A Novel Combination of Tree-Based Modeling and Monte Carlo Simulation for Assessing Risk Levels of Flyrock Induced by Mine Blasting

The purpose of this study was to develop and introduce novel techniques for predicting and simulating the flyrock phenomenon that happens in mines due to blasting. Two tree-based techniques, genetic programming (GP) and random forest (RF), were developed and applied to predict flyrock distance. Identification and measurements of important parameters affecting flyrock were done in six different quarry sites in Malaysia. An extensive database was established and used to conduct various parametric studies for both GP and RF models to obtain the best models. The performance prediction of the best GP and RF models were evaluated using several performance indices and the GP model could provide higher capacity in predicting flyrock distance. Afterward, Monte Carlo simulation (MCS) was used with the developed GP equation to analyze the flyrock risk in the studied sites. The MCS results showed that only 10% of the flyrock events will exceed 290 m. This proved that the blast safety area can be identified clearly for the blasting operations in the studied sites using MCS. It is important to note that the developed GP, RF and MCS models can be implemented in other relevant blasting environmental issues.

Automated summary

Novel techniques were developed in this study to predict and simulate the flyrock phenomenon in mines due to blasting, with GP and RF models showing good performance in predicting flyrock distance in different quarry sites in Malaysia. Monte Carlo simulation was used to analyze flyrock risk, showing that only 10% of flyrock events will exceed 290 m, which can help in identifying blast safety areas for blasting operations.