| With the increase of urban building density, and the development of High-level, high-risebuildings and underground traffic engineering construction, much more deep foundation pitengineering are constructed in urban built-up area. These engineering is often near the densebuildings, which restricts the construction of foundation pit engineering by the environment.It is necessary to predict the deformation and its influence on surrounding environment causedby the construction of foundation pit, which has important guiding significance to the designand construction of deep foundation pit for the city center area. The unloading mechanicseffect research on the deep foundation pit excavation is of great significance for effectivelyprotecting the safety of adjacent building.In these factors affecting the stability and deformation of foundation pit whose soil shearstrength parameters and constitutive model is esscential, which is closely related to the stresspath of soil. In this study, experimental research, theoretical analysis and numerical simulationwere applied to study the environment effect during the urban underground structureconstruction in combination with the deep foundation pit engineering of Guangzhou citycenter. The following research work was carried out.1. The geological partition of Guangzhou would be studied then get the adaptabilityevaluation of underground engineering according to each partition. The typical soils in thecenter of the city were taken as study objects. The SLB-1type of stress and strain controlledtriaxial shear penetration test instrument was used for testing silty clay and mucky soil in theinfluenced area of the foundation pit by the conventional triaxial compression test andconsolidated triaxial compression test. Then, the different points of soil shear strength indexand stress-strain relationship will be compared in different consolidation pressures and stresspaths. As the results of tests, initial consolidation pressure affected the strength of soil highly.The strength of soil afterk0consolidation was higher than isotropic consolidation. Whenthey broke by compression shear, breaking strengths were almost the same under loading or unloading. The axial strain of the soil pore water pressure changed relating to the average soilconsolidation pressure, soil and stress path. In normal triaxial,k0consolidation and unloadingtests, Pore water pressure with axial strain trends can be fitted by a function ofexponential decay curve. In axial unloading test, axial strain trends of pore water pressure canbe fitted by a function of two quadratic parabolic curves. Generally, pore water pressure andcritical pore water pressure rises rapidly as the large average consolidation pressure and highplasticity index2. It can be found from the test result of soil testing base on different stress path, byk0consolidation,—ε1relative cure was good linear relationship. The linearmrelationship of the relative curve of soil unloading stress path（13）—was great.According to that, the equation of nonlinear elastic model tangent modulus expression can bederived and found the method of unloading soil nonlinear elastic constitutive modelparameters. By analyzing（13）—curve, initial tangent modulus related to stress path,soil and average consolidation stress. The higher average consolidation pressure, the lowerplasticity index and the initial tangent module became larger. Axial unloaded initial tangentmodulus was the largest, the lateral unloading initial tangent modulus was medium and axialload minimum initial tangent modulus was smallest. Axial load initial tangent modulus andconfining pressure relationship can be represented by a power function. Lateral unloading,axial unloading initial tangent modulus and confining pressure related as linear relationship.3. By the help of the finite element analysis software MIDAS/GTS, the nonlinear elasticconstitutive model of soil unloading was used, from the whole construction process offoundation pit excavation, the space effect of the excavation, the size of the reduction factorof equivalent stiffness of shield tunnel and the soil elastic modulus size of the tunnel to studythe excavation effect on the deformation of the underlying shield tunnel, while theengineering example was analyzed. As can be seen from the result, when the foundation pitwas excavating, the vertical displacement increment of bottom tunnel was bigger than the horizontal displacement increment. Overall, deep foundation pit excavation mainly causes thevertical deformation. Shield tunnel deformation is mostly affected by the upper foundation pitexcavation but less by upper side excavation.4. The shield segment equivalent longitudinal stiffness and equivalent lateral stiffnesswas introduced to simplify the calculation model of shield lining segments, Based on thefinite element analysis software MIDAS/GTS, the nonlinear elastic constitutive model ofsoil unloading and unloading test parameters were used to study the impact of adjacent tunnelby the tunnel construction from the nearby space relative position of the tunnel, the size of thereduction factor of equivalent stiffness of shield tunnel and the elastic modulus size of tunnelin the soil. The results show that, the deformation of shield tunnel was impacted dramaticallyby another bottom shield tunnel which was constructing. When the angle of that two tunnelsare45°and90°, the deformation of upper shield segments are obvious symmetry and theinfluence will decrease stretching along each direction of the tunnel. When the angle isbetween45°and90°, the effects on the overlap are almost the same. When the angle equals to45°, the overlap of the tunnels is bigger that the impacted scope is larger so along eachdirection of the tunnel, the deformation declining velocity of shield segment become smaller.5. As the foundation pit sample, gangding Hotel, genetic algorithm was used forupgrading traditional BP neural network. By using genetic algorithm, the best number ofhidden layer nodes and the maximum number of cycles can be determined. By studying andtraining the superior neural network weights parameters can be found. Then it can proposedthe modeling method of deep excavation subsidence and horizontal displacement predictionbase on genetic BP neural network. The modeling error of variability of rock mass parameterscan be reduced. Taking factors in a certain range, the accuracy of predicted results can beincreased. As can be checked from the results, between four sets of predicted and testedvalues have0.25mm error of relative horizontal displacement. The largest absolute valueerrors of sedimentary displacement, relative horizontal displacement and relative sedimentarydisplacement are0.09mm,1.7%and1.57%. |
