The literature review was completed separately for each of 13 different types of commercial vehicles. In most cases, this resulted in more information on these types of vehicles than could be found by searching for categories of commercial vehicles. In the case of most types of trucks, the literature was considerably more robust for larger categories of vehicles (all trucks) or for trucks by weight class (light, medium, and heavy). Most of this literature has been discussed in Section 4.0.
Some of the individual methods and data sources identified in this section have been discussed as part of the review of state of the practice models, data sources or modeling methods. They have also been included here where they are relevant to the individual types of commercial vehicle models.
Home-based school travel is a major cause of traffic congestion. An estimated 20 to 25 percent of traffic in the a.m. peak period on local streets and roads is attributable to school commute trips (STPP, 2000). Effective school bus services will prevent parents from driving their children to school and clogging local roadways during critical peak hours.
There are several useful data sources for school bus trips. The National Highway Traffic Safety Administration (NHTSA, 2002) publishes national statistics on public school buses that are widely cited in school bus literature. The California Air Resource Board (CARB, 1995) estimated the statewide school bus population and vehicle miles traveled by counties as part of its air quality model. Schoolbusfleet.com is an information service of School Bus Fleet magazine. Each year, it conducts a school district survey and a school bus contractor survey, reporting school bus fleet size and passenger load size for top 100 school districts and top 50 school bus contractors. It also reports the number of pupils being transported by school buses and the total route mileage of school buses for each state (SBF, 2002). Another source is the Department of Education for individual states, which usually has a web page with data on school bus numbers and ridership. For example, the Department of Education for the state of Maine (http://www.state.me.us/education/const/ed5462001.htm) reported annual data regarding district-level school bus fleets including the number of buses as well as miles traveled in its school administrative units.
Home-based school is often modeled as a trip purpose for travel demand models in urban areas. However, there are different transportation modes for school trips. The treatment of school bus as a mode choice is complicated because representation of school bus routing and district definition and associated limitations would entail a substantial amount of data collection and network coding; it is also quite difficult to predict the provision of such service and its parameters in the future (PB and Urban Analytics, 1998). Development of school bus vehicle trip movements is taken from survey data and simply growth factored to the future; and there is no origin-destination-specific info on school bus trips. Pima Association of Governments, Arizona (2001) provides summary statistics describing the anticipated growth school bus travel demand for it metropolitan area, where typical weekday school bus trips are expected to increase by approximately 45 percent between 2000 and 2025. There are regional variations for the growth trend of school buses. Purvis (1994) compared results from a 1990 household travel survey to results from surveys conducted in 1965 and 1981, concluding a decline in school bus trips in the San Francisco Bay Area.
Airport shuttles and commuter vanpools are the major types with a number of available literatures in the category of fixed shuttle service. These modes offer the important benefits of cutting back travel cost and are highly significant for relieving the spatial and temporal distribution of traffic by reducing vehicle trips in heavy traffic zones (such as airports) or reducing rush-hour commute traffic.
Airport shuttle van has emerged as an important transit service competing with demand-responsive, door-to-door shared-ride service to and from airports at nearly half the price. A comparison study of six cities airport ground access is provided in Intermodal Ground Access to Airports: A Planning Guide (Bellomo-McGee, 1996). Poole and Griffin (1994) examined the success of airport shuttle services in Los Angeles airport, a pioneer on airport van transit service. Their study provided comprehensive data on the magnitude and distribution of airport shuttle trips in comparison with private automobiles, based on which they calculated the number of autos and auto trips that shuttle vans substituted for. The Office of Transportation Technology, Department of Energy (2000) conducted surveys of shuttle services in major airports, including the Baltimore/Washington International Airport, Dulles International Airport, Reagan National Airport, and Philadelphia International Airport. The annual airline passenger flow data and the magnitude of airport shuttle trips were reported for each airport; but no further data regarding the distribution of trips were collected. There are several studies on airport access model that treats airport shuttle as an access mode to airports. An Airport Passenger Ground Access Model is developed for Sacramento Regional Transit District (DKS-2002) for trips to and from the Sacramento International Airport. Instead of modeling airport as a special trip generator, airport trips are modeled as a unique trip purpose with its own generation, distribution, and time of travel characteristics. Another airport access model was developed by Cambridge Systematics (1998) for the Portland International Airport based on a set of two passenger surveys that primarily intended as a source for estimation of improved access mode choice and origin probability models.
