Investigation of relative flow characteristics of brine-saturated reservoir formation: A numerical study of the Hawkesbury formation

Highlights

• Laboratory- and field-scale numerical models were developed.

• Buckely-Leverett flow theory was applied using pre-defined partial differential equations.

• The developed laboratory-scale model simulates the experimental results reasonably well.

• The effect of CO₂ phase change on the CO₂ and brine saturation profiles was evaluated.

• The long-term interaction of CO₂ causes Hawkesbury formation’s pore structure to significantly change.

Abstract

Precise knowledge of the relative flow behaviour of CO₂ and brine during CO₂ sequestration in deep saline aquifers and its impact on the sequestration process is required to ensure the safety and efficiency of sequestration projects. This numerical study therefore aims to identify the interaction-induced relative flow behaviour of reservoir rock and its influence on the hydro-mechanical and geochemical phenomena in deep saline aquifers. COMSOL Multiphysics numerical simulator was used to develop a laboratory-scale relative flow model to simulate the CO₂ movement and brine drainage processes in brine-saturated Hawkesbury sandstone samples and finally, was extended it into a field-scale numerical model which can simulate the hydro-mechanical, mineralogical and geochemical behaviours of deep saline aquifers under CO₂ sequestration conditions. A 2-D axisymmetric pore-elastic model was developed using the pore-elastic module available in COMSOL and Buckely-Leverett flow theory was applied to the model using pre-defined partial differential equations. The proposed laboratory-scale model was first validated using experimental permeability data conducted under triaxial drained conditions and the model was then extended to predict relative flow characteristics, such as brine and CO₂ saturation and CO₂ pressure distribution along the sample length under different injection pressures, including both sub- and super-critical conditions and finally, real CO₂ sequestration processes.

According to the results, the developed laboratory-scale model simulates the experimental results reasonably well with less than 10% relative error. The numerical results also reveal that there is a considerable effect of CO₂ phase change on the final distribution of the CO₂ and brine saturation profiles. In addition, the CO₂ pressure distribution along the sample length shows a non-linear relationship between the CO₂ pressure and sample length. According to the results of the field-scale model, the long-term interaction of CO₂ causes Hawkesbury formation's pore structure to significantly change.

Introduction

Carbon dioxide (CO₂) sequestration in deep saline aquifers is a promising solution for the mitigation of anthropogenic CO₂ emissions (Perera et al., 2011, Rathnaweera et al., 2014). To date, growing awareness about global warming has made the effects of CO₂ sequestration on reservoir rock hydro-mechanical behaviour a topic of great research interest. As a result, field-scale numerical studies have been reported on the hydro-mechanical behaviour of deep saline reservoir rock upon exposure to CO₂ (Andre et al., 2007, Burton et al., 2009, Rathnaweera et al., 2015a, Rathnaweera et al., 2015b, Rathnaweera et al., 2016a). However, little attention has been given to how CO₂ injection is affected under laboratory-scale conditions. Therefore, the present study aims to address this gap by developing a laboratory-scale relative flow model capable of simulating reservoir rock flow behaviour in a CO₂ sequestration environment. To develop the simulation, the present study used the COMSOL Multiphysics simulator and pre-defined partial differential equations were employed to characterise the relative flow behaviour of the Hawkesbury formation under laboratory-scale conditions.

In relation to CO₂ sequestration in deep saline aquifers, a thorough understanding of the two-phase flow of CO₂ and aquifer brine through porous reservoir rock is important for both analytical and numerical simulations. Almost 160 years have passed since Darcy (1856) introduced his first theory, which is considered to be the starting point of the development of the scientific quantitative theory of flow through porous media. Since then, a considerable number of studies have been carried out to expand the basic flow theory introduced by Darcy to capture complex flow phenomena, including multi-phase flows, deformation of solid matrix, non-linear flow regimes, and other transport processes (mass, heat, etc.). In addition, the theory of two-phase flow through porous media has undergone extensive development in the last three decades, due to its capability of simulating many engineering applications (Dagan, 1986). Of these applications, the interaction between aqueous brine and injected CO_2, such as the dissolution of CO₂ in brine and the adsorption of CO₂ into the rock matrix, makes deep saline sequestration flow behaviour even more complex. Therefore, the identification of appropriate two-phase flow models applicable to deep saline sequestration is important for numerical and experimental investigations. To date, a considerable number of two-phase models have been proposed (Corey, 1954, Brooks and Corey, 1964, Mualem, 1976, Muller, 2011, Mitiku and Bauer, 2013, Oostrom et al., 2016), and some of these models are summarised in Table 1.

