The Study Of Shield Tunneling Parameters' Optimization And Prediction On Layer Of Silt And Silty Sand

With the rapid development of subway in China, shield tunneling for subway construction has been used more and more frequently. This paper, based on I-TS-11 tender of Suzhou's subway line 1 construction using earth pressure balance shield tunneling, studies how to select and optimize the parameters of shield tunneling especially in silt and silty sand stratum of Suzhou. The results show that to ensure minimal impact on the ground, the material used in synchronous grouting should be chosen as follow when using shield tunnel in silt and silty sand layer of Suzhou: Simultaneous grouting can use hard serous (ratio is 100Kg/m3 of cement,400Kg/m3 of fly ash,680Kg/m3 of yellow sand,400Kg/m3 of water,100Kg/m3 of bentonite). Serous of type AB (ratio of cement water mixture and sodium silicate is 1:0.7) should be used in secondary grouting. and grouting rate should be controlled at around 220%. Shield advance speed should not be too big, which would be appropriate to be controlled at around 2.5-3.0cm/min. The rate of tunneling work should correspond to the specific circumstances of construction properly. Then, the paper uses social science statistical software SPSS to do statistic analysis on the relationship between parameters of shield tunnel and surface deformation. On condition that the ratio of grouting is appropriate (135%-220%), the results show that the biggest impact exerted by shield tunneling in silt and silty sand layer of Suzhou comes from torque of shield while the smallest impact comes from the rate of shield tunneling, and impact of total thrust of shield is between the two previous factors. This paper applied neural network model to learn the parameters of tunneling in order to predict the change value of surface subsidence. The results show that the relative error between predicted value and the actual test value is controlled fewer than 5%. There are only several values having large relative error. However, the relative error of aggregate settlement amount of the same point is only about 0.77%, which can fully meet the construction requirements. Therefore, applying neural networks to predict surface subsidence through tunneling parameters is entirely feasible. Meanwhile, this paper provides an ideal prediction method especially for surface subsidence prediction of shield construction in complex area to guide the conduct of safe and effective shield construction.