The pullout resistance of anchorage which shares the load of structures is mainly mobilized by friction between soil and anchorage. The resistance peaks at small displacement and then decreases rapidly toward the residual state. This rapid drop in friction can result in abrupt destruction. Thus an expanded tension anchorage is used to increase the pullout resistance instead of friction anchorage. However, compared to the friction anchorage and plate anchorage studies, an investigation on evaluating the pullout resistance of expanded tension anchorage is in early stage.
In this study, a series of laboratory model tests for the expanded tension, plate and friction anchorages has been carried out to characterize the pullout resistance of those anchorages. The analysis of the pullout resistance - displacement relationship of each test shows that the friction and bearing resistances of expanded tension anchorage is smaller than the pullout resistance of friction anchorage and plate anchorage respectively carried out in the same test conditions. Therefore, the previously proposed ways to sum up the peak pullout resistances of friction and plate anchorages could overestimate the pullout resistance of expanded tension anchorage.
It was proposed that the pullout resistance of expanded tension anchorage could be expressed using reduction coefficients, respectively multiplied to the pullout resistances of friction and plate anchorages, carried out in the same test conditions. Then, an artificial neural network (ANN) model was utilized to reliably estimate the respective friction and plate anchorage resistances. Accordingly a reliable evaluation of reduction coefficients, with considering the most influencing factors, can make it possible to properly evaluate the pullout resistance of expanded tension anchorage.