Akademik Veri Yönetim Sistemi
2nd International Conference on Engineering and Natural Sciences, ICENS 2016, Sarajevo, Bosna-Hersek, 24 - 28 Mayıs 2016, ss.440-457
This paper summarizes the results of experimental
research and numerical model focused on determination of the behavior of
pull-out performance limits of what embedded into currently the most widespread
concrete type of Turkey as C25/30. Reinforcing bars having 14, 16 and 18 mm
diameters have been selected as the anchor rod in this study. Epoxy based three
component chemical adhesive has been used for the connection between concrete
and anchor bar. The depth of holes was in the range of 140-220 mm which had
been selected various for 14, 16 and 18 mm rebar diameters. Moreover, an
attempt to predict the pull-out capacity of chemical anchors embedded into
concrete using artificial neural networks (ANNs) is presented. The problem is
proposed to network models by means of three inputs and one output parameter.
The parameters such as reinforcement diameter, anchor depth, anchor diameter
were selected as input variables. The effect of the depths of holes and rebar
diameters on the pull-out capacity of adhesive anchors is product dependent.
Test results showed that increasing the anchor diameter and the depths of hole
have increased pull-out performance of anchors. The best algorithm for collapse
loads of concrete ais the Levenberg-Marquardt back propagation with R2
of 0.9837. The results indicated that ANNs are a useful technique for
predicting the pull-out capacity of adhesive anchors.