The 2D versus 3D imaging trade-off: The impact of over- or under-estimating small throats for simulating permeability in porous media
Abstract
Geochemical reactions in the subsurface can alter the porosity and permeability of a porous medium through mineral precipitation and dissolution. While effects on porosity are relatively well understood, changes in permeability are more difficult to estimate. In this work, pore-network modeling is used to estimate the permeability of a porous medium using pore and throat size distributions. These distributions can be determined from 2D Scanning Electron Microscopy (SEM) images of thin sections or from 3D X-ray Computed Tomography (CT) images of small cores. Each method has unique advantages as well as unique sources of error. 3D CT imaging has the advantage of reconstructing a 3D pore network without the inherent geometry-based biases of 2D images but is limited by resolutions around 1 μm. 2D SEM imaging has the advantage of higher resolution, and the ability to examine sub-grain scale variations in porosity and mineralogy, but is limited by the small size of the sample of pores that are quantified. A pore network model was created to estimate flow permeability in a sand-packed experimental column investigating reaction of sediments with caustic radioactive tank wastes in the context of the Hanford, WA site. Before, periodically during, and after reaction, 3D images of the porous medium in the column were produced using the X2B beam line facility at the National Synchrotron Light Source (NSLS) at Brookhaven National Lab. These images were interpreted using 3DMA-Rock to characterize the pore and throat size distributions. After completion of the experiment, the column was sectioned and imaged using 2D SEM in backscattered electron mode. The 2D images were interpreted using erosion-dilation to estimate the pore and throat size distributions. A bias correction was determined by comparison with the 3D image data. A special image processing method was developed to infer the pore space before reaction by digitally removing the precipitate. The different sets of pore property distributions were used to generate different network flow models, to examine permeability alterations due to reaction-induced changes in throat sizes. 3D CT images, limited to a resolution of approximately 4 microns, miss small throats present at grain-to-grain contacts. The higher resolution of SEM images captures small throats between grains, however grain surface roughness and other small scale features may be misinterpreted. Precise determination of throat distributions requires careful thresholding to distinguish flow-conducting throats from throats leading to pores that are really just surface roughness. Using the pore network model, the sensitivity of permeability to the throat size roughness threshold was evaluated. Permeabilities calculated from the 2D and 3D pore and throat size distributions are compared to determine the impact of the lower resolution 3D images missing small throats.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2011
- Bibcode:
- 2011AGUFM.H53N..02P
- Keywords:
-
- 1009 GEOCHEMISTRY / Geochemical modeling;
- 1829 HYDROLOGY / Groundwater hydrology;
- 1831 HYDROLOGY / Groundwater quality