Accounting for Commercial Vehicles in Urban Transportation Models

Literature Review

2.0 State of the Practice

2.1 Overview

Current state of the practice for urban transportation models that account for commercial vehicles range from the very general to very specific, but we have found no current example where a model attempted to estimate all types of commercial vehicles. Examples of the state of the practice are described in each of the three primary categories of commercial vehicles as an indication of the types of practice in use rather than as a comprehensive evaluation of all models of this type in use at this time.

2.2 Commercial Vehicles Moving People

There are no specific urban transportation models that include the movement of people in commercial vehicles as a general category of trips, but there are some examples of the inclusion of specific types of commercial vehicles in this category. These examples are described below.

Taxis have been represented as a mode choice in several mode choice models: Houston-Galveston, Cleveland, Las Vegas, and New York. In some cases, residents and non-residents (visitors) have separate models, but if this is not the case, then taxis would only represent trips by residents. The Las Vegas model is the only example that has been fully completed; all other models are under development. In this case, taxi trips are estimated and assigned to the highway network.

There are two different examples of urban models that predict school bus travel, but in both cases, school bus trips are estimated but school buses are not assigned or evaluated. In Tucson, school bus trips are estimated in the mode choice model (JHK & Associates, 1994). In Houston-Galveston, school bus trips will be estimated separately from the four-step model and subtracted from total trips as a post-process (Parsons Brinckerhoff, 1998).

There are at least three examples of airport access models for mode and destination choice of residents and non-residents. In San Francisco, the ACCESS model (Harvey, 1988) including mode choice for rental cars, taxi, and both on-demand and scheduled airport shuttles. In Portland, Cambridge Systematics (1998) estimated commercial vehicles such as airport shuttle services, shared ride shuttle services, and taxis in the mode and destination choice models. In Sacramento (DKS Associates, 2002), modeled airport trips as a separate trip purpose with a unique mode choice model for these trips.

There are also at least two examples of vanpool models. In Chicago, a stated preference survey of travelers was used to determine the variables that impact vanpooling as a modal choice. In Seattle, a vanpool model was developed from a survey of vanpoolers to determine the prior mode that vanpoolers were using. The results of this model was to subtract vanpoolers prior mode from other modes in the travel model and then assign vanpools as a separate class of vehicles in the multi-class assignment.

2.3 Commercial Vehicles Moving Goods

The best source of the current state of the practice for the magnitude of commercial vehicle trips is the NCHRP Synthesis 298 report on Truck Trip Generation Data (Fischer et al., 2001). This report focuses primarily on commercial vehicle trips related to urban freight distribution, but also includes some information on package and mail delivery as well as construction-related transport. The state of the practice is divided into three types of models:

• Vehicle-based truck trip generation rates used in statewide and regional travel demand models generally are estimated based on land-use categories that match up well with employment by industry sectors, corresponding to the data that MPOs typically have and/or forecast. A significant problem with this method is that these categories of land use are very broad, and trip rates vary considerably within these categories from region to region. This particularly is true for service-related trip purposes. Trip generation rates related to service industry employment lumps into a single land-use category a whole range of highly varied trip purposes with very different trip generation characteristics (and vehicle types used for trips). In these models, light trucks are sometimes loosely defined to include all commercial vehicles not included in other truck types (medium and heavy-duty trucks). When this is done, the unique characteristics of light-duty commercial vehicles (such as, rental cars, taxis, and vehicles used in outside sales) are often completely absent in the modeling process. Examples of this type of truck model can be found in Atlanta, Chicago, San Francisco, Buffalo, and Phoenix.
• Commodity-based models generally do not develop truck trip generation rates. Instead, annual commodity tonnage data are converted into daily truck trips using a payload conversion factor. The national Commodity Flow Survey and the TRANSEARCH database developed by Reebie Associates are the most commonly used sources of commodity flow data, and the national Vehicle Inventory and Use Survey (VIUS) and locally collected intercept surveys are the most commonly used sources of payload data. These methods tend to underestimate trips in urban areas because they do not account for trip chaining and local pickup and delivery activities (i.e., light trucks). Hence, commodity-based modeling approaches are likely to be less applicable to this project. Examples of this type of truck model can be found in Portland.
• Hybrid models are combinations of vehicle-based and commodity-based models. They also can include light trucks, loosely defined to include all commercial vehicles (Cambridge Systematics has developed these hybrid models in a number of cities and states). They have the same drawbacks as vehicle-based models in that the commercial vehicles are estimated using very generalist aggregate methods rather than specific data or rates for each type of commercial vehicle. Examples of this type of model can be found in Seattle and Los Angeles.

A new model for Houston proposes to stratify truck trips by mode of operation: fleet vehicles, which make numerous short trips; intermodal connections, which are fewer in both frequency and destinations; and heavy drayage, which typically connects a single origin-destination pair (Parsons Brinckerhoff, 1998).

The San Francisco model (Schlappi et al., 1993) separated 'garage-based trips' from 'link-based trips' to identify the differences between truck trips that traveled in individual round trips from the garage to the destination and other round trips that traveled in multiple destinations in a chain of trips.

In another study for the Greater Buffalo-Niagara region (Jack Faucett Associates, 1999), the internal truck trips were separated into three categories: parcel delivery trucks, mail trucks and all other trucks. The U.S. Postal Service provided data on mail trucks and the parcel delivery trucks were estimated from data on the largest carrier in the region. Other truck trips were estimated from a variety of sources.

The Southern California Association of Governments (Davidian, et al., 1999) documented that the largest fraction of truck trips in the metropolitan area were associated with light-heavy trucks serving households and service industries. Some attempts were made to develop new trip generation data using shipper/receiver surveys. The SCAG model also used aerial photographic land use data and econometric modeling techniques to develop new methods of developing distribution models for construction-related trips.

2.4 Commercial Vehicles for Services

There are no examples in the state of the practice where the service-related commercial vehicles are estimated uniquely from other commercial vehicles. In some cases, urban models have been developed to include service-related commercial vehicles in the estimation process. In the Quick Response Freight Manual (Cambridge Systematics, 1996), delivery and service vehicles are included in the definition of four-tire commercial vehicles and trip generation rates includes some trips from households to account for the source of many of these trips.