Local Grid Refinement in Basin Modelling — What for?

Carpentier, Bernard¹
 Pegaz-Fiornet, Sylvie¹
 Wolf, Sylvie¹
 Kowalewski, Isabelle¹

¹IFP, Rueil Malmaison, France.

In basin modelling, data distribution is often very heterogeneous. In some parts of the basin, very detailed information is available with numerous wells and/or 3D seismic cubes whereas, in other parts, only coarse information is available. Constructing a numerical basin model which captures the fine information implies the use of a high resolution grid on the whole basin. This, in turn, leads to large basin models which cannot be run with a compositional description of the fluid, especially when using Darcy approaches.

To overcome this problem, basin modellers use drastic simplifications of the migration schemes as percolation and ray tracing approaches and/or used "tartan" grids which can be, in some cases, a satisfying alternative.

In order to solve this problem and to provide the best compromise between the density of the data and the number of cells of the model, IFP has developed a local grid refinement algorithm for its basin mode (Temis). It enables us to build a mesh which combines several grid dimensions into a single, fully coupled, basin model. The mesh is made of large cells in areas where there is a lack of information and small cells (in X, Y and Z) where more precise information is available. This approach, which optimises the total number of cells, allows detailed simulations on target areas without any compromise on their coupling with the rest of the basin.

This paper illustrates the interest of mixing different grids into a LRG model in some exploration contexts. It also highlights the interest of LGR for reservoir engineering purposes. Indeed, it is more and more obvious that the understanding of fluid heterogeneities in a petroleum field is controlled by regional factors such as the structural evolution of the trap and/or its filling history. Consequently, there is a need for the petroleum industry to integrate petroleum system modelling and the simulation of field scale processes into a single basin scale model and LGR can be a possible answer.

Field cases will be presented. The first one illustrates the need of local high resolution grid to predict biodegradation in potential prospects. A second field case illustrates the contribution of LGR to evaluate the potential of small satellite prospects where the "payback" is very sensitive to the fluid composition and volume in place. Finally, a third case illustrates the importance of integrating the regional and structural evolution of the trap into the prediction of fluid distribution.