Vladimir Alvarado

Pore Scale and Macroscopic Displacement Mechanisms in Emulsion Flooding

The flow properties of complex fluids through porous media give rise to multiphase flow displacement mechanisms that operate at different scales, from pore-level to Darcy scale. Experiments have shown that injection of oil-in-water emulsions can be used as an effective enhanced-oil recovery (EOR) method, leading to substantial increase in the volume of oil recovered. Pore-scale flow visualization as well as core flooding results available in the literature have demonstrated that the enhanced recovery factor is regulated by the capillary number of the flow. However, the mechanisms by which additional oil is displaced during emulsion injection are still not clear. In this work, we carried out two different experiments to evaluate the effect of emulsion flooding both at pore and macro scales. Visualization of the flow through sand packed between transparent plexiglass parallel plates shows that emulsion flooding improves the pore-level displacement efficiency, leading to lower residual oil saturation. Oil recovery results during emulsion flooding in tertiary mode (after waterflooding) in parallel sandstone cores with very different absolute permeability values prove that emulsion flooding also leads to enhancement of conformance or volumetric sweep efficiency. Combined, the results presented here show that injection of emulsion offers multiscale mechanisms resulting from capillary-driven mobility control.

Effects of pore-level reaction on dispersion in porous media

The Darcy-level consequences of the transport of reactive tracers is analyzed by detailed pore-level modeling, based on a network model. Moments of the residence-time distribution of the conservative process (reversible reactions) are useful for investigation of the spreading of tracers, even when complete evaluation of the residence-time distribution is not available. We carry out simulations to show how reaction terms have to be included in the convection-dispersion equation to correctly predict the Darcy-level effects of reversible reactions at the pore-level. In the case of spatially homogeneous rate constants, the value of the dispersion coefficient corresponds to that of a nonreactive tracer. Spatial heterogeneities of the rate constants give rise to a dispersion coefficient that depends on the strength of the disorder in the reaction rates and the dispersion coefficient depends nonlinearly on the mean flow velocity. The effects of reaction can be summarized in terms of two dimensionless groups, the Damkohler number Da and the variance of the rate constant distribution. For Da⪢1, a macroscopic convection-dispersion-reaction equation offers a valid description of transport, even for spatially heterogeneous distributions of rate constants. The limit Da → 0 represents a breakdown of the macroscopic equation, though the relative error in the low-order moments of the residence-time distribution is less than 20% for 0.1 <Da < 1. A binary distribution of the rate constant at its percolation threshold yields the maximum value of the dispersion coefficient. Plots of the Darcy-level Peclet number, ULD∥, with respect to the length of the system, L, reaches an asymptotic value at a length much larger than the typical pore length. This indicates the presence of a correlation length much larger than the pore length.

Direct Current Electrorheological Stability Determination of Water-in-Crude Oil Emulsions†

Emulsion stability is a fundamental determination for separation technologies. We use the critical electric field (CEF) and viscosity changes in DC electrorheological (ER) experiments in dynamic mode to establish the level of stability of water-in-crude oil emulsions previously studied through bottle tests. The CEF value corresponds to the value of electric field at which the current reaches 95% or larger of the plateau value. Our results show that CEF can be obtained through current measurements and viscosity drops resulting from emulsion structure breakdown, although viscosity changes are not always a good proxy of stability. This implies that electrorheology cannot be uncritically used for static stability determination of the CEF value. Emulsion structure breakdown is explored through rheological characterization before and after voltage sweeps have been performed. When the electric field applied is below the CEF value, the storage and loss moduli response, as well as viscosity, as functions of frequency are recovered. However, when the electric field is greater than the CEF value, the emulsion structure breaks down irreversibly.

CO2 saturation, distribution and seismic response in two-dimensional permeability model.

