Simulation stochastique basée-objet de dépôts fluviatiles

Sophie Viseur. ( 2001 )
INPL, Nancy, France

Abstract

Heterogeneity of fluvial sediment architecture, due to the deposition along a channel system, provides potentiel reservoirs for hydrocarbons. During exploration, reservoir knowledge is composed of data types (drilling, seismic and geological knowledge). And, computer stochastic simulation processes represent now an essential toll using subsurface dataset for oil fiel managements. The goal of these stochastic simulation methods is to generate several equiprobable models of the fluvial architecture. Each model has to honor the subsurface data and the a priori geological knowledge. There are two major approaches to 3D stochastic simulation: pixel-based and object-based. Pixel-based methods distribute property values (either continuous as permeability and porosity, or discrete as lithofacies indexes) into the reservoir volume. Object-based (or boolean) methods consist in generating objects that model the different fluvial bodies and distributing them into the reservoir volume. However, the difficulties generally met in such approaches is to be able to incorporate hard (e.g. well data) and soft (e.g. : proposition data) data while generating models for statistical analysis. Consequently, the works of our PhD aims at finding a new object-based process allowing to obtain rapidly several equiprobable models of fluvial architecture while accounting for the subsurface dataset.

Download / Links

    BibTeX Reference

    @PHDTHESIS{Viseur01,
        author = { Viseur, Sophie },
         title = { Simulation stochastique basée-objet de dépôts fluviatiles },
       chapter = { 0 },
          year = { 2001 },
        school = { INPL, Nancy, France },
      abstract = { Heterogeneity of fluvial sediment architecture, due to the deposition along a channel system, provides potentiel reservoirs for hydrocarbons. During exploration, reservoir knowledge is composed of data types (drilling, seismic and geological knowledge). And, computer stochastic simulation processes represent now an essential toll using subsurface dataset for oil fiel managements.
    The goal of these stochastic simulation methods is to generate several equiprobable models of the fluvial architecture. Each model has to honor the subsurface data and the a priori geological knowledge. There are two major approaches to 3D stochastic simulation: pixel-based and object-based. Pixel-based methods distribute property values (either continuous as permeability and porosity, or discrete as lithofacies indexes) into the reservoir volume. Object-based (or boolean) methods consist in generating objects that model the different fluvial bodies and distributing them into the reservoir volume.
    However, the difficulties generally met in such approaches is to be able to incorporate hard (e.g. well data) and soft (e.g. : proposition data) data while generating models for statistical analysis. Consequently, the works of our PhD aims at finding a new object-based process allowing to obtain rapidly several equiprobable models of fluvial architecture while accounting for the subsurface dataset. }
    }