Importance and Sensitivity of Variables Defining Throw and Flyrock in Surface Blasting by Artificial Neural Network Method

Abstract

Rock breakage by explosives is followed by throw or heaving the broken material and occasional flyrock. Heaving is a desired feature of blasting for efficient mucking. However, flyrock is a rock fragment that travels beyond the designated distance from a blast in surface mines, and poses a threat to adjacent habitats. Here, we decipher the importance and sensitivity of the variables and factors used to establish the predictive regime of throw with more emphasis on flyrock. The data collected were modelled using artificial neural network approach. The importance and sensitivity of variables and factors were delineated so that they are in tune with the rationale of the outcome of the blast. A combinatory approach was devised to arrive at minimal variables and factors to reduce the statistical redundancy, and to propose a rational predictive regime for throw and flyrock in surface mines.Keywords: Artificial neural network, blasting, flyrock, throw, surface mines.BLASTING is an integral part of excavation in mines and continues to be a major method of rock fragmentation due to the economy of operation. Blasting, in addition to fragmentation, is associated with throwing the muck generated, vibrations, air overpressure and flyrock. While fragmentation and throw are desired effects, flyrock is an undesirable outcome. Flyrock is a fragment of rock that travels greater distances than desired, in comparison to throw which is limited to a few multiples of bench height. Flyrock is not only a threat to nearby habitats, but poses a challenge to miners as all sorts of 'Objects of Concern ' (OC) 1 are affected by it. Flyrock is one of the major causes of blast induced fatalities and accidents 2 . There are several reasons for flyrock which belong to the domain of rockmass including structural discontinuities 3 , blast design and explosive variables. Several attempts were made by different authors to identify the reasons for flyrock and several equations have been proposed to predict flyrock distance. However, there is a disparity between cause of flyrock and the variables identified that have been used in prediction regime 4 . Such a disparity is reflected in Tables 1 and 2 and a comparison is shown in Figure 1. RESEARCH COMMUNICATIONS CURRENT SCIENCE, VOL. 111, NO. 9, 10 NOVEMBER 2016 1525 A similar compilation of variables used in models that predict distances travelled by flyrock is shown in Table 2.As seen in Table 2, variables namely stemming length, blasthole depth, specific charge, burden and blasthole diameter emerge as the principal ones in predicting the distance which a flyrock can travel.A comparison of top seven causative factors (Table 1), and those used in predictive equations (Table 2) are given in Figure 1.From Figure 1 we infer that principally two variables, namely burden and rock, differ with regard to cause and prediction, probably due to the difficulty in assessing rock mass and burden. Other variables closely follow each other in cause and prediction citations, establishing their importance.Acco...

A critical analysis of recent research into the prediction of flyrock and related issues resulting from surface blasting activities

“…Armaghani et al (2016a) No discussion on testing methodology. Raina and Murthy (2016) Blast data were collected from mines, however, there was no specific discussion on type of data or data acquisition methodology.…”

“…The testing methodology applied by researchers now becomes critical in order to determine what these error(s) are and the magnitude(s) thereof. Monjezi, Amini Khoshalan, and Yazdian Varjani (2010) 9-16-2 Monjezi et al (2012) 9-5-2-1 Raina,Murthy, and Soni (2013) 7-20-14-8-1 Ghasemi (2014) 6-9-1 Raina and Murthy (2016) 20-16-6-4-1…”

“…8. Raina and Murthy (2016) presented a study based on the ANN method' with the aim of identifying the significance of different parameters in flyrock prediction. Blast data was collected from ten mines and analysed by means of ANN software, EasyNN-Plus © , in order to design, train, and validate a suitable ANN model.…”

Flyrock in surface mining - Limitations of current predictive models and a better alterative through modelling the aerodynamics of flyrock trajectory

“…Artificial intelligence (AI) has been increasingly applied in the form of artificial neural networks (ANNs) and fuzzy logic (Tivedi et al, 2014). Such routines have been used with up to 21 parameters describing the blast and to some degree, the properties of the rock mass to be blasted (Raina and Murthy, 2016b). The independent parameters may be combined in linear arrays, or raised to various powers, or included in various mathematical functions such as logarithms and exponentials.…”

Calculation Method of the Blasting Throwing Energy and Its Variation Affected by the Burden

“…Raina A.K. et al [33][34][35] launched a series of studies on the fly-rock problem in throwing blasting. Aiming at the difficulty in throwing rock fragments from traditional cut blasting, Zhang H et al [36] proposed and demonstrated an advanced blasting technology.…”