High-Speed Weigh-in-Motion System Calibration Practices (2008)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

The scope of this synthesis captures the state of the practice in calibrating high speed weigh-in-motion (WIM) systems. “Practice” relates to the operational procedures used by state agencies to evaluate the in situ performance of WIM systems in terms of their load measuring accuracy, rather than the equipment-specific technical details used by WIM sup- pliers and installers for obtaining these measurements. “High speed” refers to WIM sys- tems installed on the main driving lanes of highways and includes systems used for traffic data collection, enforcement screening, or a combination of both data collection and enforcement. Two main tools were used in conducting this synthesis; a thorough review of the litera- ture on the subject and an online questionnaire addressed to highway and load enforcement agencies administering WIM systems in the United States. The literature review covered the following topics: • Current standards related to WIM system calibration, • Historic WIM calibration practices in the United States, • Current WIM calibration practices in the United States, • WIM-related research in the United States, • European WIM calibration practices, and • European WIM-related research. The literature review presented in detail the ASTM E1318-02 standard for test truck WIM calibration and the provisional AASHTO standard MP 14-05 for pavement smoothness requirements in the approach to WIM systems. In addition, emphasis was given to the Euro- pean WIM specifications developed under the Cooperation in the Field of Scientific and Tech- nical (COST) study 323 research. The main source of historic WIM calibration practices was the McCall and Vodrazka’s States’ Successful Practices Weigh-in-Motion Handbook pub- lished in 1997. The source of current WIM calibration practices was documentation provided by agencies managing WIM systems and personal interviews with their personnel. These sources revealed several main differences between state practices and the ASTM E1318-02 standard, as follows: • For on-site calibration, only one agency conformed to the standard’s use of one Class 9 and one Class 5 truck. Most agencies use a single Class 9 truck. • A requirement that test trucks are equipped with air suspension systems on non-steering axles is not enforced. • Determination of calibration factors for specified speeds (for WIM systems utilizing this feature) is not performed. • A requirement that subsets of the WIM measurements meet the accuracy tolerances, in addition to the entire dataset [i.e., this is a stringent requirement introduced for the post- installation validations of WIM systems under the Long Term Pavement Performance (LTPP) Specific Pavement Study pool fund study]. SUMMARY HIGH SPEED WEIGH-IN-MOTION SYSTEM CALIBRATION PRACTICES

Overall, review of current state WIM calibration practices revealed the following: • Agencies recognize that on-site test truck WIM calibration needs to be complemented by routine WIM data quality control (QC) to ensure long-term data quality. Software for this purpose has been developed by a number of state agencies as well as the LTPP program. • Some agencies effectively use static load data of traffic stream vehicles as a means of calibrating their WIM systems. • State WIM calibration practices change and institutional knowledge is lost as agency personnel retire and new managers with different levels of expertise and resources suc- ceed them. The on-line survey questionnaire was addressed to state agencies that manage WIM sys- tems. These included departments of transportation (DOTs) and state weight and dimension enforcement agencies. The questions were organized into the following three main groups: • WIM calibration using test trucks, • WIM calibration using traffic stream trucks of known static weight, and • WIM calibration monitoring using traffic stream data QC techniques. In addition, background questions were included on the main type of WIM sensor tech- nology used, and the general methodology used for initial calibration of these systems. At the end of the questionnaire, a number of WIM inventory questions were included. Question- naires were distributed in web format to all DOTs and enforcement agencies that use WIM systems for screening. The actual number of questionnaires distributed was 53 (i.e., 50 states, District of Columbia, Puerto Rico, and one extra questionnaire used by Connecticut, which submitted separate responses for traffic data collection and research), with an additional 15 questionnaires distributed to enforcement agencies. Where necessary, the questionnaires were followed with telephone interviews to assist responders filling out the survey and to ask for any additional documentation available. The response rate to the questionnaire was 78% (for those distributed to DOTs) and 73% (for those distributed to enforcement agencies). The main findings of the questionnaire are provided in the following table. The agencies that responded to the questionnaire use one or more of the three calibration methods as summa- rized here. 2 Method Traffic Data Only Traffic Data and Enforcement Enforcement Only Test Truck Only 7 0 0 Traffic Trucks Only 2 0 6 Traffic Data QC Only 4 0 1 Test Truck and Traffic Trucks 3 0 0 Test Truck and Traffic Data QC 8 3 0 Traffic Trucks and Traffic Data QC 2 1 3 All Three Methods 4 3 1 Other 6 1 0 No Response 4 0 0

