Search

CO2 injection into shale reservoirs has been recently proposed as a promising method that can be used to enhance hydrocarbon recovery from shale reservoirs. Adsorption behavior of hydrocarbon(s)/CO2 mixtures under shale-reservoir conditions plays an important role in affecting the efficiency of CO2-enhanced hydrocarbon recovery from shale. In organic pores residing in shale reservoirs, the adsorption behavior of hydrocarbon(s)/CO2 mixtures can be significantly affected by the strong fluid/pore-wall interactions. In this work, a double-nanopore system comprising of two pores with sizes of 1 nm and 3 nm is built; then the competitive adsorption behavior of hydrocarbon(s)/CO2 mixtures (i.e., C1/nC4, C1/CO2, nC4/CO2, and C1/nC4/CO2 mixtures) is investigated in this double-nanopore system using the molecular dynamic (MD) simulations. Firstly, the competitive adsorption behavior of C1/nC4 mixture in double-nanopore system is studied with a depressurization manner. The effects of pressure and pore size distribution on competitive adsorption between hydrocarbons and CO2 are discussed. To investigate the efficiency of CO2 in replacing C1 or nC4 molecules from organic pores, dynamic distribution characteristics of C1/CO2, nC4/CO2, and C1/nC4/CO2 mixtures in the double-nanopore system are further investigated. The competitive adsorption behavior of C1/nC4 mixture indicates that, in both nanopores, as pressure decreases, adsorption of lighter hydrocarbon (i.e., C1) decreases significantly, but adsorption of heavier component (i.e., nC4) increases slightly. It suggests that as pressure decreases, the lighter hydrocarbons can be easily extracted from nanopores, while the heavier hydrocarbons may not be readily produced. Adsorption behavior of C1/CO2 indicates that CO2 can help the C1 recovery from nanopores; meanwhile, the recovery efficiency in the larger pore, (i.e., 3 nm), is much higher than that in the smaller pore (i.e., 1 nm). On the contrary, as pressure decreases, adsorption of nC4 in nC4/CO2mixtures in both nanopores is becoming stronger with the presence of CO2; the same behavior is also observed for C1/nC4/CO2 mixture. This implies that, although CO2 injection may help the recovery of lighter hydrocarbons (e.g., C1), but may not be an efficient agent for the recovery of heavier hydrocarbons (e.g., nC4).

The understanding of proppant flow through fractures is critical in evaluating the performance of hydraulic fracturing operations. As a continuation of our experimental efforts devoted to understanding how proppant flows in rough vertical fractures, in this paper, we examine the effect of injection parameters on the proppant transport in rough vertical fractures. The effects of polymer concentration, injection rate, proppant concentration, and type of proppant were investigated in detail. Experimental results show that, in general, two types of settling patterns can be observed in the proppant transport in a rough fracture: sand-dune settling pattern and tree-like settling pattern. The following conditions tend to suppress the tree-like settling behavior of proppants in the rough fracture: a higher density of the proppants, a lower viscosity of the polymer solution, and a lower injection rate. This is attributed to a higher settling velocity of the proppants in the fracture under these conditions. A sufficiently high polymer concentration is needed to enable effective proppant flow in rough fractures. In general, the relative coverage of proppants increased dramatically as the polymer concentration increased, implying that the higher viscosity of fracturing fluid could enhance the slurry's ability to place more proppant vertically into the fracture and help to maintain a better conductivity after fracturing treatment. A sufficiently high injection rate of the slurry is also needed to enable effective proppant flow in rough fractures. At certain low injection rates, the proppants carried by a low polymer solution might not exhibit a tree-like settling pattern, diminishing the effect of roughness on the proppant transport. This means that even in rough fractures, the tree-like settling pattern of the proppants did not necessarily occur for sure; the injection rate should be properly selected to enable such phenomenon. With other conditions being kept constant, a higher proppant loading led to a higher final relative coverage of the proppants in the rough fractures. But if the injection rate used for delivering the proppants is not sufficiently high, we may encounter injectivity issues; in our lab experiments, this caused the choking of the pump. The heavier proppant (ceramic proppants) in the rough fracture models tended to suppress the tree-like settling pattern that was experienced by the lighter proppant (silica sands). This is attributed to the larger density of the ceramic proppants, leading to a larger settling velocity. In order to maximize the spreading of a given proppant over a rough fracture model, we should determine the proper values of all the essential injection parameters (including polymer solution, injection rate, proppant concentration) by striking a good balance among them. The conclusions obtained in this study shed light on how to optimize slurry injection parameters to achieve an optimal proppant-filling ratio during hydraulic fracturing.

Summary Cyclic steam stimulation (CSS) is a proven effective technique for boosting oil production. Metal species can act as a catalyst for aquathermolysis reactions between heavy oil and water during the CSS process. For this paper, a series of CSS experiments with and without metal nanoparticles was conducted at temperatures up to 220°C to compare the performance of nickel and iron oxide nanoparticles in promoting aquathermolysis reactions in CSS; further, different loadings of metal nanoparticles were also tested in the CSS experiments. During the experiments, we monitored the variations of oil recovery factor, oil viscosity, gas composition, and water production. The experimental results show that both nickel and iron oxide nanoparticles can act as a catalyst for aquathermolysis reactions and reduce the viscosity of heavy oil. However, their respective catalytic effects differ significantly: nickel nanoparticles can break the C-S bond more effectively than iron oxide metal nanoparticles, thus achieving a higher ultimate oil recovery factor of CSS. The introduction of metal nanoparticles boosted oil production and increased water production from the very first cycle in the CSS process. The gas chromatography (GC) analysis and the pressure data recorded during each soaking period revealed that a higher amount of evolved gas including alkenes and hydrogen sulfide was generated in the early stage, increasing reservoir pressure and forcing more condensed water to be produced from the sandpack.

Conventional hot water flooding generally has disadvantages of high energy consumption and large capital expenses. To improve the economy of hot water flooding, a novel geothermal energy assisted heavy oil recovery method is presented. To validate this idea, a numerical simulation study is conducted to analyze the performance of the heat exchange well and heavy oil recovery process. Furthermore, the new method and conventional hot water flooding are compared to examine the economic feasibility of this new method based on the financial net present value. Finally, factors that may affect the feasibility of this new method are discussed. The simulation results show that compared with cold water flooding, the new method can increase heavy oil recovery by more than 7%. Moreover, the financial net present values of the new method at different oil prices all surpass those of the conventional water flooding, and thus the new method proposed is feasible both technically and economically. The analysis on the factors affecting the feasibility reveals that the critical oil price (the oil price when the financial net present value equals 0) of the new method decreases with increasing heavy oil reservoir thickness, decreasing oil viscosity, permeability variation coefficient and initial reservoir pressure.

