Techniques for incorporating faults in reservoir modeling systems have historically been developed first for extensional systems, and then extended to compressional systems. This approach has often led to reverse or thrust faults beings treated as special cases, requiring much manual editing of the model. The resulting 3D reservoir grids were often not acceptable to reservoir engineers due to the shape and orientation of the grid cells along the faults.
The methodology that we have developed starts with the premise that no type of fault should need to be treated as a special case. The fault modeling can handle any shape fault, and the horizon modeling properly models repeat section. Reservoir grids created from these frameworkscan treat faults as either pillar or stair-step faults; stair-step faults are often desired by the reservoir engineer to facilitate flow simulation. We have developed a new technique for indexing a 3D grid built with stair-stepped reverse faults that preserves layer connectivity across the faults. In this manner, the grid that the engineer needs is also one that the geologist can use for facies and petrophysical modeling. The asset team is able to share the model without rebuilding or modifying the basic structure for each discipline, allowing the modeling to be part of a larger workflow.
Building structural frameworks for use in reservoir modeling often consists of three main steps: modeling faults, modeling horizons, and constructing the reservoir grid. In some methods, the last two steps are combined into a single process. Existing methods have limitations in each step; some methods are stronger in fault modeling, some in horizon modeling, and some in grid building, but to date, none have ideal solutions for the unique problems of reverse or thrust faults. Compressional structures pose problems in all three steps. First, modeling thrust faults, where the dip of the fault can change from low angle to high angle, can be difficult for pillar-based fault modeling systems, as the number of nodes on the pillar is often limited to a fairly small number. This limits the number of straight-line segments on the pillar and the curved nature of the fault may not becaptured correctly. In addition, low angle intersections between thrust faults may not be correctly modeled, as these low angle intersections can distort the pillar shape. In the second step, horizon modeling, there is the problem of multi-valued horizon data. That is, because the horizon exists at two or more different depths at the same XY location, the horizon surface cannot be modeled as a single continuous surface. A method that goes directly to a 3D grid from the fault framework without creating an intermediate horizon model can avoid this problem, but this solution is not ideal as it requires the modeler to make decisions about grid parameters very early in the modeling process. It also requires a complete rebuilding of the model whenever grid parameters are changed. Finally, a 3D corner point grid cannot use stair-step gridding for the thrust faults without either creating problems in the shadow zone areas or losing layering index continuity across the faults.