Prediction of TBM penetration rate using intact and mass rock properties (case study: Zagros long tunnel, Iran)

In this paper, the correlations between the different measurements of rock brittleness (i.e., B1, B2, B3, and BI) and the penetration rate (PR) of tunnel boring machine (TBM) through Zagros long tunnel were evaluated. According to the results of simple regression analyses, there was no correlation between the penetration rate of TBM and the brittleness of B1, but stronger log-linear correlations have been observed between the penetration rate of TBM and the B2, B3, and BI brittleness indices. As part of the present study, correlations between some of rock mass classification systems (RQD, RMR, GSI, Q, and QTBM) and penetration rate of TBM have been investigated. The results of simple regression analyses showed that the first four rock mass classification systems did not exhibit a good correlation with the TBM penetration rate, but the QTBM had strong correlation with that. Multiple linear regression (MLR) analyses were applied for estimating the TBM penetration rate based on three properties of the rocks (porosity (n), brittleness of B3, and elastic modulus (E) of intact rock) and rock mass quality system (Q). Also, in this study, artificial neural network (ANN) analyses were applied on the data to develop predictive models for the penetration rate of TBM from porosity, brittleness of B3, elastic modulus, and Q values. The comparison of the models produced from ANN and MLR analyses using the coefficients of determination showed that the ANN models for predicting the TBM penetration rate were more reliable than the MLR models.