Search

Reactive transport in porous media is ubiquitous in nature and industry, with applications ranging from geothermal energy, enhanced oil recovery, soil contamination, to carbon storage, etc. A long-lasting issue in this field is reactive transport parametrizations and linking the experimental data to continuum-scale models. For this purpose, pore-scale modelling can be used as a valuable tool for upscaling purposes to decrease the existing uncertainties. In this thesis, we performed reactive transport modelling across different scales, namely pore-scale and continuum-scale for different applications of geothermal energy, low salinity waterflooding and CO2 convective mixing during carbon storage in saline aquifers. Moreover, we have experimentally studied solute transport upscaling from pore- to REV-scale. Using coupling PHREEQC geochemical module and pore-network modelling under single-phase condition, we showed how averaging over the domain to predict mineral reaction rates can lead to erroneous results, in some cases even the dominant reaction path (dissolution/precipitation) could not be determined accurately. We have also coupled oil-brine and rock-brine (sandstone) surface complexation models with multicomponent solute transport under the steady-state two-phase quasi-static condition in pore network to study the effect of hydrodynamic condition on low salinity effect. We showed that under two-phase condition the advancing fluid is divided into two flowing and stagnant zones. The contribution of stagnant zones at different injection fluid saturation can cause a delay in the low salinity effect in the continuum-scale as the low salinity water penetration in these zones is diffusion-controlled. Moreover, we experimentally showed that upscaling the solute transport from pore-scale to REV-scale depends on the process (i.e., solute loading or unloading scenarios). The unloading process usually results in a higher dispersion coefficient, while the difference diminishes in high injection flow rates. Using continuum-scale reactive transport modelling, we have studied the role of geochemistry on CO2 convective mixing during carbon storage in saline aquifers. Firstly, we investigated the signature of geochemical interactions on the process in sandstone aquifers. We coupled a representative set of fluid-fluid and sandstone-fluid reactions with an in-house convection-diffusion transport model and showed that geochemical interplay can result in earlier onset of convection and higher total stored carbon due to mineralization mechanism. We highlighted how CO2 storage may change the mineralogy of the sandstone aquifers. Next, we looked into the dynamics of the convective mixing and how the hydrodynamic dispersion and geochemistry can change it over a wide range of Rayleigh numbers. We showed that ignoring the geochemical interactions in carbonate aquifers can lead to underestimation of stored CO2 into the simulation domain. Lastly, we revisited the convective mixing scaling relations and showed how they can be modified to take into account the effects of hydrodynamic dispersion, permeability anisotropy and geochemistry. The results of these studies provide a broad perspective on how geochemistry can contribute to a wide range of environmental and chemical engineering applications. We have shown how pore-scale reactive transport simulations can work as a bridge to upscale the experimental data related to reaction rates, to be fed into the continuum-scale models. We have also investigated the crucial importance of geochemistry on solutal convective mixing in the porous media, which provided important implications for field-scale CO2 storage.

Low salinity water-flooding is a relatively new enhanced oil recovery method, proven to accelerate oil production at core and field scales, through injection of low salinity water in sandstone reservoirs. The most notable mechanism under low salinity water-flooding is wettability alteration from more oil-wet to more water-wet, known as low salinity effect (LSE). Currently, there is a limited understanding of how wettability alteration, which is a fluid-solid interface, can effect continuum scale processes. This thesis is an attempt to bridge this gap in the literature through the development of the pore-scale model. Direct numerical simulations (DNS) at the pore scale is performed using computational fluid dynamics (CFD) approach in which the Navier-Stokes's momentum equation is coupled with the volume of fluid method to represent the two-phase flow. The model is adapted to include mixing and transport of low salinity water with wettability alteration. Numerical results show that the initial water topology does not fully saturate with low salinity water under two-phase flow conditions. This is due to the presence of stagnant/dead-end regions in the water topology. Due to slow mixing of low and high salinity water in stagnant regions, these areas remain oil-wet whereas flowing regions (advection controlled) of the water topology experience wettability change from oil-wet to water-wet. Under these local mixed-wet conditions, a unique flow mechanism of water is observed, referred to as 'pull-push'. Oil internally redistributes as a result of pull-push mechanism and reduces the efficiency of low salinity waterflooding. Our results also show that the pore size distribution has an influence on the oil recovery under the tertiary mode of low salinity waterflooding and more heterogeneous pore size distribution can lead to higher oil recovery.

