A New Performance Prediction Model for Rock TBMs

Abstract

Tunnel boring machine (TBM) tunneling projects will have many challenges and the prediction of their performance, possibly the most critical one. Inaccurate predictions can drastically affect the decision making process and the scheduling of the entire project. TBM penetration prediction involves understanding of the rock fragmentation process from micro to macro scale and includes parameters of both TBM specifications and ground conditions. Various models have been proposed by researchers (Alber, Proceedings of the international conference of Eurock ’96. Balkema, Rotterdam, pp 721–725, 1996; Barton, Tunn Tunn Int 31:41–48, 1999; CSM, Computer model for TBM performance prediction, http://inside.mines.edu/UserFiles/Image/miningEngineering/EMI/pdf/tbm_performance_prediction.pdf, 2003; Hassanpour et al., Tunn Undergr Space Technol 26:595–603, 2011; Hassanpour et al., Geo Mech Geoeng Int J 4:287–297, 2009; Rostami and Ozdemir, Rapid excavation and tunneling conference proceedings, pp 793–809, 1993) since the early phases of TBM application, helped in the development and improvement of penetration prediction models. These models mostly not capable of modeling the stochastic nature of rock cutting/excavation process and ignores the uncertainties in the input parameters. Present study attempted to develop a new performance prediction model using nonlinear regression analysis on TBM tunneling data collected from previously executed tunneling projects. To stochastically model the uncertainties occurring, the input parameters are modeled as random variables and is assumed to follow normal distribution. Simulations based on randomized input of these parameters using Monte Carlo method are carried out to find the statistical distribution of the penetration rate. The distribution obtained for each project compare well with actual penetration rate.