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 43
43
CHAPTER FIVE
CASE STUDIES
This chapter documents the data management practices of Subjective Rating
a select group of transportation agencies. The case studies
include an agency that conducts most of the data collection in- In the last step, the operator verifies that the data have been
house and three agencies that contract most of the network- saved and inputs a subjective evaluation of the crack detec-
level pavement condition data collection with data collec- tion process, following a set of recommendations summa-
tion service providers and use different quality acceptance rized in Table 5. For example, if more than 90% of the sta-
approaches. tions reviewed pass the crack detection criteria and all the
data were saved in the hard drive and the network, the batch
is given a "good" rating.
MARYLAND
MDSHA uses an in-house automated system to measure Repeatability and Accuracy Examination
smoothness, rutting, and cracking, in addition to other data
such as right-of-way images, longitudinal and transverse MDSHA has also implemented a quality control program to
slopes, and GPS coordinates (66). The network-level data monitor data repeatability and the accuracy of test equipment
collection process includes: (1) data management, (2) pre- using a test loop. The experiences of Virginia and other states
processing, (3) processing, (4) quality control, (5) quality have been incorporated into the program. The test loop is
acceptance (denoted as quality assurance), (6) classification measured 20 times at the beginning of each data collection
and rating, and (7) data reduction. The quality control and season, then run once every three weeks during the season.
acceptance procedures for the automated crack detection are To analyze data accuracy of a particular test-loop run, the
discussed by Groeger et al. (66). moving average of all the previous runs, including the initial
10-run results, is considered as the reference value for that
Quality Control particular test. This test is also compared with the previous
one to check data repeatability.
As discussed in chapter four, the quality control plan includes
checks to verify that all fields are processed, reviews of section-
Quality Acceptance
level data in a search for abnormalities, and checks to verify
that all the data have been saved. The reviewer then inputs a Quality acceptance is done by a quality assurance auditor,
subjective evaluation of the crack detection process (good,
who is not the equipment operator. This process verifies that
fair, or poor).
the data collection and quality control processes have been
conducted properly. The independent auditor checks the data
Section-Level Review management spreadsheet, verifies that the data are complete,
verifies that the data have been saved and backed-up, and re-
The section-level review is conducted by looking at the total checks a random sample of 10% of the data files collected
quantity of cracking by station and searching for abnormali- implementing the same procedure used for quality control.
ties; for example, a road segment with many spikes. The oper- This sample includes any files that have comments that are
ator reviews the segments with abnormalities by manually out of the ordinary. If there is one discrepancy in a file, it is
superimposing the cracks detected with the actual pictures. At noted on the data management form. If more than two dis-
the time the plan was published, this process was applied to crepancies are detected, 50% of the file is reviewed to deter-
approximately 50% of the pictures and the goal was to recog- mine if there is a systematic error. If more than 10% of the
nize 80% of the cracks. The operator also looks at the last quality acceptance samples have discrepancies, consideration
rehabilitation date and verifies that the amount of cracking is is given to repeating the crack detection process (66). All
consistent with the age of the surface; for example, a pave- data are backed-up on the server on a daily basis and copied
ment recently rehabilitated would have little cracking. to tapes once a week.
OCR for page 44
44
TABLE 5
EXAMPLE OF QUALITY CONTROL RATING MATRIX USED BY MDSHA
QC Procedure Good Fair Poor
Stations Processed <100%
Criterion 1: Detected > 80% of Cracking >90% Stations 70%90% Stations <70% Stations
Data Saved to Hard Drive Yes No
Data Saved to Network Yes No
Source: Groeger et al. (66).
Classification and Rating match PMS specifications, and the pavement condition index
(PCI) is calculated.
After the data have been processed for crack detection, qual-
ity control, and quality acceptance, the next step in the process
is to classify and rate cracks using an automated process. Time-Series Comparisons
Cracks are classified as longitudinal and transverse, their
locations in the pavement (outside wheel path, inside wheel MDSHA also monitors network-level time-series data with
path, center, left edge, or right edge) are determined, and the the help of a software tool developed and implemented in
severity is rated as low, medium, or high using the AASHTO 2004. This quality acceptance tool checks reasonableness of
cracking protocol definition and the crack width determined the data trend. The tool is routinely used to test each data sub-
by the system. mission and includes two steps: data trend monitoring and
data quality investigation.
Data Reduction Data Trend Monitoring MDSHA uses the percentage
of the network in acceptable condition--called acceptable
The cracking data are reduced to a condition rating of 0 to rate--to monitor the network-level pavement condition. A
100 and assigned a condition state of very good, good, fair, software program summarizes the acceptable rates in terms
mediocre, or poor using a software program known as the of IRI, rutting, cracking, and friction individually by route,
MDSHA Automated Distress Analysis Tool. This tool also county, district, or statewide. A table is prepared for each of the
performs the final quality check through a suite of logic, pavement condition indicators along with the values obtained
range, and trend checks on the data and generates a progress for the last five years. As an example, Table 6 shows 2 years
report to document the pace of data collection and data pro- of the 5-year comparison for IRI for a sample of routes. For
cessing. If the checks detect any problems, the file is flagged each year, columns 1 through 5 indicate the percentage of
and a note is output to an error log. If the file passes all the pavements in each condition state and the last column the
checks, data are converted into U.S. units and reformatted to percentage of road in acceptable condition (states 1 through 3).
TABLE 6
IRI TREND MONITORING BY ROUTE
2007 2006
Route Acc. Acc.
1 2 3 4 5 1 2 3 4 5
Rate Rate
IS68E 41.04 36.57 18.16 3.23 1 95.77 30.6 37.81 25.87 4.23 1.49 94.28
IS68W 33.83 41.79 20.65 3.48 0.25 96.27 24.38 43.78 27.11 2.99 1.74 95.27
MD35N 0 33.33 58.33 4.17 4.17 91.67 0 33.33 58.33 8.33 0 91.67
MD35S 0 33.33 58.33 4.17 4.17 91.67 0 41.67 50 4.17 4.17 91.67
MD36N 12.97 40.61 29.35 7.85 9.22 82.94 9.62 39.52 34.02 9.28 7.56 83.16
MD36S 12.24 42.52 28.91 7.48 8.84 83.67 11.6 41.98 32.08 7.85 6.48 85.67
MD47N 0 11.76 58.82 23.53 5.88 70.59 0 17.65 58.82 23.53 0 76.47
MD47S 0 5.88 76.47 17.65 0 82.35 0 11.76 70.59 17.65 0 82.35
MD49E 0 9.09 72.73 9.09 9.09 81.82 0 10 80 10 0 90
MD49W 0 10 90 0 0 100 0 10 90 0 0 100
Source: W. Xiong, personal communication, 2008.
Note: Columns 15: the percentage of pavements in each condition; last column percentage of road in acceptable
condition. In bold, routes where pavement condition indicators differ more than 2% from previous year.
Acc. = acceptable.
