Application of ANNs and MVLRA for Estimation of Specific Charge in Small Tunnel

Drilling and blasting method has been used for many years in underground excavations and still is very popular because of its many advantages. Blast performance is ordinarily measured by specific charge and by explosive consumption of broken rock. The empirical models are available for estimation of specific charge and different sets of parameters. This paper presents the possibility of applying artificial neural networks (ANNs) to estimate the specific charge in various conditions of tunnel blasting. Among available existing parameters in the literature, some of the most influencing parameters are selected. After running different models, P wave, rock-quality designation (RQD), tunnel area, maximum depth of the hole, and coupling ratio (charge-to-hole diameter) are selected to estimate specific charge of tunnel blasting under various conditions. Also, conventional multi variable linear regression analysis (MVLRA) is applied to estimate specific charge. The results show that the accuracy of ANN is more than the MVLRA-based models. (c) 2012 American Society of Civil Engineers.