Get access

Cell to node projections: An assessment of error

Authors

  • Imene R. Goumiri,

    1. Department of Civil and Environmental Engineering, Princeton University, Princeton, NJ 08544, U.S.A.
    Current affiliation:
    1. Student at MATMECA, Bordeaux, France
    Search for more papers by this author
  • Jean H. Prevost

    Corresponding author
    1. Department of Civil and Environmental Engineering, Princeton University, Princeton, NJ 08544, U.S.A.
    • Department of Civil and Environmental Engineering, Princeton University, Princeton, NJ 08544, U.S.A.
    Search for more papers by this author

Abstract

Reservoir simulators typically use cell-centered finite volume schemes and do not model directly the coupling of the flow processes with the geomechanics. Coupling of geomechanics with fluid flow can be important in many cases, but introducing fully coupled geomechanical effects in those simulators is not a trivial issue, because the geomechanics is better done by using the Galerkin vertex-centered finite element methods by which the solid displacements are computed at the vertices of the cells. This creates difficulties in interfacing cell variables with nodal variables. Uncoupled or loosely coupled models are used by many researchers/practitioners by which a reservoir model is coupled to a geomechanical model by staggering in-time flow and deformation via a sophisticated interface that repeatedly calls first flow and then mechanics. The method therefore requires projection of the reservoir cell variables onto the nodes of the geomechanics Galerkin finite element mesh.

In this note, we attempt to quantify the errors associated with cell to node projection operations. For that purpose, we use a simple model of the pressure equation for a heterogeneous medium in one dimension. We are able to derive the exact analytical solution for this problem for both nodal and cell pressures. This allows us to compute the errors due to projection analytically, function of meshing refinement and permeability field variations. We compute upper and lower bounds for the errors, and analyze their magnitude for a variety of cases. We conclude that, in general, cell to node projection operations lead to substantial errors. Copyright © 2010 John Wiley & Sons, Ltd.

Ancillary