A direct pore-scale modeling technique is developed to study flow and transport in naturally-occurring tight formations. The model takes as input high-resolution x-ray micro-CT images of naturally-occurring porous media. The governing equations of flow and transport are then directly solved in the high-resolution image. Also, a highly scalable multi-GPU code is developed to handle the computational cost of the simulations. We achieve more than 1200X speedup with 54 GPUs on a distributed memory machine using this hybrid parallel code. The model is then employed to study solute transport in a tight sandstone with 1 mD permeability. We first acquire a 0.5 microns resolution image of the sample using a state-of-the-art micro-CT scanner. The image is then used by the model to simulate flow and transport at the pore level. We demonstrate how the pore-scale features of tight rocks affect solute dispersion in the medium.
Tight oil and gas resources (e.g., tight gas, shale gas, and shale oil) have emerged as significant sources of hydrocarbons in the US in recent years. The growing importance of these resources is reflected in the energy projections of EIA, with technically recoverable US shale gas and shale oil resources now estimated at 542 trillion cubic feet and 33 billion barrels, respectively . Tight formations typically exhibit geological variability on macro, micro, and nano scales that result in marked variability in hydrocarbon production. There is currently a significant gap in fundamental scientific understanding of flow and transport in tight rocks. The physics of flow in reservoirs with micro- and nano-darcy permeability and/or micro fracture networks can be significantly different from the physics of fluid flow in conventional reservoirs. Characterization of the pore networks of very fine-grained rocks and the effect of micro fractures is key to understanding flow in these systems and reducing the uncertainties in reserves estimates and reservoir performance predictions. Therefore there is a significant need for an array of powerful techniques to help develop an improved understanding of the transport phenomena in these rock systems.
Experimental techniques used to investigate flow processes relevant to hydrocarbon recovery schemes from these tight rocks are expensive and time consuming, but undoubtedly necessary. These techniques often involve the use of various imaging technologies (e.g., micro and nano computed tomography (CT) and scanning electron microscopy) as well as specialized core-flooding methods. These techniques are now being augmented with Digital Rock Physics (DRP) methods with the advent of less expensive high performance computing resources and imaging technologies. DRP couples most faithful representations of the pore space (e.g., high-resolution micro- and nano-CT images) with flow and transport physics at the pore scale to predict flow functions of tight rock samples that are often prohibitively difficult to measure in the laboratory. These technologies are usually divided into two categories:
direct pore-level models, and
pore-scale network models.
In this paper we focus on the first category and discuss how it can be used as an effective tool to shed light on some of the subtle flow and transport physics at the pore scale in a tight sandstone sample. We first present the pore-scale model use in this study. We also present the computational performance we have achieved through developing a multi-GPU code. Finally, we present the simulation results of solute dispersion in a tight sandstone.