Upscaling on the Girassol Field: Comparative Tests and Impact of Near-well Upscaling

Abstract

Abstract Girassol is a large and complex turbidite and highly faulted field. 3D high resolution seismic was shot, which gives good geological description and high level definition of heterogeneity. High quality seismic data allows constructing detailed stratigraphic model containing several tens of millions of gridblocks, which cannot be directly used for flow simulation. Reservoir properties must be upscaled on coarsen grid in dynamic flow simulation. This paper discusses the upscaling techniques appropriate for turbidite and highly faulted field such as Girassol field. Upscaling of absolute permeability and geological facies are particularly addressed. Near-well upscaling technique is illustrated. Introduction Girassol is a large and complex turbidite and highly faulted field located offshore Angola. The field is at 1400 m water depth and was discovered in 1996. With only 1000 m average cap rock burial, it is characterized by very good reservoir properties. Girassol is part of Congo fan and built during Oligocene times. The field consists of several individual sand complex deposits with strong permeability contrast between channels and levees. High quality seismic data is available which allows a very detailed geological description 1–3. The stratigraphic model, filled with detailed geological facies, contains several tens of millions of gridblocks, and cannot be directly used for flow simulation. Therefore, the reservoir properties must be upscaled for dynamic flow simulation. The main difficulty in upscaling is the calculation of equivalent absolute permeability, which has been greatly discussed in the literature 4–18. Heuristic methods are flexible and computational efficient. But they have not sufficient accuracy in most cases. Numerical methods are generally considered more accurate. But the upscaling results depend greatly on boundary conditions used in fine grid flow simulation 12–14,16,17, which leads to non-unique solutions. In general, no universal upscaling procedure can be used everywhere. The "best" upscaling procedure should adapt to geological context. In this paper, we have considered various upscaling approaches to find the most suitable procedure for the complex turbidite Girassol field. A representative zone was selected to test different heuristic and numerical upscaling methods, and to compare the effects of permeability and transmissibility upscaling. The near-well upscaling procedure 19–23 is also used in this kind of field to improve simulation accuracy. We have found that the numerical upscaling procedure with a permeameter boundary condition can be used, but that it is necessary to extend the calculation domain to improve the accuracy. Near-well upscaling is a key point to ensure a good agreement between fine grid and coarse grid simulations. In order to define the rock type in multiphase flow simulations, geological facies should be upscaled. In this paper, the most representative sandy facies is used in facies upscaling. Both single-phase and two-phase (water-oil) flows are tested. Comparing coarse grid and fine grid simulations, results are globally quite satisfactory, especially for the well production data. Upscaling methods A fine grid is used for geological modelling, while a coarse grid is used for dynamic flow simulation. Both of them are types of Corner-Point Geometry (CPG). However, we have a good topological relation between these two grids. A coarse gridblock regroups several fine gridblocks in each direction.