The methods, variables, parameters, and data sources used for estimating commercial vehicle travel must be related to the appropriate level of planning application and the resources that are available to a metropolitan planning organization (MPO). The project team identified numerous sets of commercial vehicle forecasting approaches that could be used to evaluate different categories of vehicle types and meet MPOs' varying needs and levels of travel model complexity. These methods fall into three categories ranging from simple applications requiring limited data inputs to more advanced techniques to produce more spatial and temporal detail for a more specific range of commercial vehicle types:
MPOs analyze various types of performance measures, including mobility, safety, reliability, and environmental impacts. These analyses are performed at the regional or county level, at the corridor or traffic analysis zone (TAZ) level, and at the intersection or link level, depending on the purpose of the analyses they carry out. To meet MPOs' varying needs, the project team divided the analyses levels into the following three groups:
Both the Aggregate Demand and the Network-based Quick Response Methods have been developed to address the macro analyses, with the difference being that Network-based Quick Response Methods would be more useful for commercial vehicle categories that have larger impacts on congestion or air quality. The Model Estimation Methods are primarily for the meso level of analysis because it is more useful to use local data when considering subarea or corridor-level analyses. This report does not present a method to address the micro level of analyses, leaving this subject instead for future research.
Figure 6.1 presents the modeling process for the aggregate regional VMT, network-based quick response, and Model Estimation Methods. This figure shows, in the form of a flow chart, the steps necessary to estimate impacts for commercial vehicles in urban transportation models. The Aggregate Demand Method provides a means to estimate regional impacts using nationally derived parameters and regional demographic estimates. The network-based quick response and Model Estimation Methods are both based on applying four-step planning modeling techniques to estimate trips, origins and destinations, time periods, and volumes for commercial vehicles traveling in an urban area.
Figure 6.1
Modeling Process for Three Methods to Estimate Commercial Vehicle Travel in Urban Transportation Models
At the simplest level, MPOs may be served best if the fleet size rate by land use, employment, or any other readily available data can be provided, along with the miles traveled per vehicle per day for each category of commercial vehicle. These rates and miles traveled can be used to estimate VMT. They also can be used for the traditional trip generation and trip distribution steps for estimating commercial vehicle trip tables. The primary advantage of this approach is that it extends the typical commercial vehicle forecasting procedures used by MPOs to a broader range of commercial vehicle and trip types. This technique is primarily applicable at a regional (macro) level of detail.
This is a simple procedure for using national average vehicle rates and miles traveled to project commercial vehicle traffic in the MPO's jurisdiction. It assumes the availability of demographic projections (population, employment, tourists visiting, etc.) for the year under consideration. Using existing data, the number of vehicles for a particular category of commercial vehicle is calculated as follows:
FleetSizec = VehicleRatec x SocioeconomicData
where:
The number of commercial vehicles can be estimated using the above equation. The miles traveled per vehicle per day for commercial vehicle categories are available from a variety of sources, identified in the Task 3 report. Hence, it is possible to estimate the VMT for commercial vehicle categories, as follows:
DailyVMTc = VMTperVehiclec x FleetSizec
where:
Some of the sources of data on mileage for specific commercial vehicle categories report annual vehicle mileage rather than daily vehicle mileage. This method uses these data to estimate daily vehicle miles using the same equation to determine fleet size, but then estimates daily mileage as follows:
DailyVMTc = (AverageAnnualMileagec Operating Daysc) x FleetSizec
where:
| Commercial Vehicle Category | Number of Days per Year | Assumption |
|---|---|---|
| School Bus | 180 | Weekdays from September to June |
| Fixed Shuttle Services | 365 | Every day |
| Private Transportation | 365 | Every day |
| Paratransit | 365 | Every day |
| Rental Cars | 365 | Every day |
| Package, Product and Mail Delivery | 306 | Weekdays and Saturdays, excluding holidays |
| Urban Freight Distribution, Warehouse Deliveries | 306 | Weekdays and Saturdays, excluding holidays |
| Construction Transport | 260 | Weekdays, excluding holidays |
| Safety Vehicles | 365 | Every day |
| Utility Vehicles | 260 | Weekdays, excluding holidays |
| Public Service Vehicles | 260 | Weekdays, excluding holidays |
| Business and Personal Services | 306 | Weekdays and Saturdays, excluding holidays |
The Network-based Quick Response Method applies a simplified four-step planning model where the parameters are derived from national data as default parameters. The method uses national average vehicle rates to develop vehicle trips or tours (depending on the vehicle category) generated by commercial vehicles, distributing these trips using the gravity model method, and assigning these trips to a planning model network to produce VMT. This procedure is applicable for either regional- (macro) or corridor- (meso) level detail, since the data is developed at a TAZ level and applied to a transportation planning network.
