Stochastic Evaluation of Tunneling Performance Using Discrete-Event Simulation

Estimating tunneling time and cost performance is an extremely important and challenging task for all contractors during the construction planning and cost estimating phase of tunnel projects. The complexity of tunneling operations and the variety of associated uncertainties contribute to a significant amount of risk in the estimation of tunneling time and cost. Since conventional deterministic estimating cannot measure tunneling risks, particularly geologic uncertainty, in a systematic and quantitative manner, a stochastic approach is necessary for evaluating tunneling performance. This paper presents a stochastic methodology for evaluating tunneling performance by using discrete-event simulation. Different tunnel excavation and support methods when applied to different geologic conditions define different tunneling alternatives, which change the precedence logic of tunneling operations. Thus, the construction performance for each tunneling alternative in the project must be assessed separately. To organize tunneling costs, all work items in the project are categorized into divisions based on a work breakdown structure (WBS) designed specifically for tunnel projects. Cost items within each division are then classified into fixed costs and variable costs that depend on the selected excavation-support methods and the resulting tunnel advance rates. The tunnel advance rates for a particular alternative are determined by analyzing the probabilistic network of tunneling activities associated with that alternative. The duration of tunneling activities is expressed by time-estimating equations formulated by analyzing the work at the process level. Uncertainties associated with tunneling processes are evaluated by subjective assessment using the Perry & Greig method. The probabilistic scheduling networks for different tunneling alternatives are analyzed using discrete-event simulation performed in ProbSched, a graphical probabilistic scheduling program implemented as an add-in to the STROBOSCOPE simulation system. The simulation results provide probability distributions of tunnel advance rates and tunneling unit costs for all possible alternatives, which can be used to determine optimal excavation and support policies for the project. The stochastic evaluation of tunneling performance for the Hanging Lake Project in Colorado is presented as an example application.