Long-Term Pavement Performance (LTPP)

Historical Summary and Purpose: The Long-Term Pavement Performance (LTPP) program is a 20-year research project begun in 1987 as part of the Strategic Highway Research Program (SHRP). During the early 1980s, the Transportation Research Board (TRB) of the National Research Council, under the sponsorship of the Federal Highway Administration (FHWA) and with the cooperation of the American Association of State Highway and Transportation Officials (AASHTO), undertook a study of the deterioration of the Nation's highway system.⁽¹⁾ The SHRP was established on the recommendation of this study to focus research and development activities aimed at improving highway transportation. The Long-Term Pavement Performance program was one of six key research areas identified by this study.⁽²⁾ The LTPP program is a comprehensive program to "satisfy the total range of pavement information needs" drawing on "technical knowledge of the pavements presently available and seeking to develop models that will better explain how pavements perform ... this includes specific effects on pavement performance of various design features, traffic and environment, etc." The traffic and environmental data contained in the LTPP data collection plan are of potentially extreme interest as measures of exposure for highway safety issues as well. The concept of a traffic database, later named the Central Traffic Database (CTDB), originated in 1989 when the Expert Task Group concluded that the volume of traffic and load data that would be collected over the 20 years of the LTPP program required a separate database.

Data Contents and Structure: The LTPP data are housed in seven modules. A brief description of those modules that could be of interest in highway safety studies is described below:

(1) Climatic module.

Data derived from the National Oceanic and Atmospheric Administration (NOAA) weather data. Climatic data include site-specific estimates (based on the five closest weather stations) of various temperature, precipitation, humidity, and solar data statistics on a monthly basis for each test section, as well as actual values for the weather stations.

(2) Inventory module.

Data that identify the site and describe the pavement at the time the section was chosen. Data include location, material properties, composition, construction improvements, etc.

(3) Maintenance module.

Data describing all maintenance activities associated with the site.

(4) Monitoring module.

Friction, deflection, and distress data that could be of interest in wet pavement accident studies, etc.

(5) Traffic module (Central Traffic Database [CTDB]).

Historical and monitored traffic data. Yearly estimates of volumes, axle loads, and equivalent single-axle loads are available for each site. Also, data on truck weights and distributions are available at 789 sites quarterly for 7 days. Approximately 35 percent of these sites have weigh-in-motion collectors and the rest are Automatic Vehicle Classification (AVC) counters.

Experimental Design, Sample Plan, and Location Distribution: Data are collected in four geographic regions by regional staff members. With regard to traffic data, staff members are responsible for reviewing and processing the traffic counts, classification, and weight data, as well as ensuring acceptable collection procedures. The regional offices transmit their data to the national LTPP Traffic Database. Here, the data are further scrutinized and edited and it is the responsibility of this office to decide what data are of sufficient quality to release to the general public.

Traffic data are collected on more than 789 sites on key highway routes. In addition to new traffic data collection, historic traffic data were also requested where available. There are generally two types of traffic data available -- vehicle count and classification data (Automatic Vehicle Classification [AVC] devices) and vehicle count and weight data (Weigh-in-Motion [WIM], either permanent or portable). The location of the WIM data collection may not always be exactly at the site, especially near interchanges. For the purpose of safety analyses, it is important that the researcher verify the exact location of the traffic data. These data have been of varying quality and one of the future objectives will be to back-validate some of the historic data with the new data, incorporating trends established based on the new data. Figure 1 show the geographic regions and Table 1 lists the number of locations by State for these locations. (Note: A revised table will be submitted that identifies locations that have WIM equipment and that have AVC equipment only when it is available).

Data Acquisition and Documentation: Information from the LTPP studies is available from the LTPP Information Management System (IMS), a database developed under SHRP. The pavement performance data are stored in the National Information Management System (NIMS) located at the TRB in Washington, D.C., and are updated on a regular basis. Similarly, the more detailed traffic data are housed in the CTDB and updated on a regular basis. Summary traffic data from the CTDB are periodically sent to NIMS for inclusion with the pavement performance data. These updates include corrections of previous erroneous data. Procedures and standards were established to ensure data quality, and extensive data quality checks are preformed throughout the collection and recording process. Information is also available indicating the level of data reliability. Although data are collected at the regional level and stored in Regional Information Management Systems (RIMS) and regional CTDBs, data are only released to the public after they have passed these checks and are stored in the national databases.

A guide that contains more details on the background and objectives of LTPP -- what data are collected, how to request data, data formats, and examples of reports generated -- can be found in reference 2. Complete information on how the data are collected, what quality checks are imposed, etc., can be found in other documents.

