The weakest characteristic of most scaled reservoir models is the incompetence of describing accurately the reservoir heterogeneity. This limitation has a strong impact in the adequate match of existing historical data, usually resulting in unrealistic predictions. The scaled reservoir model validation is a fundamental step toward developing a successful reservoir simulation model. The results achieved from an imprecise simulation model can be more harmful for reservoir management decisions than not having a simulation model at all.
In this paper we present a quick and efficient application designed to meet specific needs to preserve, in scaled models, the reservoir heterogeneities represented in high-resolution geostatistical realizations. The developed tool works in a systematized architecture where multi-attributes as geological, petrophysical and reservoir objects are stored in a dynamic hierarchical platform. The validation procedure, which works as a 3D visual interactive platform, includes optimization methods and transmissibility definitions using vertical windows.
We have applied the proposed tool to Dina Cretaceos and Palogrande Fields, Valle Superior del Magdalena Basin in Colombia, South America. The Dina Cretaceos Field is estimated to contain over 180 million barrels of oil in place of which 26% has been produced. The Palogrande Field is estimated to contain over 230 million barrels of oil in place with a recovery of 20.5 %. Departing from high-resolution geostatistical reservoir images, scaled models for simulation purposes were built up and validated with our tool. The flow capacity tuning procedure, a strong stratigraphic constraining, and a detailed control of sand to sand connections reduced the risk of oversimplifying the scaling up process. Therefore, a much better reservoir heterogeneity representation was achieved and a suitable history matching were possible without manipulating, deforming, or even losing the physical sense of the model.
Introduction
Increased resolution in reservoir characterization is currently driving the need for efficient and accurate upscaling techniques to correctly model reservoir heterogeneities and avoid unrealistic predictions of fluid flow in scaled models. A comprehensive review of published upscaling methods can be found in the literature.1–9 Upscaling is a technique that transforms a detailed geologic model to a coarse-grid simulation model in such a way the fluid-flow behavior in the coarse model can be duplicated.10 This is a non-trivial step in the process of preserving the geological features of the reservoir.11–16 An accurate upscaling consists of two inseparable parts: gridding and averaging.
The gridding procedure points towards capturing the global geologic features in terms of flow units, constrained by the major stratigraphic reservoir characteristics. For this purpose, the concept of Vertical Proportion Curve (VPC) can be used to define the minimal number of grid units needed to keep the vertical and areal development of the facies involved.
The averaging procedure aims to assign the appropriate grid blocks petrophysical properties to be used in the reservoir simulation. In our case, we use a hybrid approach to minimize the impact of upscaling the reservoir heterogeneity. This approach takes into account the geological features to apply suitable averages. The final upscaled model is validated against the high-resolution model by tuning the well flow capacities and defining the convenient transmissibilities using vertical windows.
Facing the stated problem demands a solid environment. In the last decade, object-oriented programming has gained popularity due to real-world attributes can be easily represented using object-oriented applied software. The code functionality is expressed by variables and methods implemented within each object, fitting the requirements needed to model abstract concepts. This environment can be perfectly used to solve the upscaling issue.
In this paper, we develop a static data integration methodology implemented into a dynamic hierarchical platform to overcome this problem. The methodology includes merging segments of static data from reservoir modeling outputs into a set of objects to represent the numerical reservoir model prior to simulation.
