Can we discretize reservoir models in chronostratigraphic space ?

in: Proc. 28th Gocad Meeting, Nancy

Abstract

Reservoir models are commonly based on stratigraphic grids. Mallet [2004] proposes an alterna- tive by associating reservoir geometry with a chronostratigraphic model, in which geological layers and heterogeneities are represented at the time of deposition. Consequently, in this space, geological layers are horizontal planes and no post-sedimentary events such as faults, folds and sedimentary hyatus have yet occured. Such models are used to model ne scale heterogeneities and to perform geostatistical algorithms. Instead of passing from the chronostratigraphic model to the reservoir model by applying local upscaling methods [Mallet, 2004], in this paper, we propose to discretize directly the ne-scale model in chronostratigraphic space. This means to create a network of nodes, (called a pipenetwork, [Vitel, 2007]) at which rock properties are stored. Each node has also a control volume and nodes are linked by pipes which are used to store the transmissibility property. Such ne model can then be upscaled globally using the method proposed by Vitel [2007]. Both control volumes and transmissibilities can indeed be computed in chronostratigraphic space using sedimentation velocity [Mallet, 2004]. Moreover, boundary conditions can be transferred from the physical space using the chronostratigraphic parameterization. In order to use this discretization, all the phenomena that could have occurred since the time of rock deposition must be handled. The sedimentation velocity enables to link the time in chronos- tratigraphic space to the actual thickness of sediments. In this article, we explain how to discretize conforming sequences. We show that it is possible to handle discontinuity surfaces such as faults and sedimentary hyatus from a theorical point of view and give some keys to implement such a method.

Download / Links

    BibTeX Reference

    @INPROCEEDINGS{241_cherpeau,
        author = { Cherpeau, Nicolas and Caumon, Guillaume },
         title = { Can we discretize reservoir models in chronostratigraphic space ? },
     booktitle = { Proc. 28th Gocad Meeting, Nancy },
          year = { 2008 },
      abstract = { Reservoir models are commonly based on stratigraphic grids. Mallet [2004] proposes an alterna-
    tive by associating reservoir geometry with a chronostratigraphic model, in which geological layers
    and heterogeneities are represented at the time of deposition. Consequently, in this space, geological
    layers are horizontal planes and no post-sedimentary events such as faults, folds and sedimentary
    hyatus have yet occured. Such models are used to model ne scale heterogeneities and to perform
    geostatistical algorithms.
    Instead of passing from the chronostratigraphic model to the reservoir model by applying local
    upscaling methods [Mallet, 2004], in this paper, we propose to discretize directly the ne-scale
    model in chronostratigraphic space. This means to create a network of nodes, (called a pipenetwork,
    [Vitel, 2007]) at which rock properties are stored. Each node has also a control volume and nodes
    are linked by pipes which are used to store the transmissibility property. Such ne model can
    then be upscaled globally using the method proposed by Vitel [2007]. Both control volumes and
    transmissibilities can indeed be computed in chronostratigraphic space using sedimentation velocity
    [Mallet, 2004]. Moreover, boundary conditions can be transferred from the physical space using the
    chronostratigraphic parameterization.
    In order to use this discretization, all the phenomena that could have occurred since the time of
    rock deposition must be handled. The sedimentation velocity enables to link the time in chronos-
    tratigraphic space to the actual thickness of sediments.
    In this article, we explain how to discretize conforming sequences. We show that it is possible
    to handle discontinuity surfaces such as faults and sedimentary hyatus from a theorical point of
    view and give some keys to implement such a method. }
    }