Prepared for possible presentation at the 1997 Annual Meeting of the Transportation Research Board and possible publication

Selection of Automated Pavement Distress Evaluation Technology: Iowa DOT Case Study

by

Omar G. Smadi
Pavement Management Specialist
Center for Transportation Research and Education

Brian R. McWaters
Pavement Engineer
Iowa Department of Transportation
800 Lincoln Way
Ames, Iowa 50010

Kevin B. Jones
Special Investigations Engineer
Iowa Department of Transportation
800 Lincoln Way
Ames, Iowa 50010

Robert L. Gumbert
County Engineer
Tama County
1002 E. 5th
Tama, Iowa 52339

Randall M. Krauel
Public Works Director and City Engineer
City of Carroll
112 E. 5th Street
Carroll, Iowa 51401

T.H. Maze
Director and Professor
Center for Transportation Research and Education

ABSTRACT

This paper describes the methodology used by Iowa to select a provider of automated pavement distress data collection services. Iowa is collecting data for the state's federal-aid-eligible highways on a statewide basis and storing the data in a statewide pavement management information system. The information will be later distributed to Iowa's cities and counties and the Iowa Department of Transportation for the pavement management of highways under each governmental entity's jurisdiction. The paper discusses the procedures and the measures used in the selection process.

Introduction

To meet the pavement management system requirements of the Intermodal Surface Transportation Efficiency Act (ISTEA) of 1991, the Iowa Department of Transportation, in cooperation with Iowa local and regional governments, developed a program of activities to support the development of a statewide pavement systems. Later, the National Highway System Designation Act of 1995 made the implementation of pavement management systems optional. However, Iowa's transportation management systems (all the ISTEA-required management systems and the required traffic monitoring system) policy committee agreed to continue the development of the statewide process but to allow regional governments the option of not participating in the statewide process.

ISTEA required that states apply a pavement management system to all federal-aid-eligible highways. In Iowa, the Iowa Department of Transportation's network of highways is only about half of the entire federal-aid-eligible system. The remainder is under the management of local governments (cities and counties). Iowa's approach to the management of the federal-aid-eligible system is to manage the pavement included in the National Highway System (NHS) portion of federal-aid-eligible highways and the non-NHS portion in two separate statewide systems. Also Iowa's approach is to centrally develop the non-NHS federal-aid-eligible highway pavement management system for the entire state. Local, regional, and Iowa state governments will use the information originating from the statewide system to assist in managing highways under each governmental entity's jurisdiction. The advantage of this approach is that statewide system supports the costs of developing a procedure for collecting data, managing the database, and calibrating pavement management software so that each governmental entity does not have to perform these functions on its own. With the pavement manage tools and pavement condition data, local, regional, and the state governments can develop their own pavement maintenance, rehabilitation, and reconstruction programs.

Iowa's program to develop and implement the statewide system is the Iowa Pavement Management Program (IPMP). One of the key objectives of the IPMP is to create an unbiased information system to support pavement management decision-making by all levels of government. In an effort to minimize the possibility of bias in the process, the IPMP is supervised by a task force with representatives from state, regional, county, and city government. The Center for Transportation Research and Education (CTRE) at Iowa State University was selected as an impartial organization to support the information system's development. In addition, the supervisory committee believes that bias can be kept to a minimum if pavement condition data collection follows standard processes and if data collection is automated, as opposed to data collection through visual inspection by inspectors.

The purpose of this paper is to describe the process used for determining the pavement condition data elements to be collected, the selection process used to select a provider of pavement data collection services, and the quality control process being used to monitor the data collection service provider. Data collection started recently and data are being collected for Iowa's non-NHS federal-aid-eligible system during 1996. The plan is to collect data for half the system each year and, therefore, the entire statewide system will not be completed until 1997. Undoubtedly, the data collection structure and the technology employed to collect pavement condition will evolve in the future. However, all parties involved in the IPMP realize that developing the information system is a long-term effort and that functions within the system will continuously be re-engineered to become more effective. The following sections describe the starting point used for the pavement condition data collection element of the IPMP.

