Structural analysis of dapped-end beams through machine learning techniques

Dapped-end beams (or half-joints) are extensively used in both reinforced concrete bridge structures and roofs and flooring for prefabricated buildings. Due to their geometric configuration, they represent discontinuity regions where high stresses concentrate. Therefore, it is crucial to evaluate the load-bearing capacity of these elements which may depend on various parameters, such as the strength of materials, the quantity and the different layouts of the used reinforcements which also significantly influence their failure mode. For this purpose, a literature review led to the creation of an experimental tests database used to conduct an in-depth statistical analysis, considering the influence of various parameters on the ultimate load capacity and failure mode of dapped-end beams. The collected database has been also used to train a regression model with supervised machine learning techniques in order to predict the ultimate capacity of dapped-end beams.