Data are released on two levels: (1) a sectional release and (2) an experimental analysis release. Data in Level 1 generally should be considered for analysis of a given test section, not comparisons across sections. These data have passed a minimum number of quality checks and, if used in analyses, should be used cautiously. Level 2 data have completed all assurance checks and are considered acceptable for analysis. Many quality control issues are still under development and consideration in an ongoing FHWA contract. Among these is the prospect of grouping sites into classifications according to the completeness of the traffic data at those sites. A classification being considered for the amount of data available is "preferred," meaning that at least 9 months of continuous data are available; "desirable" would mean that at least 6 months of continuous data are available; and "minimum" would mean that anywhere from 1 day to 6 months of data are available. Missing data can be due to lack of continuous WIM devices, equipment failure, etc. These classifications have not been set and could have changed by the time of this report. The researcher is referred to the periodic progress reports produced from this contract. The FHWA contact for this information is Kris Gupta. At this time, there is a limited amount of data available to the public, i.e., data that have passed Quality Assurance/Quality Control (QA/QC) checks. Although the plan is to have at least 50 percent of the data available by the end of 1995, the FHWA contact can best update the researcher on this.

Potential uses of the LTPP traffic data would have to focus on safety studies that are location based. For example, the question of "are double-tractor configurations overly represented in on-/off-ramp accidents as compared to singles?" might be addressed using the LTPP traffic data. First, it would be necessary to ascertain whether or not there are a sufficient number of LTPP sites with complete enough traffic data to supply enough accidents to do an adequate evaluation. Secondly, are accident histories available at these sites and over a sufficient time period? This would be the general process for examining the feasibility of using the LTPP traffic data (or any location-specific traffic database):

1. Formulate the hypothesis.

2. Determine what traffic data best represent the exposure for the data required to address the hypothesis.

3. Determine if there are sufficient sites of the type required by the hypothesis in the CTDB. How complete are the traffic data at these sites?

4. Determine whether accident histories are available and in sufficient numbers to justify the analysis.

These steps should be attainable using only a minimum amount of resources.

The only way to receive LTPP data from the national databases is to submit a complete LTPP Data Request Form to the TRB NIMS Administrator:

Penny Passikoff
National Academy of Sciences
Transportation Research Board
2101 Constitution Avenue, NW
Washington, D.C. 20418
TEL: (202) 334-3259
FAX: (202) 334-3495

Costs for obtaining the data include a $75 handling fee, media costs that depend on the type of media selected on the form, shipping costs, and any costs due to custom requests. State and Federal agencies and international participants do not have to pay the $75 handling fee.

References

(1) Rowshan, Shahed. Long-Term Pavement Performance Information Management System Data Users Guide. Federal Highway Administration, Report No. FHWA-RD-93-094, July 1993.

(2) Herman, John L.; Charlie R. Copeland; and W.O. Hadley. SHRP-LTPP Traffic Data Collection and Analysis: Five-Year Report. Texas Research and Development Foundation, Austin, TX. SHRP-P-386, 1994.

Nationwide Personal Transportation Survey (NPTS), FHWA

Purpose: The Nationwide Personal Transportation Survey (NPTS) provides nationally representative estimates of personal travel in the United States. All modes of transport are covered, including passenger cars, trucks, motorcycles, buses, trains, subways, airplanes, taxis, bicycles, and walking. The dataset includes information on demographic characteristics of the household, person-level information on the individuals participating in the survey, descriptive information on each vehicle in the household, and two levels of travel information. The first level of travel information is a detailed account of all trips taken on the survey day. The second level is information on trips longer than 121 km that occurred during the 2-week period immediately prior to the survey day. Travel information includes mode, vehicle type, road type, date of travel, time of day, trip purpose, origin and destination, elapsed time, and area type.

Source: The most recent NPTS (1990) was conducted by the Research Triangle Institute of Research Triangle Park, NC, under the sponsorship of the U.S. Department of Transportation.⁽¹⁾ A random sample of 26,172 households with telephones was selected by means of a random-digit dialing procedure, and almost 22,000 households responded. Responses were collected by means of a telephone interview. (Earlier surveys were done using in-home interviews.) Each household was assigned a 24-h travel day (defined as 4:00 a.m. on the travel day to 3:59 a.m. on the following day) and a 14-day travel period. The survey period was from March 1990 to March 1991. Person-level interviews were conducted with all household members age 5 years and older. Trip-level interviews were conducted with all household members age 13 and older. The latter respondents supplied travel information on residents 5 to 13 years of age.

Coverage: The current file (1990) is the fourth in the series; earlier NPTS files are for 1969, 1977, and 1983. All personal trips, all modes of transportation, all purposes, and all 50 States and the District of Columbia are covered. Connecticut, the New York Metropolitan Planning Organization (MPO), and the Indianapolis MPO funded oversampling in their respective areas. The file includes weight variables, so that estimates of national totals can be computed.

