Non-destructive strength evaluation of concrete: Analysis of some key factors using synthetic simulations

Abstract

Non-destructive techniques (NDT) like rebound hammer (RH) and ultrasonic pulse velocity (UPV) are widely used in conjunction with destructive techniques (core tests) for assessing the concrete strength in existing buildings. The methodology consists in fitting regression models between NDT techniques and destructive tests on a limited number of cores. The quality of the model is affected by many influencing factors such as: the number of cores, the quality of NDT measurements, the variability in concrete strength, the existence and magnitude of possible uncontrolled factors (like saturation rate) and the combination of techniques. In this paper, the effects of these factors are studied using a synthetic simulation approach in order to well understand them and consequently to develop a methodology for improving the quality of strength assessment. In order to assess the quality of fitted model and its ability to estimate strength, RMSE and r2 errors are calculated and it is found that the calculation of r2 alone may give misleading indication since r2 is very sensitive to the scattering of the explanatory variable. Another important result of the present study is that there is a critical minimal number of cores which makes the combination efficient while for a lower number the use of single technique is preferable. This number depends on the qualities of the two techniques to be combined.