Vanpooling is also a fixed-route transportation mode using commercial vehicles to carry passengers. Pratt (2000) provided an in-depth analysis on vanpool programs. He presented highly useful data on the characteristics of vanpool trips. He also describes several successful vanpool programs such as the 3M Company Employer-Based Vanpool Program, Golden Gate Vanpool Transportation Demonstration Project, Connecticut's Easy Street Vanpool Program and Pace Vanpool and Subscription Bus Programs in Suburban Chicago. The Puget Sound Region in the state of Washington leads the nation in vanpooling, providing 40 percent of the nation's public vanpools. The Washington State Department of Transportation (2000) conducted a vanpool market study to identify the market characteristics and conditions that contribute to the region's success in vanpool programs. The study assessed existing market share, based on which it projected the future growth opportunities for vanpooling. Cambridge Systematics, Inc. (2002) developed a vanpool model for Puget Sound Regional Council (Psrc) based on a vanpool survey database. The model was used to determine the auto and transit modes from which vanpoolers will shift when new vanpool services are provided. A similar vanpool model was developed by Koppelman et al. (2002) to estimate commuter behavior and future ride sharing intention based on state-preference survey data using a sample of commuters employed in the suburban Chicago area.
In general, there is quite a lot of literature regarding different aspects of taxis, however most of it is from Western Europe and other non-United States developed countries (Canada and Australia). There is a significant amount of research on the economics of taxi and demand-related issues, primarily on the effects of deregulation of taxis in different places on supply and demand. They are indirectly related to demand, with specific references to policy actions and in some cases include the effects of taxi fare on demand and supply.
There is also a lot of literature on demand responsive transport service using taxis. Most of this literature deals more with the supply side providing ideas and approaches for better utilization than with the demand to such system, but some of them address demand issues. In this regard, there is also a fair amount of literature on ITS and its contribution to taxi service, and on innovative approaches to increase taxi efficiency. Another group of references that have some relevance are mode choice models that include some treatment of taxis.
The best sources regarding taxis are the Taxi and Livery Statistics and Taxi Travel and the 1995 Nationwide Personal Transportation Survey. However, these do not provide models or methods for estimation. The Hong Kong paper by Yang et al. (2000) is probably the best example for a methodology approach to forecast taxi demand suggesting a simultaneous equation system of passenger demand, taxi utilization and level of service and present the important variables in the models including number of taxis utilized, taxi fare, disposal income. Data from New York City show 80-105 million empty taxi miles annually and a range of $1.02-$1.25 revenue per mile. Gilbert et al. (2002) provide data from few cities regarding the utilization levels of different private transportation systems.
The APTA Transit Vehicle Data Book provides data on demand responsive service for many cities. Data include vehicle type and number of vehicles for each agency. However, the only cost item provided is the amount paid to the manufacturer per vehicle. Other than the number of vehicles and their type, there is no data on the level of usage.
The APTA Public Transportation Fact Book includes data on operational levels and additional data regarding paratransit usage. For 1999, it reports 100 million unlinked passenger trips, 813 million passenger miles, an average unlinked trip length of 8.1 miles, 718 million of operated vehicles miles, 48 million of vehicle hours, 608 million of vehicle revenue miles and 41 million of vehicle revenue hours. A case study from Charlottesville, VA (Fitzgerald et al., 2000) utilized data from Section 15 and from the local paratransit provider and show an average of 5.02 paratransit trips per month in the city.
The literature regarding paratransit services either discusses some current paratransit systems (see for example Kuimil et al., 1999), reports on the results of a survey of paratransit operators (see for example the report by East-West Gateway Coordinating Council, 1996) or reports on a survey of customers (see for example Densor, 1997). Finally, there is a lot of literature with the logistics operation of paratransit dealing more with the supply side, however as they provide algorithms to better supply the demand they have some relevance to demand as well (see for example Fu, 2002 a and b). However, most of this literature is not relevant to our study.
None of the literature found describe models to support estimate of paratransit vehicles and/or usage, except the Florida one (Stasiak et al., 1998). (Note the full text of this source has not been obtained yet.)
