Prediction of TBM Disc Cutter Wear and Penetration Rate in Tunneling Through Hard and Abrasive Rock Using Multi-layer Shallow Neural Network and Response Surface Methods

Tunnel boring machine (TBM) is a popular rock cutting machine for rapid construction of tunnels. This paper mainly dwells on the excess cutter wear and low penetration rates encountered while tunnelling through hard and abrasive rock in a head race tunnel being driven for hydel power generation. Wear prediction in disc cutters with its mechanism was reviewed. Field performance data and laboratory characterization of rock were done for analyzing the causative factors. This was followed by data analysis using a multilayer shallow neural network (MSNN) for identifying the key parameters and their influence on the output parameters (cutter wear and rate of penetration). Five major process control parameters including two machine parameters, namely, thrust and torque, one design parameter, i.e., radial position of cutter and two rock parameters namely uniaxial compressive strength (UCS) and Cerchar abrasivity index (CAI) are considered in the study. Rock type is kept constant (quartzite) to analyze the influence of the machine operating parameters on the cutter penetration rate and the cutter wear. Two different scenarios were analyzed. The correlation coefficients obtained between output and target for two cases investigated were 0.927 and 0.965, respectively. Sensitivity analysis of the input parameters on the output parameter is also carried out. For validation of the result, response surface method (RSM) was used for the analysis of historical data. Both MSNN and RSM predict the influence of key variables affecting cutter wear (CW) and rate of penetration (RoP) with a good confidence. In a given rock setting, it is possible now to fix the optimal values of Thrust and Torque to control the cutter wear while maintaining an acceptable rate of TBM penetration.