Categories of Data Collection Issues

The initial work conducted on the IPMP identified five categories of data issues which must be resolved before the system could be developed. They included design and development of the pavement network inventory, project and construction history, pavement condition survey, traffic counts, and database management.

The pavement network work inventory information resides in the Iowa DOT's statewide highway database and, although the state's database is somewhat cumbersome to work with, the data are available. In all cases, however, county engineers and city public works directors were asked to verify and modify, if necessary, the information contained in the state database before it was placed in the IPMP database.

Project and construction history for the state system is available in the state's database, and city and county staff provided project and construction information for the portions of the system within their jurisdiction. Traffic data are available through the state's traffic monitoring system and will be augmented by ground counts collected by local governments within their jurisdictions.

CTRE is currently developing a statewide Geographic Information System (GIS) database which will be populated with the data elements collected. To integrate the various data elements, the database has dynamic segmentation capabilities.

The only data issue that could not be resolved through the use of internal resources is collection of pavement condition data, particularly the automated collection of pavement condition data.

Automated Pavement Condition Data Collection Issues

The IPMP supervisory task force appointed a subcommittee of state, city, and county representatives to develop the criteria for pavement condition data collection. Initially, the committee concentrated on six issues. They were the following:

1. Identifying categories of pavement types based on paving materials, where pavements in each category perform similarly, or similarly enough, to be included in the same pavement performance models. The categorization must be amply broadly to allow its application across all three levels of jurisdictions (i.e., state, city, and county roadways).
2. The requirements for pavement condition data elements (distress types) necessary to support network-level decision-making at each level of jurisdiction.
3. The requirements for pavement condition data elements (distress types) necessary to support project-level decision-making at each level of jurisdiction.
4. The coverage of the data collection (e.g., could data be collected for a representative sample, or was it necessary to have 100 percent coverage).
5. The necessary frequency of data collection (e.g., collect pavement condition data every year or less frequently).
6. The feasibility and effectiveness of using automated data collection equipment to collect condition data to support the data requirements for pavement management at each level of jurisdiction.

Condition Data Collection Issue Recommendations

The IPMP subcommittee developed recommendations to address all six of the pavement condition data collection issues. The following are their recommendations:

1. As more experience is gained in dealing with pavement managed by jurisdictions at all three levels, categories of pavements may be adjusted or refined. As a preliminary recommendation, only four types of pavement were categorized for independent treatment in the IPMP. The categories selected were 1) all types of portland cement concrete (PCC) pavement, 2) composite pavement (COMP) consisting of PCC pavement with an asphalt cement concrete (ACC) overlay, 3) ACC pavement, and 4) bituminous treated (BT) surfaces (e.g., chip and seal coat surfaces).
2. The subcommittee's requirements for data collection elements (distress types) for supporting network-level pavement management decision-making are shown in Table 1.
3. The subcommittee's requirements for data collection elements (distress types) for supporting project-level pavement management decision-making are shown in Table 2.
4. The recommended condition data collection frequency and coverage protocol were the result of compromise between data collection costs and thoroughness and currency of data. The subcommittee recommended collection in one direction for two-lane roads, with north bound and east bound as the primary directions for data collection on two-lane roads. On multi-lane highways, data will be collected in the outside lane in the north and east bound directions and the inside lane in the south and west bound directions. Test section length was selected to be 0.5 km in urban areas and 2.0 km in rural areas, regardless of the length and the location of the pavement management sections. It was also recommended that data collection frequency for the entire network would be once every two years, where half the network is covered every year. Pavement condition data will be collected for all pavements located within a regional government jurisdiction during one year.
5. The subcommittee recommended that the services of an automated distress data collection service provider be sought and agreed to serve as a vendor selection committee.

