Early warning signs for mining accidents: Detecting crackling noise

Abstract

Acoustic emission signals carry much of the information about collapsing cavities and progressing cracks in minerals. The acoustic noise signals are constituted by individual events, so-called “jerks,” which form a spectrum of “crackling noise” with well-defined characteristics. The close connection between collapsing minerals under uniaxial stress and the spectra of crackling noise has been systematically investigated over the last 5 years on small samples, but it is only in a new paper in American Mineralogist that Jiang et al. (2016) have made a major breakthrough by using much larger samples and higher stresses. They show that crackling noises change their spectral parameters from a non-critical steady state to an all-important precursor regime before samples finally collapse. They show that the noise distribution always follows a power law but that the exponents are different in the two regimes. Their work on coal and sandstone have hence shown that early warning signs can indeed be extracted from acoustic signals and thereby offer a new method to prevent mining accidents. The prediction of mining accidents is one of the great outstanding problems for mineralogists. Accidents in coal and hard rock mining are particularly severe with an alarming number of miners dying each year. Most of the deaths occur in developing countries, in China, and in rural parts of developed countries. Historically, one of the worst mining accidents occurred April 26, 1942, in the Chinese Benxihu (Honkeiko) Colliery (coal mine), killing 1549 miners. Several of these accidents are related to the collapse of mine shafts. Such collapses are not fully preventable when large tectonic movements occur, but the death toll can be massively reduced if the collapse can be predicted. A key research area for mineralogists is to identify any precursor events that can identify markers to predict the …