Reliability Method Applying To The Risk Analysis Of Foundation Pit Based On Measured Datum

With the increasing of urban population density and continuous development of China’s cities, rational development and utilization of underground space is the requirement of urban sustainable development, following more and more foundation pit projects with the characters of "deep, big, tight and hard". However, the increasing foundation pit may result in increasing accidents year by year, so the risk of foundation pit has caused wide public concern over recent years. Because being able to include many uncertain factors, and offer quantitative results of risk analysis, reliability method based on probability theory is often used for risk analysis of foundation pit. Usually it is difficult to find the right risk analysis model which is suitable for practical foundation pit project, so commonly a new response surface (RS) relationship is constructed for risk analysis of foundation pit instead of the actual calculation function. Here a new RS construction method with less workload and high precision is introduced. Based on the RS, instead of assumed random sample, measured datum at site is directly used for the risk analysis of foundation pit. With practical projects, on account of design and construction phases with greater risk, the main risk indicators, such as design safety factor and deformation, are simulated and analyzed. And further confidence interval of main risk indicators is given.In this paper, uniform design method (UD), ACE (Alternating Conditional Expectations) regression techniques, Bootstrap simulation (BS) method and Monte Carlo simulation (MCS) method are used to solve the following several problems:1. Design risk of foundation pit is analyzed through RS and BS method. With a practical project, firstly for14parameters, UD method is applied to construct60input data sets for RS constructing, and based on the finite element analysis method (FEAM), the ACE regression technique is employed for finding the RS relationship of main design indices affected by14parameters. The field measured datum is analyzed through BS techniques, and combined with RS relationship, the probability distribution of the major design indices is simulated. Further in order to conduct the designer to control the design risk, the confidence interval with a certain degree of confidence is given.2. Foundation pit deformation at various construction stages is simulated and analyzed. Firstly less input data sets which are from parameters of affecting deformation are determined by the UD method, and then through the analysis of FEAM, the foundation pit deformation at various construction stages is obtained. On this basis the ACE techniques are used to construct RS relationship at various construction stages. Based on RS, MCS method is used for simulating the probability distribution of foundation pit deformation at various construction stages. With a practical engineering, the effectiveness of these methods is proved.3. Simulation results of foundation pit risk are compared based on different simulated sample. Firstly, UD method is applied to determine the representative data sets of calculation about influencing factors of risk, after experimental output results are gotten by FEAM, ACE method is used for constructing RS. On the basis of the same RS relationship, based on measured datum at site (small sample) BS is employed for risk analysis, and based on random samples (big samples) of assumed distribution, MCS is also for risk analysis. Through comparing different simulation results, it shows that two simulation results in most cases are consistent, but there may be some differences in the individual cases, and the BS is more suitable for small samples.These methods successfully solve some risk analysis problems at the design and construction phases of foundation pit. RS method which is made up by UD method and ACE techniques has the characteristics of less computation and higher precision. The simulation results of BS based on the field measured datum and MCS based on random samples can get mutual authentication, but the BS method is more suitable for small samples. For the design and construction phases, the foundation pit risk analysis methods are given, and these methods can offer certain targeted and quantitative analysis results for different stages. Practical projects prove that the proposed methods can supply with some guidance for the risk analysis of actual project.This dissertation is part of the project sponsored by the Natural Science Foundation of China (Grant No.:51078175) and the Natural Science Foundation of Yunnan Province (Grant No.:2008E038M).