In order to improve the field productivity, the industries can create new fractured wells to reduce the interval between the initial fractures (initial fracture indicates a fracture that is induced during the initial fracturing treatment). Production from the initial fractured wells can induce stress reorientation in the vicinity of the fractures. As such, a refracturing treatment in the stress reorientation region can lead to a reoriented refracture which has a certain azimuth with respect to the initial fractures. This azimuth can even be 90° if the stress reorientation is sufficiently large. In addition to the conventional parameters (including refracture's length, refracture's conductivity, and the in-situ conditions), there are two more factors, the reorientation azimuth and the interference from the initial fracture, which can exert significant influence on the transient flow behavior of the reoriented refractures. In such case, the conventional analytical/semi-analytical models neglecting these two factors are no longer applicable to characterize the transient flow behavior from such a refracture. In this work, we develop a novel semi-analytical model to characterize the transient flow behavior of a reoriented refracture considering the interference from the initial fractures. In this model, the fractures are explicitly represented with discretized segments. We apply finite difference approximation to the initial fractures and refractures, respectively, to simulate the transient flow in the fracture system. In addition, the fluid flow in the matrix system can be characterized by Green function and Newman product method. Based on the continuity of flux and pressure, we couple the fracture flow equations with the matrix flow equations to construct a semi-analytical model. With the aid of the proposed model, we conduct a thorough investigation of the transient flow behavior of reoriented refractures. It is observed that, at the early production period, two fracture pseudo-steady-state flows can appear due to the interference from the initial fractures. The formation linear flow and bilinear flow can also be observed during the production period of a refractured well. In an anisotropic-permeability reservoir, the refracture's azimuth and position can exert a significant impact on the productivity of the fractures. There is a region in the vicinity of the initial fractures, within which an orthogonal refracture can lead to the highest field productivity.

Three-phase vapor-liquid-aqueous (VLA) equilibria can frequently appear in hydrocarbon reservoirs. The presence of the aqueous phase, which can commonly exist as a pure water phase, can dramatically increase the probability of encountering convergence issues in the three-phase equilibrium calculation algorithms. The primary reason is that there is a lack of good initial estimates of equilibrium ratios in both stability tests and flash calculations. In this research, we improve the three-phase VLA equilibrium calculation algorithm for water/hydrocarbon mixtures by developing a new scheme for initializing equilibrium ratios as well as a new procedure for three-phase equilibrium calculations. By applying this new initialization scheme, one can securely identify the aqueous-like stationary point on the tangent plane distance function. In our algorithm, convergence problems, which can frequently appear in the flash calculations of three-phase VLA equilibrium calculations, are avoided. We demonstrate the robustness and efficiency of the newly developed algorithm by showing a number of example calculations.

In this work, we develop an alternative two-phase flash algorithm with the consideration of capillary pressure at specified mole numbers, volume, and temperature. This algorithm is built upon the volume-temperature (VT) two-phase flash algorithm as proposed by Mikyska and Firoozabadi (2012), which relies on the use of the newly defined volume function and volume function coefficient. Our algorithm contains two loops: an inner loop and an outer loop. In the inner loop, the interior point method is used to solve the vapor-phase fraction by minimizing the objective function. In the outer loop, the volume fraction of the vapor phase is updated by solving the pressure equilibrium equation which simultaneously considers the capillary pressure between the two phases; herein, we first divide the interval (0, 1) to 1000 subintervals and then apply the bisection method in each subinterval. This method is used to robustly solve the pressure equilibrium equation. The capillary pressure is calculated using the Young-Laplace equation on the basis of the interfacial tension predicted by the parachor model. Eventually, knowing the volume fractions enables one to update the equilibrium ratios in the outer loop. We demonstrate the correctness and robustness of our algorithm by testing it against four fluid mixtures, including binary and ternary mixtures. The two-phase envelope obtained from the VT-flash algorithm with capillarity effect has shown excellent agreement with that calculated by the pressure-temperature (PT) flash algorithm with capillarity effect. In addition, we examine the influence of capillary pressure on the two-phase envelope in the molar-density/temperature (cT) space.

Hydrate reservoirs generally have some unique characteristics such as hydrate dissociation and reformation, permeability change, which may result in quite different pressure-transient behavior from that in conventional reservoirs. In this work, we develop the pressure-transient analysis method for Class III hydrate reservoirs. By establishing a 2D cylindrical model, we then conduct a pressure buildup test and five flow regimes are identified according to the exhibited characteristics observed from the pressure-transient curves. Furthermore, sensitivity analysis is conducted to study the effects of initial reservoir pressure, initial reservoir temperature, permeability reduction exponent, BHP (bottom-hole pressure), and production time on pressure-transient behavior. We identify five flow regimes on the pressure-transient curves. Due to the dissociation of gas hydrate, the first regime is dominated by the non-uniform distribution of permeability, and the pressure derivative curve exhibits an upward straight line. We identify a “V-shaped” valley in the second regime with a peak in the third regime, which suggests the occurrence of hydrate reformation near the wellbore. Such characteristics exhibited in the three regimes are unique to hydrate reservoirs. In addition, the methodology proposed in this study is successfully applied to analyze the actual well testing data obtained from the Malik hydrate reservoir in Canada.

SummaryProduction from a fractured vertical well will lead to a redistribution of the stress field in formations. If the induced stress changes are sufficiently large to overcome the effect of the initial horizontal-stress deviator, the direction of the minimum horizontal stress can be turned into the direction of the maximum horizontal stress within an elliptical region around the initial fracture, resulting in a stress-reversal region near the wellbore. In such cases, a refracturing treatment can create a refracture that propagates orthogonally to the initial fracture because of the stress reversal. As such, the high-pressure area of the formation can be stimulated by the refracture, and the productivity of the refractured well can be improved. In this work, we develop a semianalytical model to evaluate the performance of a refractured vertical well with an orthogonal refracture. To simulate the well performance throughout the entire production period, we divide the well production into three stages: the first stage, when the well is producing oil with the initial fracture; the second stage, when the well is shut down for the refracturing treatment; and the third stage, when the well is producing oil with both the initial fracture and the refracture. In addition, by discretizing the initial fracture and the refracture into small segments, the conductivity of the fractures can be taken into account, and the geometry of the fracture system can be captured. We use the Green-function method to analytically simulate the reservoir flow and use the finite-difference method to numerically simulate the fracture flow; therefore, a semianalytical model can be constructed by coupling the reservoir-flow equations with the fracture-flow equations. This proposed model is applied to different wellbore and reservoir conditions. The calculated results show that this proposed model is versatile because it can simulate various wellbore constraints, including the conditions of constant bottomhole pressure (BHP), varying BHP, constant production rate, and varying production rate. The permeability anisotropy of the reservoir system, as well as the nonuniform conductivity distribution along the fracture, can also be incorporated into this proposed model. In addition, we demonstrate that this proposed model can be used to simulate other types of refractured vertical wells with minor modifications.

