.png){kind=icon}
.png){kind=icon}
Return
Non-destructive strength evaluation of existing concrete structures: from empirical models to machine learning
.png){kind=icon}
PRE
AI
J
J
E
K
H
DOI：10.1016/j.measurement.2026.122824.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Core sampling provides the most direct assessment of compressive strength in existing concrete structures, but its application is limited by local damage, restricted sampling locations, and the difficulty of obtaining representative samples. Accordingly, non-destructive testing (NDT) and partially destructive methods have been widely used through empirical correlations between measured parameters and compressive strength, generally requiring calibration with core test results. Although combined NDT approaches can improve prediction accuracy by integrating methods with different physical bases and complementary sensitivities, their reliability still depends on material and environmental conditions, calibration quality, and model applicability. Recent advances in machine learning (ML) provide a promising framework for capturing complex and nonlinear relationships between NDT responses and compressive strength beyond predefined empirical equations. This review comprehensively summarizes empirical and ML-based approaches for estimating the compressive strength of existing concrete structures, particularly emphasizing how NDT inputs, and mixture-related variables contribute to prediction performance. Furthermore, future research directions are discussed to address the scarcity of standardized field data and improve field-based uncertainty assessment for the practical application of ML-integrated NDT to compressive strength evaluation.
Keywords:
Concrete
Non-destructive testing
Compressive strength
Machine learning
Journal
IF:
5.6
Papers:
1.9W
Citations:
5.4W
Organization
Cited Papers