Return to Main Page

Design of a Long Term Pavement Monitoring System for the Canadian Strategic Highway Research Program

 

Section 1

OVERVIEW OF THE PAVEMENT MONITORING PROGRAM AND ITS CONTRIBUTION TO PAVEMENT DESIGN AND ASSET MAINTENANCE

1.0 OBJECTIVES OF THE REPORT

This report has a number of objectives. In the narrowest sense, it is a report detailing how to measure the life cycle cost of a particular pavement section given the information obtained from long term pavement performance monitoring. In a broader sense, this report sets forth an integrated, comprehensive pavement monitoring, performance prediction, and pavement management system. In particular, we have endeavored to begin with the long term pavement monitoring gameplan of the Strategic Highway Research Program and augment and modify it so that it meets Canadian needs. Our objective is to develop for Canada an immediately workable system that can be used to manage Canadian pavement assets effectively beginning immediately.

1.1 RELATIONSHIP OF OUR FINDINGS TO THE STRATEGIC HIGHWAY RESEARCH PROGRAM LONG TERM PAVEMENT PERFORMANCE MONITORING PROGRAM

The analysis described in this report has considered carefully and comprehensively the prior design work done by the long term pavement performance (LTPP) program of the Strategic Highway Research Program (SHRP) in the United States. We have borrowed ideas directly from the SHRP long term pavement performance project design when we believe those ideas increase the probability of success for the Canadian program. However, we have superseded those aspects of the SHRP long term pavement performance program that we believe compromise its prospects for success. In part, we have used SHRP as a "straw man" and have attempted to improve it systematically.

Our analysis has identified two serious flaws in the SHRP design, flaws that could obviate the ultimate success of the program. Our report discusses those flaws at some length and advances specific remedies. It is not our intent to denigrate the SHRP long term pavement performance program. On the contrary, we recognize the difficulty of the challenge facing SHRP and the seminal contributions SHRP has made. However, given our mission to design an integrated, coherent, workable long term pavement performance program for Canada, we have sought to borrow from the SHRP­LTPP program where appropriate yet to supersede those portions of the SHRP­LTPP program we believe to be defective, unworkable, or limiting. We reiterate that our paramount objective is to design a coherent and integrated long term pavement performance program for Canada. We will draw from the United States SHRP program as we deem it appropriate for Canada.

We believe that certain aspects of SHRP will be adequate in supporting prudent and effective long term pavement monitoring and delivering key information to decision makers. However, we have identified and proposed alternative approaches for two areas of the SHRP­LTPP program we believe to be problematic: (1.) the statistical methodology proposed and (2.) the limitations that the proposed statistical methodology will impose on pavement management systems that should use the SHRP­LTPP information.

We have been somewhat surprised at the rather cursory and presumptive discussion in the SHRP­LTPP documentation related to the statistical methodology that will be used to process the long term pavement performance data. The challenge of monitoring and managing pavement assets if far too complex and too important to accept without serious inquiry or design modification the use of standard, prepackaged classical statistical techniques. We are confident that "plain vanilla" classical statistical techniques will not work and may in fact prove to be an Achilles heel for SHRP­LTPP as presently conceived. This report gives a good deal of attention to explaining the classical statistical approach, indicating its limitations with regard to the problem at hand, and outlining and proposing a more advanced, more state­of­the­art methodology that overcomes these shortcomings and maximizes the probability of success for Canada.

Coincidentally, we have also been surprised at the lack of focus in the SHRP documentation on the question: Who are the customers of long term pavement performance monitoring? How will the information gained from long term pavement performance monitoring benefit those customers? In what way will those customers be better off as a result of receiving an expensive stream of long term pavement deterioration information over the next fifteen or more years? We have given a great deal of attention to answering these questions in this report, and we believe we have a workable, positive answer. Our analysis of these questions revolves around the notion that better information has zero value until and unless people use that information to make different decisions than they would otherwise make. Information has value only if it motivates changes in people's decisions. Long term pavement performance data has value only if it changes the way pavements are designed and maintained.

