost. Some states maintain miles traveled as part of their registration database.
Travel data for safety service vehicles are currently limited to data collected in one commercial survey (Detroit) that included only snow plows and tow trucks. Similar travel diary data on police, fire, and rescue vehicles should be included in these surveys. This will require administering travel diary surveys to both public and private establishments for a comprehensive assessment of the number of trips, origins and destinations, services provided, departure times, and costs. If these safety vehicles are estimated as a separate category, we recommend that data on average annual or daily mileage be collected and used to estimate VMT rather than collecting travel survey data, but if these vehicles are combined into a commercial services category, it would be important to include travel surveys for safety vehicles in this group.
The primary data source for utility vehicles is the augmented Department of Motor Vehicles registration database, which contains both public and private utility vehicles. According to these data, 24 percent of utility vehicles are publicly owned. Unfortunately, the registration data are only useful to identify fleet size and do not contain the necessary travel information found in commercial vehicle surveys.
Data for public package, product, and mail delivery are maintained by the USPS. Data on routes, miles traveled, time of travel, and number of vehicles should be available in their entirety for accounting for public delivery vehicles in urban models through the Freedom of Information Act of 1966. Data that are normally withheld for privacy reasons should be obtainable due to the aggregate level of detail used in travel demand models.
Fleet size, fleet composition, and VMT by vehicle type for all urban areas should be obtained from the USPS. Special attention should be given to urban areas which contain major distribution centers. Population, household, and employment data are maintained at the five-digit and three-digit ZIP code levels by the U.S. Census. These data are available through the Census web site (http://www.census.gov/) and should be obtained for all urban areas that USPS data are available. From these data, statistically valid public package, product, and mail delivery Network-based Quick Response Method procedures can be estimated for urban areas. With this vast database, procedures for specific urban area sizes and geographic locations could be developed, as well as more advanced modeling procedures.
Data on private package, product, and mail delivery are only available through a limited number of existing commercial vehicle surveys. More specific and complete data are therefore required to develop better models. Large service providers such as United Parcel Service and FedEx maintain detailed information similar to the USPS, but these data have historically been very difficult to obtain. Travel diary surveys of vehicles in package, product, and mail firms should be conducted to collect data on number of trips, mileages, origins and destinations, time of travel, and travel costs.
The team reviewed three commercial surveys that contained the necessary travel data for private utility vehicles. If these surveys were administered to public and private establishments, then a comprehensive assessment of the number of trips, origins and destinations, services provided, departure times, and costs would allow evaluation of the Model Estimation Methods for utility service vehicles.
Currently, the Taxi Fact Book is the only source of data on taxi and limousine services. These data are robust and provide useful information for the Network-based Quick Response Methods, but do not contain data on tours that individual taxis make. This makes it difficult to differentiate between taxi travel behavior in center cities and suburban areas, where the mileages are expected to be very different.
The Model Estimation Method for estimating taxi trips would be based on travel diary surveys of taxi companies. These models would estimate taxi trips (both empty and full taxi trips) as necessary components of taxi-related travel. Trips would be based on locations where taxis pick up passengers, such as hotels, employment centers, and airports, and on locations where taxis drop off passengers, such as residences, hotels, and employment centers. The travel diary surveys would provide data on number of trips, average mileage, number of passengers, travel cost, and time period of travel to support these elements of the model.
The existing data for school buses (http://schoolbusfleet.com/t_home.cfm?CFID=5754288&CFTOKEN=24981657 from School Bus Fleet magazine) cover public school buses for the 100 largest school districts only, and include no travel diary information. These data are appropriate for Aggregate Demand analysis, but do not provide the necessary trip information for the Network-based Quick Response Method or model development modeling techniques.
School bus trips fall into two categories: trips made to transport students between home and school and school-based other trips. Home school trips follow fixed routes, beginning and ending at school bus garaging sites. School-based other trips begin and end at school and go to various recreational or educational destinations.
In urban models, school bus travel patterns are similar to public transit systems and more complicated to code than other vehicle patterns. The standard four-step modeling procedures for estimating trip generation, distribution, and assignment cannot estimate all aspects of school bus patterns accurately. In their pick-up mode, school buses travel in a loop starting from a garaging facility, making a series of pick-up stops at different residential locations, and then proceeding to school to drop off the students. Next, the same bus may repeat this pick-up/drop-off process for a different school before returning to a garaging facility. They travel mostly on local and collector roads on regularly scheduled routes and cannot detour to avoid congestion. Most automobiles, in contrast, follow a path well approximated by an equilibrium assignment not suitable for school buses.