Tight/shale reservoirs are extensively containing nanopores, and the confined space in nanopores can greatly alter the phase behavior of reservoir fluids due to the strong capillarity effect. Many researches have been recently conducted to investigate the effect of capillary pressure in nanopores on altering the oleic-vapor two-phase equilibria. Few attempts have been made to describe the effect of capillarity on the aqueous-oleic-vapor three-phase equilibria. This work proposes a new algorithm for performing three-phase pressure-temperature (P-T) flash coupled with capillary effect. This algorithm considers two capillary pressures that exist across the two interfaces dividing the three phases in a nanopore. When describing the three-phase equilibria, two types of reservoir wettability are considered: water-wet formation and oil-wet formation. In each case, distribution of the three phases in a nanopore is determined based on the spreading coefficient which refers to the spreading ability of an oleic phase over the spreading ability of an aqueous phase. Example calculations are conducted to show the robustness of the new algorithm as well as to study the effect of capillarity on the three-phase equilibria. Computation results show that the three-phase P-T envelope for a given hydrocarbons/water mixture in a nanopore can be significantly altered by capillary pressure, but how the envelope moves will depend on the wettability of the nanopore and the spreading coefficient. The general trend is that both the upper branch (i.e., the oleic-aqueous/vapor-oleic-aqueous boundary) and the lower branch (i.e., the liquid-vapor/liquid-liquid-vapor boundary) of the three-phase envelope tend to move downward. Compared to the water-wet case, the oil-wet nanopore will shift the three-phase boundaries in a much larger degree. In addition to the alteration of the three-phase envelope, the presence of capillarity will also lead to the alteration of the phase fractions and phase compositions in the nanopore. Nanofluidic experiments should be conducted in the future to verify the findings obtained from the calculations.

Summary Field studies have shown that, if an inclined fracture has a significant inclination angle from the vertical direction or the fracture has a poor growth along the inclined direction, this fracture probably cannot fully penetrate the formation, resulting in a partially penetrating inclined fracture (PPIF) in these formations. It is necessary for the petroleum industry to conduct a pressure-transient analysis on such fractures to properly understand the major mechanisms governing the oil production from them. In this work, we develop a semianalytical model to characterize the pressure-transient behavior of a finite-conductivity PPIF. We discretize the fracture into small panels, and each of these panels is treated as a plane source. The fluid flow in the fracture system is numerically characterized with a finite-difference method, whereas the fluid flow in the matrix system is analytically characterized on the basis of the Green's-function method. As such, a semianalytical model for characterizing the transient-flow behavior of a PPIF can be readily constructed by coupling the transient flow in the fracture and that in the matrix. With the aid of the proposed model, we conduct a detailed study on the transient-flow behavior of the PPIFs. Our calculation results show that a PPIF with a finite conductivity in a bounded reservoir can exhibit the following flow regimes: wellbore afterflow, fracture radial flow, bilinear flow, inclined-formation linear flow, vertical elliptical flow, vertical pseudoradial flow, inclined pseudoradial flow, horizontal-formation linear flow, horizontal elliptical flow, horizontal pseudoradial flow, and boundary-dominated flow. A negative-slope period can appear on the pressure-derivative curve, which is attributed to a converging flow near the wellbore. Even with a small dimensionless fracture conductivity, a PPIF can exhibit a horizontal-formation linear flow. In addition to PPIFs, the proposed model also can be used to simulate the pressure-transient behavior of fully penetrating vertical fractures (FPVFs), partially penetrating vertical fractures (PPVFs), fully penetrating inclined fractures (FPIFs), and horizontal fractures (HFs).

For the most nonlinear temperature-dependent volume translation techniques, there is a crossing of isotherms in the pressure-volume (PV) diagram, which leads to thermodynamic inconsistency, gives anomalous prediction of some physical properties and thereby restricts the applicability of the volume translated equation of state (EOS). In this work, a generalized nonlinear temperature-dependent volume translation model is developed for PR EOS to achieve more accurate prediction of the liquid densities for pure components. Based on a criterion we have derived in a prior work, a mathematical constraint is applied when the model parameters are being regressed to match the measured density data; by adopting this criterion, we ensure that no any crossover phenomenon occurs for the new volume translation model over a wide range of pressure and temperature (up close to 100 MPa and 1000 K). The model parameters are determined based on the regression of the density data of 17 non-polar and slightly polar substances. It is shown that the new consistent volume-translated PR EOS improves the accuracy in calculating the liquid density for the 17 substances with an overall absolute average percentage deviation of 1.44%.

Accurate description of absolute adsorption/desorption behavior for hydrocarbons on shale is of critical importance to the understanding of the fundamental mechanisms governing the storage, transport, and recovery of shale gas or shale gas condensate in shale reservoirs. By applying a thermogravimetric method, we first measure the excess adsorption/desorption isotherms of pure CH4 and n-C4H10 on shale samples over the temperature range of 303.15–393.15 K. The maximum test pressures considered for CH4 and n-C4H10 are 50 bar and 2 bar, respectively. Grand Canonical Monte Carlo (GCMC) simulations are then applied to calculate the density of the adsorption phase by considering the fluid-pore surface interactions. We use such calculated density of the adsorption phase to calibrate the excess adsorption/desorption isotherms, which enables us to eventually obtain the absolute adsorption/desorption isotherms. Such approach for estimating the density of the adsorption phase is essentially different from the commonly used approaches in which the density of the adsorption phase is considered to be independent of temperature, pressure, and pore size. The adsorption/desorption test results show that both CH4 and n-C4H10 exhibit more adsorption as temperature decreases or pressure increases. Their adsorption/desorption isotherms exhibit hysteresis phenomenon and this phenomenon weakens as temperature increases. Comparatively, the hysteresis behavior observed for n-C4H10 is more obvious than that for CH4. Compared with CH4, n-C4H10 has higher adsorption capacity under the same condition, indicating its higher affinity towards the shale with organic matters. As for the conventional approaches, the density calculated from the van der Waals constant b or the liquid hydrocarbon density can be used to reasonably well evaluate the absolute adsorption isotherms of n-C4H10 on shale, but tends to underestimate the absolute adsorption of CH4 on shale. GCMC simulations show that the density of the adsorption phase is strongly correlated with system pressure, temperature, and pore size. Compared to the conventional approaches, GCMC simulations can better capture the in-situ density of adsorption phase; on the basis of the in-situ density of adsorption phase, we can then achieve more accurate determination of the absolute adsorption isotherms of a given hydrocarbon on shale. This study raises the imperativeness of leveraging more sophisticated simulation tools (such as GCMC) for more accurate determination of absolute adsorption isotherms.

