Genetic-like Modeling of Karst Network.

in: Proc. 30th Gocad Meeting, Nancy

Abstract

Cave systems display strong spatial heterogeneities and have a huge impact on fluid flows. Providing a consistent geological model to flow simulators calls for a realistic 3D distribution of karst features. Henrion et al. [2007] and Pellerin et al. [2008] developed a stochastic method, based on ODSIM [Henrion et al., 2010], to generate 3D models of cave systems. The methodology consists in a genetic-like simulation of the development of karst network along fractures. Starting from a geological model which includes depositional pattern, fractures and petrophysical properties, a graph of connectivity is extracted [Vitel and Mallet, 2005]. This representation enables the application of a graph search algorithm to assess preferential flow paths between input and output points. However, the defined algorithm does not allow to reproduce the hierarchical organization of karst systems. In this paper, we propose a new implementation of the graph search algorithm to extract preferential flow paths reproducing the complex topology of karst networks. The approach combines a cost and a distance function to define optimal paths between the nodes of the graph of connectivity. In addition to the distance between input and output points, we suggest to use the distance to the previously extracted paths to obtain deeper connection levels between karst conduits. This results in a path geometry organizing progressively in a branching karst network. Results obtained with this algorithm are presented and perspectives are finally discussed.

Download / Links

BibTeX Reference

@inproceedings{BonneauGM2010,
 abstract = { Cave systems display strong spatial heterogeneities and have a huge impact on fluid flows. Providing a consistent geological model to flow simulators calls for a realistic 3D distribution of karst features. Henrion et al. [2007] and Pellerin et al. [2008] developed a stochastic method, based on ODSIM [Henrion et al., 2010], to generate 3D models of cave systems. The methodology consists in a genetic-like simulation of the development of karst network along fractures. Starting from a geological model which includes depositional pattern, fractures and petrophysical properties, a graph of connectivity is extracted [Vitel and Mallet, 2005]. This representation enables the application of a graph search algorithm to assess preferential flow paths between input and output points. However, the defined algorithm does not allow to reproduce the hierarchical organization of karst systems. In this paper, we propose a new implementation of the graph search algorithm to extract preferential flow paths reproducing the complex topology of karst networks.
The approach combines a cost and a distance function to define optimal paths between the nodes of the graph of connectivity. In addition to the distance between input and output points, we suggest to use the distance to the previously extracted paths to obtain deeper connection levels between karst conduits. This results in a path geometry organizing progressively in a branching karst network. Results obtained with this algorithm are presented and perspectives are finally discussed. },
 author = { Bonneau, Francois AND Henrion, Vincent AND Collon, Pauline },
 booktitle = { Proc. 30th Gocad Meeting, Nancy },
 title = { Genetic-like Modeling of Karst Network. },
 year = { 2010 }
}