Unravelling two-phase flow in geological reservoirs for CO₂ storage

Sophie Roman

Mahdi Mansouri

The present work performed at the Nanoµlab of the Institut des Sciences de la Terre d’Orléans (ISTO, France) describes the invasion mechanisms of two immiscible fluids in a porous media model and their consequences on the process of CO₂ storage in geological formations.

The research summary written by Sophie Roman is based on the peer-reviewed article “How interfacial dynamics controls drainage pore-invasion patterns in porous media” from Mahdi Mansouri-Boroujeni, Cyprien Soulaine, Mohamed Azaroual, and Sophie Roman. Their work was recently published in Advances in Water Resources 171 (2023): 104353.

Abstract

This article investigates the behaviour of immiscible two-phase flow through porous media focusing on how pore-scale processes impact macroscopic fluid front behaviour. The researchers explored pore invasion dynamics and front behaviour during drainage through analytical and experimental approaches for various viscosity ratios between both fluids and capillary numbers (the ratio of viscous forces over surface tension). A microfluidic setup with controlled pore-doublet geometry was used to isolate and examine pore-invasion mechanisms. A model based on averaged Navier-Stokes equations captured the interface dynamics and identified different invasion mechanisms correlating with front behaviour under various flow conditions. The results revealed three flow regimes: viscous, capillary, and a crossover regime, with mixed behaviour between the two. Overalls, the study predicts global front behaviour through numerical simulation and experimental results across all regimes.

Introduction | How meniscus dynamics controls invasion patterns in porous media

CO₂ storage in geological formations aims at mitigating carbon emissions. Saline aquifers have the largest identified storage potential. They are porous and permeable reservoir rocks containing brine (i.e., saline fluid) in the spaces between the rock grains (called pores).

The injection of CO₂ in deep saline aquifers results in a multiphase flow system in which complex multiphysics processes are involved. The CO₂ is injected under supercritical form (scCO₂) because the pressure and temperature in the geological formations of interest are above 73 bars, and 31°C (the critical point of CO₂), scCO₂ forms an immiscible phase with the brine, the resident fluid. Today, the different mechanisms leading to the trapping of CO₂ in rock reservoirs are well-identified. However, the effective CO₂ storage capacity worldwide is uncertain because current assessment often neglects the influence of complex coupled processes occurring during and after injection. In particular, during the displacement of two immiscible fluids, front instabilities can develop that question the validity of the classical models used to predict immiscible displacements at the reservoir scale.

There is a need for experimental studies at the scale of the pores, usually the micrometre scale, to clarify the link between the overall fluid/fluid front behaviour in porous media and pore-scale phenomena. Pore-scale phenomena that influence the fluid front behaviour are, for example, sudden pressure and velocity fluctuation of the interfaces or Haines jumps, which are abrupt movements of a fluid-fluid interface in two-phase flow after it was pinned at a pore constriction.

In this work, we use microfluidics to design simplified porous media geometry to study pore invasion mechanisms in very well-controlled conditions.

Aims

To experimentally and numerically investigate:

• pore-scale dynamics,
• key pore-invasion mechanisms,
• their contribution to defining the fluid/fluid front behaviour in porous media.

Experiment setup | Mimicking geological reservoir rocks

Our microfluidic device geometry comprises two sinusoidal channels connected at the inlet and outlet (i.e., pore-doublet geometry), they are representing successive pores and throats found in geological reservoir rocks (Fig. 1) made in PDMS (polydimethylsiloxane, a polymer widely used for microfabrication). A fluid-fluid displacement process through the PDMS microfluidic device is monitored using a high-resolution camera connected to a microscope for the two-phase flow observations.

A pressure difference in the 0 – 20 kPa range between the inlet and outlet of the microfluidic device is applied using a highly accurate OB1 – 4 Channel Microfluidic Flow controller.