If we accept the proposition that pavement performance monitoring and pavement decision making are intimately coupled, it is clear the design of pavement management systems must be intimately conjoined with the design of long term pavement performance monitoring. In other words, the design of pavement management systems to be used by provinces and states has profound implications for the design of the long term pavement performance monitoring program designed to provide data to those systems. Conversely, the type of data assembled and the way it is manipulated has major implications for the design of workable pavement management systems. Long term pavement monitoring cannot be designed independently from pavement management, nor can pavement management be designed independently from long term pavement monitoring.

We have given a good deal of attention to the coordinated design of an integrated pavement monitoring, performance prediction, and pavement management system. We view the problem as a system problem amenable only to a system solution. It is not a problem that can be solved one piece at a time, gathering data to support in some vaguely defined way some vaguely defined pavement management system of uncertain specification. In this report, we have specifically tailored our proposed pavement monitoring program and subsequent statistical manipulation so that it can support probabilistic as well as deterministic pavement management systems.

1.2 KEY OBJECTIVE: TO SUPPORT IMPROVED PAVEMENT CONSTRUCTION AND MAINTENANCE DECISIONS

Much of the SHRP documentation from the United States articulates the objective of the long term pavement monitoring program as "recalibrating the AASHTO design equations" and "extending the life of pavements by 10 percent or more." The tacit assumption in such statements is that long term pavement monitoring only has value to the pavement design decision. We would sharply dispute this tacit assumption, for we feel it dramatically understates the benefits of long term pavement performance monitoring and statistical inference. Equally importantly in our view, and largely unstated and/or neglected in the SHRP documentation, is the fact that maintenance as well as design decisions will be guided by the results of the long term pavement performance monitoring program. To ignore maintenance while considering only design is to systematically understate the benefits of the long term pavement monitoring program and to perhaps omit important design considerations that can have tremendous value in guiding maintenance decisions. Our approach painstakingly seeks to consider design and maintenance decisions as an integrated asset management unit.

To illustrate our point by analogy, for a 767 jet would the service life be more affected by maintenance actions undertaken throughout the life of the jet or by the initial design? For a Mercedes Benz, would the service life be more affected by maintenance undertaken throughout the life of the car or by the initial design? Is the service life of the Golden Gate Bridge more strongly a function of the initial design or of the intensive maintenance strategy undertaken since the bridge was erected? Empirical observation across a range of assets over a range of industries suggests that the service life (and life cycle cost) of an asset is an inextricably intertwined function of both the initial design of the asset and the maintenance strategy applied over the life of the asset. Empirical work also suggests that people tend to systematically underestimate the long term benefits of routine maintenance and underestimate the costs of rehabilitation and refurbishing. While people pay lip service to the adage "A stitch in time saves nine," people do not behave according to the adage to a sufficient degree.

To concentrate long term pavement performance monitoring on initial design decisions as the SHRP­LTPP documentation implies is to dramatically understate the value of performance monitoring and the value delivered to pavement users from the program. Performance monitoring will give invaluable information that helps people make better maintenance decisions as well as better design decisions. However, in order to take advantage of such benefits, we must give careful attention to implementing pavement management systems that can accept as input the long term pavement monitoring data and can prudently and correctly coordinate design and maintenance decisions. The approach we have designed and outlined here is intended to offer a better integrated design and maintenance strategy one that is continually guided by long term performance data as it is assembled.

In brief, long term pavement performance monitoring should not be viewed strictly as "science expedition." The mission is not one of scientific inquiry, although scientific inquiry promises to be a byproduct of the design we will set forth. We recommend a colder, harder, more pragmatic view of long term pavement performance monitoring. Our view is that the objective of long term pavement performance monitoring is to assemble better information to support better decision making in the field and deliver that better information to decision makers so they can make better decisions immediately. Highway users should not be asked to wait until all the information is in and all the science can be checked out and debated in academic circles or research laboratories. Highway users need and deserve the best possible service now.

