Using expert opinion to quantify accuracy and reliability of nondestructive evaluation on bridges
Date
2013
Journal Title
Journal ISSN
Volume Title
Abstract
Bridge inspection is an important phase in the bridge management process. In 2009, a joint American Society of Civil Engineers Structural Engineering Institute (ASCE/SEI) and AASHTO Ad-Hoc group was created to identify the issues to guarantee bridge safety and to study how current bridge practices could be improved for the future. This group recommended: "A more rational, risk-based approach to determining the appropriate inspection intervals for bridges is needed, as opposed to a set twenty-four month cycle for all bridges". The committee also recommended a wider use of NDE methods. The difficulty in increasing the use of these NDE methods is the increased costs and time spent using them. One way to deal with this is to implement risk based planning to determine the appropriate inspection frequency, scope, intensity, and methodology. To do this, though, the accuracy, reliability, bias, and cost of each test must be quantified. This study attempts to quantify these parameters for common bridge NDE methods. This was done by two methods. Firstly, a literature review was performed to determine common NDE methods being used and studied for bridge inspection and statistical data was found for these methods. To complement the literature, a four round Delphi method survey was conducted with experts in the NDE bridge field in order to develop a broader range of data that matches real life practices. All of the data was then analyzed and conclusions were drawn to quantify the accuracy, reliability, bias, and various costs incurred for common bridge NDE methods. Based on these results it can be seen that most commonly used bridge NDE methods tend to be under biased and relatively repeatable. It was shown, however, that while inspectors seem to have a pretty good relative understanding of the variability in different tests, they tend to not have as clear of an understanding of the absolute scale of the variability. Furthermore, the accuracy of commonly used bridge NDE methods tends to be relatively variable with the average test measuring a true response between 80% and 85% of the time. Lastly the costs associated with the NDE methods examined here tended to be highly variable making this measure difficult to evaluate. However, by comparing the rankings of each of the four categories that were examined for each NDE method, it is possible to correlate the cost of a method to the bias, accuracy and reliability. This could lead to a more reliable risk-based approach to bridge inspection in the future.