Crackling noise in digital and real rocks - implications for forecasting catastrophic failure in porous granular media

摘要: ‘Crackling noise’ occurs in a wide variety of systems that respond to steady-state external forcing an intermittent way, leading sudden bursts energy release similar those heard when crumpling piece paper or listening fire. In rock physics changes internal stress associated with microscopically-brittle rupture events lead acoustic emissions can be recorded on the sample boundary, and used infer state damage. Crackling noise is inherently stochastic, but population often exhibits remarkably robust scaling properties, terms source area, duration, energy, waiting time between events. Here we describe how these properties emerge evolve spontaneously fully-dynamic discrete element model sedimentary rocks subject uniaxial compression applied at constant strain rate. The elements have structural disorder real rock, this only heterogeneity. Despite stationary rate lack any time-dependent weakening processes, results are all characterized by emergent power law distributions over broad range scales, agreement experimental observation. As deformation evolves, exponents change systematically way evolution damage experiments . potential for real-time forecasting catastrophic failure obeying such rules then examined using synthetic data from laboratory tests prior volcanic eruptions. combination non-linearity constitutive irreducible stochastic component governed material heterogeneity finite sampling AE leads significant variations precision accuracy forecast time. This proportion ‘false alarms’ (forecast too early) ‘missed events’ late), as well over-optimistic assessments quality known (the ‘benefit hindsight’). becomes progressively more complex, diminishes, going ideal synthetics controlled open natural larger scales space