1.3 SMALL SAMPLE SIZE PROBLEMS WILL BESET THE CANADIAN LONG TERM PAVEMENT PERFORMANCE PROGRAM (C­LTPP)

There is no way to escape the small sample size problem in long term pavement monitoring; it is endemic. The care, cost, and time required for long term pavement monitoring program necessarily restricts the number of sites that can be considered. If we wish to estimate say fifteen or twenty parameters that prospectively affect long term pavement performance, classical statistical methods would require at least fifteen data points, i.e., fifteen sites per parameter, to achieve a bare minimum of statistical significance. To achieve a desirable level of statistical significance, one would require systematic, sustained, accurate, long­term monitoring of some 225­300 sites over the next fifteen or more years for each pavement type. Monitoring at such a broad scale is beyond the anticipated scope of the program and probably beyond the patience and financial resources of the political bodies that would be asked to provide fifteen or more years of continuous funding. Furthermore, as we shall continually reemphasize here, the world of pavements and pavement monitoring is far from ideal, necessitating the maximum conceivable number of sites. The complexity of the pavement situation all but guarantees that pavement monitoring faces perhaps the worst possible statistical situation.

The United States program has attempted to circumvent the small sample size problem in two ways. First, the U.S. program will monitor presently existing sites that are partway through their ultimate life. Second, the U.S. program will consider several different types of roadways

The approach we will recommend overcomes this difficulty by squarely addressing the small sample size problem. We have appealed to a rather unique technique for conquering the inevitable small sample size problem, a technique called Bayesian statistics. Much of this report describes the classical statistical approach (which incidentally is an integral part of the proposed SHRP long term pavement performance program), identifies its shortcomings, and shows how the Bayesian statistical approach overcomes those shortcomings with regard to the needs of long term pavement performance monitoring. We also give special attention to showing why the data acquired by the long term pavement performance program can be identical whether the classical statistical approach or the Bayesian statistical approach is selected. The difference lies not in the nature of the data gathered but rather in the calculations that are made using that data. In short, the "magic" is in the data processing, not the data assembled.

1.4 LONG TERM PAVEMENT PERFORMANCE MONITORING MUST BE DESIGNED TO SUPPORT BETTER PAVEMENT DESIGN AND MAINTENANCE DECISIONS

The flavor of the SHRP­LTPP documentation revolves around the notion, indeed the faith, that long term pavement monitoring can give

· the right data; data that is carefully controlled, uncorrupted, and understood.

· the right science. That is, the right data will allow us to identify the deterioration forces at work. For example, the right data will allow us to make the right calibration of the AASHTO design equations.

We doubt whether any monitoring program for an asset as heterogeneous, confounded, and complex as pavement can ever generate the right data and the right science. Too much can go wrong. Too much confounding can occur. Too much measurement error can occur. There are too many possible combinations of causal or independent variables. There are too many prospective combinations of design and maintenance alternatives to test them all to a sufficient degree.

The objective by which we have developed our design for Canada states the objective of long term pavement monitoring slightly differently. The system we have designed includes a pavement monitoring system and a coordinated and compatible pavement design and management system that

· help people make better pavement design and management decisions in the field.

and

· do so beginning with the very first data from the program and continuously improving thereafter.

We have commented in our own prior pavement design and management work that the only legitimate objective of gathering better data is to make better decisions. The only benefits people can claim from long term pavement performance monitoring are those embodied in future pavement design and management decisions. Gathering better data so that people feel more comfortable about the underlying deterioration science or so they can feel more comfortable doing the same thing they were going to do anyway is not a legitimate reason to fund a program as ambitious and expensive as long term pavement performance monitoring. Gathering better data is justifiable only insofar as that better data leads people to make different and better decisions sooner than they otherwise would.

We believe the C­LTPP program has the potential to help people make better decisions than they otherwise would, starting immediately with the inception of the program. With the proposed design, C­LTPP will not have to wait ten or more years for "sufficient data." To achieve immediate impact, however, requires the specific design features proposed here, i.e., the specific extensions of the SHRP­LTPP program we have conceived. It requires the integration of Bayesian statistical methods with quantitative pavement management techniques. There must be a clear and unequivocal path from long term pavement monitoring through statistical inference through pavement management decision making to the field where different and more effective strategies are executed. As the remainder of this report will show, our design provides this clear and unequivocal path.

 

(Continue)

Return to Table of Contents

Return to Main Page