Various fluid pairs, PDMS coating, and pressure differences are considered to vary the viscosity of the fluids, the contact angle and surface tension between the fluids and the PDMS, and the flow regime. These parameters are chosen to be representative of displacement conditions for CO₂ sequestration in subsurface environments. Fluid properties are presented in Table 1.

For all experiments, the fluid initially saturating the microfluidic device is the wetting fluid (i.e., the fluid that has an affinity with the PDMS and thus the ability to maintain contact with its surface), and the injected fluid is the non-wetting one (i.e., the fluid that does not wet the surface of PDMS).

Image processing

Sequences of images of the displacement of one fluid by the other are recorded. Image processing techniques are applied using a self-developed MATLAB^® code to separate fluid phases during two-phase flow and to track the fluid/fluid interfaces. We are also using micro-PIV (Particle Image Velocimetry) technique to measure the velocity field of the wetting fluid using the PIVlab tool.

Key findings | Two-phase flow regimes and identification of the crossover regime

It is known that  M<1 for low values of Ca, the displacement process is governed by capillary forces; in this case, the pore size distribution governs the invasion pattern called capillary fingering. For greater values of Ca, viscous fingering appears because of the lower viscosity of the invading fluid and of the influence of viscous forces. For high values of Ca and M the fluid front is stable.

In this work, we have characterized these different flow regimes for highly controlled flow conditions. Importantly we identified a new flow regime called the crossover flow regime. In Figure 2, the classification of the three flow regimes based on the values of Ca and M is shown, as well as the main characteristic of the two-phase flow patterns.

For the viscous flow regime, viscous forces control the displacement that occurs in both channels with a stable front; it does not stop at pore throats. This flow regime is also characterized by residual wetting fluid remaining at the curvature of the pores (the interface does not expand laterally over the whole pore), resulting in lower displacement efficiency of the wetting fluid.

When capillary forces are dominant – i.e., for the capillary flow regime – only one of the channels is invaded, the one for which the interface reaches the invasion threshold pressure first. Haines jumps are the dominant pore-invasion mechanisms in this case – i.e., the interface gets pinned at a pore constriction before invading the pore abruptly. During a jump, the interface can reach a velocity up to 30 times higher than the mean velocity of the flow (figure 3, b). In addition, micro-PIV measurements in the channel that is not being invaded (channel 2 in Figure 3) showed a backward movement of the wetting fluid linked to jumps of the interface in channel 1.

These flow instabilities can have consequences on mixing, dissipation or reaction mechanisms that are not addressed yet. The crossover flow regime newly identified presents a mixed behaviour between the capillary and viscous two-phase flow regimes. Indeed, in this case, both channels are invaded, but not at the same rate. Pore-by-pore invasion is the main process, as observed for the capillary flow regime. After each jump of the interface, the flow is highly disturbed, and we observed cooperative pore filling, meaning that invasion in one channel might provoke (or delay) invasion in the adjacent channel, thus causing different pore-invasion dynamics from one pore to another. The displacement efficiency is higher for the crossover flow regime.

Invasion patterns for the different regimes can be seen in the videos of the experiments.

The experiments are also confronted with a numerical model based on section-averaged Navier–Stokes equations that are able to predict the global two-phase flow and the interface dynamics.

Conclusions | A new pore-doublet microfluidic setup to study two-phase flow through porous media

In this work, we performed a numerical and experimental investigation of front behaviour and pore invasions dynamics mimicking the injection of CO₂ in porous reservoir rocks for long-term storage of this greenhouse gas.

We use a microfluidic setup that includes a geometry made of successive pores and throats to isolate and explore pore-invasion mechanisms. This setup will be used now to study residual trapping in the case when the invading fluid breaks up at a constriction and form disconnected bubbles or ganglia, a phenomenon called snap-off.

A similar setup can also be used to study the trapping of CO₂ by dissolution in the aqueous phase. Finally, the numerical model is continuously improved and confronted with experiments to better predicts two-phase flow processes in porous media.