Predictive Probability of Detection Curves for Concrete Compressive Strength Assessment Using the Rebound Hammer Test

The rebound hammer test is a widely known, rapid, cost-efficient non-destructive method, commonly used for assessing in-place concrete compressive strength. The rebound hammer test uses empirical correlations to provide indirect estimations of compressive strength based on surface hardness. Concrete degradation is reflected by a change in rebound numbers. However, rebound numbers are highly sensitive to factors such as material variability and surface conditions, leading to high scatter in compressive strength estimates and limiting the overall reliability of the method itself for the detection of compressive strength loss. Well-established reliability assessment tools for non-destructive testing, such as the probability of detection (POD), have seen limited application in this technique. Conventional POD would require extensive experimental data across a range of strengths, rendering its implementation impractical. The so-called predictive probability of detection (P-POD) is a model-assisted method of demonstrated capability for assessing the reliability of non-destructive testing methods using only data from a single reference dataset. In this paper, the P-POD method is applied to rebound hammer testing for the first time. A reference curve, derived from over 2300 data points, is used for deriving an empirical model and for validating the accuracy of the P-POD method. The resulting absolute prediction errors are between 0.22 and 16.55%. The paper addresses key challenges like deviations from methodological assumptions and the lack of analytical models. The findings translate into a new framework for evaluating the reliability of rebound hammer tests, with significant potential to enhance confidence in non-destructive compressive strength measurements.