ABSTRACT
The p-y load transfer method is the most widely used for the design of monopile foundations. Recently, many CPT based p-y models were proposed experimentally or numerically to overcome the uncertainty in deriving stiffness and strength parameters of soil from in-situ testing. This paper aims to evaluate the performance of existing CPT based p-y models for predicting the monopile behavior, by conducting a series of centrifuge tests in dense Geba sand. Four different diameter (D = 0.9 m, 1.8 m, 1.25 m and 2.5 m) monopiles with two aspect ratios (L/D) of 3.6 and 5 were studied. The lateral load was applied at two different loading eccentricities of 0.67L and L above the ground surface. It was found that the model of Novello (1999) model gave the best overall prediction (at both small and large deflection) for the tests in this study, while the model of Dyson and Randolph (2001) model always underestimate the pile response. Compared with the Novello (1999), the model of Suryasentana and Lehane (2014) and Li et al. (2014) will overestimate the pile response at large deflection, but, can well predict the pile response at small deflection. In all cases, the API p-y model significantly overestimates the response with a factor more than 2. The experimental observations in this study proves the feasibility and reliability of developing CPT based p-y model for monopile.
