Automated Procedure to Identify Concrete Defects from Impact-Echo Data

In this study, a procedure for interpreting impact-echo data in an automated, simple manner for detecting defects in concrete bridge decks is presented. Such a procedure is needed because it can be challenging for inexperienced impact-echo users to correctly distinguish between sound and defective concrete. This data interpretation procedure was developed considering the statistical nature of impact-echo data in a manner to allow impact-echo users of all skill levels to understand and implement the procedure. The developed method predominantly relies on conducting segmented linear regression analysis of the cumulative probabilities of an impact-echo data set to identify frequency thresholds distinguishing sound concrete from defective concrete. The accuracy of this method was validated using two case studies of five slab specimens and a full-scale bridge deck, each containing various typical defects. The developed procedure was found to be able to predict the condition of the slab specimens containing shallow delaminations without human assistance within 3.1 percentage points of the maximum attainable accuracy. It was also able to correctly predict the condition of the full-scale bridge deck containing delaminations, voids, corrosion damage, concrete deterioration, and poorly constructed concrete within 3.5 percentage points of the maximum attainable accuracy.

Automated summary

In this study, a procedure for interpreting impact-echo data in an automated, simple manner for detecting defects in concrete bridge decks is presented. The developed procedure relies on statistical analysis of the data to identify frequency thresholds distinguishing sound concrete from defective concrete. The accuracy of this method was validated using case studies, and it showed high predictive capabilities for detecting different types of defects.