Stochastic mechanics-based Bayesian method for calibrating geotechnical parameters of Shanghai deep soft clay using CPTU data

The construction of urban deep underground engineering requires scientific calculation. The reasonable constitutive model and accurate geotechnical parameters are the most important support for soil mechanics. The modified Cam-clay (MCC) model is a simple one as its constants are simple and intuitive, and is widely used in geotechnical engineering. However, it's different to get precise parameters in deep soft clay by the indoor tests due to sampling disturbance, test error and other unavoidable factors. Considering the prior information, laboratory test data, and piezocone penetration test (CPTU) data, the key geotechnical parameters are treated as random variables, the stochastic mechanics-Bayesian method is proposed to calibrate the key geotechnical parameters of the deep soil layers, such as the critical state stress ratio, compression coefficient, rebound coefficient, and overconsolidation ratio. Based on the Suzhou River deep drainage and storage pipeline system in Shanghai and the foundation pit of Yunling, the deep soil layer ⑧ at the base plate is taken as the research object. Firstly, the mechanical conversion between the CPTU data (i.e., cone tip resistance, lateral friction stress and pore pressure) and the limit expansion pressure of cylindrical cavity is derived in the MCC model. The mechanical conversion for the CPTU data is verified by using the test data from Bothkennar geotechnical test site in Scotland. Secondly, the quadratic response surface without being crossed between the key geotechnical parameters and the CPTU data can be established by regression. The Markov-chain Monte Carlo (MCMC) sampling method will obtain the posterior distributions of the key geotechnical parameters. Finally, The numercial calculation for the foundation pit is carried out with the mean values of the posterior geotechnical parameters. The results show that the uncertainties of the geotechnical parameters are significantly reduced, and the results obtained by the numerical simulation of the foundation pit using the mean values of geotechnical parameters are closer to the monitoring values. It is proved that the stochastic mechanics- based Bayesian method is effective and efficient.