There are very limited data available on rental cars. A rental car survey was conducted for the RIAC garage near the T.F. Green airport in Rhode Island. It provided useful data on annual car rentals, average daily rentals, and maximum daily rental by different rental companies. Forecasts for 2005, 2010, and 2020 were also available in the T.F. Green airport master plan (Landrum & Brown, Inc, 2002). District Five of Florida Department of Transportation (1999) has conducted a rental car survey to obtain data necessary for the development of generation rates for tourist and visitors to enhance the Florida Standard Urban Transportation Model Structure models. Since there are a lot of rental companies located in or near airport, a promising source of rental car data is airport rental car revenue reports. For example, Palm Beach airport releases a monthly report on rental car gross revenue and passenger number during September 2001 and October 2002 (http://www.pbia.org/news.htm).
In terms of modeling, most airport access models treat driving as one mode, without distinguishing the difference between driving with a rental car or a driver-owned car. However, the Airport Passenger Ground Access Model for Sacramento's Downtown/Natomas/Airport Corridor is an exception (Sacramento Regional Transit District, 2002). In this model, driving with a rental car is modeled as a separate mode along five other airport access modes including drive and park. Renting cost, time to/from the terminal, in-vehicle time and as well as parking cost at destination are the attributes being modeled for the rental car mode.
Some of the literature discusses trends in different areas; however most of these are not from United States sources. Some discuss the problems, planning implications and general trends in urban freight related to some level to delivery services.
There is a lot of theoretical work on the supply side including different routing problems that are of great relevance to delivery services but are not relevant for this study. A lot of these articles deal with routing optimization for delivery services and algorithms for handling issues related to the traveling salesman and the Chinese postman. There is also a new and increasing literature about the effect of e commerce on freight. Some of this literature also partially addresses the delivery industry.
Morlok et al. (2000) reviewed the revenues and goods delivered by the different major delivery companies in the United States. In 1997, the four carriers that account for well over 90 percent of the United States parcel activity - Airborne, Federal Express (FedEx), United Parcel Service (UPS), and the U.S. Postal Service - had $37.9 billion in transportation revenue. In the U.S. DOT 1977 Commodity Flow Survey, only 3.2 percent of the value of goods shipped went via parcel carriers. But by the latest survey, in 1997, that percentage had grown to 12.3 percent. The survey in 1977 was far less comprehensive than the more recent survey, so it probably overstated the relative importance of parcel carriers.
There are also a lot of articles about FedEx and UPS in newspaper-type journals like 'traffic word' and 'American shipper' that existing databases do not provide abstracts for.
On-demand urban freight distribution from warehouses to various businesses and consumers has become increasingly competitive in terms of delivery times. As a result of this, urban freight flows to, from, and within an urban area has increased exponentially causing more congestion, and air and noise pollution. This has led to the centralization of warehouses increasing transportation distances. However, in order to improve logistical performance, a better freight transportation system needs to be developed to increase shipment frequencies and handle smaller shipment sizes that usually lead to an increase in the number of stops per hour.
Most evaluations and economic assessments of transportation proposals and policies omit a valuation of the time spent in transit for individual items or loads of freight. Knowing about delays, and indeed the practical value of reliability, is useful to shippers and receivers, but this information does not necessarily appear directly in vehicle operating costs and person travel times. As a result, benefits generated by improvements from road investment and traffic management may be understated, and expenditure decisions may be biased towards passenger movements. In order to capture the transit time, an Australian survey of freight shippers was used along with stated preference methods and multinomial logit models to estimate the value of such factors. The estimated value of $1.40 per hour per pallet for metropolitan multi-drop freight services, potentially a substantial value not currently tracked consistently or utilized in transport evaluation procedures in Australia, illustrates the significance of these results (Wigan, 1999).
The 2003 AirCargo World, an online commonwealth business media publication, provides information on warehouses at airports in the United States that store air cargo shipments (Air Cargo World, 2003). The information useful in deriving the magnitude of warehouse delivery commercial vehicles is the warehouse space in square feet, and tonnage of shipment handled in tons on an annual basis.
In a recent study in the Netherlands, a new approach for modeling and evaluation of urban goods distribution called the 'GoodTrip' was developed that estimates good flows, urban freight traffic and its impacts (Boerkamps, 2000). In this model, the logistical chain links activities of consumers, supermarkets, hypermarkets, distribution centers and producers. Based on consumer demand, the GoodTrip model calculates the volume per goods type in every zone. The goods flows in the logistical chain are determined by the spatial distribution of activities and the market shares of each activity type - consumer, supermarket, hypermarket, distribution center, etc. This attraction constraint calculation starts with consumers and ends at the producers or at the city borders. A vehicle loading algorithm then assigns the goods flows to vehicles. A shortest route algorithm assigns all tours of each transportation mode to the corresponding infrastructure networks. This results in logistical indicators, vehicle mileage, network loads, emissions, and finally energy use of urban freight distribution.
