Prediction quality in simulated anchor pull-out tests

The efficient and permanently safe design of fastening systems requires a thorough understanding of all involved load carrying mechanisms and their potential change in course of time due to deterioration and material ageing. The latter are determined by the involved materials - concrete, steel, and, depending on the system, adhesive mortars. The objective of life-time design and assessment requires models with truly predictive capabilities which can only be realized if all relevant mechanisms are captured. In this contribution we focus on concrete cone failure, one of the dominant failure mechanism for bonded as well as mechanical anchors in tension and shear, where in the later case the localization of the concrete also depends on the material properties of the used adhesive mortar. After calibrating a well-established discrete particle model, LDPM, based on standard uniaxial compression and three point bending tests a blind prediction study on anchor pull-out and shear tests is performed. Experimentally obtained load-displacement data, DIC strain measurements and photogrammetrically obtained crack patterns are used to evaluate the predictive quality of the investigated models. In case of bonded anchors one set of data (‘confined’ tests) is used to calibrate the bond law while an independent second set allowing cone failure serves for the model validation