The majority of agencies perform test truck WIM calibrations on a routine basis at intervals ranging from 6 to 24 months, with most doing so every 12 months. Most agencies use a single Class 9 truck, whereas a small percentage uses two Class 9 trucks. Others use a Class 5, 6, 7, or 10 truck, either alone or with a Class 9 truck. Although a large majority of these agencies report considering pavement roughness, only about 25% does so objectively [i.e., 11.1% per- form the straight edge/circular plate test described by ASTM E1318-02, 3.6% simulate this test using software that accepts the pavement profile as input, and 14.8% simply use the Inter- national Roughness Index (IRI)]. Even fewer agencies objectively consider the structural condition of the foundation of the sensors. Most agencies use fixed weigh scales for obtain- ing the static loads of test trucks, although more than 40% of agencies using WIM for both data collection and enforcement screening use portable scales. Most agencies perform the sta- tic measurements only once. About half of the WIM systems, regardless of application, per- form test truck runs using the site median traffic speed, whereas the remainder uses either the posted speed limit or multiple speeds. The majority of agencies administering dual-use WIM systems use multiple test speeds. The majority of agencies using WIM traffic for data col- lection only conduct 10 test runs per vehicle speed, whereas agencies using WIM for either traffic data and enforcement or enforcement alone conduct three test runs per vehicle speed. Responders indicated that overall, 87% of agencies carry out calibration calculations on site. The method for computing calibration factors is equally split among agency software, ven- dor software, and short-hand calculations. For combined-use WIM systems, the majority of the agencies use short-hand calculations, whereas for enforcement-screening-only systems, about two-thirds of the agencies use vendor software. The main load data elements for which WIM errors are computed are the gross vehicle weight (GVW), the individual axle loads, and the tandem axle loads. The majority of agencies compute calibration factors by setting the mean GVW equal to zero, or by setting a combination of the mean GVW and the mean axle load errors equal to zero. Few agencies compute calibration factors by minimizing the least square errors between WIM and static axle loads through zero-intercept regression. Depend- ing on the WIM application, up to 67% of the agencies report deriving speed-specific cali- bration factors, although a significant percentage of them reports inputting their average value in all speed bins after calibration. Agencies that use traffic stream vehicles of known static weight for WIM calibration obtain static weights largely by permanent static scales at truck inspection stations. Only about one-third of agencies perform these calibrations on a routine basis at intervals ranging from 1 to 12 months. Most agencies do so only when there is an indication of calibration drift. There is roughly an equal division between the methods used for selecting the number of traffic stream vehicles used. Where a fixed number of vehicles is specified, it varies between 1 and 100, with an average number of 40 vehicles. Where a fixed time interval is used, it ranges between 1 and 168 hours, with the majority of agencies using data collected over a 1- to 4-hour period. The type of vehicles included in this sample varies; the majority of agencies using WIM for traffic data or traffic data and enforcement favor selecting vehicles in certain classes regardless of speed, whereas the majority of agencies using WIM for enforcement screening only use a random selection of truck classes. Axle spacing is measured mostly by manual means. The responses to the question as to where error calculations are performed varies; some agencies always do so at the site, whereas others do so at the office. Interestingly, enforcement agencies are more likely to perform the error and calibration computations at the site, which is explained by their ready access to static scale data. The actual method used for performing the calcula- tions varies but most often vendor software is used. The most common traffic load elements for which errors are computed are GVW, individual axle loads, and tandem axle loads. The most commonly used approach for computing calibration factors for traffic data WIM systems is by setting the mean GVW to zero. For traffic data and enforcement and enforcement-only WIM systems, the most common calibration approach is to set the combined errors for GVW and individual axle loads to zero. About 16% of the agencies that operate traffic data WIM use regression for computing calibration factors. Most agencies do not compute multiple calibration factors corresponding to different traffic speeds. 3

Many agencies use traffic stream WIM data analysis techniques to monitor the calibration status of WIM systems. An indication by such monitoring that the calibration of a system is changing is used as a trigger for performing an on-site validation/re-calibration using one of the two methods described earlier. Some agencies however use this approach as the only WIM calibration method. The majority of agencies that operates dual-use WIM systems download data automatically, whereas the majority of agencies that manages enforcement screening WIM systems do so manually. The actual WIM data QC analysis frequency ranges from daily to monthly, or it is decided on the basis of personnel availability or perceived calibration need. It is performed by a combination of manual or automated means or a combination of the two. Typically, an initial series of data checks is made to determine any operational prob- lems (e.g., vehicle errors, system errors, and unclassified vehicles). With few exceptions, almost all the agencies that responded believe that WIM data QC is capable of identifying sys- tem operational problems. The agencies were asked which traffic stream vehicle types are used and which of their characteristics are monitored in performing calibration monitoring. A large majority of agencies, regardless of WIM data application, focus their traffic stream WIM data analysis on Class 9 trucks or, more specifically, on 3S2 trucks. The most common load-related truck properties being monitored are the following: • Steering axle load, • Left-side/right-side wheel loads of the steering axle, • GVW for empty versus loaded trucks, and • GVW by vehicle speed. Interestingly, the steering axle load standard deviation (SD) and the GVW SD are monitored mostly by agencies that manage enforcement screening WIM systems. The most common dis- tance measure monitored is the axle spacing of the tractor tandem axles of 3S2 trucks and, less frequently, the total wheelbase versus the sum of the axle space data. Approximately 5% of the agencies that responded indicated that they take no further action when WIM data QC suggests calibration drift. 4