Scientific Approach to Reduce Ground Vibration in Mines Due to Blasting

Abstract

With more and more people coming out of poverty every year, the number of consumers for industrial goods is rapidly increasing. This increase in demand directly affects the source of the industrial goods. The mining sector is showing a slower growth rate than earlier because of various environmental concerns and restrictions coming from governments. This puts impact on the production rate though demand keeps increasing. The easiest and cheapest way of extracting minerals from ground is blasting. Blasting has been in use as a practice of extraction ever since industrialization started. Although it is an easier method, the energy from blasting does cause a number of damage to the environment. As much as 80% of the blast energy goes to waste in the form of heat, noise etc. With such a small fraction of energy only being used, the rest damages the surrounding living and non-living creatures. This is why there have been rigorous researches all over the world dedicated to optimization of the blast designs so as to use the maximum power of blasting in fragmentation and waste the to be minimum. Therefore, it is essential to predict the vibration considering parameters of a blasting operation. There have been many strides in predicting the ground vibration using various regression methods and empirical methods. The empirical methods have become a standard for easy prediction of the ground vibration and design of benches. But the limitation of these empirical methods is that it has been calculated from the blast vibration studies conducted in a particular condition, which may not be suitable for all kind of strata conditions. The ground vibration changes with local strata conditions. Therefore, this paper aims to establish a stringent and novel method to predict the ground vibration using neural networks with the help of various blasting parameters. This paper further show that even with a small sample of data, the accuracy of the ANN is much more reliable than the prediction done with the empirical/conventional predictors currently in use.