Strengths: The NPTS file is the only source for national data on personal travel. Sample sizes are large, with 22,317 households, 48,385 persons, 35,152 licensed drivers, and 41,178 vehicles in the most recent sample. The survey design includes both driver and passenger travel, so vehicle occupancy rates can be analyzed. NPTS files are now available for 1969, 1977, 1983, and 1990, allowing trends over a period of 21 years to be analyzed. Efforts were made to maintain comparability of the major elements of the survey over that period. Travel can be broken down by region and for households in certain metropolitan statistical areas. Detailed information is available on the socioeconomic status of the household; age, gender, and other characteristics of the travelers; purpose of trip; type, make, and model of vehicle; and time, distance, and duration of travel. Interviews are conducted using computer-assisted telephone interviewing techniques, so many inconsistencies could be identified during the interview and addressed by the respondent.

Limitations: Road type is available only for a small subset of day trips. Sample sizes for commercial vehicles are small--the focus of the survey was on personal travel--so the NPTS is not useful for truck travel. The focus of NPTS is on national travel. It is possible to estimate the travel for regions of the country and for certain States and Metropolitan Sampling Areas (MSAs), but estimates for individual local areas, MSAs, or States may not be based on large enough sample sizes and may be imprecise. Households without telephones could not be included in the sample because the sampling procedure was based on a random-digit dialing procedure. In addition, the data are all self-reported.

Sampling Errors: Sampling errors can be calculated using appropriate software. See the User's Guide.

Access: The data are contained in six hierarchical files and can be obtained either as an EBCDIC file (similar to plain ASCII) or formatted for the SAS statistical analysis package. The files can be obtained on magnetic tape through the Volpe National Transportation Systems Center, Cambridge, MA, (617) 494-2450.

References

(1) User's Guide for the Public Use Tapes: 1990 Nationwide Personal Transportation Survey, December 1991, Report No. FHWA-PL-92-007.

National Truck Trip Information Survey (NTTIS), UMTRI

Purpose: The National Truck Trip Information Survey (NTTIS) provides national estimates of truck travel that can be cross-classified by truck configuration and loading, road type, area type, and time of day. Details on truck configuration and loading include cabstyle, number of trailers (if any), number of axles for each unit, empty weight and length for each unit, cargo body style, cargo type for each unit, and cargo weight for each unit. Road type is divided into three categories: limited access, U.S. and State numbered routes, and other roads. Area is classified using Federal Highway Administration definitions of urban or rural. The time of operation is classified as either day or night.

Source: The NTTIS was conducted by the Center for National Truck Statistics, part of the University of Michigan Transportation Research Institute (UMTRI).⁽¹⁾ The work was supported primarily by the Motor Vehicle Manufacturers Association, the Western Highway Institute, the Engine Manufacturers Association, and the American Trucking Associations. An initial sample of 8,144 trucks was drawn from registration files maintained by the R.L. Polk Company. The sampling frame was stratified by State and within each State, and by whether the truck appeared to be a tractor, straight truck, or unknown type. An interval selection procedure with a random start was used to draw the sample. Interviewers contacted current owners and operators of the vehicles by telephone to obtain a general description of the vehicle and company that operated it. Questions included estimates of annual travel that were checked against estimates from the TIUS.

A subsample of approximately 5,000 trucks was drawn for the travel survey. On four randomly selected days over a year, each truck was surveyed as to its use for the previous 24-h period. The survey method was to essentially follow the truck for 24 h. Survey staff collected information on the actual route the vehicle followed, cargo carried (if any) and where it was loaded or unloaded, and a complete description of the truck's configuration. The route was then followed on a map and the mileage was classified by road type, time of day, and urban/rural. All data were subject to extensive editing to ensure accuracy. To the extent possible and where necessary, inconsistencies and inaccuracies were cleared up by more phone calls to survey respondents.

Coverage: The NTTIS was a one-time survey. The sampling frame was trucks registered in the United States in 1983. The phone survey to collect the initial vehicle description and then the follow-up calls for trip information took place between November 1985 and February 1987. The file covers all medium and heavy trucks (GVWR > 4536 kg) registered in the United States, except for trucks owned by any level of government.

Strengths: Travel estimates can be cross-classified by truck configuration, loading, and operating environment -- a level of detail unmatched in any other file of travel data.⁽²⁾ It is possible, for example, to compare the travel of loaded and unloaded two-axle tank trailers on limited-access roads in urban areas at night. All data were carefully reviewed by editors experienced with the trucking industry. Ambiguous or unusual responses were clarified, where possible, with respondents. It is expected that the data are as accurate as is feasible.