Polymers used in enhanced oil recovery (EOR) help to control the mobility ratio between oil and aqueous phases and as a result, polymer flooding improves sweep efficiency in reservoirs. However, the conventional wisdom is that polymer flooding does not have considerable effect on pore-level displacement because pressure forces would not be enough to overcome trapping caused by capillary forces. Recently, both coreflood experiments and field data suggest that injecting viscoelastic polymers, such as hydrolyzed polyacrylamide (HPAM), can result in lower residual oil saturation. The hypothesis is that the polymer elasticity provides several pore-level mechanisms for oil mobilization that are generally not significant for purely-viscous fluids. Both experiments and modeling need to be performed to investigate the effect of polymer elasticity on residual oil saturation. Pore-scale modeling and micro-fluidic experiments can be used to investigate pore-level physics, and then used to upscale to the macro-scale. The objective of this work is to understand the effect of polymer elasticity on apparent viscosity and residual oil saturation in porous media. Single- and multi-phase pore-level computational fluid dynamics (CFD) modeling for viscoelastic polymer flow is performed to investigate the dominant mechanisms at the pore level to mobilize trapped oil. Several interesting results are found from the CFD results. First, the elasticity of the polymer results in an increase in normal stress at the pore-level; therefore, the normal stresses exerted on a static oil droplet are significant and not negligible as for a purely-viscous fluid. The CFD results show that viscoelastic fluid exerts additional forces on the oil-phase which may help mobilize trapped oil out of the porous medium. Second, due to the elasticity of polymer, the viscoelastic polymer has some level of pulling effect; while passing above a dead-end pore it can pull out the trapped oil phase and then mobilize it. However, both CFD modeling and micro-fluidic experiments show the pulling-effect is not likely the main mechanism to reduce oil saturation at pore-level. Third, dynamic CFD simulations show less deformation of the oil phase while viscoelastic polymer is displacing fluid compared to purely viscous fluid. It may justify the hypothesis that polymer elasticity resists against snap-off mechanism. As a result, when viscoelastic polymer displaces the oil ganglia, the oil phase does not snap off, and the oil phase remains connected, and therefore easier to move in porous media compared to disconnected oil. For single phase flow, a closed-form flow equation has been developed based on CFD modeling in converging/diverging ducts representative of pore throats. The pore-level equations were substituted into a pore-network model and validated against experimental data. Good agreement is observed. This study reveals important findings about the effect of polymer elasticity to reduce the residual oil saturation; however, more experiments and simulations are recommended to fully-understand the mobilization mechanisms and take advantage of them to optimize the polymer-flooding process in the field.

Abstract: Enhanced oil recovery (EOR) from naturally fractured reservoirs (NFR's) involves the injection of external agents into the fracture networks to extract the resident crude oil in the matrix. Characteristically, these agents displace the oil by diffusion (if miscible) or spontaneous imbibition (if immiscible). The present work performs two-dimensional simulations of these processes at the pore-scale and extends the results to core-scale by using the Lattice Boltzmann method (LBM). Simulations of miscible recovery of kerosene and light mineral oil by injection of pentane are performed to observe the displacement profiles. The LBM model is then extended to simulate the co-current and counter-current spontaneous imbibition of water in a kerosene saturated water-wet sand pack. The oil-water interface development is analyzed at the pore-scale, and the saturation profiles are compared with experiments from literature. Two-phase relative permeabilities of the displacement processes are determined, and their dependence on various physical parameters is studied.

The increased production of nanomaterials in recent years has been unprecedented. Given their potential toxicity, understanding the mechanisms controlling the transport of nanoparticles in the subsurface is important. In this study, a technique was developed for using a Laser Scanning Cytometer (LSC) to visualize and quantify the stable attachment of nano-scale particles. Experiments using three different size particles, 510 nm, 210 nm and 57 nm, in conjunction with a flow cell system containing saturated glass beads under varied injection duration, solution chemistry, Darcy velocity and solids content were performed. A technique for using the LSC data to develop spatial distributions of attached particles was developed. The ability to provide quantifiable data and a spatial distribution of nanoparticle attachment at the pore-scale is unique and provides direct insight into the fundamental mechanisms governing nanoparticle transport. The experimental results show attachment decreases with decreasing particle size. The increase in injection duration for the 510 nm particles indicates a likely maximum retention capacity (Smax). Blocking effects are observed for the 57 nm particles in which attached particles block the available attachment sites and slow the rate of attachment. Secondary minimum attachment plays a minor role for the attachment of both the 510 nm and 57 nm particles and is independent of particle size. Only about 10% of the attachment is attributed to secondary minimum attachment. Change of Darcy velocity has no profound influence on the attachment of the 57 nm particles indicating diffusion-dominated attachment. Diffusion control is further confirmed by the spatial distributions of attached 57 nm particles showing attaching on downstream glass bead areas. Investigations of initial solids content reveal the importance of particle (aqueous) - particle (attached) interactions. For the 510 nm and 210 nm particles, there exists a critical initial solids content above which the attachment decreases with increasing initial solids content. This trend does not occur for the 57 nm particles which exhibit increasing attachment with increasing solids content due to much weaker repulsive interaction energy. Advisor: Yusong Li