Field Calibration of Non-Destructive Testing for Post-Fire Residual Strength of Reinforced Concrete

Abstract

Background: Non-destructive testing (NDT) using Ultrasonic Pulse Velocity (UPV) and Rebound Hammer (RN) is widely used for post-fire concrete assessment, but standard calibrations derived from laboratory specimens are unreliable under field conditions. A further challenge arises when a fire-damaged building contains multiple concrete grades (e.g., M20 and M25).

Objective: This study recalibrates single-method and combined (SonReb) NDT equations using field data from real fire-damaged buildings, tests whether M20 and M25 share a common NDT- strength slope, and quantifies between-building variability.

Methods: UPV, RN, and destructive core tests were conducted on 81 members (beams, slabs, columns) across 13 fire-damaged government buildings in Nepal (M20: n=64; M25: n=17). Calibrations were derived at pooled, M20, and M25 levels using ordinary least squares, evaluated with leave-one-out cross-validation (R²_LOOCV), and tested for slope equality via interaction terms. Between-building variance was partitioned using Intra Class Correlation (ICC).

Results: UPV (r=0.64) and RN (r=0.72) correlated with core strength. Slope-equality tests showed no significant grade interaction (p≥0.79), supporting a common slope with grade-specific intercepts. However, the M25 RN model failed cross-validation (R2LOOCV=−0.12). The M20 SonReb model performed best (R²_LOOCV=0.60; RMSE=4.49 MPa). Between-building differences explained 63% of strength variation (ICC=0.63).

Conclusion: NDT is suitable for comparative ranking of fire-damaged members but not for absolute strength estimation without site-specific recalibration using a limited number of destructive cores. The common-slope, grade-intercept approach is statistically defensible for mixed-grade buildings.