Click for next page ( 13

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 12
The error associated with applying the load spectra calculated using TWRG data is caused by the error sources that affect site-specific load distributions (i.e., the equipment error and the sampling error associated with less than complete annual data being collected at each of the TWRG WIM sites) and by two additional error sources: the error associated with using an average condition for several roads to represent a specific road and the error associated with assigning a given road to a specific TWRG. These two errors are considerably larger than the first error. Statewide Load Distribution Data The last type of axle-load distribution, statewide, should be used only to provide load esti- mates when the state highway agency has little knowledge of the loads trucks will carry on the roadway being designed. This means that the agency has little confidence in its ability to predict the TWRG for the pavement section. The above-cited analyses indicate that use of con- temporaneous statewide load data produces MAPEs of about 25 percent for combination trucks.4 Statewide load distributions are obtained (for each VC) by combining the data collected from all WIM sites in a state. These distributions (most likely moderate distributions) then serve to represent average conditions that can be used whenever something better is not available. Fig- ure 2.2 illustrates a possible statewide distribution for tandem axles of Class 9 trucks. (This distribution was obtained by averaging the three load distributions shown in Figure 2.1.) Statewide distributions are the least reliable load spectra for pavement design. Their use is an acknowledgment that the state has little idea about the vehicle weights being carried on the pavement in question. However, because these distributions represent average load condi- tions, they will be poor representations for pavements that experience very heavy or very light loading conditions. Thus, the pavement design will be reasonable if not optimal. 2.3 Level 1 WIM Sites Level 1 WIM sites are sites at which design-lane WIM data have been collected for some period of time. These include sites at which WIM data have been collected continuously for a year or more, as well as sites at which data have only been collected for one or more short periods of time. Data collected at these sites are used by TrafLoad to produce one or more sets of load spectra for each site. All data should be quality checked and collected using accurately cali- brated WIM equipment. The following subsections describe the way in which TrafLoad uses WIM data and other infor- mation required by TrafLoad. The discussions include recommendations for developing a WIM data-collection system that will produce data that can be used effectively by TrafLoad. 4 Ibid., Table 4.5. 2-12

OCR for page 12
Figure 2.2 A Typical Statewide Load Distribution for FHWA Class 9 Trucks Fraction of Axles in Weight Range 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 Axle Weight Groupa a Each group is identified by the maximum weight, in thousands of pounds, for the group. WIM Inputs to TrafLoad For each Level 1 site, TrafLoad requires a current load spectra dataset. For some sites, TrafLoad should also be provided with a seasonal load spectra dataset. Each of these datasets consists of separate sets of load spectra by month and DOW, and there is a maximum of 84 sets of load spectra for each dataset. TrafLoad takes advantage of the resulting disaggregate sets of load spectra in two ways. First, for any month for which all seven DOW sets of load spectra are available, the monthly load spectra are developed by averaging the seven sets of load spectra in a way that avoids overweighting or underweighting the load spectra collected on any particular day of the week.5 For all months for which all seven DOW load spectra are available, this form of averaging 5 For each VC, the averaging procedure uses the monthly average DOW traffic volumes for that class to obtain weighted averages of the seven DOW load spectra for the class. The procedure is described in Section 3.3 of Part 4, which is available online at, and is based on the AASHTO procedure for summarizing traffic data (American Association of State Highway and Transportation Officials, AASHTO Guidelines for Traffic Data Programs, 1992, Chapter 5: Summarizing Traffic Data). 2-13

OCR for page 12
produces monthly load spectra that have no DOW bias. To improve the probability of obtain- ing complete sets of seven DOW load spectra, it is recommended that, if practical, WIM data from continuously monitored sites be collected and stored for at least 2 weeks of each month or, even better, for the entire month. (Collecting data for the entire month has the additional benefit of avoiding time-of-month bias, though this type of bias is usually much smaller than DOW bias.) Secondly, for a site and month for which only a partial set of DOW load spectra are available (i.e., load spectra for 1 to 6 days of the week), TrafLoad uses information about how pavement damage factors for each VC vary by DOW to produce estimated load spectra for the month that combine the available DOW load spectra with information about the extent to which the load spectra for the other days of the week are likely to be more or less damaging than the available load spectra. (See Part 4, Section 3.3, available online at detail.asp?id=4403.) The seasonal and current load spectra datasets are discussed further in the following subsections. Continuous WIM Sites and Seasonal Load Spectra Datasets Seasonal load spectra datasets are obtained from data collected at some or all of the continuous WIM sites in the states. The most significant use of data from these datasets is the develop- ment of monthly adjustment factors. These factors are used to adjust the load spectra obtained at other WIM sites for a given month so that they are reasonably representative of the load spectra for another month for which reliable WIM data are unavailable. If load spectra are available by month and DOW, data from seasonal load spectra datasets are also used to produce DOW adjustment ratios that are used in developing a set of monthly load spectra from a set of DOW load spectra for a month for which one or more DOW load spectra are missing. Each seasonal load spectra dataset must contain data for each month of the year, with a full week of data for at least one of these months (and preferably for all months). Ideally, this con- dition is met by defining a seasonal load spectra dataset to consist of 12 consecutive months of data. However, to allow for outages of WIM equipment, the data-collection period represented by a seasonal load spectra dataset may be extended to cover a period of up to 24 months. If, for any month of the year, the resulting dataset contains 2 months of data, only the data for the later month are used. Seasonal load spectra datasets are developed using data collected from a single lane (or design lane) using WIM equipment that has been consistently calibrated over the entire collection period. Each such dataset should be collected over a period during which seasonal changes in axle weights are believed to be reasonably representative of those that currently occur at the site. Month-to-month variations in calibration should be minimized, as should any secular upward or downward drift in calibration. TrafLoad will interpret any such variations or drift as representing actual changes in axle weights, and, as a result, it will assume that similar vari- ations in axle weights also occur at sites at which only 1 or 2 months of WIM data have been collected. 2-14