Generally, the injection of CO₂ into deep saline aquifers can be characterised by mass conservation equations for the three components, water, salt (NaCl) and CO₂. Ignoring non-isothermal effects, chemical reactions and the effects of mechanical stress, Pruess et al. (1999) introduced a three-phase system, water-salt-CO₂, following the mass conservation equation:

$$\frac{d}{dt}\int M^{K}d{V}_n=\int F^K.nd{\Gamma }_n+\int q^{K}d{V}_n$$

In addition, some initial simulation studies of CO₂ injection into deep saline aquifers have been conducted with existing petroleum reservoir simulators (van der Meer, 1992, Korbol and Kaddour, 1995, Weir et al., 1995, Law, 1995, Law and Bachu, 1996, Yang and Long, 2011, Dongyan et al., 2015, Jamshidian et al., 2015). However, as mentioned earlier, a small number of studies have been undertaken on the application of these simulators to problems with small rock domains. Tang et al. (2002) conducted a numerical study to investigate the coupled behaviour of flow, stress and damage in rock failure. They proposed a flow-stress-damage (FSD) coupling model for heterogeneous rocks that takes into account the growth of existing fractures and the formation of new fractures. The simulation results of this study revealed that the nature of fluid flow in rocks varies from material to material, and strongly depends upon the heterogeneity of the rock. Kueper and Frind (1991) performed two-phase flow simulation to characterise the migration properties of a dense, non-aqueous phase liquid through heterogeneous porous rock using input parameters from laboratory measurements of capillary pressure-saturation curves. The results of this study showed the migration characteristics of non-wetting liquid strongly depend on the capillary properties of the porous medium and are also influenced by the fluid's physical properties. Mercer and Cohen (1990) performed a numerical study on non-aqueous phase liquids and found that these non-aqueous liquids behave differently from dissolved solutes in the sub-surface. Watanabe et al. (2008) developed a numerical model to simulate the fracture permeability of artificially-developed granite. The model was basically used to characterise the fluid flow behaviour through rock fractures under confining pressures. According to these researchers, numerically-obtained fracture permeability is important for understanding sub-surface flow, especially the fluid flow behaviour in aperture structures under different confining pressures. Although there are a number of studies on rock fluid interaction under laboratory conditions, a small number of laboratory-scale numerical studies have set out to identify the CO₂ sequestration effects, but their research is limited. For example, Andre et al. (2014) presented a coupled numerical approach to represent the drying mechanisms of brine-saturated reservoir rocks under laboratory conditions. According to their results, the capillary properties of reservoir rock prevent the sudden evaporation of irreducible water. Moreover, they showed that the capillary force effect can be minimised by maintaining a sufficient CO₂ injection rate. In addition, the results also revealed that brine salinity in the aquifer has a major impact on reservoir rock porosity and permeability.

A review of various two-phase relative permeability-saturation-capillary pressure models by Oostrom et al. (2016) revealed the applicability of those models to the simulation of deep saline sequestration. Six published numerical multiphase concepts, including van Genuchten-Mualem, Brooks-Corey-Burdine, van Genuchten-Corey, van Genuchten-hybrid Mualem-Corey, van Genuchten-endpoint power law and Brooks-Corey-variable Corey were analysed using data from four well-characterised sandstones (Berea, Paaratte, Tuscaloosa, Mt. Simon). According to their results, the plume extension and saturation distribution of each sandstone are highly dependent on the nature of the relative permeability-saturation-capillary pressure model used. Moreover, the van Genuchten-Mualem model over-estimated the CO₂ relative permeability, resulting in considerably large plume distributions during and after injection. In addition, the van Genuchten-Corey model predicted the smallest plume distribution due to the under-estimation of the aqueous phase relative permeability. Of the six models, the van Genuchten-hybrid Mualem-Corey and Buckley-Leverett models were found to be the best fit to the experimental results. Therefore, in order to have accurate flow estimations, it is necessary to select a reliable relative permeability relation when modelling deep saline sequestration process.

Therefore, the present study selected the Buckley-Leverett flow concept to develop a laboratory-scale model using the COMSOL Multiphysics simulator and the model was then validated using experimental data. The relative flow behaviour of brine-saturated Hawkesbury sandstone under triaxial laboratory conditions was simulated by injecting CO₂ at a constant pressure at one end of the sample and the drainage characteristics of both CO₂ and brine were evaluated. In addition, the influence of CO₂ injection pressure, including the phase behaviour of CO₂ on the relative flow characteristics of reservoir rock, was also investigated. The sample description and model concept and the governing equations developed are presented in the next sections.