Summary The Peng-Robinson equation of state (PR-EOS) (Robinson and Peng 1978) with capillary effect has been used extensively to describe the phase behavior of hydrocarbons under nanoconfinement in shale reservoirs. In nanopores, surface adsorption may be significant, and molecular distribution is heterogeneous. Although the PR-EOS cannot consider these effects, statistical thermodynamic approaches such as density-functional theory (DFT) can explicitly consider the intermolecular and fluid/surface interactions. In this work, we compare the phase behavior of pure hydrocarbons and mixtures in nanopores from the PR-EOS with capillary effect and engineering DFT. We apply the Young-Laplace (YL) equation, assuming zero contact angle to calculate the capillary pressure in the PR-EOS with capillary effect. On the other hand, we extend the PR-EOS to inhomogeneous conditions with weighted-density approximation (WDA) in engineering DFT. For pure components, both approaches predict that the dewpoint temperature increases in hydrocarbon-wet nanopores. Although engineering DFT predicts that the confined dewpoint temperature approaches bulk saturation point when pore size approaches 30 nm, the saturation point obtained from the PR-EOS with capillary effect approaches bulk only when the pore size is as large as 1000 nm. With engineering DFT, the critical points in nanopores deviate from those in bulk, but no change is observed from the PR-EOS with capillary-effect model. The difference on the dewpoint temperature between the PR-EOS with capillary effect and engineering DFT decreases as the system pressure approaches the critical pressure. At low-pressure conditions, the PR-EOS with capillary-effect model becomes unreliable. For binary mixtures, both approaches predict that the lower dewpoint decreases and the upper dewpoint increases. More interestingly, phase transition can still occur when the system temperature is higher than the bulk cricondentherm point. Engineering DFT predicts that the confined lower dewpoint approaches bulk when pore size approaches 20 nm, whereas the dewpoint obtained from the PR-EOS with capillary effect approaches bulk only when the pore size is as large as 100 nm. This work illustrates that assuming homogeneous distributions in nanopores may not be appropriate to predict the phase behavior of hydrocarbons under nanoconfinement.

Summary Solvent/thermal hybrid methods have been proposed recently to enhance heavy-oil recovery and to overcome the shortcomings that are encountered when either method is solely applied. One of the methods for this hybridization is to combine electromagnetic (EM) heating and solvent injection to facilitate heavy-oil production by gravity drainage. This approach has several advantages including reduced CO2 emissions, decreased water consumption, and appropriateness for water-hostile reservoirs. We are currently lacking any mathematical model for better understanding, designing, and optimizing this hybrid technique, which is partly attributed to this technique still being in its infancy. We propose a semianalytical model to predict the oil-flow rate resulting from the combined EM heating and solvent-assisted gravity drainage. The model first calculates the temperature distribution within the EM-excited zone caused by the radiation-dominated EM heating. Using different attenuation coefficients within and beyond the vapor chamber, the model can properly describe the corresponding temperature responses in these regions. Next, an average temperature of the chamber edge contributed by EM heating is used to estimate the temperature-dependent properties, such as vapor/liquid equilibrium ratios (K-values), heavy-oil/solvent-mixture viscosity, and solvent diffusivity. Subsequently, a 1D diffusion equation is used to calculate the solvent-concentration distribution ahead of the chamber edge. Eventually, the oil-flow rate is evaluated with the calculated temperature and solvent distributions ahead of the chamber edge. The proposed model is validated against the experimental results obtained in our previous study, and the predicted oil-flow rate agrees reasonably well with the experimental data. The proposed model can efficiently predict the oil-flow rate of this hybrid process. We conduct sensitivity analyses to examine the effect of major influential factors on the performance of this hybrid technique, including EM heating powers, solvent types, solvent-injection pressures, and initial reservoir temperatures. The modeling results demonstrate that a higher EM heating power, a heavier solvent, and a higher solvent-injection pressure could accelerate the oil-recovery rate, but tend to lower the net present value (NPV) and increase the energy consumption. In summary, the newly proposed model provides an efficient tool to understand, design, and optimize the combined technique of EM heating and solvent-assisted gravity drainage.

Multiphase equilibria for water/hydrocarbon mixtures are frequently encountered in hydrocarbon reservoirs. The presence of water in these mixtures can lead to a higher number of equilibrating phases, increasing the complexity of the multiphase split calculations. It is a common approach to treating water as the bulk free phase and performing only two-phase split calculations on the hydrocarbon-rich liquid phase and vapor phase. The free-water flash algorithm uses a different approach; it considers the effect of water presence on the overall phase equilibrium of water/hydrocarbon mixtures, albeit also assuming the aqueous phase to be pure water. The free-water algorithm might be less accurate in some cases where the solubility of methane in the aqueous phase cannot be neglected. In this study, a modified version of the free-water flash method previously developed by the authors, i.e., the so-called augmented free-water flash, is extended to perform three-phase vapor-oleic-aqueous (VOA) flash calculations for water/hydrocarbons mixtures on the basis of the assumption that only the existence of water and methane is considered in the aqueous phase. The flash package incorporating this augmented free-water method can handle the single-phase, two-phase, or three-phase equilibrium calculations. Example calculations made on two water/hydrocarbons mixtures demonstrate that the phase compositions and phase mole fractions calculated by augmented free-water method provide better predictions compared with the traditional free-water method since the solubility of methane is considered in the aqueous phase. Our new algorithm is also shown to be computationally more efficient than the conventional full three-phase flash algorithm. Therefore, the augmented free-water approach strikes a good balance between computational efficiency and prediction accuracy.

