Fluid Distribution in Immiscible Two-Phase Fluid Displacement for CO₂ Storage

The injection of CO2 in deep geological formations such as saline aquifers has been identified as
a promising solution for reducing CO2 emissions. Following injection into a saline aquifer, the
CO2 invades the host rock and pushes out the resident brine by immiscible two-phase fluid
displacement. Successful implementation of CO2 storage, therefore, relies heavily on
understanding and being able to predict multiphase fluid displacement patterns and mechanisms
at the pore scale. CO2 invasion patterns and fluid displacement mechanisms affect CO2 storage
efficiency and depend on factors such as injection rate, mobility ratio, interfacial tension and the
properties of the porous media.
Multi-phase fluid flow phenomena at the pore scale can be observed and studied experimentally
using micromodels, which are two-dimensional (2D) microfluidic devices with pore network
patterns etched/imprinted into materials such as silicon, glass, poly-methyl-methacrylate
(PMMA), poly-dimethyl-siloxane (PDMS) and polyester resin (1). Micromodels can be used as
2D representations of natural porous media to visualize displacement mechanisms of two- or threephase
flow (2). Advances in microfabrication techniques have made it possible to precisely
fabricate a diverse range of pore network patterns (1). Recently, we have demonstrated the use of
rapid laser manufacturing of microfluidic devices from glass substrates (3).
Accordingly, the objective of this work is to obtain a detailed understanding of pore-scale fluid
displacement phenomena combining advanced manufacturing tools, visualization experiments and
numerical modelling. In this study, we use 2D micromodels (Figure 1a) fabricated on borosilicate
glass substrates using a picosecond laser (3). Glass is chosen over materials such as silicon and
polymers, as it has a unique combination of desirable properties, including high transparency,
hardness, thermal stability, surface stability, chemical inertness and resistance to acids (3).
Traditional methods for manufacturing microfluidic devices from glass are complex, multi-step
processes that involve coupling techniques such as photolithography, etching and bonding. As a
result these methods are time-consuming and require an assortment of expensive tools (3). Laser
micromachining and welding is an advantageous fabrication technique as it allows rapid
prototyping of micromodels and only one tool (a picosecond laser) is required for the entire
micromodel fabrication process.
Figure 1 a) Laser-fabricated glass micromodel b) Preferential fluid flow in glass micromodel
A visualization setup is used to visualize immiscible two-phase flow experiments in the fabricated
micromodels (Figure 1b). The set up includes a syringe pump, a camera mounted on a translation
stage and a uniform light source. The tests conducted cover capillary numbers ranging from
9.5×10-6 to 1.9×10-5 to represent typical ratios of the capillary to viscous forces at reservoir
conditions. Increasing the flowrate by a factor of 10 resulted in an increase in the amount of
resident non-wetting phase (air) displaced from the triangular reservoir by the invading wetting
phase (water) from 45% to 70%. More trapping of the non-wetting phase was observed at lower
flowrates.
During the laser-fabrication process, surface defects and heterogeneities were introduced resulting
in “trenches” with different widths on opposite sides of the micromodel. The surface morphology
of the micromodel was found to have a strong influence on the flow patterns observed. The wetting
fluid preferred to invade the wall with the narrower conduit and not the other as shown in Figure
1b (the invading water appears as the lighter grey phase, while air is the darker phase in Figure
1b). Fluid invasion patterns were however reproducible at the flowrates studied.
Acknowledgements
This project has received funding from the European Research Council (ERC) under the European
Union’s Horizon 2020 research and innovation programme (MILEPOST, Grant agreement no.:
695070). This paper reflects only the authors’ view and ERC is not responsible for any use that
may be made of the information it contains.