Further research is needed to develop either the quick-response or an model development procedure to include school buses in urban models, but the Aggregate Demand Method is considered adequate for estimating school buses, given their low overall impact on the transportation system. There are two other options for including school buses in urban models. The first is to develop a trip chain model which will consider school bus trips as a loop from a garaging facility or school to different residential locations and then back to the same or a different school. For school-based other trips, it will be necessary to collect data from school districts and to estimate trips per student, trips per teacher, or any other variables. This method would require travel surveys of school districts to identify the travel patterns of home school and school-based bus trips. Travel times for buses could be estimated using GIS techniques, rather than by building school bus networks of all routes and schedules. Even this method would only be able to construct generalized school bus tours. The actual school bus routes could be established based on the business practices of the school districts (opening and closing hours of various schools, eligibility requirements by distance from school, fee for service, etc.) which are beyond the scope of any model to consider.
Currently, the Airport Ground Access Planning Guide is the only source of data for shuttle services. Even this source is limited because it considers data from only 29 airports. No shuttle services outside of airports were considered and no data sources distinguished between fixed schedule and demand response shuttle services.
In order to more accurately predict the impact of shuttle services on the urban transportation network, agencies should obtain information on population and the number of tourists and hotel rooms in the area. In addition, agencies should have an accurate count of the total number of passengers at airports and other stations where shuttle services are provided. Agencies also should obtain information on the number of businesses providing shuttle services, the size of their fleets, their routes, the number of trips they make annually, and the average length of each trip.
In addition to the data requirements detailed above, it is recommended that agencies conduct travel diary surveys of a sample of shuttle service providers. This sample can be extrapolated to the expanded set of shuttle service vehicles to determine the trips for shuttle services. Once trips have been estimated, trip distribution can be estimated and shuttle service vehicles assigned to the transportation network using the same survey data.
There is clearly a correlation between shuttle services, taxis and rental cars within any urban area, where the demand for shuttle services and taxis will be in part based on the cost and need for rental cars. In cities where parking costs are high, transit services are good and the airport is located far from the central business district (such as New York), the taxi and shuttle services are more widely used than rental cars. In cities where the parking costs are low, transit services are not good and the airport is located closer to the central business district (such as Orlando, Tampa, Palm Springs, and San Jose), then rental cars are used more widely than taxis and shuttle services. We observed this correlation by reviewing the Airport Ground Access Planning Guide for mode splits to and from 29 airports, but expect that the correlation extends to other parts of the urban area as well.
Since the mid-1980s, the number of paratransit and social service systems across the United States has increased significantly. However, paratransit contributes only 0.006 percent of total VMT, which is insignificant compared to other categories of commercial vehicles.
The Federal Transit Administration (FTA) collects and disseminates data on the state of mass transportation via the National Transit Database (NTD) program. Over 600 of the nation's transportation providers submit data to the NTD annually. However, the FTA data include only those systems that receive FTA funds and are therefore required to report their data to FTA. Paratransit systems that do not receive FTA funds, such as church service buses, are not required to submit their data but may do so voluntarily.
The NTD database includes the number of paratransit vehicles, vehicle miles, vehicle hours, passenger miles, and passenger trip information. This is sufficient for estimating parameters for some of the Network-based Quick Response Method parameters, but is not sufficient for estimating parameters for the Model Estimation Methods. Based on this analysis, paratransit trip attractions are highly correlated to population over the age of 60 and total employment, which can be estimated from the NTD and the Census Bureau database. However, trips cannot be estimated using these data. The project team tried to correlate trips and employment, but the statistical fit (based on the R-square value) was not satisfactory. For estimating paratransit trips, data are needed on the destinations of paratransit users, such as specific types of employment (for example, government, education, and medical services). These parameters could be developed with travel surveys of paratransit vehicles, which identify number of trips, origins and destinations, trip departure times, and travel costs. It is possible that paratransit trips could be modeled using data from existing travel surveys, since most paratransit users are residents, but these samples may be too small or limited due to the fact that household surveys do not include group quarters, such as retirement or group homes.
The future research and data development needs focus on filling the data gaps identified in the research on accounting for commercial vehicles in urban transportation models. While a significant amount of data was identified to support the development of model parameters for commercial vehicle in urban transportation planning models (that is planned in Phase II), it is also clear that a more consistent methodology for collecting data across vehicle categories would greatly enhance the accuracy and reliability of models developed from these data. Existing commercial vehicle surveys are very limited in understanding the travel behavior of commercial vehicles and exclude a significant number of commercial vehicle categories, but do contribute to our ability to understand simple trip characteristics of available commercial vehicle categories.
