Summary. An experimental method is proposed for quantification of oil in effluent samples from displacement tests with liquids. On a mass basis. this new procedure is offered as an alternative to the classic volumetric approach. A known amount of a suitable organic solvent with good selective solubility and density contrast relative to the oil phase is thoroughly mixed with the effluent sample. After centrifugation, the oil/solvent mixture is analyzed for density. Subsequently, the oil mass fraction in the pseudobinary mixture (oil/solvent) is obtained through a previously determined correlation of this parameter vs. density. With these data-amount of solvent and oil mass fraction-plus the definition of the latter in a binary mixture, one arrives at the actual mass of oil in the effluent sample. The method is applied to a crude oil from Miranga field in the Reconcave basin (northeastern Brazil). Accuracy is thoroughly analyzed and compared with that of the volumetric approach. The method is also tested under such severe conditions as emulsified oil production. Finally, recommendations are made to implement the method in continuous oil-recovery monitoring, which would be a major step in the full automation of laboratory displacement tests. Introduction Displacement tests with liquids are conducted in the laboratory either for recovery studies or simply in relative permeability determinations. In both applications, the effluent from the displacement is most often made up of oil and water phases. Frequently, at least one of these phases is a fairly stable emulsion. Normally, material-balance calculations are carried out on a volumetric basis through the visual observation of the effluent sample previously submitted to centrifugation. Typical sampling flasks consist of 15-cm centrifuge tubes graduated in 0.1-cm increments (+/-0.05 cm uncertainty). Displacement tests representative of EOR experiments, for example, use plugs or series of plugs 3.8 cm [1.5 in.] in diameter and 30 to 50 cm [12 to 20 in.] in length with porosities in the range of 20 to 30%. Under these circumstances, total PV, s are on the order of 100 cm . On the other hand, cumulative fluid injection in these kinds of tests reaches up to 4 to 5 PV's. If sample size is limited to 5% PV, a complete run may yield 100 or more samples to be analyzed for oil and water. Regardless of the evaluation method applied, the numbers above call immediate attention to error accumulation and its effect on final oil recovery. An additional source of error specific of the volu-metric approach is the eventual emulsification of the effluent. Standard centrifugation is hardly effective in surfactant-type stabilized emulsions. In spite of its well-known limitations, the assessment of oil recovery on a volumetric basis is widely applied because of its simplic ty and straightforward execution. To the best of our knowledge, however, its accuracy has never been thoroughly investigated. The following have been proposed as alternative methods:oil/water on-line separators with selective wettability membranes 1;oil/water on-line acoustically monitored separators: andoil-saturation scanning devices that use microwave, ultra-high frequency (UHF), gamma ray, and X-ray attenuation and, more recently, computed tomography. While the first approach is still highly specific with respect to the oil/water system, the second has been tested with success on in-house-developed units and a commercial apparatus recently has been offered to the public- The third, although regarded by tome as the ultimate material-balance and saturation profile tool, requires fairly sophisticated equipment along with highly trained personnel. Our proposition is to present an inexpensive alternative to the classic volumetric approach to oil-recovery calculations in laboratory displacement tests. The new procedure, based on the solvent-extraction (SE) principle, is shown to be simple and yet more ac curate than the conventional volumetric approach. In parallel, the study includes the treatment of a specific problem very common to EOR experiments, that of emulsified oil production. Both the volumetric and the SE methods are tested with simulated effluents from water, alkaline, and surfactant flooding processes. The final analysis to demonstrate the effect of error prop-agation was conducted on a complete set of samples from an actua water/alkali displacement test. Experimental Procedure Modern electronic density meters rely on the change of the natural frequency of a hollow oscillator when filled with different liquids or gases. There is a linear relationship between the differences in the square of the change in natural frequencies and the densities of any two fluids separately filling the oscillator cell. One such apparatus, which can measure density with an accuracy of up to five significant figures, was used throughout this work. Other features of the instrument are the precise temperature control, short measuring intervals, and small sample sizes (- 1 cm). Furthermore, by means of a built-in processor, the instrument may be adjusted to read solute mass fraction in a binary mixture directly. The only requirement is a linear density/frequency behavior in the range of interest. These instrument features encouraged us to experiment with an SE technique and the corresponding correlation of density vs. oil-in-solvent mass fraction to quantify the mass of oil in effluent samples from displacement tests. The first requirement in attempting this task was to identify a proper solvent with enough contrast in density relative to crude oils in general to take advantage of the instrument's accuracy. Our choice was chloroform, with a listed density of 1.489 g/cm and water solubility of only 0.82 wt% (at 20 deg. C [68 deg. F]). The next step was to determine the correlation of density vs. reservoir oil mass fraction in the mixture with chloroform. Throughout this study, dead crude oil from Miranga field in the Reconcave basin (northeastern Brazil) was used. Its density was 0.8191 g/cm at reservoir temperature (60 deg. C [140 deg. F]). SPERE P. 1057^