For the purposes of this simplified four-step planning model, the project team assumes that commercial vehicle travel does not include trips from outside the region. This assumption is based on the understanding that long-haul movement of commercial vehicles (i.e., tractor-trailers) carrying freight would be estimated using commodity flow forecasting methods separately from this process to estimate commercial vehicles within an urban area. Therefore, the simplified four-step planning model process does not include any external travel. If this process is applied at a corridor level, then trips from the region that pass into, out of, or through the study area must be included as external travel.
The simplified four-step planning model process can be applied by individual vehicle category or by group of commercial vehicles. The parameters, methods, and data sources for these models are described in Section 3.0 by category and by group, respectively. The Quick Response Freight Manual6-1 provides a similar approach for the development of commercial vehicle trips carrying freight, which is the second group of commercial vehicles.
The advanced method applies a simplified four-step planning model where the parameters are derived from local survey data. Model Estimation Methods rely on more detailed data sources than network-based quick response techniques. The surveys required are establishment surveys for specific industries, as described below:
The four-step planning model components may be similar in structure to the network-based quick response techniques, or they may be developed to be more sophisticated if the data supports this. These models are more resource-intensive than the network-based quick response techniques, but provide greater flexibility in terms of capabilities and accuracy for a specific region. Advanced models developed for macro-level analysis can be used for meso-level (corridor) analysis, but would only be appropriate if the regional models already were are developed.
A number of modeling methodologies currently being researched will advance the state-of-the-art for forecasting commercial vehicles. These are described briefly below.
Tour-based models estimate the number of "tours" that an individual commercial vehicle will make from when the vehicle leaves the garage to when it returns to the same garage. A number of individual trips typically comprise each tour. Model estimation requires a tour-based commercial vehicle survey; these are the same type of establishment surveys recommended for use in developing the advanced four-step models. The surveys should include public and private establishments (retail, service, manufacturing, and government). If establishments include movement of people (rental cars, taxis, etc.), then this could cover all commercial vehicles.
Tour-based models are estimated by type of establishment (such as manufacturing and construction, etc.). These establishment models predict the number and types of vehicles (light, medium, heavy), the purpose of each trip on a tour (service, goods, other, return to establishment) and the location of the stops for every trip on a tour. These methods can account for a mixture of vehicles providing service and moving goods as well as empty vehicles returning to the establishment directly. One example of this type of tour-based model was estimated for retail and service delivery vehicles in Calgary.6-2
Supply chain models estimate the supply chain from distributor to warehouses to retailer to buyer. These supply chains can then be converted into the number of commercial vehicles required to support the supply of goods from the distributor to the buyer, including any intermediate storage locations. Supply chain models only represent the movement of goods and possibly services but would not be appropriate to model the movement of people in commercial vehicles. Supply chain models are typically estimated by type of supply chain (just in time, inventory, etc.) and product. One example of this type of supply chain model is the GoodTrip model developed for the City of Groningen, Netherlands.6-3
Integrated models estimate the personal and commercial vehicles from an integrated model of land use and demographics. In this definition, integrated models would include personal travel, commercial travel, and forecasting of households and businesses within a single modeling framework. These models will predict the movement of people, goods, and services by design. These integrated models can predict the demand and supply of each vehicle type - for example, the demand for school buses as a function of the number of children in each household and the supply for school buses as a function of the size and population of the school district. Integrated models also can predict the location and need for new schools as a function of the growth in households and changes in lifestyles (i.e., decisions to have children). One example of this type of integrated model is the Oregon 2nd Generation Land Use Transport Model.6-4
All three methods described in Section 5.0 were applied separately for each of the 12 individual categories of commercial vehicles and documented the results in the Methods, Parameters, and Data sources report.6-5 Methods and data sources were similar for many categories, but there are differences in application based on available data and expectations for certain causal relationships with demographic variables.
