Reliability of the cone penetration test (CPT) in estimating the undrained shear strength of clay is investigated in this thesis. For clays, the undrained shear strength from CPT results is usually derived by empirical correlations between cone resistance and undrained shear strength from laboratory tests or field vane tests, expressed in terms of cone factor and function of overburden pressure. The uncertainties of the undrained shear strength are caused by data scatter of the cone factor in the correlation, model error of the cone factor, effect of anisotropy, and spatial variability of cone resistance. Among these uncertainties, it is revealed that the most serious one is the data scatter of the cone factor in the correlation. Between the laboratory test and the field vane test used for correlations, it is found that the field vane test are more reliable.
In consideration of the above uncertainties, the failure probabilities of a gravity platform in the North Sea are calculated as an example problem, and it is concluded that the estimation of the strength of clay with CPT alone might not be enough, and it is recommended that some other more reliable tests are accompanied along with CPT.