19.12.14 KB. Ok to add published version to spiral after 6 months embargo - expires 15 March 201

We present a particle‐based method to simulate carbonate dissolution at the pore scale directly on the voxels of three‐dimensional micro‐CT images. The flow field is computed on the images by solving the incompressible Navier‐Stokes equations. Rock‐fluid interaction is modeled using a three‐step approach: solute advection, diffusion, and reaction. Advection is simulated with a semianalytical pore‐scale streamline tracing algorithm, diffusion by random walk is superimposed, while the reaction rate is defined by the flux of particles through the pore‐solid interface. We derive a relationship between the local particle flux and the independently measured batch calcite dissolution rate. We validate our method against a dynamic imaging experiment where a Ketton oolite is imaged during CO2‐saturated brine injection at reservoir conditions. The image‐calculated increases in porosity and permeability are predicted accurately, and the spatial distribution of the dissolution front is correctly replicated. The experiments and simulations are performed at a high flow rate, in the uniform dissolution regime – Pe ≫ 1 and PeDa ≪ 1—thus extending the reaction throughout the sample. Transport is advection dominated, and dissolution is limited to regions with significant inflow of solute. We show that the sample‐averaged reaction rate is 1 order of magnitude lower than that measured in batch reactors. This decrease is the result of restrictions imposed on the flux of solute to the solid surface by the heterogeneous flow field, at the millimeter scale.

No abstract

No abstract

X-ray microtomography is routinely used to image the three-dimensional pore space of sedimentary rocks. Flow and transport properties can then be simulated directly in such images. Advective transport in porous media is frequently simulated using streamlines. We present a novel streamline tracing algorithm based on a substantial development of the most widely used method (the Pollock algorithm) employed for macroscale (Darcy) flow, making it consistent with solutions of the Navier-Stokes equation with no flow at solid boundaries. We use this new algorithm to calculate breakthrough curves and timeof-flight distributions for advection-dominated transport in two three-dimensional images of sedimentary rocks containing up to 10 9 voxels: a sandstone and a carbonate. We show that our approach provides a more accurate description of flow, particularly when only a few image voxels span each pore. Therefore, it is better suited to capture anomalous (non-Fickian) transport behaviour than the standard Pollock method.

No abstract

We study the influence of the pore‐space geometry on sample‐averaged dissolution rates in millimeter‐scale carbonate samples undergoing reaction‐controlled mineral dissolution upon the injection of a CO2‐saturated brine. The representation of the pore space is obtained directly from micro‐CT images with a resolution of a few microns. Simulations are performed with a particle tracking approach on images of three porous rocks of increasing pore‐space complexity: a bead pack, a Ketton oolite, and an Estaillades limestone. Reactive transport is simulated with a hybrid approach that combines a Lagrangian method for transport and reaction with the Eulerian flow field obtained by solving the incompressible Navier‐Stokes equations directly on the voxels of three‐dimensional images. Particle advection is performed with a semianalytical streamline method and diffusion is simulated via a random walk. Mineral dissolution is defined in terms of the particle flux through the pore‐solid interface, which can be related analytically to the batch (intrinsic) reaction rate. The impact of the flow heterogeneity on reactive transport is illustrated in a series of simulations performed at different flow rates. The average dissolution rates depend on both the heterogeneity of the sample and on the flow rate. The most heterogeneous rock may exhibit a decrease of up to two orders of magnitude in the sample‐averaged reaction rates in comparison with the batch rate. Furthermore, we provide new insights for the dissolution regime that would be traditionally characterized as uniform. In most cases, at the pore‐scale, dissolution preferentially enlarges fast‐flow channels which greatly restricts the effective surface available for reaction.

The influence of parameter choice in the Monte Carlo simulation of zero-field-cooled-field-cooled magnetization curves of granular systems is analyzed. The main simulation techniques are summarized and compared, in terms of the determination of macroscopic quantities usually associated with nanoscopic details of the sample.