Carbon dioxide capture and storage (CCS) has been actively researched as a strategy to mitigate CO(2) emissions into the atmosphere. The three components in CCS are monitoring, verification, and accounting (MVA). Seismic monitoring technologies can meet the requirements of MVA, but they require a quantitative relationships between multiphase saturation distributions and wave propagation elastic properties. One of the main obstacles for quantitative MVA activities arises from the nature of the saturation distribution, typically classified anywhere from homogeneous to patchy. The emerging saturation distribution, in turn, regulates the relationship between compressional velocity and saturation. In this work, we carry out multiphase flow simulations in a 2-D aquifer model with a log-normal absolute permeability distribution and a capillary pressure function parametrized by permeability. The heterogeneity level is tuned by assigning the value of the Dykstra-Parson (DP) coefficient, which sets the variance of the log-normal horizontal permeability distribution in the entire domain. Vertical permeability is a 10th of the horizontal value in each gridcell. We show that despite apparent differences in saturation distribution among different realizations, CO(2) trapping and the V(p)-S(w) Rock Physics relationship are mostly functions of the DP coefficient. When the results are compared with the well accepted limits, Gassmann-Wood (homogeneous) (A Text Book of Sound; G. Bell and Suns LTD: London, 1941) and Gassmann-Hill (patchy) models, the V(p)-S(w) relationship never reaches the upper bound, that is, patchy model curve, even at the highest heterogeneity level in the model.

Dispersion of paramagnetic tracers in bead packs by T1, mapping : Experiments and simulations

NMR imaging was used to study dispersion in 6 mm bead pack. T1 maps were employed to measure the rate of axial spreading of paramagnetic tracers (GdCl3) inside the bead pack in the range of flow rate from 0.015 mL/s to 0.175 mL/s. From the T1 maps, tracer concentration profiles were obtained, which yielded dimensionless axial dispersion coefficient and mean transit time. Spatial variations in the dispersion coefficient were observed at flow rates above 0.08 mL/s. We hypothesized that the observed spatial oscillations in the dispersion coefficient arise from the spatial variations of the velocity distribution. To validate this mechanism we showed by simulation that similar dispersion coefficient variation occur in a layered network.

Dynamic Network Model of Mobility Control in Emulsion Flow Through Porous Media

Modeling the flow of emulsion in porous media is extremely challenging due to the complex nature of the associated flows and multiscale phenomena. At the pore scale, the dispersed phase size can be of the same order of magnitude of the pore length scale and therefore effective viscosity models do not apply. A physically meaningful macroscopic flow model must incorporate the transport of the dispersed phase through the porous material and the changes on flow resistance due to drop deformation as it flows through pore throats. In this work, we present a dynamic capillary network model that uses experimentally determined pore-level constitutive relationships between flow rate and pressure drop in constricted capillaries to obtain representative transient macroscopic flow behavior emerging from microscopic emulsion flow at the pore level. A parametric analysis is conducted to study the effect of dispersed phase droplet size and capillary number on the flow response to both emulsion and alternating water/emulsion flooding in porous media. The results clearly show that emulsion flooding changes the continuous-phase mobility and consequently flow paths through the porous media, and how the intensity of mobility control can be tuned by the emulsion characteristics.

Flow of Oil‐Water Emulsion Through Constricted Capillary Tubes

The flow of oil‐in‐water emulsions through a constricted capillary tube was analyzed by experiments and theory. The experiments consisted of flow visualization and pressure drop measurements of the flow. A number of different emulsions were prepared using synthetic oils and deionized water. The average drop size varied from smaller to larger than the neck radius. Fluid mobility, defined as flow rate over pressure drop, was used to quantify the magnitude of the pore‐blocking caused by drops larger than the constriction radius. The effect of the interfacial tension and viscosity ratio between the two phases on the changes of the local mobility was determined by solving the free surface flow of an infinite oil drop immersed in water flowing through a constricted capillary tube by Finite Element Method.

