APPROCHE PSEUDO-GENETIQUE POUR LA SIMULATION STOCHASTIQUE DE LA GEOMETRIE 3D DE RESEAUX FRACTURES ET KARSTIQUES

Vincent Henrion. ( 2011 )
Institut National Polytechnique de Lorraine

Abstract

Fractures and karstic networks are known to signi cantly a ect ow paths and therefore raise speci c issues in a wide variety of geoscience fi elds. The common question beyond these problems is to determine whether there is a network of fractures and/or karstic conduits and if yes what are its geometrical and hydraulic characteristics. Characterization and modeling of these features is a challenge for it usually displays complex geometries and heterogeneous spatial distribution. Moreover, in most cases, neither fracture and karst nor their host environment can be observed or described with certainty at all scales and location of relevance. For these reasons, fractures and karstic networks are usually integrated into 3D geological model through a probabilistic framework. Stochastic object- or pixel-based simulations are commonly performed to generate 3D models of fractures and karst but failed to reproduce the whole complexity of these natural objects and 3D models often lack geological realism. To address the issues related to fracture and karst modeling, we present two genetic-like approaches. The motivation of this work is to constrain the stochastic simulation of fractures and karsts by geometrical and heuristic rules which mimic the physical processes governing their formation. The resulting fracture and karst models display similar characteristics as those of natural pattern.

Download / Links

BibTeX Reference

@PHDTHESIS{henrionth2011,
    author = { Henrion, Vincent },
     title = { APPROCHE PSEUDO-GENETIQUE POUR LA SIMULATION STOCHASTIQUE DE LA GEOMETRIE 3D DE RESEAUX FRACTURES ET KARSTIQUES },
      year = { 2011 },
    school = { Institut National Polytechnique de Lorraine },
  abstract = { Fractures and karstic networks are known to signicantly aect ow paths and therefore raise
specic issues in a wide variety of geoscience fields. The common question beyond these problems is
to determine whether there is a network of fractures and/or karstic conduits and if yes what are its
geometrical and hydraulic characteristics. Characterization and modeling of these features is a challenge
for it usually displays complex geometries and heterogeneous spatial distribution. Moreover, in most cases,
neither fracture and karst nor their host environment can be observed or described with certainty at all
scales and location of relevance. For these reasons, fractures and karstic networks are usually integrated
into 3D geological model through a probabilistic framework. Stochastic object- or pixel-based simulations
are commonly performed to generate 3D models of fractures and karst but failed to reproduce the whole
complexity of these natural objects and 3D models often lack geological realism.
To address the issues related to fracture and karst modeling, we present two genetic-like approaches.
The motivation of this work is to constrain the stochastic simulation of fractures and karsts by geometrical
and heuristic rules which mimic the physical processes governing their formation. The resulting fracture
and karst models display similar characteristics as those of natural pattern. }
}