The travel behavior characteristics that are described for the Aggregate Demand Method include the fleet size, the vehicle trips, and the vehicle miles traveled for each commercial vehicle category. For each travel behavior characteristic, we provide a description of appropriate methods and in most cases, an estimate of the parameter for these travel behavior characteristics. These estimates are derived from available data, which is frequently limited in sample size, and not recommended to represent a national default value for the parameter estimates. For detailed information about the application of these methods to individual categories of vehicles, please review the Methods, Parameters, and Data sources report.
The travel behavior characteristics that are described for the network-based quick response method include the trips produced and attracted to a traffic analysis zone, the trips distributed between traffic analysis zones, the vehicle type, the time of day, the vehicle occupancy (for commercial vehicles moving people) and the characteristics of trip assignment. Again, we provided a description of the methods and an estimate of the parameter, but with information on the limited sample size available to estimate these parameters.
Following the evaluation of forecasting commercial vehicle travel by vehicle category (12), the same travel behavior characteristics of the factor analysis and network-based quick response methods were evaluated by groups of commercial vehicles. These groups represent aggregations of the commercial vehicle categories based on the primary purpose of the commercial vehicle: to move people, to move goods or to provide services. The methods and parameters developed for these groups of commercial vehicles are intended for use by metropolitan planning organizations that do not need or want to segregate commercial vehicles into 12 categories. It also is intended to work in coordination with the commercial vehicle categories for agencies that require additional detail for some groups but not others. For example, agencies may decide that it is useful to have more accuracy for commercial vehicles moving goods and providing services, but that this additional accuracy is not required for commercial vehicles moving people.
About 2.4 percent of total vehicle miles traveled in urban areas in the United States each year are attributable to vehicles in these five categories. Rental cars, which make up 80 percent of vehicles in the commercial passenger group, account for fully 2.0 percent of total VMT in the United States, while school buses, taxis, and shuttle and paratransit services account for about 0.4 percent of VMT.
The size of the commercial passenger vehicles by category appears to be related to different tradeoffs of service. If the rental car market is higher than average is one city, then the shuttle service and taxi market may be smaller than average, and vice versa. These tradeoffs were apparent in the airport data obtained for the study.6-6 Paratransit and school bus categories are more independent categories based on resident populations.
Current urban transportation models do not include the "commercial passenger vehicle" as a separate trip purpose. However, several metropolitan planning organizations have attempted to include specific categories of these commercial vehicle in their models, primarily based on mode choice patterns. For example: the Las Vegas model is the only one that considers taxis as a separate mode in the mode choice model and assigns them to the highway network; the Tucson and Houston-Galveston models predict school bus travel, but do not assign or evaluate these trips; the San Francisco model includes mode choice for rental cars, taxis, and airport shuttles; the Portland model includes shuttle services and taxis in the mode and destination choice models; and the Sacramento model includes airport trips as a separate trip purpose. In addition, a number of models deal separately with the development of trip tables for taxi trips (sometimes combined with 'truck' trips) and their assignment to the network using procedures akin to the network-based quick response methods. Also, all urban models based on local survey data can be presumed to include rental cars used by residents with all trips made using privately owned passenger cars.
