No abstract

The targeted reservoir for foam mobility control is usually layered or heterogeneous. However, a major limitation of existing foam models is that there are no dependencies of the foam modeling parameters on permeability, even though the permeability is accounted inherently only through changes in gas-water capillary pressure and shear rate. This results in considerable errors in predicting the foam mobility at largely varying permeabilities, which prevents users from simulating correctly the conformance achievable with the help of foam in heterogeneous reservoirs. Developing a foam simulator with systematic permeability-dependencies of foam properties is a key enabler for the rigorous simulation of foam floods in the field. An advanced population-balance foam model has been developed with reasonable physical mechanisms associated with the effect of permeability on the bubble density, foam generation and stability in porous media. The derivations indicate that the gas viscosity scaling constant increases with permeability exponentially, while the upper limit of foam texture, the foam generation coefficient, and the limiting capillary pressure decrease exponentially as the permeability increases. All these factors collectively affect the foam mobility. The upper limit of foam texture and the foam generation coefficient share the same power-law exponent with permeability because of the similar fundament. As a result, three additional power-law exponents are needed to correlate with permeability in the new model. To verify the correlations of the parameters with the permeability change, an automated regression program was applied to fit the resistance factors of several groups of foam flood experiments with foam quality scans at different permeabilities. The newly developed permeability-dependency functions showed its great competency in matching all the experimental data in a wide range of permeability. The optimized parameters are largely consistent with the theoretical exponents of the power-law functions of the aforementioned physical properties correlated to permeability, but also suggest extra modifications. In particular, the exponent for the limiting capillary pressure is about -0.5, which is equivalent to that the limiting water saturation is approximately independent of the permeability according to the Leverett J-function. As a result, the new functions of permeability dependencies for the foam-model parameters in the population-balance model enables the foam modeling with only a single input of foam parameters at a referenced permeability. A 2D layered reservoir case was used to test the new permeability functions, which shows the significant difference in terms of the oil recovery and the injector BHP between whether considering the permeability effect or not. This paper proposed, for the first time, a systematic methodology to account for the critical permeability effect to simulate foam flooding in heterogeneous reservoirs. This is a key advance in consideration of the major limitation of existing reservoir simulators using fixed or ad-hoc foam-model parameters throughout the entire reservoir. This new model enables the reservoir engineers to simulate and optimize the foam performance in real fields with better accuracy of foam physics in porous media.

No abstract

Recovering oil from oil-wet matrix in fractured carbonate rocks is highly challenging. Recent experiments have indicated that ultra-low-interfacial-tension (ULIFT) foam flood could significantly boost the oil recovery from such rocks. However, there is limited information available about the foam and the microemulsion transport in the fractured system to extract the oil from low permeability matrix. Adaptation of this technology in the field would not be possible without a good understanding of the process. The aim of this work is to model and history match the ULIFT foam flood in fractured carbonate cores for further gaining insight into the complex four-phase flow. The model was set up based on a group of experiments using cores split lengthwise to simulate axially confined fractures. Pre-generated foam was tested in this system due to the lack of in-situ generation of foam in the straight fracture at the core scale. Various foam coalescence mechanisms, with/without oil, were modeled, and a dynamic-texture population-balance foam model was developed for this purpose. Our model incorporates the effects of oil and permeability as well as the coexistence of foam and microemulsion on the foam apparent viscosity. The model is able to reasonably well history match both the oil recoveries and the total pressure drops of the ULIFT foam floods in fractured carbonate cores. More impressively, the modeling results agree very well with the pressure gradient of each section of the core, indicating that the spatial variation and distribution of the foam texture are largely captured. The simulation results also show that the pre-generated foam greatly resists the fluid flow in the fracture close to the injector side and enhances the diversion of injected fluids into the matrix layers, leading to improved oil displacement. The resulting oil crossflow from the matrix to the fracture destabilizes the foam at the foam front thereby slowing the transportation of foam in the fracture. Additional case studies suggest that significantly more oil can be recovered if the foam destabilization by oil could be reduced/mitigated. Test results disclosed in this paper demonstrate for the first time the successful modeling and history-match of ULIFT foam floods in fractured rocks. Valuable insight into this complex process has been gained through this innovative research. This is of great value with respect to the further optimization of the corefloods, the design of the surfactant formulation, and the feasibility of applying this new technology to the field scale.

Among the many parameters needed to optimize a polymer flood is the choice of polymer viscosity, mobility ratio and polymer slug size that should be injected to maximize oil recovery. In this paper, a new polymer flooding experimental study is addressed to answer two questions. Firstly, considering a given crude oil, what optimal polymer solution viscosity should be injected? And secondly how much polymer solution should be injected during the polymer flood to maximise recovery? Experiments were carried out using 1D homogeneous Bentheimer cores of similar properties. The cores were oil flooded using crude oil (µo = 120cP at T=60°C) and aged to obtain intermediate wet conditions. The polymer was a partially hydrolysed polyacrylamide (HPAM) dissolved in a moderate salinity brine. The polymer solutions were prepared at different concentrations (from 1500ppm to 3000ppm) to cover a large range of viscosity ratio (Rμ=μoμp from 2 to 18) which correspond to end-point mobility ratios of 0.5 and 5.4, respectively. Corefloods results show as expected, that the polymer is more efficient in terms of oil recovery when viscosity ratio is low. In line with polymer flooding theory, we observed at intermediate wettability conditions, that a maximum oil recovery is reached at M =1 (Rµ = 5) and that oil recovery did not increase when reducing the ratio to M = 0.5 (Rµ= 2). However, when considering aspects such as polymer mass required, injectivity concerns and flow stability, we observe two favorable conditions, corresponding Rµ= 5 and Rµ = 10, for a mobility ratios of 1 and 7, respectively. Different polymer slug sizes were injected in the cores at the above conditions (Rµ = 5 and Rµ = 10) followed by water flooding (chase water). Both injections (polymer and water injection) were carried out at same flow rate to minimize miscible viscous fingering at the rear of the polymer slug. Results show that an optimal polymer slug size exists for which one can obtain the same microscopic oil recovery than that of continuous polymer injections at Rµ=5 and Rµ=10, an important finding that can impact the economic viability of the process. In conclusion, our experimental study shows that at 1D scale, optimal values of viscosity ratio, polymer slug size and polymer mass injected lead to the same maximum oil recovery obtained by continuous polymer injection. A necessary starting point before upscaling a polymer flood and studying the impact of heterogeneities.