Research On The Rules And Prediction Of Ground Deformation Induced By Tunnelling With Shallow Tunnelling Method Under Water-rich Soft Stratum

At present,newly built shallow buried tunnels often need to be excavated in water-rich soft stratum environment with high groundwater level.Compared with dry area,the ground deformation caused by the Shallow Tunnelling Method(STM)and the influences of many factors on the ground deformation are not clear in the water-rich soft stratum,which will undoubtedly increase the risk and difficulty during construction.In this paper,the rules and prediction method of the ground deformation induced by STM tunnelling in water-rich soft stratum are successively studied by laboratory model test,field monitoring,numerical simulation and machine learning model.The main research work and achievements are as follows:(1)The mechanism of ground deformation induced by soft ground tunnelling is qualitatively studied by laboratory model test.Three different conditions are considered in the model test:the tunnel heading is under no reinforcement,reinforced with pipe roof and reinforced with "pipe roof + grouting".The stress release rule,the longitudinal surface settlement and the development pattern of the soft soil in front of the tunnel heading are studied under different reinforcement conditions.Moreover,the results from other similar model tests with sand are used for comparison,and the differences between the deformation characteristics in soft soil and sand are revealed.(2)The rules of surface settlement under groundwater seepage are quantitatively analyzed by field monitoring data.Taking the north section of the Hangzhou Zizhi Tunnel and the overlapped section of the Line 5 and the Line 7 in Shenzhen Metro as two project backgrounds,the three stages of the development of surface settlement and the influencing range of the rapid settlement stage are fiurstly summarized.The ratio of final crown settlement to surface settlement is then given.Finally,empirical formulas are used to fit the transverse and longitudinal surface settlement trough.The influence of groundwater seepage on surface settlement is revealed from the three aspects mentioned above.(3)Considering that there is a certain deviation between the monitoring data of surface settlement and the fitting result using the Peck formula,a modified Peck formula is proposed to describe the shape of transverse surface settlement trough in water-rich soft stratum.Field monitoring data and subsequent numerical results are used to verify the applicability of the modified formula.(4)Based on the abovementioned two tunnel projects,three-dimensional fluid-mechanical coupled numerical analyses are carried out to analyze the acting mechanism of groundwater seepage on the ground deformation in water-rich soft stratum.The modelling and simulation processes of the numerical model are firstly elaborated.The computed results of surface settlement are then compared with the field monitoring data to demonstrate the in:fluence of groundwater seepage on the development of surface settlement.Finally,through investigating the connection between the pore pressure distribution and the development of ground deformation,the acting mechanism of groundwater seepage are deeply investigated and the characteristics of ground deformation are presented.(5)Parametric analyses are carried out on twin tunnels with small clear spacing.The influence of the five parameters in relation to seepage on the pore pressure distribution,the surface settlement and the internal force of the lining are investigated,including the permeability of the stratum,the permeability ratio of the lining and its surrounding stratum,the pre-reinforcement measure,the excavation sequence and the distance between the tunnel headings of twin tunnels.The simulation results provide reasonable guidances for STM in water-rich soft stratum(6)Based on machine learning model,a method is proposed for predicting the surface settlement above the tunnel.The sample data are firstly selected and quantified according to the hydrogeological conditions,tunnel geometry,construction conditions,and the surface settlement above the central axis of the tunnel.Then the training sample data and the prediction sample data are allocated according to the proportion of 8:2.Through the model parameter optimization and genetic algorithm optimization,the BP neural network model and support vector machine regression model with high prediction precision are established respectively.The feasibility of the machine learning model is proved.Finally,the sensitivity of the two machine learning models is analyzed,and the impact of neglecting seepage on the accuracy of settlement prediction is discussed.