Abstract: |
Multiple-Point Statistics (MPS) simulation has emerged recently as a practical facies modeling technique to simulate complex geological patterns, such as sinuous channels, that cannot be modeled using (two-point statistics) variogram-based techniques. MPS simulation consists of extracting patterns from training images that describe the type of facies heterogeneity expected in the subsurface, and then reproducing similar patterns while honoring all conditioning data collected from the field under study. The next step in conventional reservoir modeling workflows consists of populating porosity, permeability and water saturation within the facies elements simulated by MPS. This step is typically performed using traditional variogram-based techniques such as Sequential Gaussian Simulation (SGS). However, in contrast to object-based modeling where intra-geobody coordinates can be computed while geobody objects get simulated, pixel-based methods such as MPS do not record the position of the cells relative to the simulated facies geobody that they belong to, thus pixel-based facies models cannot directly provide SGS with the required information for reproducing petrophysical trends within individual facies geobodies, for example decreasing porosity from base to top, or from axis to margin in turbidite channels. Yet intra-geobody petrophysical property trends are commonly observed, particularly in clastic environments; they create preferential flow paths that may have a significant impact on reservoir performance.
In this paper, we propose two alternative workflows to simulate petrophysical property trends within facies geobodies modeled using MPS. The first solution requires generating a training image of the petrophysical property to be modeled within a particular facies, and then the original MPS simulation program SNESIM is modified to be able to simulate continuous variables by discretizing the range of (continuous) petrophysical property values into a limited number of categories. The second solution consists of post processing the MPS facies model to calculate the distance from each cell to the closest boundary (a distance function) of the geobody to which the cell belongs, either in 3D or along a particular plane or direction. Then a function of that distance, which can account for facies geobody overlapping, is used as a local trend in SGS with locally varying mean. |
