Multiscale Reconstruction of Compositional Transport
Time: 12:45 – 13:30, December 6
Designing strategies for efficient oil production from reservoirs relies heavily on reservoir simulation studies, which in-turn is based on various nonlinear formulations. It is therefore very important to develop a robust simulation model that captures the flow of various components present in different phases and the associated thermodynamic and chemical interactions. A compositional formulation is a reliable option for understanding these complex subsurface processes. However, this type of model has a great computational cost, since the number of equations (nc) that needs to be solved in each grid block increases proportionally with the number of components employed.
Through this research work, we propose a robust methodology to determine the solution of a compositional transport problem. It is a two stage approach with two different prolongation operators been defined, with each having different computational complexities. In the first stage, a fine scale prolongation operator is implemented on a modified conservation equation with the objective of reconstructing the leading and trailing shock positions in space. Once the position of shocks are identified, the solution lying in the regions outside the shock can be solved on a coarse-scale mesh, since the structure of the transport solution outside of the two-phase region is relatively simple. Later the fine scale projection of this coarse solution is carried out using a constant prolongation operator.
Subsequently, the solution lying in the two phase region is determined based on the “Compositional Space Parameterization” (CSP) technique. The CSP approach enhances the EOS computations, by replacing flash calculation with interpolations in the parameter space of the problem. The phase behaviour of gas-injection processes is predominantly controlled by the properties of two key tie-lines that extend through the initial and injection compositions, and hence it is convenient to parameterize the problem based on these two tie-lines. By this way, the solution for ‘nc’ components lying in between the leading and trailing shocks is reconstructed by solving just two equations on the basis of parameterization technique. The proposed reconstruction strategy results in coarsening of the compositional problem in space as well as representation, thereby the simulation time is appreciably reduced by several folds without significant loss in accuracy.