Performance prediction of hard rock TBM tunneling

The rock breakage mechanisms induced by tunnel boring machines (TBM) are analyzed. A series of TBM performance prediction models developed since 1970s are introduced, which include the single factor prediction model, CSM (Colorado School of Mines) model, NTNU (Norwegian University of Science and Technology) model, QTBM model, probabilistic model and neuro-fuzzy model. The single factor prediction model only utilizes one parameter of the rock material properties, such as rock uniaxial compressive strength, Brazilian tensile strength or the total hardness, to predict the TBM penetration rate. Based on the database of measured cutting force generated in the Linear Cutting Machine (LCM) with disc cutter in different rock types, CSM proposed a prediction model of multiple variable regression analysis. In this model, the rock uniaxial compressive strength and tensile strength are taken in consideration. NTNU model is based on extensive laboratory tests, TBM performance data and geological mapping data. The model can predict net penetration rate, cutter life, machine utilization and excavation cost step by step. Rock drillability, porosity, joint spacing and orientation are taken into account in its penetration rate prediction model. The probabilistic model is based on a large database of over 600 public domain case histories. QTBM is originated form the Q system, and the difference is some new parameters that are relevant to the TBM performance are considered. The neuro-fuzzy model is a black box model, which may not consider the interaction between TBM and rock mass that still is regarded as a dynamic, uncertain, complex and non-linear and ill-defined process.