Genetic full waveform inversion to characterise fractures

Understanding the physical properties of fractures, location, orientation, and geometry is of primary importance in the comprehension of geo-resources and geo-hazards. In doing so seismic waves can be an effective tool for non-invasive investigation of fracture parameters.

We have developed a novel stochastic Full Waveform Fracture Inversion employing Genetic Algorithms (GA-FWFI) to quantify mechanical properties of fractures (fracture-specific stiffnesses) in a synthetic sample. We focus on the inversion of transmitted waves, simulating ultrasonic laboratory data, on a specimen containing 6 horizontal fractures orthogonal to the wave propagation. The inversion results show a good convergence to the solution which best fits the observed data within the maximum number of generations and number of models set.