Research On The Reliability Of Bearing Capacity Of Deep Mixing Composite Foundation

Deep mixing composite foundation are used extensively, but deep mixing composite foundation –design is semi-empirical status. Design with empirical formula is simple and feasible. In addition, it has been submitted to adopt probability –based limit method into composite foundation-design, along with the development of reliability research on soil foundation.To meet this case, this paper predicted the ultimate bearing capacity of collected data using three predicting model and neural network. Reliability index of projects is calculated by JC and Monte Carlo method. And the same time, the resistance partial factor is determined utilizing optimization method. Composite modulus calculated by method of weighted mean and loading test are compared, and the relationship between them is derived. This paper completes several jobs as follow:1. Draw out that neural network method is fit for expecting the ultimate capacity of deep mixing composite foundation in Tianjin, and the result is best.2. A limit state function is established and reliability is analyzed. In the analyzing of resistance, three extensively used distribution: normal distribution, log- normal distribution, extreme-value type I distribution are not rejected by the ratio of test data to design data. Reliability index is influenced by the variance of basic parameters, resistance distribution, and ratio of live load to dead load.3. Suggesting reliability index of deep mixing composite foundation in Tianjin is 4.2,and corresponding resistance partial factor is 2.016.4. The relationship between composite modulus EC calculated using method of weighted mean and ESP calculated according loading test data is derived, that is ESP =0.33 EC +3.6.