The magnitude and distribution of commercial vehicles involved in on-demand product and package delivery is often triggered by tele-shopping or e commerce. Tele-shopping or shopping on-line has led to the rapid growth in the e commerce industry for business-to-business (B2B) and business-to-consumer (B2C) transactions. This in turn lead to more sources of product supply and more package delivery with an emphasis on fast, guaranteed delivery that leads to trucks carrying only partial loads. Also an increase in computerized logistics has paved the way for a variety of goods to be in trucks, warehouses, and stores everywhere.
A recent study (Niles, 2001) indicates that about 32 million households purchased $10 billion worth in goods on-line in the year 2000 in United States. The number of Internet household users was estimated to grow up to 44 million in 2002. Though the volume of Internet shopping is still very small in comparison to $2 trillion dollars of annual retail spending, it is approaching 10 percent of catalog shopping volume where people order by mail or phone. One research firm estimates that online shopping will account for six percent of all United States retail sales in 2003.
A research team from University of Texas, Austin, (Whinston, 2001) investigated the productivity of numerous dot-coms and made a distinction between two types of dot-com companies - digital and physical. Digital dot-coms are Internet-based companies such as Yahoo, EBay and America Online, whose products and services are digital in nature and are delivered directly over the Internet. By contrast, the physical dot-coms sell physical products (e.g., books, CDs, toys) that are shipped to consumers. It was found that nearly 80 percent of the physical dot-coms held and managed inventory, and handled packaging and shipping processes by themselves (B2C), citing customer service excellence as the primary reason. By contrast, the digital products companies manage inventory directly through their websites and related applications and mostly B2B transactions. Some of the vital statistics of typical dot-coms are shown below (Whinston, 2001):
| Variable | 1998 | 1999 | Change |
|---|---|---|---|
| Average revenue | $7,658,801 | $22,533,672 | +194.0% |
| Average number employees | 126 | 225 | +78.0% |
| Revenue per employee | $60,614 | $100,255 | +65.0% |
| Average gross income | $3,574,871 | $9,205,196 | +157.0% |
| Gross profit margin | 46.7% | 40.9% | -5.8% |
| Variable | 1998 | 1999 | Change |
|---|---|---|---|
| Average revenue | $11,739,416 | $48,456,957 | +313.0% |
| Average number employees | 129 | 353 | +174.0% |
| Revenue per employee | $91,003 | $137,272 | +51.0% |
| Average gross income | $2,641,952 | $10,765,578 | +307.0% |
| Gross profit margin | 17.8% | 21.0% | +3.2% |
The U.S. Census Bureau provides information on a quarterly basis about e commerce in the United States (U.S. Census, 2002). The nationwide statistics include e commerce value of shipments in the manufacturing sector, e commerce sales in the merchant wholesale trade and retail trade, e commerce revenue in service industries, and e commerce sales by merchandise line through e shopping and mail-order.
In order to get an estimate of the number of commercial vehicles that are involved in on-demand package delivery, a set of factors have to be developed to convert e commerce revenue and sales to trucks. The potential variables that would enter into developing these factors are number of household Internet users in a typical urban area, e commerce revenue generated per household, and percentage of e tailing when compared to total retail sales.
This category of commercial vehicles is primarily involved in transporting construction equipment and materials to the construction site, and is usually heavy trucks. The magnitude of this class of commercial vehicles in an urban area largely depends on the size and duration of the construction project. For instance, the construction of a multi-storied building will involve transporting large construction equipment as well as huge amounts of building material, whereas, a highway project will most likely include transporting large quantities of material than equipment. The distribution of construction-related commercial vehicles will largely depend on the location of the construction site and the proximity of quarries and warehouses that supply building material.
