Developing Statistical Models for Solving Tunnel Boring Machine Performance Problem

Abstract

AbstractThe efficiency of tunnel boring machines (TBMs) in tunnelling projects has great importance for civil and geotechnical industries. A reliable and applicable model for predicting TBM performance is of interest and necessity in any tunnelling project before construction and even ordering TBM machine. In this study, a series of statistical-based models/equations, i.e., simple regression, linear, and non-linear multiple regression (LMR and NLMR) models were developed to predict TBM performance including advance rate, AR, and penetration rate, PR. The most effective parameters on TBM performance based on different categories of rock material, rock mass, and machine properties were selected and used. Results obtained by simple regression models showed that they are not good enough for receiving a suitable accuracy in predicting TBM PR/AR. In addition, LMR and NLMR equations received a higher performance prediction compared to simple regression models. A coefficient of determination of about 0.6 confirmed a suitable and applicable accuracy level for the developed LMR and NLMR equations in estimating TBM PR/AR.KeywordsTBM performanceSimple regressionLMRNLMRPrediction, tunnelling project