Stochastic Simulation of Karsts using Gopy: comparison with observed ones using automated extraction of geometrical and topological parameters.

A. Fournillon and Sophie Viseur and S. Abelard and A. Arfib and Jean Borgomano. ( 2011 )
in: Proc. 31st Gocad Meeting, Nancy

Abstract

Karstic features represent a major challenge in carbonate reservoir characterisation. Indeed, they are hardly observable and they have a great influence on fluid flow. To obtain conditioned and plausible representations of the entire karst networks, stochastic simulation techniques must be proposed. In this paper, a geostatistical method for simulating karstic networks is presented. It consists in decomposing karstic networks into sets of unidirectional element families, which are characterised by variograms and proportion trends to reproduce the connectivity between the different elements. This technique has been implemented in Gocad thanks to the Gopy interface. Moreover, a procedure is proposed to check the reliability of the generated structures. The study of accessible karsts is a key to obtain geometrical and topological parameters that can be compared between the simulated an observed karsts. All over the world, cavers extensively explore and map karstic networks. The surveys they make can provide useful analogues of unexplored karst. Thus, an automated method (KNIT) has been also developed to extract geometrical and topological parameters from 2D maps. The obtained parameters can be computed from both the simulated and the observed karsts to be finally statistically compared. In this paper, both a simulation algorithm is proposed to generate karst networks and a checking process, which allows comparison between simulated and observed karst networks. These procedure are based on the Gopy. To illustrate these methods, they have been applied on the Var area (SouthEast of France) where the cavers have listed more than 2000 caves. Among these caves, less than 1800 are mapped and less than 500 have a horizontal development exceeding 20m. The 2D maps of these horizontal caves represent the used database. Simultaneously, stochastic simulations of the karst networks have been generated and served as support for the same parameter extractions using both Gopy and KNIT. A statistical analysis of the different obtained parameters is finally presented.

Download / Links

BibTeX Reference

@inproceedings{FournillonGM2011,
 abstract = { Karstic features represent a major challenge in carbonate reservoir characterisation. Indeed, they are hardly observable and they have a great influence on fluid flow. To obtain conditioned and plausible representations of the entire karst networks, stochastic simulation techniques must be proposed. In this paper, a geostatistical method for simulating karstic networks is presented. It consists in decomposing karstic networks into sets of unidirectional element families, which are characterised by variograms and proportion trends to reproduce the connectivity between the different elements. This technique has been implemented in Gocad thanks to the Gopy interface.
Moreover, a procedure is proposed to check the reliability of the generated structures. The study of accessible karsts is a key to obtain geometrical and topological parameters that can be compared between the simulated an observed karsts. All over the world, cavers extensively explore and map karstic networks. The surveys they make can provide useful analogues of unexplored karst. Thus, an automated method (KNIT) has been also developed to extract geometrical and topological parameters from 2D maps. The obtained parameters can be computed from both the simulated and the observed karsts to be finally statistically compared.
In this paper, both a simulation algorithm is proposed to generate karst networks and a checking process, which allows comparison between simulated and observed karst networks. These procedure are based on the Gopy. To illustrate these methods, they have been applied on the Var area (SouthEast of France) where the cavers have listed more than 2000 caves. Among these caves, less than 1800 are mapped and less than 500 have a horizontal development exceeding 20m. The 2D maps of these horizontal caves represent the used database. Simultaneously, stochastic simulations of the karst networks have been generated and served as support for the same parameter extractions using both Gopy and KNIT. A statistical analysis of the different obtained parameters is finally presented. },
 author = { Fournillon, A. AND Viseur, Sophie AND Abelard, S. AND Arfib, A. AND Borgomano, Jean },
 booktitle = { Proc. 31st Gocad Meeting, Nancy },
 title = { Stochastic Simulation of Karsts using Gopy: comparison with observed ones using automated extraction of geometrical and topological parameters. },
 year = { 2011 }
}