Francois Bonneau

Research Topic:

• Multiscale characterization of mechanical discontinuities.
  
• Stochastic simulation of Discrete Fracture Networks driven by geological and geomechanical concepts.

Keywords: Stochastic Simulation, Multiscale Characterization, Discrete Fracture Network, Optimization, Connectivity, Percolation, Numerical proxy.

My pages in research databases:

• researchgate
• ORCID

PhD Topic: Integrating indirect data in stochastic Discrete Fracture Network simulation.

The stochastic simulation of discrete fracture network is based on the sampling of distribution law that describe the
geometry of natural fracture networks. It generally simulates each fracture by selecting the position, the geometry and the dimensions of a planar object. The general organization of the discrete fracture network emerges from this stationary and stochastic process.

This thesis explores a pseudo genetic and stochastic approach using rules that drive the seeding and the propagation of non planar objects, and allow the emergence of a fractal organization. The simulation mimics the natural fracturing process by considering the constraint accumulation zone and the shadow zone associated to each fracture already simulated in the fracture seeding, growth and linkage. We explore the impact of the method on the fractal dimension of discrete fracture network models, and we quantify its impact on both the connectivity and the percolation threshold.

The validation of the stochastic approach is based on the realism of models both in terms of geometry and impact
on its physical behaviour. Our approach constrain the geometry of discrete fracture networks at fracture and at fracture network scales using statistic distribution laws and mechanical concepts. Mechanical discontinuities can also be described by indirect data that quantify the response of the fractured rock volume to dynamic or mechanical stimulation. We propose an efficient way to take into account flow information recorded from tracer tests and microseismic events that trigger after a hydraulic stimulation. The method is integrated during the stochastic simulation in order to remove the need of an optimization process that may be time consuming or may impact the fractal organization of the network.