Publication: Correlating rock mass quality with tbm performance in pahang-selangor raw water transfer tunnel
Date
2023-08-01
Authors
Sylvanus Sebbeh-Newton
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The rock mass classification system and the numerical modelling are the two main design tools widely used in tunnel stability investigations. Subjective uncertainties is a major challenge with the use of rock classification systems, thus the need for data-driven classifiers. Also, ignoring uncertainties in the post-peak rock mass strength parameters in numerical modelling may lead to underestimation of probability of failure. The purpose of this study is to account for the uncertainties in these two tunnel design tools. To address the subjectivity in the classification system specifically the JH system, four machine learning models: Support Vector Machine (SVM), K-nearest neighborhood (KNN), Random forest (RF), and extremely randomized trees (ERT) were used to develop data-driven rock mass classifiers for the Pahang-Selangor Raw Water Tunnel (PSRWT) in Malaysia. The entire database used for this research consists of 180 rock mass data and 79,813 TBM operating data points. This dataset represents 11.6 km of the tunnel from chainage 6.85 km to chainage 18.59 km. Nine (9) Tunnel Boring Machine (TBM) operating parameters; rate of penetration, cutterhead torque, cutterhead thrust force, cutterhead revolution per minute, stroke speed, boring pressure, pitching, and motor amps were used as input variables. The results indicate that ERT has the potential to correctly identify rock masses and also has a good generalization ability . To also address the uncertainties in the numerical techniques, the point estimate method (PEM) incorporated in RS2 was used to model two case scenarios. In case 1, uncertainties in both peak and post-peak rock mass strength parameters were considered whereas in case 2, only uncertainties in the peak parameters were considered. A comparison of the two cases revealed that the probability of failure was significantly underestimated when the uncertainties in the post-peak rock mass strength parameters were ignored.