Statistical analysis for evaluating concrete strength of existing structure using non-destructive and destructive test

Destructive tests for evaluating concrete compressive strength are costly and challenging in certain instances. Using a rebound hammer (RH) and ultrasonic pulse velocity (UPV), i.e., non-destructive methods for strength evaluation, proved more beneficial in all senses. However, calibrating the model between non-destructive testing (NDT) and compressive strength is essential for estimating strength. The reliability of this calibration is a crucial task that leads to selecting a minimal number of cores to be taken out (core) from a structure. The present study aims to identify and optimize the on-site reliability model. Extensive data from 275 core samples were obtained from the Construction Diagnostic Centre, Pune (India), which RH and UPV examined. The cores are taken from thirty existing RCC structures built between 1975 and 2005. The Root Mean Square Error (RMSE) and the coefficient of determination (R²) for single method (SM) and combined method (CM) are used to investigate the total number of cores needed for calibration. According to RMSEpred and R²pred, at least 6–8 cores are required to achieve the correct prediction phase with a CM rather than using SM. The CM leads to more reliable results than an SM with the least RMSE and higher R² values by analyzing 100 iterations for each number of cores (NC). Also, the CM shows more reliable results than the SM in the fitting and prediction phase. As a reasonable number of samples, 9 cores must be considered to converge for an SM, compared to 6–8 cores necessary for CM to estimate the strength precisely.