Estimation of Model Uncertainties for Reliability-Based Design of Cantilever Walls in Sand

Because of our geotechnical heritage that is steeped in empirical calibrations, model uncertainties can be significant. Even a simple estimate of the average model bias is crucial for reliability-based design. If the model is conservative, it is obvious that the probabilities of failure calculated subsequently will be biased, because those design situations that belong to the safe domain will be assigned incorrectly to the failure domain, as a result of the built-in conservatism. This paper presents a critical evaluation of model factors for limit equilibrium analysis of cantilever retaining walls in sand. A total of 20 tests were collected from the literature to calibrate a theoretical model. It is important to note that there is no unique way of defining the model factor. Detailed statistical analyses showed that developing a regression equation using the theoretical embedment depth as the predictor variable can produce a significantly higher coefficient of determination, in contrast to other plausible methods. In addition, it was also shown that reasonable variations in theoretical details produce less effect on the degree of model uncertainty than variations in the definition of the model factor. This highlights the importance of choosing the definition of the model factor carefully.