A three-dimensional, three-phase black oil simulator and a three-dimensional single-phase simulator for compressible and slightly compressible transport formulations are developed for modeling the flow regimes and flow geometries around sealing/non-sealing flow barriers in hydrocarbon reservoirs. A local grid refinement technique is utilized in these numerical simulation models to accurately capture the flow regimes and flow geometries.
In previous studies, a similar local grid refinement algorithm was applied to study impermeable flow barriers, water coning problems, and waterflood front tracking in the presence of horizontal wells. The previous application on impermeable flow barriers was only capable of capturing flow characteristics around the impermeable barriers placed orthogonal to the flow domain in two dimensions. In this study, however, inclined faults, which are more commonly found in nature, can be modeled and more accurately represented in three-dimensional flow domains. The stationary characteristic of sealing/non-sealing faults permits the use of static local grid refinement technique.
A system of faults in a large reservoir is considered in all of the cases studied. Sections of the reservoir are solved individually in the presence of geological discontinuities. During the implementation phase of the algorithmic approach an efficient flexible index file is developed and implemented for solving the system of equations. The results from the static local grid refinement technique indicate a much higher level of accuracy over the results generated using a coarse grid approximation around faulted zones of the reservoir. This level of accuracy is characterized by comparing the results generated by a completely fine grid and/or by a conventional refinement protocol.
The proposed model is capable of providing a better representation of faulted reservoir architectures using significantly less CPU time. The proposed model can also be used in pressure transient analysis of wells located near nonsealing fault structures.
The most powerful tool to predict reservoir behavior under various conditions is reservoir simulator. Mainly, reservoir simulators are utilized for studying future production schemes of the reservoir. Therefore, the model that solves mathematical equations should accurately reflect the physical characteristics of the reservoir.
Smaller grid block sizes may need to be used to generate more accurate solutions from the finite difference equations. However, as the number of grid blocks is increased, so are the computational time, and computer storage. This will translate into an increased cost of study. Therefore, a fine grid system yields more accurate solutions than a coarse grid system but takes much more amount of computer time than the coarse grid system for the same reservoir. In a local grid refinement technique only certain coarse blocks are subdivided into smaller grid blocks. Applications of local grid refinement include near well-bore domain studies, water coning problem, and representation of geological discontinuities such as faults and fractures.
The first level of discretization of a reservoir is called coarse grid representation. In most of the reservoir simulation studies, the scale of the base grid system is relatively coarse. If all the coarse grid blocks are subdivided into smaller grid blocks, a fine grid system will be generated.