Benchmark on the discretizations schemes used to run fluid flow simulations on unstructured grids

Margaux Raguenel and Pierre Samier and Gilles Darche. ( 2020 )
in: 2020 RING Meeting, ASGA

Abstract

To understand the behaviour of natural hydrocarbon resources", we run multiphasic multifluid (water, oil, gas) reservoir flow simulations in the subsurface, ie porous media. Different levels of complexity are addressed during these simulations. On the one hand, geological structures, such as discontinuities or rock physical properties variations, must be represented at different scales (from m to km). On the other hand, the integration of wells into the model is mandatory (at the meter scale). The industry standard is currently to use structured hexahedral grids, that do not have the flexibility to represent faithfully the geometry of the reservoir but that enable fast and representative simulations. For the last decades, significant efforts have been made to develop unstructured grids that better account for the geological structural 3D features and complex well patterns in reservoir models. However, even if several small or/and theoretical studies have been conducted, only few complete industrial simulation studies have been done with these technologies. \\ The use of unstructured grids calls indeed for several adaptations in flow simulators both in theory and in practice. One key point is the computation of the transmissibilities between cells and of well connection factors, with discretization schemes adapted to unstructured grids. \\ In this paper we present the workflow that has been developed to run flow simulations on unstructured grids. We address key points for the adaptation of the classical structured workflow, especially when transferring the grid characteristics to the simulator," and for the computation of the transmissibilities. Several discretization schemes to compute these transmissibilities are compared by running fluid flow simulations.

Download / Links

BibTeX Reference

@inproceedings{RAGUENEL_RM2020,
 abstract = { To understand the behaviour of natural hydrocarbon resources", we run multiphasic multifluid (water, oil, gas) reservoir flow simulations in the subsurface, ie porous media. Different levels of complexity are addressed during these simulations. On the one hand, geological structures, such as discontinuities or rock physical properties variations, must be represented at different scales (from m to km). On the other hand, the integration of wells into the model is mandatory (at the meter scale). The industry standard is currently to use structured hexahedral grids, that do not have the flexibility to represent faithfully the geometry of the reservoir but that enable fast and representative simulations. For the last decades, significant efforts have been made to develop unstructured grids that better account for the geological structural 3D features and complex well patterns in reservoir models. However, even if several small or/and theoretical studies have been conducted, only few complete industrial simulation studies have been done with these technologies. \\ The use of unstructured grids calls indeed for several adaptations in flow simulators both in theory and in practice. One key point is the computation of the transmissibilities between cells and of well connection factors, with discretization schemes adapted to unstructured grids. \\ In this paper we present the workflow that has been developed to run flow simulations on unstructured grids. We address key points for the adaptation of the classical structured workflow, especially when transferring the grid characteristics to the simulator," and for the computation of the transmissibilities. Several discretization schemes to compute these transmissibilities are compared by running fluid flow simulations. },
 author = { Raguenel, Margaux AND Samier, Pierre AND Darche, Gilles },
 booktitle = { 2020 RING Meeting },
 publisher = { ASGA },
 title = { Benchmark on the discretizations schemes used to run fluid flow simulations on unstructured grids },
 year = { 2020 }
}