Prediction of leakage into Lunner tunnel based on discrete fracture flow models

Abstract

As part of the project “Tunnels for the citizen”, sub-project B “Environmental concerns”, coordinated by the Norwegian Road Authority (“Statens vegvesen”), a discrete fracture network model was used to investigate the hydrogeological conditions before and after construction of the Lunner tunnel. A limited model was considered which covers an area of 550 m x 550 m comprising the transition zone from two rock types (hornfels and syenite) where potential problems were foreseen. Available data from site investigation performed by Statens vegvesen and NGU was used to build the model. First, large scale features which could be geologically mapped were represented deterministically. Smaller scale features which could only be characterised in a statistical sense from boreholes observation in terms of orientation, dip, length, density were used to stochastically generate discrete fracture networks through which water flows. Saturated transient and steady state calculations were performed to predict the amount of leakage into the tunnel during construction. Only linear groundwater flow was considered, with a constant recharge from precipitation. Due to the uncertainty related to crucial input parameters such as fracture length and fracture hydraulic properties, a parametric analysis was carried out to investigate the range of variation in the model predictions. The results from the modelling give a three dimensional picture of the groundwater level after tunnel excavation. They outline the interrelation between tunnel and main conductive faults in the establishment of a lowered water table. Due to tunnel excavation, a rapid drawdown is established above the tunnel and propagates into the rock mass along conductive fault zones. Injection of the faulted zone contributes to a drastic reduction in leakage rates in the whole tunnel, although locally the water inflow increased on both sides of the injection interval through secondary fracture sets. The work presented in this report contributed to: • assess the capabilities of discrete fracture network models generally, and more specifically their application to modelling of groundwater flow around tunnels in fractured rock masses • test the commercial software Napsac used for the purpose of the analyses • carry out a blind prediction of the effect associated with tunnel excavation in a potentially sensitive area, based on data collected during pre-investigation work • evaluate the results from discrete fracture modelling and the sensitivity to input parameters. Of particular interest were the correlations between tunnel leakage, pore pressure changes and groundwater drawdown, which could be used to define acceptance criteria for tunnel leakage based on the vulnerability of vegetation and water sources.