Investigations into CO₂ immiscible displacements from pore scale to core scale

Abstract

The flow of immiscible displacement in porous media is of great importance in many applications of subsurface transports, such as geological sequestration of CO₂ in saline aquifers and depleted oil/gas fields, enhanced oil recovery (EOR) and removal of nonaqueous phase liquids (NAPLs) from aquifer sediments. In this work, the investigation of immiscible displacement focuses on drainage process, in which wetting fluid residing in porous media is displaced by a non-wetting fluid. Fingering led by unstable displacement is one of the major reasons limiting the efficiency in drainage process. For example, in CO₂ sequestration, fingering of low-viscosity gaseous or supercritical CO₂ limits their access to reservoir rocks and thus only a fraction of reservoir is occupied by injected CO₂. In enhanced oil recovery by CO₂ or water, less than half of original oil in place is recovered due to viscous fingers. The immiscible displacement process in CO₂ injection into the geological formations and reservoirs involves complex cross-scale flow from pore scale to core scale and eventually reservoir scale. Pore-scale flow can provide the most fundamental information of microscopic factors that have significant effects on macroscopic process, which cannot be captured by current continuum-scale models. Simulating a reservoir-scale flow with pore-scale modelling is not possible as it requires enormous computation. Instead, core-scale model is a brilliant option to characterize the flow dynamics of the reservoir to obtain upscaled flow functions (e.g., relative permeability and capillary pressure) for reservoir-scale models. A solid understanding of immiscible displacement from both pore-scale and core-scale perspectives is critically important in evaluating the impacts and efficiencies of displacement processes and providing accurate flow information for the reservoir model.

In this PhD project, immiscible displacements were investigated in the core sample and pore-scale network to evaluate the effect of key factors (namely fluid pressure, temperature, CO₂ phase and injection flowrate) on crucial flow parameters, such as differential pressure, flow patterns, phase saturation, relative permeability and displacement efficiency. The study sheds more light on the impact of capillary and viscous forces on multiphase flow characteristics and explores the conditions where capillary or viscous forces dominate the flow. In addition, immiscible displacements in homogenous and physical rock networks are compared by numerical simulation.

CO₂-water core flooding experiments were carried out in a sandstone under different subsurface temperature and pressure conditions by considering the effects of CO₂ phase, pressure and CO₂ injection rate. The pressure fluctuation behaviours, capillary displacement pressure, relative permeability and water recovery are investigated for gas CO₂-water, supercritical CO₂-water and liquid CO₂-water displacements. Pressure fluctuations during the displacements are analysed by wavelet decomposition method. The pressure fluctuations are affected by the CO₂ phase but is almost independent of injection rate. The capillary displacement pressure is quantified by the pressure jump occurring at the beginning of CO₂ invasion into the core sample, which agrees well with the value estimated based on Kozeny model. The relative permeability is calculated by the JBN method, and it is found that liquid CO₂-water displacement has a higher relative permeability. Uncertainty of the relative permeability caused by the assumption of incompressible fluids is quantified.

Flow instability in immiscible displacements is predominately affected by the interactions between capillary and viscous forces, which can determine the displacement efficiency and CO₂ storage capacity. CO₂-water and CO₂-oil core flooding experiments with various viscosity ratios and capillary numbers were implemented to understand the interplay of capillary and viscous effects during the displacements by analysis of the pressure behaviours, and the experimental results are further demonstrated by macroscopic capillary number and the Log Nca-Log M phase diagram. In viscous-dominated displacement, differential pressure evidently depends on the injection rate and the pressure decline curve is fitted by a power function. The exponent of the function is found to be significantly larger at the crossover between capillary-dominated and viscous-dominated regions. In capillary-dominated displacement, the pressure profile is characterized by a pressure jump at the beginning and intermittent fluctuations during displacement. Further analysis by wavelet decomposition indicates a transition point existing in standard deviation of pressure fluctuations when the displacement is transformed from capillary-dominated to viscous-dominated. The experiment results agree well with the macroscopic capillary number, which characterizes the interaction between capillary and viscous forces at a critical value of 𝑁ᵐᵃᶜʳᵒ꜀ₐ~1.

For further understanding of the effects of capillary and viscous forces on the immiscible displacements occurring on the scale of individual pores and of how these processes determine the invasion patterns, CO₂-water and water-oil immiscible displacements were conducted in the pore-scale network with physical rock structures. Viscosity ratio and capillary number for the displacements were varied by employing different fluid pairs and injection flowrates. The pressure behaviour, flow pattern, invading phase saturation, trapped oil cluster size and macroscopic capillary number were investigated for displacements transforming from viscous-dominated to capillary-controlled. The behaviour of differential pressure is similar to that in core flooding experiments. A transition point exists on the relationship between standard deviation of pressure fluctuations and capillary number when displacement varies from viscous-dominated to capillary-controlled, which is also in accord with the results from core flooding experiments except for the value of the transition point. The non-wetting phase saturation shows great dependence on the capillary number. When capillary force dominates the flow, the non-wetting phase saturation hardly increases with increasing the amount of injection after breakthrough while it increases greatly when viscous force is the dominating force. With increasing capillary number, the trapped oil cluster size become smaller while the interfacial length between two fluids gets larger. The calculated value of macroscopic capillary number accurately predicts the viscous-dominated flow with 𝑁ᵐᵃᶜʳᵒ꜀ₐ~1, but the value maybe underestimated for capillary-controlled flow due to the underestimation of trapped oil cluster size.

Another key issue considered in the project is the effect of pore structure on the immiscible displacement. A direct numerical simulation method is employed by COMSOL Multiphysics 5.6 software to simulate the immiscible displacement in two microfluidic networks, one with a homogeneous structure and the other with a physical rock structure. The numerical model is first validated by immiscible displacement in a single capillary. The simulated results are in good agreement with the theoretical solution calculated based on Darcy-Weisbach equation. Then, flow dynamics, such as flow patterns, differential pressure and phase saturation were investigated in these two networks when the displacement is transformed from capillary-controlled to viscous-dominated. Early breakthrough resulted by viscous fingering is observed in both networks and more obviously in physical rock network. Capillary fingering is not clearly detected, properly due to the limitation of the network scale. In both networks, the flow patterns become compact at equilibrium compared with that at breakthrough in displacements with high injection flowrate (or capillary number) while they remain almost unchanged for displacements with low injection flowrate. The behaviours of differential pressure are similar in these two networks other than pressure achieving equilibrium more quickly in homogeneous network under the same injection flowrate condition. The pressure behaviours are in accord with the results from pore-scale experiments except obvious inertia effect at the beginning of displacements. Transition points are discovered on the relationships between pressure fluctuations and capillary number in both networks and are nearly the same, which are around Log Nca = -3.2. The dynamic, breakthrough-time, equilibrium-time saturation of non-wetting phase is evaluated. It is discovered that the saturation is larger in homogeneous network at high injection flowrates while it is about the same at low injection flowrates. The trend of saturation in physical rock network is in good agreement with pore-scale experimental result.