Phase behavior of shale fluids in small pores is not well understood. One complexity comes from the fact that sorption of components by organic-rich shale can be significant and selective. In an attempt to elucidate the effect of sorption on phase behavior of shale fluids, we present a new experimental method that can be used to measure the bubble-point pressures of N2/n-C4H10 mixtures in the presence of an actual shale sample. Pressure/volume (P/V) isotherms for a given mixture were firstly measured in a PVT cell. Then, the measurements of the P/V isotherms for the same mixture were repeated in a partially confined space by opening a valve between the PVT cell and a shale container. The so-called partially confined space consists of the pore space inside the shale sample, the bulk space in the PVT cell and in the connecting tubing, and the non-cementing pore spaces among the shale particles. Results show that, the measured bubble-point pressure of the N2/n-C4H10 mixture in the partially confined space was higher than the corresponding bubble-point pressure in the bulk space. A detailed analysis indicates that, when loaded in the partially confined space, n-C4H10 exhibits a higher level of sorption capacity on the shale sample than N2, resulting in a higher concentration of N2 left in the free fluid than that in the initial mixture, which is the so-called selective sorption. The higher concentration of N2 led to the higher bubble-point pressure as observed in the measurements. The increase of the bubble-point pressure due to the selective sorption was observed to be greater at a lower temperature. This is because the sorption of n-C4H10 relative to that of N2 is more significant at a lower temperature. A higher temperature did not lead to a higher increment in the bubble-point pressure likely because bubble-point is more sensitive to composition than to temperature for these mixtures at the conditions tested. This emphasizes the importance of considering sorption in phase behavior for small pores.

Conventional production optimization for waterflooding reservoirs relies on full-scale reservoir simulators and repeated simulation runs to calulate gradient information. These factors lead to expensive computational cost. Different from the conventional optimization, in this paper, a novel waterflooding optimization method is developed by running a data-driven interwell numerical simulation model with flow-path tracking (INSIM-FPT) in three dimensions. With the application of flow-path tracking strategy, the more-accurate calculations of dynamic allocation factors and control pore volumes can be obtained from INSIM-FT-3D model. Moreover, based on the INSIM model, we define producer-centered production efficiency of wells with multiple perforations and propose a rapid waterflooding optimization method. The optimized well production rates and injection rates are obtained with the help of the oil cut and the existing injector-centered allocation factors derived from INSIM model. The new optimization method requires only one time forward run without repeated iterative calculations. And the optimization is completely based on reservoir-engineering points of view, making it easy to explain oil increment mechanisms. Finally, a synthetic case and the other large-scale field case is established to test its optimization performance.

Based on the free-water assumption that the water-rich liquid phase contains only pure water, we develop two simple free-water Rachford-Rice methods: 1) a two-phase free-water Rachford-Rice method where the phase fractions can be analytically solved; and 2) a three-phase free-water Rachford-Rice method where there is only one unknown in the objective function (i.e. the vapour-phase fraction) that is used for solving the phase fractions. Combining these two Rachford-Rice methods, a new free-water flash algorithm is developed to perform multiphase flash calculations where single-phase equilibria, two-phase equilibria, and three-phase vapour-liquid-aqueous equilibria can be considered. In this free-water flash algorithm, we first test if the mixture is stable; if the mixture is found to be unstable, we directly initiate the three-phase free-water flash. A set of criteria is developed for one to properly switch from a three-phase free-water flash to either a two-phase free-water flash or a conventional two-phase flash, depending on whether a water-rich phase is present in the two-phase equilibrium. We also develop efficient and robust methods for initializing the equilibrium ratios for the two-phase flashes. The negative flash is allowed in the flash calculation algorithms. A number of example calculations are carried out to demonstrate the robustness of the newly developed algorithm. A good agreement can be achieved between the flash results obtained by the new flash algorithm and those obtained by the conventional full three-phase flash algorithm. This article is protected by copyright. All rights reserved

Late cycles of cyclic steam stimulation (CSS) are characterized with a decreasing heavy-oil recovery and an increasing water cut. Nickel nanoparticles can be used to promote aquathermolysis reactions between water and heavy oil in steam injection processes, thereby increasing the recovery factor. In this paper, detailed investigations were performed to determine the optimum operational parameters and answers to the following questions: (1) What is the optimum concentration of nickel nanoparticles for promoting aquathermolysis under high steam temperature?; (2) Can we improve oil recovery at lower steam temperatures with the presence of nickel nanoparticles?; (3) What effect does the penetration depth of nickel nanoparticles have on the final oil recovery? CSS experiments were conducted between temperatures 150°C and 220°C. Steam generated under these temperatures was injected into sand packs saturated with Mexican heavy oil. Powder-form nickel nanoparticle was introduced into this process to boost the oil production. In an attempt to obtain the optimum concentration, different concentrations were tested. Next, oil sands without any nanoparticle additives were first added into the cylinder. Then, only one third of the sand pack was mixed with nickel nanoparticle near the injection port. Experiments were executed to study the effects of temperature, nickel concentrations, and nanoparticle penetration depth on the ultimate oil recovery and produced oil-water ratios after each cycle. Produced oil quality and emulsion formation were evaluated with gas-chromatography (GC) analysis, viscosity measurements, saturates-asphaltenes-resins-aromatics (SARA) test, and microscopic analysis of the effluents. Experimental results show that the best concentration of nickel nanoparticles, which gives the highest ultimate oil recovery factor, is 0.20 wt% of initial oil in place (IOIP) under 220°C, while the nickel concentration of 0.05 wt% provides the highest recovery factors at the early stages. A lower temperature of 150°C provides a much lower recovery factor than 220°C, which is mainly due to a lower level of aquathermolysis reactions at 150°C. By analyzing the compositions of produced oil and gas samples with GC and SARA, we confirm that (1) the major reaction mechanism during the aquathermolysis reaction is the breakage of C-S bond, (2) the nickel nanoparticles can act as catalyst for the aquathermolysis reaction, and (3) the catalytic effect becomes less remarkable from cycle to cycle. One run of experiment to test the effect of particle penetration depth revealed that nickel nanoparticle distributed near the injection port greatly contributed to the ultimate recovery factor.

Summary Steam/solvent coinjection has been studied as a potential method to improve the efficiency of conventional steam-assisted gravity drainage (SAGD) for bitumen recovery. This research is part of an experimental program for phase behavior of Athabasca-bitumen/solvent mixtures. This paper presents a new set of experimental data for phase equilibrium, viscosity, density, and asphaltene precipitation for 11 mixtures of Athabasca bitumen with n-hexane and 10 mixtures of the same bitumen with n-octane. Phase-boundary measurements were conducted at temperatures up to 160°C and pressures up to 10 MPa. The bitumen sample used in this research was studied in our previous research, in which the same bitumen was not effectively diluted by n-butane because of the coexistence of a butane-rich liquid with a bitumen-rich liquid phase. In this research, the liquid/liquid separation of hydrocarbons was not observed for n-hexane/bitumen (HB) and n-octane/bitumen (OB) mixtures for the range of temperatures and pressures tested, even at solvent concentrations higher than 90 mol%. This observation indicates that the amount of solvent available near the edge of a steam chamber is expected to be entirely used for bitumen dilution beyond the chamber edge in coinjection of steam with heavier hydrocarbon solvents, such as n-hexane and n-octane. Experiments for asphaltene precipitation at atmospheric pressure showed a larger amount of precipitates with n-hexane than with n-octane at a given solvent concentration higher than 50 wt%. For solvent concentrations less than 50 wt%, no asphaltene precipitation was observed for both solvents with the bitumen sample tested in this research.

