NDT-Driven Data Classification Approach for Predicting In situ Compressive Strength of Bamboo-Reinforced Concrete

Bamboo reinforcement is widely available in all tropical and subtropical regions and offers renewable alternative to steel reinforcement in concrete structures. It is mainly used in low-cost housing and temporary structures due to its ability to reduce construction cost and enhance the strength of unreinforced section. For sustainable construction, researchers are gradually exploring non-conventional reinforced concrete to determine its strength and mechanical properties using NDT techniques, however, correlation models for BRC concrete are still lacking. This study aims to established a correlation between results obtained from NDT - Schmidt rebound hammer (RS) test and - Ultrasonic pulse velocity (Ѵ) test and destructive testing on BRC to obtain its compressive strength (f _BRC ), using machine learning techniques (ML) in Matlab. While correlation data for conventional concrete are widely available, these models cannot be directly applied to BRC due to significant density variation. An experimental program involved casting 225 BRC specimens: beams measuring 700 mm × 150 mm × 150 mm and cubes of 15cm × 15 cm all using 30 grade concrete. The beams include 3.5% sand-coated longitudinal bamboo reinforcement with a nominal cover of 20 mm. RS and Ѵ data were collected from beams, while destructive test was conducted on companion cube specimens. The collected data was than imported into MATLAB Simulink, where supervised ML was applied using the Classification Learner app. The support vector mechanism (SVM) effectively correlated R _S and f _BRC while decision tree classifier classified Ѵ data set. The correlation equation was then developed to predict actual off _BRC .