Computational & Technology Resources
an online resource for computational,
engineering & technology publications
PROCEEDINGS OF THE SEVENTH INTERNATIONAL CONFERENCE ON THE APPLICATION OF ARTIFICIAL INTELLIGENCE TO CIVIL AND STRUCTURAL ENGINEERING
Edited by: B.H.V. Topping
Neural Network Modelling of Single Adhesive Anchors under Tensile Loading
A.F. Ashour+ and S.S.S. Sakla*
+School of Engineering, Design and Technology, University of Bradford, United Kingdom
A.F. Ashour, S.S.S. Sakla, "Neural Network Modelling of Single Adhesive Anchors under Tensile Loading", in B.H.V. Topping, (Editor), "Proceedings of the Seventh International Conference on the Application of Artificial Intelligence to Civil and Structural Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 54, 2003. doi:10.4203/ccp.78.54
Keywords: adhesives, anchors, fasteners, concrete, embedment, neural networks, prediction, capacity, database.
Bonded anchors are increasingly employed as structural fastenings to hardened concrete. They can be classified as adhesive bonded or grouted anchors depending on the bonding agent, anchor type and hole diameter. An adhesive anchor is installed using a reinforcing bar or threaded rod inserted in a drilled hole in hardened concrete using a polymer-based bonding agent including epoxies, vinylesters and polyesters. Typically, the drilled hole diameter is only about 10 to 25% larger than the anchor diameter. On the other hand, a grouted anchor is a threaded rod, headed bolt, deformed reinforcing bar inserted in a drilled hole filled with a cementitious or polymer grout. In this case, the diameter of the pre-drilled hole is at least 150% larger than that of the fastener.
The behaviour of single adhesive anchors has been the subject of many experimental and theoretical investigations in recent years [1,2,3,4]. Many of these investigations aimed at developing reliable design models to predict the tensile capacity of this wide-spread type of fasteners. The problem of predicting the tensile capacity of single adhesive anchors is characterized by the large number of factors that affect their behaviour [3,4]. The large number of variables governing the behaviour of adhesive anchors has made it difficult to develop a generalized formula for the prediction of their tensile capacity. In addition, some of these parameters, such as chemical resin type, resin system, anchor bolt type and cleaning conditions of the drilled hole are difficult to quantify in design models. As a result, most specifications recommend that the performance of these anchors be determined by product-specific and condition-specific testing .
A multilayered feed-forward neural network trained with a back-propagation algorithm was constructed using 10 design variables as network inputs and the tensile capacity of adhesive anchors as the only output. The proposed ANN model was trained and tested using the comprehensive worldwide adhesive anchor database compiled by the ACI committee 355. The ability of the proposed ANN model to predict the tensile capacity of adhesive anchors was examined against other test results not used as part of the training data. The trained ANN model predicted the tensile capacity of the testing data set within an average error of 1.6% of the experimental results. The predictions obtained from the trained ANN showed that the tensile capacity of adhesive anchors is linearly proportional to the anchor diameter and embedment depth as suggested by the uniform bond stress model. The effect of the concrete compressive strength on the tensile capacity of the adhesive anchors is not significant and product dependent. Overall, the uniform bond stress model is the most suitable one among those available in the literature for predicting the effect of different parameters on the tensile capacity of adhesive anchors.
purchase the full-text of this paper (price £20)