Calibration of model uncertainties in base heave stability for wide excavations in clay

In this study, the statistical properties for the model factor [the ratio of actual factor of safety (FS) to the calculated FS] of three popular base heave calculation methods are calibrated based on real case histories with wide excavation. The statistical properties of concern include the mean value and coefficient of variation (COV) of the model factor. It is found that the mean values of the model factors for the modified Terzaghi׳s and Bjerrum-Eide׳s methods are close to 1, whereas that for the slip circle method is significantly greater than 1. The COVs associated with the modified Terzaghi׳s and Bjerrum-Eide׳s methods are relatively low, while that associated with the slip circle method is relatively high. Based on the calibration results, reliability-based design charts are provided. It is concluded that the required factor of safety (FS) depends on two factors: (a) the target reliability index and (b) the COV of the calculated FS. For the case where this COV=10%, the required FS in codes for the modified Terzaghi׳s and slip circle methods (FS=1.5 and 1.2, respectively) seems reasonable. In contrast, the required FS for Bjerrum-Eide׳s method (FS=1.2) seems insufficient.