A simple equilibrium chemical model is presented for continuous, linear, alkaline waterflooding of acid oils. The unique feature of the theory is that the chemistry of the acid hydrolysis to produce surfactants is included, but only for a single acid species. The in-situ produced surfactant is presumed to alter the oil/water fractional flow curves depending on its local concentration. Alkali adsorption lag is accounted for by base ion exchange with the reservoir rock. The effect of varying acid number, mobility ratio, and injected pH is investigated for secondary and tertiary alkaline flooding. Since the surface-active agent is produced in-situ, a continuous alkaline flood behaves similar to a displacement with a surfactant pulse. This surfactant-pulse behavior strands otherwise mobile oil. It also leads to delayed and reduced enhanced oil recovery for adverse mobility ratios, especially in the tertiary mode. Caustic ion exchange significantly delays enhanced oil production at low injected pH. New, experimental tertiary caustic displacements are presented for Ranger-zone oil in Wilmington sands. Tertiary oil recovery is observed once mobility control is established. Qualitative agreement is found between the chemical displacement model and the experimental displacement results. Introduction Use of alkaline agents to enhance oil recovery has considerable economic impetus. Hence, significant effort has been directed toward understanding and applying the process. To date, however, little progress has been made toward quantifying the alkaline flooding technique with a chemical displacement model. Part of the reason why simulation models have not been forthcoming for alkali recovery schemes is the wide divergence of opinion on the governing principles. Currently, there are at least eight postulated recovery mechanisms. As classified by Johnson and Radke and Somerton, these include emulsification with entrainment, emulsification with entrapment, emulsification (i.e., spontaneous or shear induced) with coalescence, wettability reversal (i.e., oil-wet to water-wet or water-wet to oil-wet), wettability gradients, oil-phase swelling (i.e., from water-in-oil emulsions), disruption of rigid films, and low interfacial tensions. The contradictions among these mechanisms apparently reside in the chemical sensitivity of the crude oil and the reservoir rock to reaction with hydroxide. Different crude oils in different reservoir rock can lead to widely disparate behavior upon contact with alkali under varying environments such as temperature, salinity, hardness concentration, and pH. The alkaline process remains one of the most complicated and least understood. It is not surprising that there is no consensus on how to design a high-pH flood for successful oil recovery. One theme, however, does unify all present understanding. The crude oil must contain acidic components, so that a finite acid number (i.e., the milligrams of potassium hydroxide required to neutralize 1 gram of oil) is necessary. Acid species in the oil react with hydroxide to produce salts, which must be surface active. It is not alkali per se that enhances oil recovery, but rather the hydrolyzed surfactant products. Therefore, a high acid number is not a sufficient recovery criterion, because not all the hydrolyzed acid species will be interfacially active. That acid crude oils can produce surfactants upon contact with alkali is well documented. The alkali technique must be distinguished from all others by the fundamental basis that the chemicals promoting oil recovery are generated in situ by saponification. SPEJ P. 245^

Summary Alkaline waterflooding is complicated because the surfactant species are generated in sits from acidic components in the crude oil rather than injected externally. We previously outlined an equilibrium displacement theory that captured the essential features of the process. Assumption of local equilibrium, however, does not allow for a careful examination of the complex transport and kinetic phenomena that occur. Our non-equilibrium theory models the pertinent mass-transfer and kinetic resistances affecting interfacial tensions (IFT's) during displacement and oil-recovery efficiency. The model shows that when natural acid in a trapped oil blob contacts alkali, it diffuses to the interface, adsorbs, reacts with alkali, desorbs, and convects into the bulk aqueous phase. We quantify these transport steps, calculate concentration profiles in the oil and flowing aqueous phases during a linear displacement, and determine how transient tension behaves as a function of time and distance. Mass-transfer resistances are insignificant, but sorption resistances at the oil/water interface affect the transient evolution of IFT during alkaline displacement. Interfacial sorption barriers are modeled with first-order kinetics. Desorption resistances may be quite large, so low IFT's may not be established in typical laboratory-scale cores. Reversible or irreversible surfactant adsorption on reservoir rock is detrimental to alkaline water-flooding. Rock adsorption reduces the amount of surfactant available to the oil/water interface and may raise the IFT to levels that are ineffective in mobilizing residual oil globules. Most importantly, we provide a conceptual framework to design a successful alkaline waterflood. Introduction In the caustic oil recovery process, alkali contacts crude oil containing natural acids. Once the oil is bathed in alkali, the indigenous acids migrate to the oil/water interface and react with the alkali to produce organic salts, some of which are water-soluble and surface-active. The IFT may be reduced to a level that permits the mobilization of residual oil globules. Because many, if not most, acidic oils are viscous, significant mass-transfer resistances may occur in this extraction process. Diffusion resistances, however, are not the only, nor the most important, resistances in the alkaline recovery scheme. Recently, Rubin and Radke argued that the dynamic-tension minima for acidic oils in alkali are caused by significant sorption barriers at the oil/water interface. They present both theoretical and experimental evidence for this assertion. Unfortunately, these concepts are not directly applicable to the alkaline displacement process. This paper shows the application of their ideas to alkaline flooding of acidic crude oils. We formulate a quantitative description of linear, tertiary, alkaline waterflooding, including all relevant mass-transfer and interfacial-sorption resistances. The proposed chromatographic model permits the calculation of proposed chromatographic model permits the calculation of dynamic IFT's and, hence, of critical capillary numbers for displacement of tertiary oil globules as a function of time and distance on continual injection of an alkaline solution. Essentially, the theory is one of nonequilibrium oil displacement. Therefore, we scale the alkaline displacement process-that is, our calculations permit an estimate of the process-that is, our calculations permit an estimate of the time and distance scales necessary to achieve local equilibrium. If times and distances are long enough, local equilibrium is attained. Our model must reduce to its equilibrium subcase, which is outlined in Ref. 1. Our objective is to investigate the transient phenomena and, in particular, the transient tensions and accompanying transient capillary numbers that occur during alkaline displacement of oil. Before detailing the theory, we briefly outline the transport steps that are thought to occur during alkaline flooding.

