Review of Empirical and Intelligent Techniques for Evaluating Rock Fragmentation Induced by Blasting

Abstract

Rock fragmentation, or the fragment size distribution of blasted rock of bench blasting, is crucial in excavation of any civil or mining project. The blasting operation plays a pivotal role in the overall economics of opencast mines. The blasting affects all the downstream operations, i.e. loading, transport, crushing, and milling operations. Prediction of rock fragmentation is important for practicing blasting engineer. It is well known that the rock fragmentation depends upon blast design parameters such as stiffness ratio, powder factor, and maximum charge per delay. Measurement of blast fragmentation is vital for deciding efficiency of blasting. Various blast fragmentation measurements are evolving from sieve analysis to image analysis. Challenge still remains accuracy of fragmentation vis-a-vis time and cost required for measurement and analysis of fragment size and distribution. During initial era, various empirical equations were developed for predicting fragment size based on blast design parameters. During the last decade, various machine learning (ML) models such as artificial neural network and support vector machine have been proposed for prediction of rock fragmentation. These ML models were reviewed in this study and their advantages and disadvantages were discussed. In addition, practical applications of the ML techniques for civil and mining engineers will be described in detail. This study is a useful source for those who are interested to do further research in the field of rock fragmentation induced by blasting. Theory-based or physics-based ML is a new corridor of ML techniques, which are able to bring the concept of different theories behind rock fragmentation into modeling part to have a more generalized and accurate predictive techniques.