Experimental and Numerical Studies of Gas/Oil Multicontact Miscible Displacements in Homogeneous Porous Media

Abstract

Abstract Both vaporizing and condensing miscible gas floods are being conducted in a large number of reservoirs worldwide. The performance of these gas floods is usually determined via a combination of laboratory analysis and compositional simulation. However the ability of numerical simulation to correctly predict the progress of a multi-contact miscible displacement has not been fully verified. This paper investigatesthe physical processes occurring during multi-contact miscible displacementthe ability of the Coats correlation to model the changes in gas-oil relative permeabilities with interfacial tension (IFT) and the associated inaccuracies in predicted oil recoverythe importance of gas/oil disequilibrium on the prediction of oil recovery and gas-oil ratio These objectives are achieved by using a commercial compositional simulator to predict the behaviour of multicontact miscible (MCM) floods in well-characterised, beadpack experiments. Both condensing and vaporizing drives are investigated using a ternary liquid system that exhibits an upper critical point at ambient conditions. Relative permeabilities as a function of IFT were obtained independently from displacement experiments. Glass beadpacks permitted the visualisation of the displacements whilst the effluent profiles in terms of composition and phase volume versus time were measured using a calibration with refractive index of the fluids. Condensing and vaporizing gas-drive displacements were very efficient with about 90% of the oil in place recovered at one pore volume injected. Compared to immiscible displacements, breakthrough and total recoveries were increased by 8% and 20% respectively. However, in both condensing and vaporising drives, the compositional simulation (that assumes equilibrium conditions) over-predicted oil recovery. Further simulations identified that these errors are due to inaccuracies in Coats' correlation describing the behaviour of relative permeabilities as miscibility is approached and/or the fact that the produced fluids were not in compositional equilibrium. Introduction Miscible gas injection is being performed in a large number of reservoirs worldwide with the aim of improving oil recovery1. In most cases the injected gas is not initially miscible with the oil: miscibility is achieved by the net transfer of components from the oil into the gas (a vaporizing gas drive) or from the gas to the oil (a condensing gas drive). Compositional numerical simulation is usually used to predict the performance of these recovery schemes on the basis of equation of state properties determined from regression on data obtained from laboratory experiments. The accuracy of these predictions is critically dependent upon the validity of the assumptions used in these simulations. The most critical of these assumptions is that of instantaneous local thermodynamic equilibrium in each grid block. However, whilst the ability of numerical simulation to predict first contact miscible and immiscible displacements has been evaluated by comparison with experiment in a number of studies2–7, there are very few studies investigating the compositional simulation of multi-contact miscible experiments8–11. Moreover there are virtually no studies which use the simulator to predict both phase and flow behaviour directly from the experimental properties. Most studies resort to some degree of history matching either to model the phase behaviour of multi-component hydrocarbons through the use of pseudo-components8–11 or to model the influence of unknown heterogeneities in the cores or rock slabs8–11.