Risk Analysis Of Tunnel Project Construction

Tunneling is characterized by high degree of risk due to a prior limited understanding of the complex hydro geological condition and immature construction technique and poor management. With the rapid progress of the construction technique and accumulation of the construction skills, although the safety of tunnel construction has been promoted, the accidents such as collapse, large water inflow, excessive deformation and rock fall has happened now and then. These accidents always lead to a high rise of cost and delay of the project timescale, what is worse; it may lead to injury of the workers, and a threat to the surface traffic, buildings and the life of passengers. The primary reason of the accidents is we never know the ground conditions well before excavation. A well understanding of the ground conditions of tunnels is vital to complete the tunnels smoothly and safely.A risk analysis model is presented in this article to control the tunnel risk. It contains a geology prediction part and a construction strategy decision part, the former part is based on the Markov Process and the Neural Networks, it uses the parameters observed during tunneling to predict and update the ground conditions in the excavation face and ahead in a probabilistic way and effectively complements the ground conditions of specific locations along the tunnel alignment obtained from the in site equipment. The latter part is used to choose the optimal construction strategy to reduce the risk and the construction cost as much as possible according to the ground condition offered by the ground prediction model. The model is applied to a case study successfully,409training data is used to train the model, while347test data is used to confirm the model. The result shows that the predicted ground conditions have a good agreement with the real ground conditions, especially in the complete soil and rock which reach a high accuracy of93.37percent and73.28percent, respectively. While in the mixed condition, the performance is not that good, this is probably because the mixed condition has both the properties of the soil and rock. The strategy decision model decides to use the open construction strategy in the30.5percent of the test data set and the conservative construction strategy in the remains. It will reduce17.15percent of the cost compared to use the conservative strategy in all place.The model effectively complements the ground conditions of specific locations along the tunnel alignment obtained from the in site equipments and enhances the understanding of the tunnel ground condition, and that it reduces the construction risks and cuts the costs in some points. Despite the fact that more efforts are needed in the aspects of ground conditions definition, the utility functions determination and the application of the model to other tunnels, it does not hinder the model to become a powerful tool for designers and constructors to mitigate the construction risks and reduce the conservative measures during design and construction, furthermore, it provides some experience for other researchers to study the risks associated with the tunnel construction.