Table 1: Vital Network-Level Data Elements by Pavement Type and Jurisdiction

	Pavement Type
Jurisdiction	PCC	COMP	ACC	BT
	Ride	Ride	Ride	Ride
COUNTY SYSTEM	Joint Distress & Transverse Cracking	Joint Distress & Transverse Cracking	Block/Alligator Cracking Transverse Cracking	Potholes
	Ride	Ride	Ride	Ride
CITY SYSTEM	Joint Distress & Transverse Cracking Patching	Joint Distress & Transverse Cracking Potholes	Transverse/Longitudinal Cracking Block/Alligator Cracking	Potholes
	Ride	Ride	Ride	Ride
STATE SYSTEM	Joint Distress & Transverse Cracking	Joint Distress & Transverse Cracking Rutting	Transverse Cracking Rutting	Rutting

Table 2: Project-Level Data Elements by Pavement Type and Jurisdiction

	Pavement Type
Jurisdiction	PCC	COMP	ACC	BT
	Ride	Ride	Ride	Ride
	Joint Distress	T-Cracking	T-Cracking	Potholes
	T-Cracking	Rutting	Block/Alligator	Patching
COUNTY SYSTEM	Faulting ASR/D-crack	Bleeding	Rutting Bleeding	Block/Alligator T-Cracking Rutting Bleeding
	Ride	T/L-Cracking	T/L-Cracking	Potholes
	Joint Distress	Ride	Ride	Block/Alligator
CITY	T/L-Cracking	Potholes	Block/Alligator	Ride
SYSTEM	Faulting	Patching	Potholes	Patching
	Patching	Rutting	Patching	Rutting
	ASR/D-Cracking		Rutting	Bleeding
			Bleeding
	Ride	Ride	Ride	Ride
	Joint Distress	T-Cracking	T-Cracking	Potholes
STATE	T-Cracking	Rutting	Block/Alligator	Patching
SYSTEM	Structure	Structure	Structure	Block/Alligator
	Patching	Patching	Patching	T-Cracking
	Faulting	Bleeding	Rutting	Rutting
	ASR/D-Cracking		Bleeding	Bleeding

Selection of Automated Pavement Condition Data Collection Service Provider

Based on the frequency and coverage scheme developed for pavement condition data collection, an estimate was developed for the number of kilometers to be tested categorized by urban and rural areas and by two-lane and multi-lane highway, and the number of test sections to be surveyed. This estimate was used to estimate the level of work involved collecting data for the statewide network. The steps leading to the selection of a vendor were:

1. Five prospective vendors were contacted and given the estimated level of work required for data collection (e.g., kilometers covered, number of test sections, etc.). The five vendors were asked to provide general information on their firm and the technology they use for data collection, and to provide a general estimate of the cost of collecting data for the entire statewide system.
2. The Federal Highway Administration (FHWA) conducted two tests of automated distress data collection equipment in Texas in 1993 and North Carolina in 1994. It was hoped that the results of these two tests could be used to steer the selection of a vendor in Iowa. Unfortunately, neither FHWA test provided results which were amply conclusive to provide a basis for a selection decision. Following the analysis of the FHWA results, it was decided to ask the prospective vendors to provide a demonstration of their technology in Iowa to familiarize state, city, and county personnel with their equipment and its capabilities.
3. The next step was to further investigate potential vendors. This involved further discussion with vendors regarding the technology they employ, their availability to collect data in Iowa, types of working relationships they encourage with clients (e.g., purchase of the data collection service only, purchase of the equipment only, or some mixture of a equipment purchase and service contract), their experience with other clients, and the names of client contacts. Following the interviews with the vendors and discussion with other clients, two of the five vendors were dropped from further consideration. The reasons for dropping the two vendors were that one did not have experience with similar clients, and the cost of the other's technology was prohibitive.
4. The three remaining vendors were all invited to Ames, Iowa, to compete in a test of their ability to measure distress on the same pavement sections. The competition involved test sections of 0.5 km in length. In total there were four ACC and four PCC sections in a range of conditions. Prior to the vendors surveying the test sections, the sections were manually inspected and categorized based on the Strategic Highway Research Program's (SHRP) distress manual. The criteria used for manually evaluating the distress in the test sections are listed in Table 3. The three vendors surveyed the same test sections, and the vendors' reduced data were compared to the results of the manual inspection.
5. The initial analysis of each vendor's evaluation of the test sections indicated that no one vendor has superior accuracy in measuring all distress types. For example, one vendor may have been able to accurately identify the extent of transverse cracking but not the extent of D-cracking. Because none of vendors was the most accurate in all categories, the committee developed weights for both the relative importance of each distress type and other contract performance measures (e.g., cost) on a scale from one to five. Next, the committee ranked each vendor's performance in all sixteen areas on a scale from one to ten. The weights were multiplied by the rankings of vendor performance and totaled for each vendor to derive a score for each vendor. The vendor with the highest sum was selected.

