• Author / Uploaded
• loralara

Citation preview

INTRODUCTION Upscaling has become an increasingly important tool in recent years for converting highly detailed geological models to simulation grids. Upscaling is needed to bridge the gap between these two scales. Given a fine-scale reservoir description and a simulation grid, an upscaling algorithm assigns suitable values for porosity, permeability, and other flow functions to cells on the coarse simulation grid. Many possible choices of upscaling approach exist. Upscaling calculations use Darcy's law equation

TUTORIAL 3A Upscaling Relative Permeability Curves The objective was to take three cells out of the high permeability zone in model TUT2C (which were created in tutorial 2) and split it into 150 x 1 x 20 cells which have similar dimensions to a core sample in the X and Z directions. The original rock relative permeability data was used in these fine scale cells, and once the Eclipse simulation of a water flood in this fine scale model has been completed, the Pseudo program was used to generate pseudo relative permeability curves assuming high & low permeability values. The resulted pseudo relative permeability curves replaced the original rock curves in the coarser scale model (ROCK.DATA) & (PSEUDO.DATA). By comparing the two simulations using the pseudo and rock relative permeability, it was obvious that the pseudo relative permeability curves produced a better reservoir description. The PSEUDO model showed a better recovery & sweep efficiency, hence a higher value of total field oil production compared with the ROCK model, (Figure 1). Also, by looking at the Floviz saturation profiles (graphs 1 to 4) for both models, an early water breakthrough was observed in the ROCK model whereas when using the PSEUDO model, a later water breakthrough occurred (Figure 2). So, by using the pseudo rel. perm. curves a much better detailed reservoir description was obtained by observing the change in Pc values over the fine scale (shorter water front), which in turn reduced the effect of numerical dispersion. In order to make pseudos more representative, we might have to inc«AddressBlock»rease include more Pc values so that we can represent the detailed movement of fluids inside the reservoir. Also, if we play with

the transmissibility between layers and vertical to horizontal permeability ration TUTORIAL 3B Lab-scale model ACROSS The model is a 20x20x1cm slab of rock with alternating 2cm layers of 500 and 50md, and 2cm thick buffer zones on either end; grid cells are 1cm x 1cm x 1cm. The model is designed to illustrate flow processes operating at the scale of sediment lamination. Layers 2cm thick with a 10:1 permeability contrast are typical of a good reservoir sandstone. The model is positioned vertically so that gravity, viscous and capillary forces are all operating. The objective is to understand flow behavior at the small scale and to calculate pseudofunctions for use in upscaling calculations, but the model could also represent a (rather large) core analysis experiment. We calculated the injection rate Q as follows, , The time T to flood 1 pore volume

, so TSTEP was set to 20*1

By looking at the pressure & saturation distribution profiles (Figures 3 and 4), it's obvious that the higher pressure drop takes place at the high permeability layer (viscous dominated fluid flow). Also, the most bypassed capillary oil exists in the high perm. layers because of the low capillary pressure. The high water saturation exists in the low perm. layers because most of the oil has been swept by water. When opening the ACROSS model in FloViz it shows that the high perm. layers has the viscous force as the dominant force whereas in the low perm. layers the capillary force is the dominant force. Also, there is no significant gravitational effects because gravity has to have continues high perm. layers along the flow in order to be able to see its effect. In this case the high perm. layers are interrupted by the low perm. layers, the matter which cuts down the gravity effect. TUTORIAL 3C Lab-scale model ALONG The model is positioned horizontally along the flow direction and again the objective is to understand the flow behavior and the effects of gravity, viscous and capillary forces in the water flooding process. TUTORIAL 3D

Lab-scale model UNIFORM The model is now changed so that it represents two large blocks with low & high permeabilities respectively. When the oil recovery efficiency (ROE for region 2) was plotted for the three models ACROSS, ALONG & UNIFORM, there were differences between them (Figure 5). The best ROE was obtained by the ALONG model & the worst was the ACROSS model. The water flooding works best in the low perm. layers leaving the capillary trapped oil in the high permeability layers. At the ACROSS model the varying permeability layers are perpendicular to the flow direction, this makes it difficult to the water to sweep most of the oil because of the discontinuity of permeability across the flow. Whereas in the ALONG model the layers are positioned along with the flow direction which makes it easier to sweep the low perm. layers first and by continuing the injection eventually most of the capillary trapped oil will be recovered too. If we set the capillary pressure to zero, we would get a better sweep efficiency and better recovery because we would have a piston like displacement.