Electromagnetic (EM) heating holds a significant potential for recovering heavy oil and bitumen since it can reduce carbon emission and avoid excessive water usage, and is applicable for water hostile reservoirs. Combining solvent injection with EM heating might further reduce the energy intensity of the process. The merits of using solvent in EM heating include diluting heavy oil and thereby increasing its mobility, serving as a heat carrier by reinforcing heat convection in porous media and facilitating gravity drainage by forming a vapor chamber. In this study, we design a new experimental setup that can be used to examine the performance of combined EM heating and solvent-assisted gravity drainage for heavy oil recovery. With such experimental setup, we conduct a series of laboratory experiments to investigate the performances and mechanisms of combined EM heating and solvent-assisted gravity drainage for heavy oil recovery. During experiments, sand pack samples contained in Buchner filter funnel are placed in a microwave oven. Solvent injection can be initiated together with EM heating to simulate this hybrid process. Temperatures of the sand pack and oil recoveries are simultaneously recorded. We investigate the effects of influential factors on the process performances, including EM heating power, initial water saturation, solvent types (n-hexane and n-octane), and combination strategies of EM heating and solvent injection (simultaneous or alternate means). Experimental results show that combined EM heating and solvent-assisted gravity drainage could effectively enhance heavy oil recovery; the recovery factors of EM heating only, alternate EM heating and n-hexane injection, and alternate EM heating and n-octane injection are 12.37%, 61.18%, and, 83.59%, respectively. Moderate initial water saturation increases the heating speed and provides a higher oil recovery. The effect of solvent addition, affected by the solvent concentration in heavy oil, varies at different EM heating powers. Alternate EM heating and solvent injection is more cost effective due to the lower energy consumption and higher oil recovery compared to the simultaneous EM heating and solvent injection. A lowered asphaltene fraction in recovered oil has also been found in this hybrid process.

Summary Several interwell connectivity models such as multiple linear regression (MLR) and the capacitance model (CM) have been proposed to model waterflooding performance in high-permeability reservoirs on the basis of observed production data. However, the existing methods are not effective at characterizing the behavior of transient flows that are prevalent in low-permeability reservoirs. This paper presents a novel dynamic waterflooding model that is based on linear dynamical systems (LDSs) to characterize the injection/production relationships in an oil field during both stationary and nonstationary production phases. We leverage a state-space model (SSM), in which the changing rates of control volumes between injector/producer pairs in the reservoir of interest serve as time-varying hidden states, depending on the reservoir condition. Thus, the model can better characterize the transient dynamics in low-permeability reservoirs. We propose a self-learning procedure for the model to train its parameters as well as the evolution of the hidden states only on the basis of past observations of injection and production rates. We tested the LDS method in comparison with the state-of-the-art CM method in a wide range of synthetic reservoir models including both high-permeability and low-permeability reservoirs, as well as various dynamic scenarios involving varying bottomhole pressure (BHP) of producers, injector shut-ins, and reservoirs of larger scales. We also tested LDS on the real production data collected from Changqing oil field containing low-permeability formations. Testing results demonstrate that an LDS significantly outperforms CM in terms of modeling and predicting waterflooding performance in low-permeability reservoirs and various dynamic scenarios.

Volume translation is widely adopted in cubic equation of state (CEOS) to achieve more accurate density predictions for pure compounds and mixtures. Previous research reveals that a temperature-dependent volume-translated EOS could result in crossing of pressure-volume isotherms for a pure compound, which leads to an anomalous behavior that the predicted molar volume for a pure component can possibly be lower at a higher temperature at an isobaric condition. Such crossover phenomenon fails to consistently predict the thermodynamic properties of a pure compound, e.g., giving a negative isobaric thermal expansivity. In this study, we develop a concise criterion to judge whether a volume-translated EOS will result in crossover issues, and if so, the extent of the temperature and pressure range over which the crossover phenomenon occurs. The recently proposed volume translations are evaluated on the basis of the developed criterion. For the various types of temperature-dependent volume translations, we obtain the specific temperature/pressure conditions over which there is certainly no crossover phenomenon.

Steam-solvent coinjection has been studied and pilot-tested as a potential method to improve steam-assisted gravity drainage (SAGD) for bitumen recovery. Reliable design of coinjection requires reliable PVT data for bitumen/solvent/water mixtures, which are scarce and fragmentary in the literature. The main objective of this research was to present a new set of PVT and multiphase data for n-butane/Athabasca-bitumen/water mixtures at pressures up to 10 MPa and temperatures up to 160°C. Experiments were conducted by use of a conventional PVT apparatus. The data presented include multiphase equilibria up to four coexisting phases and liquid densities for 100% bitumen, two mixtures of n-butane/bitumen, and one mixture of n-butane/bitumen/water. A single equation-of-state model was developed to correlate all experimental data obtained, and used to interpret complex multiphase behavior observed for the highly size-asymmetric polar mixtures. Liquid-liquid separation of hydrocarbons was experimentally observed at the n-butane concentration of 97 mol% in the n-butane/bitumen system with/without water, for a wide range of temperatures at operating pressures for expanding-solvent-SAGD (ES-SAGD). This may indicate the limited solubility of n-butane in bitumen even when a high level of accumulation of n-butane takes place near a chamber edge in ES-SAGD for Athabasca bitumen. The multiphase transition that involves appearance/disappearance of the vapor phase was observed to occur near the vapor pressure of n-butane or its extension. Such phase transition occurs at a higher pressure in the presence of water, due to its vapor pressure, than in the absence of water at a given temperature. This is the first time four coexisting phases are reported for n-butane/Athabasca-bitumen/water mixtures at temperature-pressure conditions relevant to ES-SAGD.