Limitations: Data are all self-reported, although subject to careful evaluation and consistency checking. Given the frequent contact between interview staff and respondents, and the ability to check responses, it is felt that the data are not systematically biased. Estimates from the file are all national. It is not possible to retrieve travel information for particular routes or even particular States. Moreover, by 1995, the file is clearly dated. There have been several important changes in the trucking industry since 1987 -- for example, an increasing reliance on multiple-trailer trucks -- that the file cannot reflect.

Sampling Errors: All sampling strata variables are included in the analysis file. Sampling errors can be calculated with appropriate software.

Access: The NTTIS file is a hierarchical dataset consisting of three parts: (1) a truck file with data describing the power unit, (2) a tractor trip file with data on trips by tractors, and (3) a straight truck file with comparable information about straight truck trips. The trip files contain one record for each trip taken by a survey vehicle on a survey day. Access to the data is provided through the Center for National Truck Statistics at UMTRI. Contact Kenneth L. Campbell or Daniel Blower at (313) 764-0248.

References

(1) Blower, Daniel and Leslie C. Pettis. National Truck Trip Information Survey. University of Michigan Transportation Research Institute, Ann Arbor, MI, Report No. UMTRI-88-11, March 1988.

(2) Massie, Dawn L.; Kenneth L. Campbell; and Daniel F. Blower. "Large-Truck Travel Estimates From the National Truck Trip Information Survey." Transportation Research Record No. 1407, Large-Vehicle Safety Research. Transportation Research Board, Washington, D.C., 1993, pp. 42-49.

Operational Exposure Data Sources

Historical Summary and Purpose: Researchers in the field of highway operations are often in need of exposure data in the form of both quantity of traffic and traffic congestion. Several researchers at Texas Transportation Institute were queried as to their knowledge of these data sources and the following reports resulted:

Kevin Balke's understanding is that the State of Texas (and probably others) has an extensive traffic monitoring program. His personal experience included collecting ADT volumes on many arterials and highways in major cities every 4 years. These studies were managed by local MPOs and these counts were published in a report. The Texas Department of Transportation maintains permanent count stations. A map is published annually with the AADT volumes displayed by location. However, none of this has been automated -- this seems to be the major drawback in most operations study data sources. And, of course, there is the State roadway inventory file to which operations researchers often turn. Gerald Ullman relies on these State roadway inventory databases, as well as the State's ATR stations. With regard to urban area operations, some cities have systematic count programs and some do not, according to Ray Krammes. Dallas, for example, has a machine count program. Specific personnel in each city would serve as the contact for obtaining this information (in Dallas, it would be Ken Melston). State highway departments would probably be the best source for this information. In Dallas, the initial goal was to have manual counts on every 1.6-km segment of arterial road every 3 years. However, lack of funding seriously reduced this effort. Dallas still collects much of the data and stores 24-h and peak counts in a computer program and publishes two reports every January -- one that lists the most recent count on each link and one that lists historical data, i.e., all counts on all links. Fifteen-minute counts could also be attained on paper copy. The only other city in the North Texas region that has some count data is Fort Worth. Most cities in the Metroplex do counts only on an ad hoc basis and generally hire consultants to do this work. In a review of Texas cities, this was generally the case (Austin, Houston, etc.). The counts are done on an ad hoc, nonsystematic basis for specific purposes.

It may be possible to design a highway safety research project using some of these site-specific count data. For example, Dallas would appear to have sufficient count data to address a particular urban problem. Consider the comparison of accident severities as a function of congestion -- peak vs. off-peak times, weekend vs. weekdays, etc., or issues such as alcohol-related crashes in urban areas by time of day. However, due to the erratic nature of the data collection, one must be concerned about what biases such non-systematic data collection might be introducing into the safety analysis. Also, the fact that most data sources appear to be unautomated, at least in Texas, is a serious drawback.

For the most part, it appeared that operations researchers are interested primarily in very site-specific data and rely on ad hoc, often manual, procedures for obtaining exposure information. However, when they are interested in more global issues, they rely heavily on the Highway Performance Monitoring System (HPMS), described separately.

Residential Transportation Energy Consumption Survey (RTECS)

Historical Summary and Purpose: The Residential Transportation Energy Consumption Survey (RTECS) is a survey designed and administered by the Energy Information Administration (EIA). The objective of the survey is to obtain information on vehicles used for personal transportation in the United States. It is a companion survey to the Residential Energy Consumption Survey (RECS).

