UNCERTAINTY OF MOBILIZED UNDRAINED SHEAR STRENGTH

抄録

A method for analyzing the uncertainties involving the use of empirical relationships in design is developed. It is applied to evaluate the reliability of the relationship s_u(mobilized)=0.22σ'_p(measured), whereby the pertinent uncertainties are analyzed using extensive laboratory and field observed data. On the average, the relationship s_u(mobilized)=0.22σ'_p (measured) is unbiased and the error implicit in the relationship is about 15%. When cone values are used to infer s'_p, additional uncertainties on the mobilized undrained strength will incur resulting mainly from the calibration uncertainty between σ'_p and the cone values. Alternatively, vane tests may be performed to infer the mobilized strength using Bjerrum correction factor. Results show that uncertainty level associated with using the vane tests is smaller than that using the cone values, provided that similar scatter is observed between measured vane and cone values at a given site. However, prediction of the mobilized strength is even more accurate if direct measurements can be made on the preconsolidation pressure. The proposed probabilistic method also assesses the benefit of additional tests besides comparing the accuracy of using various soil parameters for predicting mobilized undrained strength. Such information is essential for planning cost-effective site characterization program.