Building centroidal Voronoi tessellations for flow simulation in reservoirs using flow information.

in: Proc. 30th Gocad Meeting, Nancy

Abstract

Simulating oil, gas and water flows in petroleum reservoirs is an important issue in performance forecasting for reservoir. For this, a simulation grid built in a geomodeler is used to discretize and solve the flow equations in the reservoir. As the last step is time and resource consuming, the number of grid cells is limited and is commonly much smaller than that of the geological fine grid representing high-resolution features. Therefore, the simulation process sensu-stricto can be performed only after the coarse grid generation (called gridding), and the transfer of properties from the fine grid to the coarse grid (called upscaling). The main challenges of gridding for flow simulation are 1) to best approximate the input geological model, 2) to minimize numerical errors during simulation process due to coarse cell shape and orientation. Centroidal Voronoi Tessellations (CVT) are very interesting because they guarantee a certain mesh quality and they can reduce numerical errors during flow simulation. Furthermore, algorithms exist to automatically improve CVT from a global point of view, taking into account geometrical information present in the fine scale geomodel, such as density and local anisotropy. We propose a new method to generate a simulation grid, based on variational CVT algorithms and flow information, such as permeability, velocity or vorticity. After briefly explaining the properties of CVT and the generation algorithms, we apply the methodology to 2D and 3D examples to generate Voronoi meshes whose cells are mostly quadrilaterals and hexahedra respectively.

Download / Links

BibTeX Reference

@INPROCEEDINGS{MerlandGM2010,
    author = { Merland, Romain and Levy, Bruno and Caumon, Guillaume and Collon, Pauline },
     title = { Building centroidal Voronoi tessellations for flow simulation in reservoirs using flow information. },
 booktitle = { Proc. 30th Gocad Meeting, Nancy },
      year = { 2010 },
  abstract = { Simulating oil, gas and water flows in petroleum reservoirs is an important issue in performance forecasting for reservoir. For this, a simulation grid built in a geomodeler is used to discretize and solve the flow equations in the reservoir. As the last step is time and resource consuming, the number of grid cells is limited and is commonly much smaller than that of the geological fine grid representing high-resolution features. Therefore, the simulation process sensu-stricto can be performed only after the coarse grid generation (called gridding), and the transfer of properties from the fine grid to the coarse grid (called upscaling). The main challenges of gridding for flow simulation are 1) to best approximate the input geological model, 2) to minimize numerical errors during simulation process due to coarse cell shape and orientation. Centroidal Voronoi Tessellations (CVT) are very interesting because they guarantee a certain mesh quality and they can reduce numerical errors during flow simulation. Furthermore, algorithms exist to automatically improve CVT from a global point of view, taking into account geometrical information present in the fine scale geomodel, such as density and local anisotropy. We propose a new method to generate a simulation grid, based on variational CVT algorithms and flow information, such as permeability, velocity or vorticity. After briefly explaining the properties of CVT and the generation algorithms, we apply the methodology to 2D and 3D examples to generate Voronoi meshes whose cells are mostly quadrilaterals and hexahedra respectively. }
}