OCR for page 12
Seasonal Load Spectra Factor Groups To allow the application of seasonal adjustments to load spectra obtained from various WIM sites, each state must establish a set of seasonal load spectra factor groups and assign every WIM site in the state to one of these groups. Furthermore, each of these groups must contain at least one WIM site for which a seasonal load spectra dataset exists. For each of these groups, the software will use the seasonal load spectra datasets to develop a set of monthly adjustment factors. The assignment of WIM sites to seasonal load spectra factor groups is based on how average pavement damage per vehicle varies over the course of a year. A convenient unidimensional measure of this last quantity (and one that is used by TrafLoad) is average 18,000-pound ESALs per vehicle (average ESALs per vehicle, or AEPV). The WIM sites that are assigned to a given load spectra factor group should have values of AEPV that exhibit similar seasonal variations; i.e., the peaks and valleys in the AEPV should occur at the same time of year, and the ratios between the maximum and minimum values of AEPV should be similar. A reasonable starting point for the development of seasonal load spectra factor groups would be to distinguish three sets of functional systems: urban (U); rural Interstate (RI); and rural other (RO). In addition, any set of roads that has its own set of seasonally varying weight limits (such as spring-thaw restrictions or higher weights during winter freeze) should be assigned to a separate factor group or to separate factor groups. In establishing these factor groups, consideration might also be given to the way in which the month of the heaviest truck loadings varies regionally (e.g., as a result of differences in harvest season). As observed above, each seasonal load spectra factor group must include at least one WIM site for which a seasonal load spectra dataset exists. When establishing seasonal load spectra fac- tor groups, a state is likely to use its own WIM sites for this purpose. However, this is not a requirement. Any WIM site at which the seasonal variation in truck weights is believed to be typical of the seasonal load spectra factor group can be used. WIM VC Groups The seasonal adjustments to the load spectra are performed separately for each of several user- defined VC groups. These VC groups are referred to as WIM VC groups to distinguish them from the VC groups (Type 1 VC groups) that are used for factoring classification counts. (See Section 3.3.) The principal purpose of allowing the use of WIM VC groups is to make sure that, for any VC, the load spectra adjustments are based on a reasonable amount of data. For this purpose, any uncommon VC should be grouped with a more common class. In particular, any VC that is not commonly observed on a daily basis at all WIM sites that are used for seasonal load spec- tra datasets should be grouped with a more common class. However, because of a limitation in the current adjustment procedure, it is recommended that all common VCs be assigned to separate WIM VC groups. 2-15

OCR for page 12
Current Load Spectra Datasets Current load spectra datasets can be obtained from most or all WIM sites. There are two prin- cipal uses for a current load spectra dataset obtained for the design lane at a particular site: It can be used in designing pavement for that site; and It can be used in the development of a set of load spectra for a TWRG to which the site is assigned. (The development and use of load spectra for TWRGs is discussed in Section 2.4.) Since the current load spectra dataset for a particular site may be used for designing pavement for that site, it is important that this dataset provide as good a set of estimates of the current load spectra at the site as practical. Accordingly, a current load spectra dataset should contain data collected using calibrated WIM equipment over a recent period when the axle loads are believed to be representative of the current axle loads observed at the site. The data-collection period may be up to 24 months long. If a seasonal load spectra dataset exists for a particular site, it may also be used as the current load spectra dataset for the site. Alternatively, a separate dataset that is believed to better reflect current conditions may be used. If a current load spectra dataset contains less than 12 months of data, the software uses monthly adjustment factors to impute load spectra for the missing month. For a particular site, the monthly adjustment factors that are used normally6 are the ones developed for the sea- sonal load spectra factor group to which the site belongs. In developing a current load spectra dataset for a particular site, it is important to minimize calibration error. To the extent practical, use of portable sensors should be minimized, since their effects on the road profile adversely affect data accuracy. There is no minimum on the number of days of data incorporated into a current load spectra dataset. However, increasing the amount of data collected decreases the amount of imputa- tion required to produce a complete set of monthly load spectra and increases the reliability of the resulting values. Imputation is unnecessary when data are collected continuously over the course of a year, and relatively little imputation is required when data are collected over 1-week periods in 3 or 4 months uniformly spread over the year. Using data that are collected in only 1 month places appreciably greater reliance on the imputation procedure. (However, even in this case, the software's imputation procedure is likely to produce a better represen- tation of the seasonally varying pavement loads than would the simple alternative of using the load spectra that are observed in 1 month to represent the load spectra that occur in the other 11 months of the year.) 6 In the special case of a site for which a separate seasonal load spectra dataset exists, the monthly adjust- ment factors used are the ones developed directly from that seasonal load spectra dataset. TrafLoad also allows the user to associate any site for which a seasonal load spectra dataset does not exist with a site for which such a dataset does exist and at which seasonal and DOW variations in load are believed to be very similar to those for the site in question. If such an association is made, daily and monthly adjustment factors developed using data from the associated site will be applied to data from the site in question in the absence of factors developed from a seasonal load spectra dataset for that site. 2-16