The first RTECS was done in 1978 and has been repeated triennially since 1985. The most recent survey for which published data are available is 1991. The following discussion relates to the 1991 survey. A survey was done in 1994, but the data are not available as of the date of this publication. The survey has been done five times. The RTECS is a follow-up survey and companion to the RECS. The RECS collects data on the households and includes preliminary information on the vehicles available to the household, while the RTECS consists of three stages in which additional data are collected on the vehicles available and the use of the vehicles by members of the household.

The data collected in the RTECS and RECS may have applicability in different areas of highway safety research. Primary data elements of interest in highway safety are the estimates of vehicle-miles of travel and the motor vehicle stock available to households for personal travel. These data elements may be linked to characteristics of households to allow computations concerning the amount of exposure (both vehicle-miles of travel and vehicle type) for similar households. Since the primary driver of each vehicle in sampled households was identified, as well as the age of the driver, the vehicle-miles of travel and vehicle used by age of primary driver may be estimated by implication. Since the data were not collected for trips by individuals within the household, the use of these estimates of exposure for different age groups may be questionable. It does appear the data are disaggregate enough for computing vehicle-miles of travel for households stratified by different household characteristics. This would provide a means for the estimation of exposure for those households and the applicability of those estimates to specific regions where similar stratifications of households could be obtained.

Data Contents and Structure: Household data collected in the RECS through personal interview that may be of interest in highway safety research include the following:

• Census region and division where household was located.

• Urban status of the household location (whether urban or rural area).

• Number of persons in the household.

• Data on the household composition (e.g., number with/without children, age of householder, etc.).

• Race of householder.

• 1990 family income (these were reported in nine different ranges).

• Number of drivers in household.

• Age and sex of primary driver for each vehicle in household.

• Average number of vehicles available to household during the year.

• Model year and vehicle type for vehicles available.

• Whether vehicle was used for commuting to and from work.

For the household data collected, data on the number of vehicles available and the vehicle-miles of travel for those vehicles were obtained. Vehicular data were not collected in the RTECS for motorcycles, bicycles, all-terrain vehicles (ATVs), and other related vehicles.

Experimental Design, Sample Plan, and Location Distribution: The focus of the RTECS is to obtain data on the vehicle-miles of travel, motor vehicle stock, and vehicle fuel consumption and expenditure data. Its companion survey (RECS) collects data on household energy consumption and expenditure. The sampling units in both the RECS and RTECS are households, with the universe being all housing units occupied as the primary residence in the 50 States and the District of Columbia. The sample of households selected in the 1991 RTECS was based on the 1990 RECS. The 1990 RECS was a multistage probability sample that incorporated a rotating panel to allow the observation of changes in energy use over time for households that fall in successive panels.

The 1990 RECS initial sample consisted of 6,757 units. Of these units, 848 were found to be ineligible for reasons such as the dwelling being uninhabitable, currently vacant, or used for seasonal occupancy. Energy-related data were collected from 4,828 households via telephone interviews, and an additional 267 units were surveyed through a mail follow-up, for a total of 5,095 responding households. The RTECS sample of households was selected from the 5,095 housing units that responded to the 1990 RECS survey. The number of RECS housing units selected for the RTECS survey was 3,045. Of those units, 2,842 were contacted by telephone and 200 were identified as households that had to be contacted by mail. The number for contact by mail was subsequently increased to 485 due to an increased number of households with unlisted or disconnected telephones.

The RTECS data collection effort consists of four phases, with the first phase being done in conjunction with the RECS. The first phase (during the RECS personal interviews) collected data on the household's vehicle stock, the vehicle identification numbers (VIN) of the vehicles, and initial odometer reading for each vehicle. The subsequent three phases were conducted at the beginning of the year (B-O-Y), mid-year (M-Y), and the end of the year (E-O-Y). These data collection efforts were done by telephone interview and, where this was not possible, the data were collected via a mail questionnaire. The B-O-Y and E-O-Y phases updated the data on the vehicle stock and collected data on the vehicle characteristics (including the vehicle make, model and model year, the vehicle odometer readings, and VIN). The M-Y phase was an inventory update where respondents were asked to complete a vehicle update worksheet and keep it for use during the telephone interview or mail it back if the household was classified as a no-telephone household.

The data collected during the RTECS allow for the computation of actual vehicle-miles of travel from the recorded odometer readings. These data represent total travel between the two points in time (i.e., B-O-Y and E-O-Y). Data were also collected on the disposition of vehicles and acquisition of new vehicles during the survey period.

