Predicting the peak particle velocity from rock blasting operations using Bayesian approach

Abstract

AbstractMeasuring the blast-induced ground vibration at blasting sites is very important, to plan and avoid adverse effects of blasting in terms of the peak particle velocity (PPV). However, the measurement of PPV often requires time, cost, and logistic commitment, which may not be economical for small-scale mining operations. This has prompted the development of numerous regression equations in the literature to estimate PPV from a relatively easier to estimate scaled distance (SD) measurement. With numerous regression equations available in the literature, there is a challenge of how to select the appropriate model for a specific blasting site, more so that rocks behave differently from site to site because of different geological processes that rocks are subjected to. This study develops a method that selects appropriate models for specific blasting sites by comparing the evidence and occurrence probability of different regression models. The appropriate model is the model with the highest evidence and occurrence probability given the available blasting site SD data. The selected model is then integrated with prior knowledge and available blasting SD data in Bayesian framework for probabilistic characterization of PPV. The SD and PPV data at the opencast coal mine, Jharia coalfield in the Dhanbad district of Jharkhand, India, is used to illustrate and validate the approach. The mean and standard deviation of simulated PPV samples from the proposed approach are 12.38 mm/s and 7.36 mm/s, respectively, which are close to the mean of 12.03 mm/s and standard deviation of 9.24 mm/s estimated from the measured PPV at the site. In addition, the probability distribution of the simulated PPV samples is consistent with the probability distribution of the measured PPV at the blasting site.