Click for next page ( 46

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 45
45 Similar summaries are prepared by the county, district, and for the entire state. If the acceptable rate for any of the pavement condition indicators differs more than 2% (routes MD47N and MD49E in the example) from the previous year, the record is highlighted for further investigation. Data Quality Investigation A data quality investigation is required for those sections in which the trend analysis indi- cates a potential data quality problem. This investigation aims to identify the reason for the suspicious rate change, determine if there was a problem, and, if there is one, find a solution to fix it or to prevent it from happening again. Historic treatment a information, test and equipment event records, pavement images, and weather conditions during testing are collected and analyzed to determine which factors may have contributed to the suspicious condition variation. If test operation or equip- ment condition is identified as a concern, a notice is sent to the data collection staff requesting that the data be re-collected or suggesting modifications to the data collection procedures. Data Collection Equipment Comparison After replacing its automated data collection equipment, MDSHA conducted a data comparison study to evaluate the b consistency of the data collected between the old and new FIGURE 20 VDOT yearly pavement condition devices. The two systems were used to collect data on a comparisons (96): (a) Pavement Condition Index; 250-mile loop. The smoothness (IRI), cracking, and rutting (b) Smoothness (IRI, in./mi). values for a sample of one hundred 0.1-mile segments were compared. The comparison showed that the two systems pro- duced similar IRI data, but statistically different rutting and shows the comparison for the overall PCI and IRI for 1996 cracking measurements. Cracking data were collected from and 1997 after removing all sections that received preserva- pavement images using a proprietary automated cracking tion treatments. The PCI plot pointed out a deficiency in the detection software tool. rating procedure used in 1997, which overestimated the PCI for the pavements in poor condition. The IRI plot also sug- To resolve the cracking data consistency problem, MDSHA gests a problem, because the smoothness was lower in 1997 initiated a study to compare the results of the two systems with than in 1996; this was attributed to the switch from ultrasonic reference ratings determined visually from the same pictures sensors to laser sensors. collected with the data collection systems for the same 100 seg- ments. This ground truth determination is critical for hardware The network-level comparison prompted a review of the and software calibration to improve data accuracy. pavement data collection approach, which helped enhance data quality requirements in successive years and establish formal quality assurance/quality control processes. Most sig- VIRGINIA nificantly, VDOT defined the following vision statement for VDOT has used different pavement distress data collection data collection "to collect pavement condition data with suf- methodologies over the past 15 years. These changes have ficient detail and accuracy to model deterioration and per- resulted in a continuous improvement process through which form multiyear planning with the PMS. Data variability for the department has gained significant experience and devel- each data element must be smaller than the year-to-year change oped sophisticated quality control and assurance procedures. in that element." VDOT collects data over 0.1-mile- (161-m)-long manage- ment units. The study also prompted the agency to require the cali- bration of smoothness measuring equipment against a ref- erence device and its verification against VDOT equip- Background ment, and pilot testing of a sub-network during the data collection contract inception phase. It also provided the Larson et al. (96) presents some interesting approaches for data that was used to develop precision (12%) and bias comparing time-history pavement condition data. Figure 20 (5%) criteria for the PCI.

OCR for page 45
46 Current Data Collection Practice conditions. Data collected by VDOT were used as reference values. The sites were used to establish the service provider's In 2005, after a formal solicitation process, VDOT contracted precision and bias, which in turn were compared with the with a service provider to collect, process, and deliver network- ones required in the RFP. level pavement condition data (51). The equipment specified included digital pavement imaging to a resolution of at least For calibration of the pavement distress measurements, 2 mm, laser measurements of longitudinal and transverse the service provider used an automated crack detection rating profiles, and automated or semi-automated distress quantifi- process and semi-automated ratings of the additional dis- cation. The potential service providers were required to pro- tresses. The reference distress surveys were conducted by vide documentation of their quality control plans for all aspects VDOT staff and the independent third party using the equip- of the project, ranging from equipment calibration through ment collected images. This effort also served to train all dis- data delivery. The selected service provider had an estab- tress raters, unify criteria, and made the necessary adjust- lished quality control plan, but added an outside third party ments to the process. Comparisons were made based on the to provide an independent verification and validation of the overall pavement condition index; the allowable difference data before delivery to VDOT for this project. The service was 10 points. provider-supplied quality process flow diagram (Figure 21) outlined the flow of data collection, data processing, quality control, independent validation and verification, and data Independent Verification and Validation acceptance processes. The verification and validation of the pavement distress data by an independent quality auditor was performed after Initial Calibration the service provider had completed all in-house quality control reviews and believed the data were ready for sub- The calibration of the service provider's longitudinal pro- mittal to VDOT. Acceptance criteria require that 95% of the file, transverse profile, and pavement distress measurement data checked fall within plus or minus 10 index points of processes was done using 13 known-location control sections. the third-party data. The third party evaluated a 10% ran- The control sites varied in length, smoothness, and distress dom sample of the pavement deliverables. This process Production Data Data Processing Collection Control Site - Semi-Auto Start up Process - Verification Sites Adequate - Automatic Internal QA - Control Sites - Image Quality (VDOT) - QA - Field QC - SOP NO NO Deliverables Deliverables Deliverables - Data - QC Report - QA Report ` - Report - Documents NO NO Independent Validation Deliver to VDOT &Verification - Deliverable Files via Batch PMS Database - 5% Data Review Pass ftp site Acceptance AMS Database - Data Completeness IV&V - Images via portable (VDOT) Video Database - Index Limits hard drive Deliverables Deliverables - IV&V Report - QA IV&V Report - Deliverable Table s - 0.1 mi Delivery Table - Homogeneous sections delivery table FIGURE 21 VDOT quality process flow diagram [after Shekharan et al. (51)].