Highlight
Image-Based Finite element Model for the Simulation of Fluid Flow in Porous Media
Achievement/Results
IGERT Fellow Nathan Lane and IGERT professors (Thompson from Chemical Engineering, Willson from Civil Engineering and White from Petroleum Engineering) are studying disordered porous materials by computer simulation of fluid flow inside their interconnected microstructures (see Figure 1). These studies will improve our understanding of processes such as oil and gas recovery, design of advanced sterilization membranes, and the transport of pathogens in groundwater. Because these materials exhibit micro- and nano-scale structures, traditional analysis has required indirect and/or continuum scale approaches. However, modern 3D imaging techniques such as synchrotron microtomography provide quantitative maps of the internal structure of materials.
Our work focuses on direct modelling of fluid flow using tomography data, which in turn can lead to a better understanding of processes in natural materials and improved design of man-made materials. In this work we are developing finite element (FEM) based pore-scale models, which offer an alternative to the widely used Lattice-Boltzmann Method (LBM). The LBM is well suited to modelling the often complicated internal structures of porous materials, but the adaptive nature of FEM models allows for improved flexibility in specifying parameters such as resolution, local refinement, and integration into multiscale frameworks. It also gives the user more control over the computational requirements (time, memory, and number of processors) of any given simulation.
Our recent work has emphasised understanding how to reduce minimum computational requirements while still obtaining consistent simulation results. The image-based FEM model has been used to study a sample of Berea sandstone, the internal structure having been imaged by synchrotron x-ray microtomography. For this study the size of the computational problem was varied while the size of the Berea sample (dimensions of the 3D image) was held constant. A procedure to intelligently reduce the size of the computational problem was shown to preserve the internal structure of the sandstone resulting in a 55% reduction in the computational requirement to achieve results matching a designated standard. Imaging such as the flow depicted in Figure 2 helps us understand the impact of mesh coarsening or refinement by comparison of the higher-resolution patterns (shown in purple) against the more computationally efficient model (shown in white).
Address Goals
The development of an FEM model provides a computationally cost effective alternative to more common LBM approaches for simulating fluid flow in real porous materials. The flexibility of FEM approaches is well suited to multiphysics, and multi-scale problems. This flexibility is helping engineers apply computer models to a more diverse and complicated range of problems. The research utilizes the Department of Energy’s (DOE) National Synchrotron Light Source (NSLS) at the Brookhaven National Laboratory, and the Advanced Photon Source (APS) at the Argonne National Laboratory.
