Assessing the load carrying capacity of concrete anchor bolts using non-destructive tests and artificial multilayer neural network

Abstract

Concrete anchor bolts play a key role in architecture, construction, mechanical and petrochemical industries. Although much work in the past has been conducted to evaluate the pull-out strength of these bolts using destructive tests, so for very little progress has been achieved in developing a non-destructive testing method that can be used to estimate their pull-out strength. In this regards the presented research work details a truly non-destructive testing procedure that can be adopted to estimate the pull-out strength of concrete anchor bolts. In the presented experimental research work the author details the procedure of combining the Schmidt hammer rebound number along with the ultra-sonic pulse velocity test to identify anchor bolts that have defects and thus lead to lower pull-out strength. 48 anchor bolts of 16 mm and 20 mm diameter with embedment lengths of 50 mm and 70 mm were tested. Ultrasonic pulse velocity readings were recorded before and after the application of Schmidt hammer rebound reading. It was observed during experimentation that samples which showed lower rebound values also depicted reduced ultrasonic pulse velocity for the fixed path length. Since the anchor bolt load carrying strength depends on its diameter, embedment length, its alignment and bond between the steel and concrete, hence by combing the two non-destructive testing methods the researchers were able to identify faculty anchor bolts and successfully access their pull-out strength. Furthermore, an artificial neural network (ANN) was developed to conduct sensitivity analysis of the various parameters effecting the load capacity of anchor bolts in order the identify the most influential factors. A high level of accuracy was achieved between the experimental and predicted load capacity using the developed ANN.