Bayesian uncertainty analysis of grouted anchors

This research proposes a data-driven, Bayesian framework to obtain reliability and pullout capacity estimates of individual grouted anchors, using suitability and investigation test data.Two covariance modeling approaches were investigated, the first quantifying the total uncertainty of model and data, the second specifying measurement/observation uncertainty and model uncertainty separately.Both methods deliver plausible predictions, providing, with the probability of failure an additional criterion on which grouted anchors can be assessed.Additionally, the estimated pull-out capacity distributions from suitability tests were in line with measured pull-out capacities obtained from investigation tests, verifying the approach.