Is That Normal? Fundamental Observations For Best Practice Blast Vibration Analysis

摘要

The scaled-distance model for blast vibration analysis is the standard method employed throughout the surface mining and quarrying industries to model Peak Particle Velocity (PPV) data. Although empirical, the method is widely accepted and through the simplicity of its application and ease of graphical verification is set to remain a popular option. The validity of the technique relies principally on the fundamental parametric assumption that blasting data is normally distributed. Throughout the evolution of scaled-distance modelling, this is an issue that has most often been presumed rather than proven (most likely due to the tediousness of the calculations involved). In the few cases when normality has been given some degree of consideration, the result has been the product of a simple visual inspection of frequency distributions: clearly unsatisfactory. However recently the soundness of this assumption has been brought into question, with obvious connotations for blast modelling in general. This paper considers this fundamental issue and seeks to investigate the normality hypothesis within the blasting context through an in-depth statistical analysis of what possibly can be described as an 'ideal' blasting dataset. This being a rare case study concerning an opencast coal site in England, where for the entire life of the site every single blast was monitored and recorded through the use of permanent monitoring stations. This has provided a unique 'total population' dataset of considerable size from which the assumptions of normality are put to the test. Through the application of systematic parametric tests, the results illustrate: 1. The quantity and quality of data required for normality to be achieved 2. The real effects of model instability, skewness and kurtosis on permitted charge weights 3. The critical observations required for proving normality 4. The relationship between proven normality and model stability. In conclusion, the key element in achieving blasting data normality is shown to be fundamentally related to the way the data is collected in the first place and adds further credence to the importance of correct monitoring technique and the adherence to strict protocols.