The reliability of a tunnel is an ever-present concern for both tunnel managers and daily commuters. Water ingress of a tunnel was identified as a hazardous element for tunnels by previous work. An example metro tunnel was investigated in this study, in which water ingress is considered to be the major failure criterion. This thesis aims to provide a time-dependent reliability prediction for the selected tunnel based on water ingress. Water ingress rate was measured during site inspection in order to understand the water ingress condition of the tunnel. An analytical method was developed as a tool to achieve a more efficient onsite water ingress rate estimation. Using this analytical method, the measurements of the water ingress rate were extrapolated to the entire tunnel.
It was found from tunnel inspections that the cracking condition of the concrete lining is a major contributing factor for water ingress. Water ingress through cracked concrete was studied as a combination of the mechanisms of water seeping through a porous material and water inflow through cracked material. The crack connectivity was defined to describe the crack interconnection within a concrete. The crack connectivity of the selected tunnel was determined from the inspected crack patterns and generalised using Monte Carlo simulation. A relationship between the crack connectivity and the hydraulic conductivity of a tunnel lining was established analytically.
The hydraulic conductivity of the tunnel lining is a key parameter for tunnel water ingress calculations. Previous studies of water ingress into a tunnel are mainly based the assumption that the hydraulic conductivity of the lining is homogeneous. The hydraulic conductivity of a cracked concrete is mainly determined by the crack width according to previous research. The crack width in concrete is difficult to measure onsite when the crack is covered by precipitated mineral matters. An analytical method was applied to evaluate the overall crack width of the tunnel lining under the assumption that it is fully homogeneous. The probabilistic distribution of crack width was calculated when the hydraulic gradient and the rock mass properties are known. The tunnel lining investigated in this study has a thickness of more than 550 mm and the crack pattern through the tunnel lining appears to be variable (based on existing ground penetrating radar data) and could cause inhomogeneous hydraulic conductivity. An analytical solution for water ingress was developed for a tunnel with inhomogeneous lining under a steady state assumption. The results show that the inhomogeneity of the lining affects the rate of water ingress significantly when the inhomogeneity degree is high.
The cracking condition of the selected tunnel has been monitored for three years with eight inspections by the tunnel operator. The crack inspection data was evaluated in this study in a time-dependent way. Based on the cracking data and the newly developed water ingress analytical model, a time-dependent reliability calculation for the selected tunnel was developed. The hydraulic parameters including water table, rock mass property and crack formation were considered probabilistically. The service failure probability of each section of the selected tunnel was calculated time-dependently. The results show that the remaining service life of the selected tunnel varies by sections. The most critical section is predicted to possibly fail in eight years while some other sections do not show signs of increased water inflow.