In a recent diversion dam replacement project in Northern California, the magnitude and directional variation of trips associated with construction of the proposed project was assessed based on an analysis of vehicle trip generation using an estimate of the required construction-related workforce (State Water Resources Control Board, 2002). The construction of the proposed project was expected to occur over an estimated eight-month period with a total construction workforce of 40 workers. The implementation of the proposed project would also generate several daily heavy truck trips for transporting material and equipment over the eight-month construction period. The following table provides an estimate of the total number of construction relation vehicle trips that would be generated by the proposed project:
| Vehicle Origin | Distribution of Local Workforce | Average Daily Workforce1 | Average Daily Vehicle Trips2 | Daily Peak-Hour Vehicle Trips3 |
|---|---|---|---|---|
| Reno | (Reno) | (Reno) | (Reno) | (Reno) |
| Construction workers | 100% | 40 | 80 | 40 |
| Heavy Trucks | 100% | 10-20 | 20-40 | 3-6 |
| Total | 100% | 50-60 | 100-120 | 43-46 |
1 Average daily workforce includes 100 percent of the construction workers and an estimate of the average daily number of heavy truck trips generated by the proposed project over the eight-month construction period
2 Vehicles and trucks accessing the construction area generate two daily trips (one inbound and one outbound)
3 Peak-hour trip generation is based on 50 percent of the resultant daily passenger vehicle generation and 15 percent of the daily heavy truck generation
Assuming a worst-case scenario, the transportation analysis assumes that each of the 40 workers would drive a separate vehicle to the project construction area, making two trips per day or one round trip from home to the construction area and back. Construction equipment and materials deliveries would occur throughout the day. Therefore, construction of the proposed project would result in a total of approximately 100-120 vehicle trips per day on average an estimated 43-46 total vehicle trips per day during the peak a.m. or p.m. period. Additionally, it is estimated that construction-related activities would include the use of several types of equipment including backhoes, scrapers, water trucks, pickup trucks, and front loaders. It is assumed that equipment would be stored on construction area and would not result in a substantial increase in the overall daily project trip generation. Parking for construction personnel and visitors would be provided in an area on or adjacent to the project construction area.
The American Road and Transportation Builders Association (ARTBA) publishes a monthly report in ARTBA's Transportation Builder Magazine that highlights price trends, employment data and contract awards for the transportation construction industry at the state-level in the United States (Economics and Research, 2002). The number and value of contracts by state are provided in this magazine for airports, bridges and tunnels, docks, piers and wharves, highways, and railways. It also provides transportation construction contractor employment data by transportation sector and the prices of building materials like concrete and asphalt.
The magnitude of construction equipment delivery commercial vehicles can be derived from the number and value of contracts awarded to various statewide construction projects. Urban area statistics can be estimated by assuming a set of factors per construction contract award in a similar way to the diversion dam replacement project mentioned above. ARTBA provides employment data per construction contract which will give an indication of the workforce needed for the proposed project, and the value of contract awarded will give an estimate of the size and duration of the project. In addition to this, assumptions have to be made in order to convert state-level statistics to a typical urban area.
Information regarding public utilities appears either in newspaper articles or in local authorities' web sites regarding their trash collection programs. Numerous local authorities have such web sites; however most of them do not describe the coverage, and the vehicles utilized. They are primarily directed to provide residents with instructions and time of collections. However, some web sites contain information on coverage, frequency of collection and time of collection enabling some estimate of trash collection vehicle utilization. Other public utilities that may be covered in such sites include leaf collection and snow removal. For example, the County of Fairfax currently provides refuse collection and recycling service to approximately 39,000 residential homes per week; as well as seasonal vacuum leaf collection service to about 18,000 of these residences (http://www.co.fairfax.va.us/gov/dpwes/trash/collection_cc-main.html). The County provides household refuse and recycling collection once a week to homes located in established Sanitary Districts. The County also provides vacuum collection of leaves from the curb three times each season. The City of Columbus Department of Public Service, Refuse Collection Division collects trash from more than 280,000 households (http://refuse.ci.columbus.oh.us/). The City uses trucks with mechanical arms to collect trash from 98 percent of residents once a week.
This category includes vehicles for Police, Fire, Building Inspections, and Tow Trucks. Providing public safety services has become one of the primary responsibilities of urban governments and most of the vehicles under this category are either owned or managed by them. Many of the sources in this category are related to speed, response time, or shortest route to emergencies, and there are not many articles that estimate vehicle trip rates, vehicle miles of travel, or any other model-related parameters. Thus, we have come across several articles on emergency vehicles routing and route optimizations, and only a very few of them are useful for our study.
