Abstract
Subseismic-scale structural heterogeneities and their lateral variability make the characterization of fluid dynamics in carbonate reservoirs an open challenge. Even a precise reconstruction of fracture networks does not guarantee a comprehensive knowledge of preferential flow paths. Current characterization of flow parameters largely relies on 0.01 - 0.1 m scale laboratory experiments, sample plug measurements, and modeling. However, a large degree of interpolation prevents these methods from fully reproducing realistic flow conditions. The goal of this dissertation is to use 4D GPR in a reservoir-scale infiltration experiment, to assess the role of small-scale structural heterogeneities (such as deformation bands and open fractures) in controlling fluid migration in carbonate reservoirs. The purpose is to use this information to visualize and quantify the influence of these structural features in unsaturated domains, either as fluid conduits or fluid baffles. For the first time, a full-resolution 4D GPR experiment was conducted, over a 20 x 20 m area at a fractured carbonate analog, by injecting 2952 liters of water into a fractured host rock to track, quantify, and monitor at reservoir-scale the dynamic evolution of an infiltrated water mass. Water content changes computed on 3D volumes show that open fracture planes are preferential flow paths, while, on the contrary, deformation bands present sealing properties for cross-fracture fluid migration. In the early stages of the infiltration, the deformation bands act as fluid baffles, especially in fully-saturated conditions, while towards the end of the infiltration process, values of water content changes are similar for porous host rock and deformation bands. At later stages of the infiltration experiment, the fluid migration mechanism switches from gravity-driven to capillary-driven. A static 3D model of the surveyed area was used as input for a dynamic simulation. To reproduce workflows currently used in standard dynamic simulation, the original model was simplified in terms of precision and resolution. By comparing 4D GPR results and standard dynamic fluid flow simulation, the study gives insights for building realistic flow models that include structural heterogeneities. This is necessary condition to improve reservoir kinematic studies, to conduct efficient residual fluid recovery, and to reduce uncertainties when upscaling from plug to field scale. The 4D GPR experiment described in this study is applicable to every field environment and can be used to successfully characterize the reservoir quality of a fractured analog. The results of this dissertation represent a bridge between laboratory sample measurements and field-scale reservoir evaluation, giving insights on the influence of small-scale structural heterogeneities on fluid flow.