We develop a method to determine the effective diffusion coefficient for each individual component of a gas mixture in a non-volatile liquid (e.g., heavy oil) at high pressures with compositional analysis. Theoretically, a multi-component one-way diffusion model is coupled with the volume-translated Peng-Robinson equation of state to quantify the mass transfer between gas and liquid (e.g., heavy oil). Experimentally, the diffusion tests have been conducted with a PVT setup for one pure CO2-heavy oil system and one C3H8–CO2-heavy oil system under constant temperature and pressure, respectively. Both the gas-phase volume and liquid-phase swelling effect are simultaneously recorded during the measurement. As for the C3H8–CO2-heavy oil system, the gas chromatography method is employed to measure compositions of the gas phase at the beginning and end of the diffusion measurement, respectively. The effective diffusion coefficients are then determined by minimizing the discrepancy between the measured and calculated gas-phase composition at the end of diffusion measurement. The newly developed technique can quantify the contributions of each component of mixture to the bulk mass transfer from gas into liquid. The effective diffusion coefficient of C3H8 in the C3H8–CO2 mixture at 3945 ± 20 kPa and 293.85 K, i.e., $$18.19 \times 10^{ - 10} {\text{m}}^{ 2} / {\text{s}}$$ , is found to be much higher than CO2 at 3950 ± 18 kPa and 293.85 K, i.e., $$8.68 \times 10^{ - 10} {\text{m}}^{ 2} / {\text{s}}$$ . In comparison with pure CO2, the presence of C3H8 in the C3H8–CO2 mixture contributes to a faster diffusion of CO2 from the gas phase into heavy oil and consequently a larger swelling factor of heavy oil.

Surfactant-alternating-gas injection, where foam generation can be expected, is a feasible approach for improving the sweep efficiency of solvent flooding by increasing the apparent viscosity of gas solvent. The foam stability plays an important role in this process. In this study, the foam stability and mobility of solvent/surfactant/pseudo-heavy-oil system under reservoir condition has been examined experimentally. The following factors are considered in the static foam stability experiments: surfactant concentration, salinity, temperature, the presence of pure n-C16H34 as pseudo-heavy oil, and two types of gases (CO2 and C3H8). During the static experiments, foam is generated by sparging C3H8 or CO2 at a constant flow rate through the surfactant Triton X-100 solution, and then stirring the mixture with a magnetic stirrer. The foam stability is subsequently evaluated in terms of its height change as a function of time. It is found that the increasing surfactant concentration contributes to an increase in foam stability; foam stability is insensitive to surfactant concentration when the surfactant concentration is above the threshold CMC. As for the C3H8 foam, an elevated temperature is detrimental to foam stability. At a higher temperature, the effective critical micelle concentration (CMC) of surfactant also increases. In general, the stability of C3H8 foam is negatively affected by an increasing salinity. As n-C16H34 is added to the solution, foam stability deteriorates. In order to study the effect of solvent type on foam stability, CO2 is used to replace C3H8 for generating foam, with the consideration of different salinity and temperature conditions. In comparison, CO2 foam only exhibits several minutes of half-life at any conditions, which is much shorter than the C3H8 foam. Additional flow experiments are also conducted by alternately injecting C3H8 and Triton X-100 solution into a glass beadpack. The pressure difference during the experiments is recorded so as to obtain the mobility of the dynamic flow in the porous media. It is found that alternating C3H8 and Triton X-100 solution results in a larger pressure difference, hence leading to more mobility reduction than alternating C3H8 and water through the porous media.

We investigate the natural gas flow behavior in the Dengying carbonate reservoir of Gaoshiti-Moxi Block in China. Outcrops observation, petrophysical properties measurement, and formation micro-resistivity image logging (FMI) all show that the vugs distribute in a stratified manner with characteristic length ranging from 0.2 to 4 cm. The horizontal permeability is significantly larger than the vertical permeability due to the stratified distribution of vugs and fractures. We build a representative conceptual reservoir model to describe this fracture-vuggy type carbonate reservoir. Both Darcy equation and Stokes-Brinkman equation are applied to simulate natural gas flow in this type of reservoir. Simulation results show that natural gas mainly flows along the free channel formed by low-dip fractures and flat ellipsoid vugs. Compared with the Darcy method, the Stokes-Brinkman method provides a more accurate prediction of gas flow behavior in the fracture-vug carbonate reservoir.

The fracture-surface characteristics (such as roughness and fractal dimensions) may greatly affect the proppant transport during hydraulic fracturing operation. Few researches have focused on investigating the proppant transport in vertical fracture with actual surface characteristics. As a continuation of our previous study (Huang et al. 2017), we qualitatively investigatethe migration of proppants in rough and vertical fractures by considering the effects of surface characteristics and rock type on the instantaneous transport and areal spreading of proppant in the fractures. We fractured different types of tight rocks (including limestone, marble, tight sandstone, and granite) with Brazilian test and molded them to manufacture 20×20cm transparent replicas with an aperture of 1 mm. We characterized the surface characteristics of these rock samples with different fractal dimensions. Subsequently, dyed fracturing fluid with or without proppant loading was injected into the rough vertical fracture. In each test, we monitored the inlet pressure continuously while the proppants were being transported in the fracture. The process was videotaped to monitor the proppant distribution in the rough fracture. Different from our previous study (Huang et al. 2017), a higher injection rate is used in this present study. The experimental results obtained in this study further consolidate the many findings reported in our recent study (Huang et al. 2017): in rough and narrow fracture, the surface roughness plays a pivotal role in affecting how proppants settle in the fracture as well as where the proppants settle in the fracture. Roughness of the vertical fractures tends to significantly enhance the vertical placement of proppants in the fracture, leading to a much higher proppant-filling ratio in a rough fracture than in a smooth fracture. Interestingly, in addition to the bridging effect observed in Huang et al. (2017), a previously formed proppants cluster can be broken up under a higher-rate slurry flow. The bridging of proppants and its subsequent breaking up can recursively occur during the high-rate slurry flow, resulting in fluctuations in the proppant filling ratios as well as fluctuations in the pressure profiles recorded in the inlet of the fracture model. The roughness of fracture models not only affects how much area of the fracture is being occupied by the proppants in the fracture, but also affects how tightly the proppants are filling up the fracture. Different types of rock have different surface characteristics, leading to the observed differences with regard to how the proppants migrate, settle down and fill up the fractures. No definite correlation could be established between any of the fractal numbers and the relative coverage of proppants in the fracture. More experiments, however, need to be conducted to reach more concrete conclusions in this regard.

