Probabilistic modelling of tunnel excavation processes using a Bayesian network

点击翻译按钮获取中文摘要

The proposed research is a part of a larger effort by Hatch to optimize and complete the construction of a 16 km long backup tunnel for the hydroelectric facility providing power to Rio Tinto Alcan's aluminum smelter at Kitimat BC. The research will address how to use historical designs, geological mapping, and excavation records from tunnels constructed at the site in the 1950's and the 1990's to determine the best current tunnelling technologies and to predict the tunnelling advance rates and costs. This research will help Hatch provide risk-informed recommendations to Rio Tinto Alcan concerning appropriate tunnelling methods and costs, as well as identifying areas where excavation difficulties could occur. The proposed research will use a Bayesian Network framework where parameters that affect tunnel excavation can be evaluated and used to make more accurate schedule and cost predictions during the tunnelling project. The Bayesian Network model that will be developed will combine a geological prediction model (to predict geological conditions ahead of the tunnel face) with a construction strategy decision model to minimize risk. A Bayesian Network has the ability to combine expert knowledge with observed data, express dependencies among variables, and the ability to learn and refine causal relationships. This research should provide a powerful tool that planners and engineers can use to systematically assess and mitigate the inherent risks associated with tunnel construction.