Snap-off in constricted capillary with elastic interface

Snap-off of bubbles and drops in constricted capillaries occurs in many different situations, from bio-fluid to multiphase flow in porous media. The breakup process has been extensively analyzed both by theory and experiments, but most work has been limited to pure interfaces, at which the surface stress is isotropic and fully defined by the interfacial tension and interface curvature. Complex interfaces may present viscous and elastic behavior leading to a complex stress state that may change the dynamics of the interfacedeformation and breakup. We extend the available asymptotic model based on lubrication approximation to include elastic interfacial stress. Drop breakup time is determined as a function of the capillary geometry and liquidproperties, including the interfacial elastic modulus. Results show that the interfacial elasticity has a stabilizing effect by slowing down the growth of the liquid collar, leading to a larger break-up time. This stabilizing effect has been observed experimentally in different, but related flows [Alvarado et al., “Interfacial visco-elasticity of crude oil-brine: An alternative EOR mechanism in smart waterflooding,” in SPE-169127 Improved Oil Recovery Symposium (Society of Petroleum Engineers, 2014)].

Time-lapse Monitoring Carbon Sequestrated Brine Aquifers- a Feasibility Study

Substantial research efforts are now underway on injecting (sequestrating) carbon dioxide (CO2) into deep saline aquifers. These sequestration efforts require remote monitoring using available geophysical tools to ensure that the sequestrated CO2 is in place and does not disturb the geological integrity of the surrounding rocks. Since seismic method is the only accepted geophysical tool that can potentially image detailed subsurface information to large depths, here we develop a monitoring strategy using seismic data alone. Fluid substitution at different saturations of CO2 in a brine filled aquifer and comparing its elastic properties with the original indicates that the formation density will play the key role in successful monitoring of carbon-sequestrated aquifers. As multicomponent seismic data are more sensitive to subsurface density variations than vertical (P-wave) component data, we believe that multicomponent seismic data are necessary for obtaining an accurate subsurface presequestration model. Multicomponent data are however more expensive than conventional (P-wave) data. Therefore, acquiring multicomponent data both for baseline and for successive monitoring surveys is not cost-effective. Since above-normal pore pressure due to sequestration may fracture the overlying formations, we investigate if microseismic events generated from these fractures could be used for monitoring. Inducing microseismic events with different fault-plane source mechanisms and computing passive seismic responses from them, we find that these computed responses are sensitive to the fracture fault plane geometry, and passive seismic data could be a potential monitoring tool. We conclude that if multicomponent seismic data are acquired prior to sequestration as a baseline survey and inverted for an accurate presequestration elastic earth model, we can then use passive seismic data for subsequent monitoring. This strategy, in turn, may provide a cost-effective way to monitor carbon sequestrated deep saline aquifers.

Stochastic-Perturbation Analysis of a One-Dimensional Dispersion-Reaction Equation: Effects of Spatially-Varying Reaction Rates

We carry out a stochastic-perturbation analysis of a one-dimensional convection–dispersion-reaction equation for reversible first-order reactions. The Damköhler number, Da, is distributed randomly from a distribution that has an exponentially decaying correlation function, controlled by a correlation length, ξ. Zeroth- and first-order approximations of the dispersion coefficient, D are computed from moments of the residence-time distribution obtained by solving a one-dimensional network model, in which each unit of the network represents a Darcy-level transport unit, and the solution of the transfer function in zeroth- and first-order approximations of the transport equation. In the zeroth-order approximation, the dispersion coefficient is calculated using the convection–dispersion-reaction equation with constant parameters, that is, perturbation corrections to the local equation are ignored. This zeroth-order dispersion coefficient is a linear function of the variance of the Damköhler number, 〈(ΔDa)2〉. A similar result was reported in a two-dimensional network simulation. The zeroth-order approximation does not give accurate predictions of mixing or spreading of a plume when Damköhler numbers, Da ≪ 1 and its variance, 〈(ΔDa)2〉 > 0.25 〈Da2〉. On the other hand, the first-order theory leads to a dispersion coefficient that is independent of the reaction parameters and to equations that do accurately predict mixing and spreading for Damköhler numbers and variances in the range √〈(ΔDa)2〉/〈Da〉≤0.3