Table 3: Distresses Measured in Test Sections

PCC Pavements
Joints - D-Cracking	Extent: Severity:	Number of joints with D-Cracking SHRP definition for Moderate and High
Joints - Spalling	Extent: Severity:	Number of spalling joints SHRP definition for Moderate and High
Transverse Cracking	Extent: Severity:	Number of transverse cracks SHRP definition for Moderate and High
Patching	Extent: Severity:	Area and number of patches Distress or no distress
ACC Pavements
Transverse Cracking	Extent: Severity:	Number of transverse cracks SHRP definition of Low, Moderate, and High
Longitudinal Cracking	Extent: Severity:	Length of longitudinal cracks Sharp definition for Moderate and High
	Determine extent and severity for wheel path and non-wheel path
Block Cracking	Extent: Severity:	Area of block cracking SHRP definition of Moderate and High
	Change in the definition of the piece size from 0.1 to 1.0 sq. meter
Alligator Cracking	Extent: Severity:	Area of alligator cracking SHRP definition for Moderate and High
Potholes	Extent: Severity:	Number of potholes Distress or no distress
Patching	Extent: Severity:	Area and number of patches Distress or no distress

Automated Data Collection Quality Control

Beyond the quality control measures taken by the vendor, the IPMP developed its own quality control measures. The IPMP quality control measures involve first identifying standards for performance with which to compare the performance of the measurements made throughout the state as the vendor conducts its survey. Quality control followed the steps listed below:

1. A set of control sites were selected to validate and calibrate the vendor's equipment. The results of the vendor's measurement of roughness, rutting, and surface distress were then compared to those determined through the manual surveys to determine benchmarks for the accuracy of the vendor's pavement condition measurements. The pavement distresses included for performance benchmarking are longitudinal, transverse, and alligator cracking for ACC pavements and transverse cracking and D-cracking for PCC pavements.
2. Control sections were established throughout the state at regular intervals during the data collection period and manually surveyed. The results of the vendor's evaluation will be compared to the results of the manual survey to determine if the accuracy of the vendor's measures are within the performance benchmarks. In addition to the control sections established for the IPMP, the vendor's measurements will be compared to the distress measurements made at SHRP pavement test sections on the Iowa DOT network.
3. Finally, random selected test sections will be sampled to determine if the vendor's performance is conforming to performance benchmarks.

Once the data have been reduced by the vendor and validated, the pavement condition will be loaded into a IPMP GIS database. Because the database has dynamic segmentation capabilities, pavement condition data and other pavement management can automatically be summarized by pavement management sections. CTRE will then make data sets, aggregated to the pavement management section, available to Iowa state, cities, counties, and regional governments for highways within their jurisdictions. Each jurisdiction will then be able to use the data for its own project- and network-level analyses using the IPMP selected pavement software, another software package selected by the jurisdiction, or a manual pavement management procedure.

The IPMP will select and purchase software for both project- and network-level analyses and make the package available for use by Iowa jurisdictions. Clearly, there is an advantage for regional and local governmental agencies to adopt the selected package since the IPMP will be responsible for calibrating the model and training of their staff.

Conclusions

The process described in this paper for the evaluation and selection of automated distress equipment has worked well for the IPMP. The iterative process of becoming more familiar with the technology at each step allowed members of the selection committee to become knowledgeable about automated distress surveying functions and more comfortable making selection decisions. Also, involving representative from all levels of governments was successful in developing support for the process from the various constituents. The same selection process will be used to select project- and network-level pavement management software supported by the IPMP.

Disclaimer

The opinions expressed and conclusions presented are those of the authors and do not necessarily reflect those of the organizations they represent.