Three-dimensional Reconstruction And Simulation Of Porous Media

As common phenomena occur in the nature, fluid flow in porous media is of importance involving with human social and even organic activities, such as oil recovery and biofluids in human body. The porous media contains extremely complex micro-scale structures and the fluid flow in porous media is subject to Darcy's law. The mechanism of fluid flow in porous media is not only dependent on the fluid property itself but also a function of the solid matrix. Hence it is necessary to build up an effective description of rock matrix at the pore scale and provide a set of corresponding theories of displacement mechanism.It is impossible to use only a couple of parameters to equation the porous media due to its high irregularity and complicated topology. Therefore, the fundamental research should focus on extract an equivalent model where the pore scale fluid flow can be simulated through with its physical substances. The current available models are mostly based on regular topology. We propose to build models from real rock images in 3D.We focus on two aspects of reconstructing effective 3D rock model, which are 3D image acquisition and equivalent pore network extraction.1. We use SEM to obtain the high resolution image in 2D of porous media. The experiments show that the pore boundaries are clearly observed. The TEM image gives a further opportunity to study pore structure with depth information since pore space and grains are distinguished semi three dimensionally.2. The 2D images are processed using sigma function filter that keeps the boundary whilst removes the noise effectively.3. We reconstruct a 3D representative from 2D image. First, the model of a bundle of capillaries is built to equivalent the 2D image characteristics. Different reconstruction methods have been compared. The pore structure is characterized using multiple-point statistical method obtained a better reproduce of the long range connection of the pore space.4. We use synchrotron X-ray tomography to obtain the real 3D image of porous rocks. It is proved to be a highly effective method as a non-destructive imaging means. Filters need to be applied for the post processing to remove the noise and distinguish the pore space and solid phase.5. An algorithm of pore network reconstruction from arbitrary 3D image of porous media has been developed. A maximal ball method has been adopted and approved evidently appealing to distinguish pores and throats in the pore space.Based on the traditional multiple-point geo-statistical method, we develop a pore space reconstruction method using multi-point templates. This method is of advantage to reproduce long range connectivity and specific pattern occurring in its 2D origin. We differentiate our method from the traditional ways by initializing the sampling points. Multi templates are used to better capture the anisotropy in large system. Besides, the deducted size of a single template can improve the computation efficiency. We reached a good agreement on Berea sandstone reconstruction.We adopt the concept of maximal ball to extract the equivalent pore networks. The method can be applied to arbitrary 3D images. A few sphere packing models and the Berea pore image have been analyzed and benchmarked against the process based method. The results from our models again are approved with good agreements and applicable to the study of porous media.The theoretical solution of single phase permeability and effective permeability has been computed against the results from pore network simulations. The comparison shows that the pore scale model provides an effective solution with physics fundamentals.