Washington State DOT conducted a survey of 5,000 tow truck trips from August 2000 to January 2001 (Nee et al., 2001). This survey includes:
These data are available and could be used to estimate the tow truck demand in urban area. A theoretical model was developed by Daskin (1984) to estimate the locations, dispatching, and routings for emergency services vehicles. This article formulates a multi-objective model that simultaneously determines the number of vehicles to deploy and their locations, and identifies the appropriate dispatch policy and the routes vehicles should use in traveling to emergency locations.
In Illinois, a linear model was used to estimate the number of police officers required for traffic services (Raub, 1987). The dependent variable in this model was the time annually spent policing and the independent variables were volumes in vehicle miles and population.
There is an efficient routing of service vehicles process using GeoRoute software. GeoRoute is an arc routing software package that includes a route optimization module based on the GENIUS traveling salesman problem heuristic, and many companies including USPS use this software for solving their routing optimization problems. The author has used GeoRoute software and optimized routes for garbage collection and for snow plowing operations.
Plumber, electrician, telephone/cable installation, and repair service vehicles are under the trades and services category. Vehicles under this category are on-demand, service-types and a visible number of them can be seen on urban streets. Although these vehicles make up a small percentage of all vehicles, they have a significant effect on safety, environmental and operations efficiency. In most cases they stop, blocking a lane, and make other drivers slow down. Unfortunately, there were not too many sources for literature available in this category.
Sears and Roebucks hired an outside consultant, AMEC, and conducted a survey in 2001 for their carry-in repair network (Docherty et al., 1999). Sears maintains cost and operational data for their products. AMEC collected and organized these data for analysis. They used GIS to visualize the special distribution of data on product demand, unit repair and unit transport cost. They also used network optimization program SAILS (Strategic Analysis of Integrated Logistics Systems) for this study. SAILS uses a mixed integer linear programmed approach, specially designed to solve large-scale multi-echelon network optimization problems. The echelons for this study are: repair facilities, repair facilities warehouses, cross-deck facilities, and access points.
GeoRoute is a multi-purpose graphical tool for applications requiring a network representation of streets in urban and rural areas (Gendreau et al., 1995). The network requires information about street-to-street connectivity, one-way streets, street types, and illegal turns at intersections. This structure is adapted to routing and scheduling problems for various types of delivery and public works vehicles. GeoRoute links to real-time traffic information and the routing algorithm considers two kinds of costs in calculating the cost of a trip: Drive and Turn. 'Drive costs' are based on cost per unit distance of travel. For example, the average speed on a freeway is faster than the average speed on a downtown city street. Turn costs are not directly related to distance, but are based on the turn angle. For example, it typically takes longer to make a U turn than to bear slightly to the right. GeoRoute is an arc routing software package that includes a route optimization module based on the GENIUS traveling salesman problem heuristic. In addition, GeoRoute can support multimodal transportation routing when it is included in the map-data.
Another method for computer-based routing and scheduling methods to improve the management of daily movements of vehicles and crew in regional service activities were reviewed (Beetle, 1989). Computer software and its application by a major commercial banking institution in Philadelphia have been presented as a case study. At that time the bank deployed about 200 automatic teller machines and the crews of the bank had to visit all the machines everyday. Computer programs were written and traffic was assigned to the highway network which consisted of nodes and links.
Under the outside sales and services category, we were looking for literature on realtors, door-to-door sales, and public services relations vehicles. In this category, we also could not find many useful articles.
Jim McLaughlin and Scott Greene (2000) at Los Angeles County Metropolitan Transportation Authority (LAMTA) concentrated on the MTA's role in coordinating health and human services transportation. A list of about 30 public transportation programs is discussed in the article. MTA and Access Services, Inc. (ASI) estimated that 1,300 dedicated vehicles and 1,400 taxis are operated within the County by 200 public and non-profit organizations. In 1999, there were a total of 4,204,270 trips, 6.8 miles trip length, and operating costs equal $62,300,868.
Federal office buildings in Washington, D.C. are typical of many large office complexes, particularly those of state governments. Federal warehouse operations have characteristics similar to those of large distribution centers. Spielberg and Smith (1981) wrote about the results of a survey of goods and service vehicle trips to federal facilities in the Washington metropolitan area. By using a combination of onsite observation and driver interviews, data on arrival and departure times, vehicle characteristics, trip purpose, origin of trips, and nature and size of load were obtained, analyzed, and used to develop planning guidelines.
