Stochastic modeling of groundwater drawdown response induced by tunnel drainage

Tunnel designers persistently experience issues regarding the impact of tunnel drainage on groundwater that can affect the safety of tunnels. From an environment-oriented approach, tunnel drainage affects the groundwater table and leads to alterations in the ecological environment in the long term. Accordingly, novel methods have been established to estimate drainage-induced drawdown concerning groundwater balance and environmental factors. However, the existing methods ignore the spatial characteristics of hydrogeological parameters and complex topography, leading to over-simplification of the groundwater flow; moreover, they lack a strategy to represent drawdown uncertainty. This paper presents a stochastic modeling method for inferring drainage-induced drawdown predictions conditional on tunnel inflow estimates. The method uses a numerical groundwater model to include the rock mass properties, fault zones, and complex topography of the tunnel area. The stochastic model processes the outcomes of the numerical model conditional on the absolute error of groundwater inflow based on Bayes' theorem. The contributions of the proposed method are as follows: (1) The method considers complex topography such as an inclined water table, fault zones, and rock type distribution. Accordingly, the groundwater simulation results indicate that the spatial distribution of rock types determines the magnitude of drawdown. Furthermore, the drawdown response was irregular and variable along the tunnel axis. (2) The Bayesian method allows the prediction of drawdown estimates with minimum data before tunnel construction with approximate drainage. The predictions can be updated using additional data during construction. (3) The application of the predictions is made flexible by arranging absolute error intervals, in which the decision-maker may consider the uncertainty of the parameters on the predicted solutions. The stochastic model was applied to a tunnel project in a poorly sampled ungauged mountainous catchment. The method provides information on parts of the tunnel that have the most significant potential impact with minimal input. The proposed method can be extended for long-term environmental impact assessment of tunnels by including surface features such as soil type, land use, and land cover. •Drawdown estimation was conducted using an environment-oriented approach.•The data scarcity problem was addressed using a stochastic modeling method.•The stochastic model estimat