Quality of Data: The data collected in the RECS and RTECS appear to be of relative high quality. Since the surveys produce estimates based on randomly chosen subsets of the entire population of occupied housing units, the estimates will always differ from the true population values and will include sources of nonsampling and sampling errors. The following sections discuss various sources of potential error in estimates produced from these surveys:

Noncovered Residential Vehicles. Since the sample of households surveyed in the RTECS were selected from the RECS, any household excluded from the RECS would not be represented in the RTECS, and the subsequent survey data would not include vehicles available to those households. Specifically, those families or individuals not included in the RECS were those living in group quarters such as college dormitories, military barracks, or large boarding houses; those living in recreational or other types of vehicles; and those with no fixed address. The effect of these exclusions is an underestimation of the total number of vehicles and related data.

Date of Reference for Survey. Since the survey design requires households to be followed for an entire year, changes in household structure and composition may not be accurately reflected. For example, the survey sample may have an overrepresentation of older established households and an underrepresentation of new households or families. Resulting estimates of vehicles and related data may have a negative bias induced by established households separating and only one portion being followed in the RTECS, vehicles acquired by household members that leave the household are not captured in the survey, and the total estimated households (used for expansion) is based on the July 1991 Current Population Survey (Bureau of the Census).

Item Nonresponse. Item nonresponse refers to the inability to collect full information when respondents either do not know the answer or refuse to answer selected questions. It can also occur when an interviewer fails to ask a question or record an answer. In the RTECS, item nonresponses were imputed to provide an estimate of the most probable response. Three techniques were used: hot-decking, predictive mean matching, and regression.

Hot-decking is a technique by which a household is randomly selected and its response to the missing data item is used as the response for the household with the missing item. The items imputed in the RTECS by this method were pre-1975 vehicle characteristics and fuel grade. Household demographic items, such as family income and ethnic background, were also imputed by this method for the RECS.

Predictive mean matching was used for imputing changes in vehicle stock for households not followed for the complete duration of the RTECS. In the 1991 RTECS, 26 percent (i.e., 795 households) were not followed for the entire year and imputations were computed to estimate the number that acquired and/or disposed of vehicles during the year. For households with no vehicles that were lost, a hot-deck procedure was used to impute the changes in vehicle stock.

Multiple regressions were used to impute annual vehicle-miles of travel for those vehicles that were imputed as being acquired. Linear and multiple regressions were also used for estimated annual mileage for vehicles where two odometer readings were not obtained in the survey. For 26 percent (i.e., 1,576) of the sample vehicles, no odometer span was available. An estimate of the annual vehicle-miles of travel had been obtained from the respondent during the RECS interview. Vehicle-miles of travel were imputed from a regression on the estimate of vehicle-miles of travel obtained in the RECS interview. For an additional 19 percent (i.e., 1,150) of the sample vehicles, no odometer span was available and an estimate of annual vehicle-miles of travel was not obtained in the RECS interview. Estimates of vehicle-miles of travel for these sample vehicles were imputed using a multiple regression using number of drivers, household income, age of household head, type of vehicle, and use of vehicle on the job as independent variables. This same method was used for imputing the vehicle-miles of travel for vehicles that were imputed as being acquired and/or disposed. Various other adjustments to the vehicle-miles of travel data were necessary to put each in terms of the same time period. Data from the Federal Highway Administration on monthly vehicle-miles of travel were used for this purpose.

Potential Problems: The RTECS data provide reasonable estimates of vehicle-miles of travel for households and vehicle types. These data will produce reasonable estimates of exposure relative to household estimates and estimates by vehicle type. However, the data do not include travel by motorcycles, bicycles, all-terrain vehicles, or similar types of vehicles, which may be critical in safety analyses. In addition, the data do not relate vehicle-miles of travel to person-miles of travel. The data are collected for vehicles and related to the households that own or have those vehicles available. While the exposure may be computed for vehicles in terms of type and vehicle-miles of travel, the data do not indicate the number of persons that may be in the vehicle on an average basis. Other data sources on average vehicle occupancy would have to be used to impute that estimate. The use of the data to compute exposure estimates by age of individuals would have to be based on the implication of primary driver for each vehicle in the survey. This is a relatively weak implication and is not considered an accurate estimate. Thus, it is not considered appropriate to use data from this source for estimating exposure for persons by age.

Data Acquisition and Documentation: Data from the RTECS and RECS are available in a variety of media. The following published reports may be purchased from the Government Printing Office (GPO):

• Household Vehicles Energy Consumption 1991; December 1993, DOE/EIA-0464(91) (No GPO Stock No.).
• Household Vehicles Energy Consumption 1988; February 1990, DOE/EIA-0464(88), GPO Stock No. 061-003-00652-3.
• Residential Transportation Energy Consumption Survey: Consumption Patterns of Household Vehicles, 1985; April 1987, DOE/EIA-0464(85), GPO Stock No. 061-003-00521-7.
• Residential Transportation Energy Consumption Survey: Consumption Patterns of Household Vehicles, 1983; January 1985, DOE/EIA-0464(83), GPO Stock No. 061-003-00420-2.
• Residential Transportation Energy Consumption Survey: Consumption Patterns of Household Vehicles, Supplement: January 1981 to September 1981; February 1973, DOE/EIA-0328, GPO Stock No. 061-003-00297-8.
• Residential Transportation Energy Consumption Survey: Consumption Patterns of Household Vehicles, June 1979 to December 1980; April 1982, DOE/EIA-0319 (No GPO Stock No.).