Peng-Robinson equation of state (PR-EOS) with capillary effect has been extensively used to describe the phase behavior of hydrocarbons under nano-confinement in shale reservoirs. In nanopores, surface adsorption may be significant and molecular distribution is heterogeneous. While PR-EOS cannot take into account these effects, statistical thermodynamic approaches such as density functional theory (DFT) can explicitly consider the intermolecular and fluid-surface interactions. In this work, we compare the phase behavior of pure hydrocarbons and mixtures in nanopores from PR-EOS with capillary effect and engineering DFT. We apply the Young-Laplace (YL) equation assuming zero contact angle to calculate the capillary pressure in PR-EOS with capillary effect. On the other hand, we extend the PR-EOS to inhomogeneous conditions by using weighted density approximation (WDA) in engineering DFT. For pure components, both approaches predict that the dew-point temperature increases in hydrocarbon-wet nanopores. While engineering DFT predicts that the confined dew-point temperature approaches bulk saturation point when pore size approaches 20 nm, the saturation point obtained from PR-EOS with capillary effect approaches bulk only when the pore size is as large as 1 μm. With engineering DFT, the critical points in nanopores deviate from that in bulk, but no change is observed from PR-EOS with capillary effect model. The difference between PR-EOS with capillary effect and engineering DFT on the dew-point temperature decreases as the system pressure approaches the critical pressure. At low pressure conditions, PR-EOS with capillary effect model becomes unreliable. For binary mixtures, both approaches predict that the lower dew-point decreases and the upper dew-point increases. More interestingly, phase transition can still occur when the system temperature is higher than the bulk cricondentherm point. Engineering DFT predict that the confined lower dew-point approaches bulk when pore size approaches 20 nm, whereas the dew-point obtained from PR-EOS with capillary effect approaches bulk only when the pore size is as large as 100 nm. This work illustrates that assuming homogeneous distributions in nanopores may not be applicable to predict the phase behavior of hydrocarbons under nano-confinement.

Multiphase isenthalpic flash calculations are often required in compositional simulations of steam-based enhanced oil recovery methods. These flash calculations are challenging in the narrow-boiling regions and in the determination of the correct number of existing phases. Based on the free-water assumption that the aqueous phase is pure water, a robust and efficient algorithm is proposed to perform isenthalpic three-phase flash calculations in this work. Multiphase equilibria can be considered by this algorithm, including single-phase equilibria, two-phase equilibria, and three-phase vapor-liquid-aqueous equilibria. Isenthalpic flash is a type of flash calculation conducted at given pressure and enthalpy for a feed mixture. In the proposed algorithm, assuming the feed is stable, the temperature is first determined by solving the energy conservation equation. Then the stability test on the feed mixture is conducted at the calculated temperature and the given pressure. If the mixture is found unstable, the two-phase and three-phase vapor-liquid-aqueous isenthalpic flash calculations can be simultaneously initiated without resorting to stability tests. To achieve simultaneous flashes, the outer loop is used to update the temperature by solving the energy conservation equation. The inner loop is used to obtain phase fractions and compositions by performing a three-phase free-water isothermal flash. Note that a two-phase isothermal flash will be initiated if an open feasible region in the phase fractions appears in any iteration during the three-phase isothermal flash or any of the ultimately calculated phase fractions from the three-phase flash do not belong to [0,1]. Negative flash is allowed in the three-phase free-water isothermal flash. A number of example calculations for water/hydrocarbon mixtures are carried out to test the robustness of the proposed algorithm. At low to medium pressures, a good agreement can be achieved between the results obtained by this algorithm and those obtained by the conventional algorithm. This algorithm performs well for the narrow-boiling regions, for example, the three-phase vapor-liquid-aqueous equilibrium region encountered for the water/hydrocarbon mixtures. During the iteration, the new algorithm can readily handle the appearance and disappearance of phases in the inner loop as temperature updates in the outer loop. The number of stability tests involved in the new algorithm is significantly reduced, helping to boost its computational efficiency.

Phase behavior of reservoir fluids in nano-pores of shale can be quite different from that in a bulk space. Its accurate description can lead to more accurate estimation of hydrocarbons in place in shale, as well as better understanding of the mechanisms governing hydrocarbon recovery from shale. This paper presents both experimental and modeling studies of the coupled effects of competitive adsorption, capillary pressure and pore size distribution on the phase behavior of fluid mixtures in partially confined spaces. Experimentally, we measured the pressure/volume (PV) relationships along isothermal temperatures for fluid mixtures (N2/n-C4H10 and CH4/n-C4H10 binaries) that were contained in bulk spaces (PVT cell) and in partially confined spaces, respectively. To make the so-called partially confined space, the PVT cell was connected to a container holding a shale core sample. The partially confined space consists of the pore space in the shale core and the bulk space in the PVT cell. Theoretically, we developed a phase-behavior model for partially confined fluids by considering competitive adsorption, capillary pressure and pore size distribution of the shale core. In this model, the bulk PVT space is considered to be a capillary tube with an infinite diameter, while the shale sample is deemed to comprise of a series of capillary tubes with various diameters. Test results show that, at the same temperature, the bubble point pressure of N2/n-C4H10 mixture in the partially confined space is higher than that in the bulk space, while the bubble point pressure of CH4/n-C4H10 mixture in the partially confined space is smaller than that in the bulk space. When a fluid mixture comes into contact with a shale sample, the individual components in the mixture may be preferentially adsorbed onto the shale sample, leading to the competitive adsorption phenomenon. The competitive adsorption of gas mixture onto shale can change the initial mixture composition, and thus affect the phase behavior of the mixture. The theoretical model can be properly tuned to yield bubble point pressures that are well matching the measured ones.

Electromagnetic (EM) heating holds a large potential in heavy oil recovery since it can reduce carbon emission and avoid excessive water usage, and is applicable for water hostile reservoirs such as shale oil reservoirs. Combining solvent injection and EM heating might further reduce the energy intensity of the process. The merits of using solvent in EM heating include diluting heavy oil and thereby increasing its mobility, serving as a heat carrier by reinforcing heat convection in porous media and facilitating gravity drainage by forming a vapor chamber. Detailed experimental investigations are needed to investigate the mechanism of such a complex process and to specify the most influential factors of this hybrid and expensive process to determine optimal operational conditions. In this study, we conduct a series of laboratory experiments to investigate the mechanisms of combined EM heating and solvent assisted gravity drainage for heavy oil recovery. During experiments, sand pack samples contained in Buchner filter funnel are placed in a microwave oven. Solvent injection can be initiated together with EM heating to simulate the hybrid process of combined EM heating and solvent assisted gravity drainage. We investigate the effects of influential factors on the process efficiency, including initial water saturation, solvent types (n-hexane and n-octane), introduction methods of solvents (injection or premixed with oil), combination strategies of solvent injection and EM heating (simultaneous or alternate means), and EM heating power. Temperatures of the sand pack and oil recoveries are simultaneously recorded. Experimental results show that combined EM heating and solvent assisted gravity drainage could effectively enhance heavy oil recovery compared with EM heating or solvent use alone. A higher heating power provides a faster temperature rise and earlier oil production in the sand pack. Moderate initial water saturation could increase the heating speed, leading to a higher oil recovery. Solvent injection can further enhance the viscosity reduction and swelling effect of heavy oil due to EM heating. Compared with n-hexane, n-octane provides higher vertical displacement efficiency and oil recovery under the same experimental conditions. Alternate EMH and solvent injection is more cost effective due to the lower energy consumption compared to the simultaneous EM heating and solvent injection.