A series of laboratory tests was conducted to investigate the in-situ steamdrive process and its effects on fluid displacement in porous media. The experiments show that in-situ hot-water flash may recover up to 60% of moveable oil and that the residual oil saturation after flash increases with lower initial water saturations. Conventional simulation with external-drive relative permeabilities led to under-prediction of oil recovery. IntroductionCyclic steaming through hydraulic or natural fractures is often used for oil recovery from low-permeability reservoirs (e.g., diatomite). The efficiency of the process is controlled by fluid and heat interactions between fracture and matrix. During injection and soak, steam condenses in the fracture and hot water imbibes into the matrix. 1 During production, water pressure in the hot matrix often drops to less than the vapor pressure, causing in-situ boiling. In-situ boiling of hot water also occurs in "flash-driven" steamfloods in nonfractured reservoirs. 2 The displacement process associated with waterto-steam phase change in porous media is called "in-situ steamdrive." The in-situ process is different than the external steamdrive process in which the displacement occurs by steam injection.The physics of water vaporization in porous media has been the subject of a number of papers for application in geothermal reservoirs, nuclear waste disposal and thermal oil recovery. Thermodynamic properties of vapor and liquid in porous media are different than "flat-surface" conditions. The degree of superheat and vaporpressure lowering is greater for smaller pore radii. 3 The water retained by surface adsorption when the pore is filling (for pore radii in the range of 2-35Å) and by capillary condensation (up to 120Å) is known to be the main reason for vapor-pressure lowering in porous media. 4-8 Recent work on adsorption/desorption of water with network models shows that access of the vapor phase to large pores is often blocked by small pores during the desorption process. 9 Comparison of water-vapor desorption and conventional capillary pressure measurements at low-wetting-phase saturation shows that adsorbed films can be a significant part of the liquid saturation. [10][11] Currently, cyclic-steam performance in low-permeability, heavyoil reservoirs is modeled with conventional thermal compositional simulators with flat-surface properties. Relative permeabilities are commonly measured by methods where gas is injected "externally." This may result in poor prediction if the relative permeabilities representing the flow process are different from those in conventional steamdrive. Subsequent errors in recovery prediction of cyclicsteam processes in fractured low-permeability reservoirs have not been quantified in the literature before.Because the source of water vapor in the in-situ steamdrive process is the formation water, the residual oil saturation after the flash process may be affected by the initial water saturation, its distribution, and the rate of pressure decline. Unlike ...

TX 75083-3836, U.S.A., fax 01-214-952-9435.Abstract Laboratory evaluation of watertlood behavior of vugular carbonates is subject to significant uncertainty. Even basic parameters such as residual oil saturation are difficult to estimate and depend on experimental conditions. We use X-ray Computed Tomography (CT) scanning to characterize the porosity, permeability, and waterflood displacement behavior in rocks from two carbonate settings. We find that the porosity and permeability distribution are dissimilar. This causes the dependence of their residual oil saturation on water flow rate to be very different. We attempt conventional methods to extract meaningful information from the data but find them inadequate.