Very little research has focused on paratransit vehicles and no models have been developed to estimate the demand for these trips. Similarly, although rental cars contribute a significant percentage of VMT on U.S. roads, the project team could find no models that estimate the demand for rental cars specifically. A few visitor models (San Francisco, Honolulu, and Las Vegas) predict the mode share of auto trips, but the percentage of these trips attributable to rental cars is not considered.
About 3.5 percent of the total vehicle miles traveled each year in urban areas in the United States are attributable to vehicles in these three categories. Urban freight vehicles alone contribute 2.7 percent of total urban area VMT, while package delivery and construction contribute 0.2 and 0.6 percent, respectively. This category does NOT include the related movement of intercity freight to, from, or through urban areas, which is forecast using other techniques.
Urban transportation models typically include "commercial freight vehicles" in a goods movement model. Some of these goods movement models are vehicle-based truck models (Atlanta, Chicago, San Francisco, Buffalo, and Phoenix), some are commodity-based models (Portland) and some are hybrid models (Seattle and Los Angeles). These truck models include trucks from the commercial service vehicles category as well as intercity freight trucks traveling to, from or through an urban area. Most of these models identify trucks by weight class or type (light-, medium-, and heavy-duty) rather than by purpose (package delivery, urban freight, and construction).
Public service vehicles are publicly owned. Business and personal service vehicles are privately owned. Safety and utility vehicles may be either publicly or privately owned.
About 5.9 percent of the total vehicle miles traveled in the urban areas in the United States each year is attributable to vehicles in these three categories. Business and personal service vehicles alone contribute 3.6 percent of the total VMT in urban areas across the nation, while public service vehicles contribute 1.6 percent of the total VMT and safety and utility vehicles contribute 0.4 percent each.
Urban transportation models currently do not include any commercial service vehicles specifically, although some models have identified a commercial vehicle trip purpose that is based on a fixed factor of personal non-home-based travel. Some truck models also include delivery and service vehicles that are four-tire commercial vehicles, based on the inclusion of these vehicles in the Quick Response Freight Manual.6-7
The average weighted impact of commercial vehicles on VMT is 11.8 percent of total VMT. The impact of individual categories on VMT does not necessarily sum to this average weighted impact of all commercial vehicles on VMT because of the different weights from each category. The range of the impact of VMT by category also is helpful to understand the potential impact, since this identifies the potential range of the impact of commercial vehicles on the transportation system. These data are summarized by vehicle group below:
6-1. Cambridge Systematics, Inc., Quick Response Freight Manual, Final Report, prepared for the U.S. Department of Transportation and the U.S. Environmental Protection Agency, DOT-T-97-10, September 1996.
6-2. Hunt, John Douglas, Stefan, Kevin J., and Abraham, John E., Modeling Retail and Service Delivery Commercial Movement Choice Behavior in Calgary, 10th International Conference on Travel Behavior Research, August 2003.
6-3. Boerkamps, J. and Binsbergen, A., GoodTrip - A New Approach for Modeling and Evaluation of Urban Goods Distribution, Delft University of Technology and the Netherlands Research School for Transport, 2000.
6-4. Oregon Department of Transportation, Oregon Model Improvement Program, http://www.odot.state.or.us/tddtpau/modeling.html, 2002.
6-5. Cambridge Systematics, Inc., Accounting for Commercial Vehicles in Urban Transportation Models: Methods, Parameters, and Data Sources, January 2003.
6-6. Cambridge Systematics, Inc., Accounting for Commercial Vehicles in Urban Transportation Models: Magnitude and Distribution, Appendix D, November 2003.
6-7. Cambridge Systematics, Inc., Quick Response Freight Manual, Final Report, prepared for the U.S. Department of Transportation and the U.S. Environmental Protection Agency, DOT-T-97-10, September 1996.
