Development of a new empirical fragmentation model using rock mass properties, blasthole parameters, and powder factor

Abstract

Rock fragmentation as one of the most important blasting results plays an indispensable role in subsequent stages such as secondary breakage, loading, hauling, crushing and grinding, and relevant energy consumption. Since several rock mass properties, blasthole parameters, and powder factor can affect the fragmentation results, development of an accurate model to predict fragment size has long been a complicated subject. In this study, rock mass properties, blasthole parameters, powder factor (q), and fragment size distribution of blasted rock using image analysis technique were determined for several blasting operations in different zones of Sungun open pit copper mine, Rashakan limestone mine, Soufian limestone mine, and Golgohar open pit iron mine. The median fragment size (X50) varied from 10.4 to 32.1 cm, and the predicted X50 by the modified Kuz–Ram models was significantly different from the X50 of the results. Because only a coefficient has been changed from one model to another and the impact of the combination of the parameters on X50 has not yet been well investigated, the impact of individual essential parameters such as blasthole diameter (ϕh), charge per blasthole (Q), rock mass properties, and q on the X50, X80, and uniformity index (n) was originally analyzed to clarify how their combination affects the fragment size of blasted rocks. ϕh and Q had a similar or a parallel effect on the fragmentation results. Two types of relations between X50 with the new combination of Q, q, SANFO, and blastability index (BI) and X50 with the combination of ϕh, q, SANFO, and BI with acceptable correlations were obtained and the correlations also increased when the parameter of joint plane spacing (JPS) was adjusted in the BI system. At the end, a new empirical model was achieved to predict X50 as a function of combination of Q, q, SANFO and adjusted BI with higher correlation (R2 = 0.865), less root mean square error (RMSE = 3.3 cm), and less coefficient of variation (CV = 19.9%) with actual field results.