An Experimental and Modeling Study of Miscibility Relationship and Displacement Behavior for a Rich-Gas/Crude-Oil System

Abstract

Summary This paper presents an experimental study that uses many slim-tube displacements to characterize the effect of hydrocarbon solvent composition onmiscibility development. It was found that the minimum enrichment requirement can be estimated reliably with a previously proposed pseudocritical temperature, T, concept. previously proposed pseudocritical temperature, T, concept. Tuning equation-of-state (EOS) parameters to match these results required incorporating rate and recoveries from selected slim-tube displacement tests in addition to conventional PVT data. Introduction The operation of an enriched-gas injection project often involves blending one or more natural gas liquid (NGL) streams with dry gases to obtain an optimum solvent. The composition of such NGL's (ethane, propane, and some higher-carbon-number components) is subject to frequent change over the life of the flood because of source and supply availability. Therefore, a procedure is needed to estimate miscibility conditions between the potential injection solvents and the reservoir fluid. Benham et al. 1 developed a graphical technique where the minimum enrichment requirement can be estimated from the temperature, pressure, average molecular weight of intermediate hydrocarbons in the displacing fluid, and C5+ molecular weight of the reservoir oil. This approach, however, often results in minimum miscibility requirements that are too conservative. Rutherford conducted a series of displacements in sandpacks and found that miscibility development with a reservoir oil at constant pressure was a function only of the injected gas T . Rutherford studied a limited range of NGL fluids, consisting primarily of ethane and propanemixtures. His results were primarily of ethane and propane mixtures. His results were extended by Jacobson to systems containing CO2 and H2S.Rutherford's TPI method is used in this study to develop a miscibility correlation over a broad range of hydrocarbon species and enrichment levels. Accurate EOS oil recovery predictions require a fluid description with parameters tuned to match certain laboratory data. Typically, these experiments include such routine measurements as differential vaporization analysis (DVA)for black-oil properties and static PVT tests on selected mixtures. In some applications, the resulting fluid description may be adequate for predicting the performance of dynamic slim-tube and coreflood tests. Kremesec and performance of dynamic slim-tube and coreflood tests. Kremesec and Sebastian, however, showed that fluid descriptions based on static PVT data alone were inadequate for predicting the results of PVT data alone were inadequate for predicting the results of dynamic tests for a vaporizing CO2/Oil system. They concluded that matching the multiple-contact minimum miscibility pressure (MMP)can improve the accuracy of such predictions significantly. Negahban and Kremesec observed that this EOS tuning strategy can correctly identify miscible and immiscible regimes and slim-tube rates and recoveries over a wide range of displacement pressures. One objective of this study is to extend theseinvestigators' findings to rich-gas condensing drives by investigating whether fluid descriptions based on only static PVT tests can predict multiple-contact displacement behavior in these systems. A further objective is to determine the additional information required to improve fluid descriptions if the PVT-based descriptions are inadequate. Experimental Studies and Simulation Requirements To develop a miscibility correlation based on the T concept for the reservoir oil in this study, we conducted 25 slim-tube tests at 135F and 2,000psi using an apparatus previously described. Table 1 shows reservoir oil properties and NGL mixture compositions with as few as one hydrocarbon component (NGL 1) and with multiple hydrocarbon components and additional species (NGL's 2 through 5). These NGL's were diluted with methane to give injection solvents with enrichment levels ranging from 35 to 51 mol % C21. The hydrocarbon components in these solvents ranged from two to seven, and some contained up to 5 mol% CO2 or nitrogen. Sight-glass observations were recorded in all slim-tube tests to provide supporting information for the displacement mechanism. Slimtube tests to measure MMP also were conducted with Solvent A(Table 1) over a 1,650- to 3,000-psi pressure range. For fluid description development, DVA, saturation pressure, and p-x diagram measurements and constant-composition-expansion (CCE) tests were conducted with the reservoir oil and Solvent A. In addition, we compared actual and predicted displacement mechanisms using Tiffin et al.'s results from a coreflood experiment conducted with the same reservoir oil and Solvent A to verify the fluid description further. Amoco's version of the Redlich-Kwong EOS was used for fluid property and phase-equilibrium calculations. Slim-tube and property and phase-equilibrium calculations. Slim-tube and coreflood simulations were performed with a previously described generalized compositional model. Other details of the slim-tube and coreflood displacement modeling were also reported previously. previously. Results and Discussion Miscibility Correlation. Using Rutherford's method, we plotted slim-tube oilrecovery measured at 1.2 HCPV of solvent injection vs. solvent T for all NGL streams (Fig. 1). T is defined as the mole-weighted average of critical temperature T of all drive gas components corrected to CO2 and nitrogen T as discussed later. The general behavior in these experiments suggested that the data could be fitted reasonably well with two straight lines intersecting at T- 464R. We observed that displacement efficiency initially increased sharply with increasing T, followed by a gradual leveling off at higher enrichment levels. For solvents with T greater than 464R, slim-tube recoveries were highand sight-glass observations confirmed a miscible condensing-gas-drive mechanism. These tests were characterized by an early gas breakthrough, followed by a gradual color change in the two-phase transition zone. For solvents with T less than 464R, injected PV at gas breakthrough decreased and the transition zone became broader, resulting in lower oil recoveries. These displacements showed immiscible behavior despite a prolonged oil production period after 1.2 HCPV of solvent injection. Fig. 1 also shows data from all other supporting slim-tube experiments conducted with the same recombined reservoir oil. The results further substantiate the choice of TPC as a correlating parameter.

Deterioration of Performance of Mixing Rules in Phase Behavior Modeling of High-Density Reservoir Fluids

“…The minimum data set therefore required to tune a selected EOS would not only be the conventional PVT data, but also a well-characterized compositional analysis, multiple contact vapor-liquid phase equilibrium data, and swelling test (Kassack and Hagen, 1985;Mansoori et al, 1990). Other tests may also be required for modeling some specific processes as mentioned earlier.…”

Phase behaviour of heavy petroleum fractions in pure propane and n-butane and with methanol as co-solvent

“…For example, Hemptinne et al [9] have used the Cubic Plus Association (CPA) model for phase equilibrium calculations of petroleum containing systems. Fazlali et al [10] combined the Peng-Robinson equation of state and the equation derived by Mansoori et al [11] to predict the density of petroleum mixtures. Sun et al [12] used the Statistical Associating Fluid Theory-Variable Range (SAFT-VR) with the semi-continuous thermodynamics in modelling the phase behaviour of crude oils.…”