Estimating Interfacial Curvature To Assess The Impact Of Ostwald Ripening On The Stability Of Residually Trapped CO2

Abstract:

The stability of residually trapped CO₂ is often taken for granted in the simulation studies used for predicting the long-term fate of CO₂ in geological storage reservoirs. Ostwald ripening is one of the mechanisms that could potentially remobilize residually trapped CO₂. This would cause the gradual growth of ganglia with low capillary pressures, at the expense of ganglia with higher capillary pressure. Ostwald ripening will be driven by differences in capillary pressure between ganglia, and subsequent diffusion of dissolved CO₂ through the aqueous phase. Therefore, a critical question is to understand the distribution of capillary pressure in isolated ganglia. The goal of this study is to develop reliable methods for estimating capillary pressure of individual ganglia of gases that are trapped during imbibition. Multi-resolution X-ray microtomography datasets from air-water spontaneous imbibition experiments in sintered glass beads and sandstone samples with voxel sizes varying from 0.64 to 4.44 µm were acquired at the Advanced Light Source, in Lawrence Berkeley National Laboratory. A series of computational techniques to estimate curvature at the interface between two immiscible fluids in porous were developed. In a first step, isosurfaces are extracted using resources from MATLAB’s Image Processing Toolbox or the Avizo software suite resulting in a triagonal mesh representing the considered surfaces. A second step is to identify and separate the interfaces between each of the three phases. The mesh is then smoothed and its curvature is estimated.

The sensitivity of results to different curvature estimation and smoothing techniques is studied. Estimating curvature on unsmoothed meshes shows a high degree of sensitivity to the resolution of the images, as well as the method chosen to calculate curvature. When the mesh is smoothed using a heat diffusion method however, curvature estimation using different methods and resolutions converges, as verified by calculating the standard deviation of the point-by-point difference in curvature values.

Three-dimensional visualization of a curvature calculation comparison on the surface of a disconnected ganglion in sintered glass beads (top/bottom: low/high resolution, left/right: raw/smoothed mesh)