Enduring effect of permeability texture for enhancing accuracy and reducing uncertainty of reservoir fluid flow through porous media

Abstract

Modeling reservoir permeability is one of the crucial tasks in reservoir simulation studies. Traditionally, it is done by kriging-based methods. More rigorous modeling of the permeability results in more reliable outputs of the reservoir models. Recently, a new category of geostatistical methods has been used for this purpose, namely multiple point statistics (MPS). By this new category of permeability modeling methods, one is able to predict the heterogeneity of the reservoir permeability as a continuous variable. These methods consider the direction of property variation in addition to the distances of known locations of the property. In this study, the reservoir performance of a modified version of the SPE 10 solution project as a pioneer case is used for investigating the efficiency of these methods and paralleling them with the kriging-based one. In this way, the permeability texture concept is introduced by applying some MPS methods. This study is accomplished in the conditions of real reservoir dimensions and velocities for the whole reservoir life. A continuous training image is used as the input of calculation for the permeability modeling. The results show that the detailed permeability of the reservoir as a continuous variable makes the reservoir simulation show the same fluid front movement and flooding behavior of the reservoir similar to the reference case with the same permeability heterogeneity. Some MPS methods enable the reservoir simulation to reproduce the fluid flow complexities such as bypassing and oil trapping during water flooding similar to the reference case. Accordingly, total oil production is predicted with higher accuracy and lower uncertainty. All studied cases are identical except for the permeability texture. Even histograms and variograms of permeabilities for the studied reservoir are quite similar, but the performance of the reservoir shows that kriging-based method results have slightly less accuracy than some MPS methods. Meanwhile, it results in lower uncertainty in outputs for this water flooding case performance.

Characterization of Subsurface Hydrogeological Structures With Convolutional Conditional Neural Processes on Limited Training Data

“…As a typical hydrogeological modeling method, stochastic simulation approaches refer to obtain a set of models that can express the uncertainty of hydrological phenomena (Graham & Talay, 2013; Tartakovsky, 2007; Wing et al., 2020; Zuo et al., 2020). It is widely used not only in geophysical inversion (Hermans et al., 2015), geological disaster (Huang et al., 2018), but also shows good performance in water resources assessment (Sidiropoulos et al., 2015) and petroleum exploration (Azamifard et al., 2020). Geostatistical simulations are a common category of these methods to reproduce the characteristics of complex hydrological phenomena, which can be divided into two‐point geostatistics and multiple‐point geostatistics (MPSs).…”