The above documents are not the only ones available, but are considered to represent those report data that are of interest to highway safety engineers. In addition to the published reports, data tapes and diskettes may be ordered directly from the National Technical Information Service (NTIS). Information on how to order these may be obtained by telephoning NTIS at (703) 487-4807, FAX number (703) 321-8547. Detailed technical questions on topics of interest to highway safety engineers may be addressed to the following:

RTECS Manager
Ronald Lambrecht
(202) 586-4962

Vehicle-Miles of Travel
John Pearson
(202) 586-6160

Trends in Household Vehicle Stock
Ronald Lambrecht
(202) 586-4962

References

(1) Household Vehicles Energy Consumption 1991; December 1993, DOE/EIA-0464(91) (No GPO Stock No.).

Truck Inventory and Use Survey (TIUS), Bureau of the Census

Purpose: The Truck Inventory and Use Survey (TIUS) is one of a number of economic censuses performed by the U.S. Bureau of the Census. It is designed to provide information on the population and use of trucks for government, business, industry, and the general public. The TIUS is conducted every 5 years. The most recent data year currently available is 1992.

The TIUS provides annualized estimates of the primary uses of trucks. Data include a physical description of the truck (axle count, cabstyle, cargo body style, overall length, empty weight, typical loaded weight, maximum loaded weight); a general description of the industry in which the vehicle is used; and a breakdown of the vehicle's use over the course of a year. For example, respondents report any placarded hazardous materials carried in the vehicle and then estimate the percentage of the total annual travel in which hazardous materials were carried. Similarly, respondents estimate the proportion of annual travel accumulated off-road, less than 80.5 km from the truck's home base, 80.5 to 321.9 km from base, and more than 321.9 km from base.

The TIUS is useful for estimating broad categories of annual truck use. Given the way the data are reported, however, it is not possible to break down or cross-classify travel estimates by road type, area type, or any other feature of the operating environment. It is also not possible to estimate travel by State, month, or season.

Source: The TIUS is a stratified probability sample of trucks registered in the 50 States and the District of Columbia. Within each State, trucks are stratified by body style. Within each stratum, a fixed number of trucks are sampled randomly. Roughly 3,000 trucks are sampled per State. Survey forms are then mailed to the registered owners of the sampled trucks. By law, the surveys must be completed and returned. The data are all self-reported and are all estimates of use for a particular year. Reports are subject to computer editing. Apparently erroneous responses are reviewed and corrected, if possible.

Coverage: The sampling frame for the TIUS covers all vehicles registered as trucks in the 50 States and the District of Columbia. This includes pickups, small vans, and other utility vehicles registered as trucks. The file excludes vehicles owned by any unit of government, passenger vehicles, ambulances, buses, and motor homes. Vehicles used exclusively off-road do not have to be registered, and thus are also excluded.

Strengths: The TIUS has a very large sample size. Roughly 154,000 vehicles were selected for the survey in 1992. Nearly 132,000 trucks are represented in the file. Estimates of population totals and annual travel from the TIUS have been compared with estimates generated by other techniques (e.g., NTTIS; for a description of NTTIS, see the discussion in an earlier section) and are in general agreement. Data collection procedures and survey questions have been fairly stable for a number of surveys, so comparisons among survey years are valid.

Limitations: The main limitation in the use of the TIUS file for safety-related exposure data is that the data represent typical or primary use only. Consequently, configurations that represent secondary use, such as bobtails or doubles, are not represented at all or are under-estimated. There is very little ability to cross-classify the travel estimates by operational characteristics that are known to be associated with differences in accident-involvement risk. For example, straight trucks do a large share of their travel in urban areas and on non-limited-access roads. Tractor-semitrailer combinations accumulate a much larger fraction of their travel on limited-access roads, which are typically the safest in the highway system. The TIUS data do not provide any means of controlling for such environmental confounding factors.

Sampling Errors: Variables representing the sampling strata are not released with the file, so it is not possible to calculate sampling errors for particular estimates. However, the published Census of Transportation includes an appendix with equations for approximating relative standard errors.

Access: Available on CD-ROM from the Bureau of Transportation Statistics and from Customer Services, Bureau of the Census, Washington, D.C. 20233. The data are the raw records from the survey, modified to limit the possibility of identifying particular individuals or businesses.

