Parameters affecting the sustained load performance of adhesive anchors and fitting approaches for their assessment

Bonded anchors are connecting elements that allow the transfer of loads from load‐bearing elements to the base structure by means of adhesion. Due to their critical structural role, sustained load requirements have been introduced for their qualification. This involves performing sustained load tests at defined load levels for a specific period of time. The assessment follows a power‐law extrapolation approach in combination with a displacement criterion derived from the short‐term tests to failure, where a prediction model is used to extrapolate the measured displacement to the target service life of the anchor. The models used to determine the displacement are generally conservative and allow a safe design. However, accurately predicting the service life of anchors remains a challenge due to the different factors affecting their mechanical behavior. These factors include post‐curing of the adhesive in long‐term tests and the embedment depth, which affects the bond stress distribution. An alternative approach uses a regression model to extrapolate the service life of the anchor based on stress level vs. time‐to‐failure plots. While this method provides direct insights into the load capacity for a specific lifetime, the independent variable used in the regression models is typically the failure time, despite tests being load‐controlled and the time recorded as the measured outcome. In this research, the sensitivity of the regression model to the parameters affecting the behavior of bonded anchors is investigated in the stress vs. time‐to‐failure method. This is done by means of an experimental campaign consisting of a series of tensile short‐ and long‐term tests. The experimental results are used to evaluate the lifetime prediction of the anchors by means of a non‐linear least squares fitting analysis with a sigmoid function model and considering the failure time and the load level, respectively, as independent variables. Additionally, the effect of fixing one of the fitting parameters, representing the viscoelastic limit of the system, is investigated by setting it at different values. The results show that performing the analysis with the load level as the independent variable and fixing the viscoelastic limit at 0% tends to result in a more conservative fit. The fitting model can be used for all the anchor sets investigated, regardless of the anchor geometry, adhesive product, and its curing time.