Evaluating undrained shear-strength of marine clay using free-fall penetration considering uncertainty: an experimental and Bayesian analysis study

Free-fall penetration (FFP) testing has drawn wide attention in evaluating in-situ undrained shear-strength (Su) of marine clay because it can save time and reduce costs. However, model parameters for interpreting Su are empirical and usually determined based on engineering experience, leading to great uncertainties in Su results. In this study, FFP model tests were conducted using conical, hemispherical and T-shaped penetrometers to gather data. Subsequently, Bayesian theory was employed to ascertain the optimal model parameters. The penetrometer motion state and influencing factors of penetration depth are analysed and discussed. By considering prior knowledge and inherent variability of Su, optimal model parameters for the three types of penetrometers are obtained using maximum likelihood estimation method. The reasons for the different model parameters of three type penetrometers are discussed. In order to assess the effectiveness of the proposed models, field tests were conducted to gather in-situ data. To reduce uncertainty inherent in limited amount of data, a Bayesian updating approach combined with Markov chain Monte Carlo method were used to evaluate posterior distribution of Su of the three penetrometers. Results show that updated Su samples were consistent with those determined by vane shear tests, which verified the reliability of proposed model parameters.