State Weigh-in-Motion (WIM) and Automatic Vehicle Counting (AVC) Devices

Historical Summary and Purpose: Truck weighing equipment is required for meeting a wide variety of public, private, and institutional needs. In the public sector, there are two major functional areas of application of these devices: data collection and enforcement. Statistically representative truck weight data are collected and used as the primary basis for engineering analyses and decisions related to planning, funding, design, operation, maintenance, and management of highway facilities. Measurements of the weights of individual trucks are needed to provide enforcement agencies with the capability to protect the highway infrastructure from damage due to unexpectedly high loads. In both data collection and enforcement, it is necessary to weigh large numbers of individual trucks.

A weigh-in-motion (WIM) system is used to attempt to approximate the gross weight of a vehicle or the portion of the vehicle weight carried by a wheel, an axle, or a group of axles by measuring, during a short time interval, the vertical component of dynamic (continually changing) force that is applied to a smooth, level road surface by the tires of the moving vehicle. Although the weight of a vehicle does not change as it moves over the surface of the road, the dynamic force applied to the roadway surface by a rolling tire on a vehicle varies dramatically when the tire/wheel mass accelerates vertically. This acceleration can be induced by roughness in the road surface and/or by an out-of-round or out-of-balance wheel/tire assembly.

Data Contents and Structure: WIM data are collected in the United States by the States under three programs. One is specified and required by the FHWA under the provisions of its Traffic Monitoring Guide (TMG). The States have designated and collected data at approximately 1,400 WIM sites in the United States. The data are stored as individual truck records by the individual States and are transmitted to FHWA.

Additional WIM data are obtained under the Long-Term Pavement Performance monitoring aspect of the Strategic Highway Research Program. Data are acquired quarterly for 7 continuous days at 777 sites throughout the United States and are transmitted to regional SHRP contractors.

The last type of WIM data is collected at truck weight enforcement stations during the weighing and sorting of trucks to determine whether they exceed legal limits. These data are not normally retained.

Each State is required to submit vehicle classification and truck weight data to the FHWA either annually or quarterly. Where continuous weigh-in-motion data are available, 1 week of data per quarter is required. These data provide input to national databases that are maintained by the FHWA. These databases include the Traffic Volume Trends System and the Vehicle Travel Information System. The Traffic Volume Trends System is a database management system that is based on state-supplied ATR data. The Vehicle Travel Information System is a microcomputer database management system that validates, summarizes, and maintains vehicle classification and truck weight study data. Tables 1 through 3 contain State-by-State information on the number of WIM sites, type of equipment, level of monitoring, the existence of historical data, and monitoring frequency. Level of monitoring refers to the amount of data collected. The preferred, minimum, etc. categories are the ones described in the LTPP traffic data, although these may not be the levels adopted by the CTDB.

Experimental Design, Sample Plan, and Location Distribution: Each State determined their own experimental design and determined the number and location of the sites based on differing economic and policy-making factors. When using WIM data from any State for highway safety evaluation purposes, the researcher should contact the respective State's DOT and request specific information regarding site-selection criteria.

Potential uses of the WIM databases must be location-oriented, similar to the ones described for the LTPP WIM.

Data Acquisition and Documentation: Data from the national database must be requested from the FHWA directly. These data include: station description data, traffic volume data, vehicle classification data, and truck weight data. Each type of data has its own individualized record format. All data files are in ASCII flat files.

Individual State data can be requested of the individual State DOTs. The formats will vary. For example, Illinois currently has 18 active WIM sites dispersed throughout the State. The WIM system has not consistently provided the necessary data to the national database due to hardware and/or software problems. Illinois DOT collects data biweekly and stores all data that are required by the FHWA. The data are processed and kept on the mainframe computer in a hexadecimal format. Their data on the continuous count ATR network are located at 21 sites. These data provide vehicle count and classification data and are kept on personal computers in ASCII format.

Washington State DOT has 41 active WIM sites -- 5 use bending plates and the rest use piezoelectric sensors. The sites are continuous monitoring sites and the data are downloaded weekly. The data provide the standard vehicle classification and truck weight data required by the FHWA. The data are converted by the State from 13-bin to 4-bin format for storage on a mainframe computer. Data from 1990 to the present are available.

Reference

(1) Parsons, Brinckerhoff, Quade & Douglas, Inc. And URS Consultants, Inc. Pavement Damage Factors Derived From Weigh-In-Motion Data Measured by Portable vs. Permanent Systems. Florida Department of Transportation Statistics Office, Traffic and Roadway Data General Consultant Task Work Order Number 4, Sub-Task 3.2, December 1993.

---