Drilling and blasting is an old and known method of excavation, both for underground and surface applications. Over the years, many researchers have investigated on different aspects of this method while determination of specific charge has been one of the most challenging topic. Many specifications of rock and explosive material are influencing the amount of required charge for each application and the conventional methods can not easily incorporate all of them. This research focuses on using "artificial neural networks" for prediction of specific charge based on certain input parameters. Results show that the method is very powerful to do so with high accuracy. However, increase of number of data can significantly improve the performance of the model.


Drilling and blasting (DAB) is one of oldest methods for rock excavation in underground and surface structure. This method is vastly used in Iran for mining as well as many civil engineering tunnels e.g. road tunnels, water tunnels, underground power planets, etc. Existence of large mountain chain in Iran necessitates a lot of tunnels, in different shape and size, for various applications. DAB method is more suitable for most cases, comparing to mechanized excavation, due to its significant flexibility, low investment cost and not requiring high technology. Results of any blasting operation are affected by interaction between explosive materials and rock mass. Thus, knowledge of rock parameters can lead to optimization of blast results and powder factor. Parameters affecting blast results may be categorized as followings [1]:

  • Explosive specifications

  • Rock mass specifications

  • Tunnel and blast hole geometry

Many models have been proposed for prediction of specific charge; most of which are mainly developed using a technique for regression analysis.

A few decades ago the idea of artificial intelligence system was presented. Since then, the system was employed to solve many problems and could even serve to daily industrial problems. Artificial neural networks (ANNs), as one of the powerful tools in this system, have been able to bring advantages for solving engineering problems. Application of ANNs, as a pattern recognizer for nonlinear behavior prediction of specific charge in underground excavations, forms the core of this research. Using suitable input parameters, could lead to a reliable ANNs models for accurate prediction of specific charge in tunneling.


The data used in this research was measured in filed by Chakraborty et.al in India [2]. As seen in Table 1 these data are from four cases, field investigations were conducted by Chakraborty in inclined drifts of a coal mine, development galleries of two metal mines and a tunnel of a hydro-electric project. The sites are listed in Table 1.

Certain methodologies were adopted for obtaining data; reader is referred to Chakraborty et.al [2] for more information.

(Table in full paper)

3.1 Relevant parameters

Using a specific type of explosive, rock mass characteristics and geometry of tunnel and blast holes are the main parameters, which affect the specific charge.

This content is only available via PDF.
You can access this article if you purchase or spend a download.