Linear Classifiers for Prediction of Squeezing Conditions in Tunnels

摘要

Empirical approaches play a significant part in the prediction of squeezing conditions in tunnels or caverns. On analysis of collected squeezing data from published literature, no empirical squeezing equations have been found for tunnels with a depth of more than 850 m. In this paper, an attempt has been made to generate linear classifications using rock quality index Q, rock mass number N, overburden height and tunnel span through a comparative study. 234 tunnel sections, especially of the Himalayan region, with depth up to 1900 m have been considered for the analysis. In this method, a demarcation line has been proposed to classify squeezing and non-squeezing conditions and it also allows computing probabilities of squeezing in combination with rock mass quality and tunnel depth. Further, developed equations have been compared with the existing four empirical squeezing equations in the case of the Rohtang road tunnel. The enhanced equations show better results to predict squeezing conditions for tunnels with overburden depth up to 1900 m in comparison to predictive competencies of previously existing criteria. The developed linear classifications are for deep and shallow seated tunnels in complex geological conditions using overburden height, tunnel span, rock quality index Q, and rock mass number N. The developed equations are useful for tunnel designers during the planning and support analysis of underground structures where very less physico-mechanical information are available for squeezing probability. It also suggests that the influence of tunnel depth on squeezing occurrence is a non-linear function.