Reactive flow during water flooding - from pore to core

In this work, we have used real rock materials to investigate reactive flows at the pore scale. Experiments using real materials instead of proxy materials can answer many relevant questions, but are often more challenging to construct. This is also what we experienced when constructing experiments with calcite (chalk proxy) and chalk; the chalk experiments were the most challenging and time-consuming.The main conclusion of this project is that a combination of microfluidic experiments and pore scale simulations is a promising path to more knowledge about the mechanisms responsible for changing the microscopic sweep efficiency. Detailed comparison between microscale experiments using real rocks and numerical simulation reveal faulty assumptions and numerical weaknesses that are corrected before applying simulations to real world situations.

In the numerical one-phase reaction simulations, we see that the detailed pore scale geometry is not important since we are in a reaction-limited regime (small concentration gradients), and the spatial resolution of the pore geometries used in these studies can be reduced (i.e. larger voxels). However, in two-phase simulations we have observed that the microscopic oil-water distribution in the pores are affected both by the surface chemistry and the pore geometry. Hence, the chemical alteration of the reservoir rock is a function of both the oil-water distribution and the history of the chemical environment experienced by the rock.

Method development resulted in working microfluidic chalk devices for averaged one and two- phase flow properties and for mineralogical processes. The intense effort that was put into this development only resulted in proofs of concept. The use of these new devices with chalk is difficult but feasible. We did not obtain in-situ observations of single chalk pores (< 1 μm), only averaged properties over larger volumes, about (100 μm)3. The mechanical weakness of chalk sets a lower limit on the sample thickness. Using the same production technology for other rocks with higher strength should result in thinner samples and observations closer to the pore scale. The conclusion is that our chalk devices a) are promising for small scale (10 micro- meter to 10 cm) sweep efficiency studies, and b) are not the best route for pore scale observations and comparisons with numerical pore scale simulations. The emphasis that was put on chalk microfluidics left no time to study flow-through-crystal devices that would allow 1:1 comparisons with numerical simulations of real chalk on the pore scale.

The highlighted results are:

1. Development of a calcite microfluidic system, from scratch, and the performance of high resolution in-situ measurements of dissolution rates.
2. Verification of the numerical model by comparing to experimental calcite dissolution data in a 1:1 comparison, where all the input parameters were the same.
3. Comparison with experiments prompted the development of a new numerical technique for handling advection of species concentrations in systems with large variations in the pore geometry (such as in real chalk pore geometries). This improved the concentration accuracy by a factor of more than 1000.
4. Implementation of wettability changes in the numerical model, from first principles, by calculating changes in surface energies due to the development of diffuse double layers near the mineral surface. This model was used to study the effect on oil recovery by adding sodium sulfate to the injection water. This did not significantly change the relative permeability curves, but had an observable effect on the local configuration of the residual oil in 2D invasion simulations.