Abstract:

For the high CO2 condensate gas reservoir in Libra block, the fluid phase behavior in the reservoir is very complex due to the injection of CO2 in the later stage of reservoir formation, which brings great challenges to the prediction of oil ring volume. Through phase equilibrium calculation model, component gradient distribution theory, and visual injection experiment, this paper describes the oil-gas phase behavior of the reservoir during CO2 injection, reveals the dynamic change of oil ring volume, and puts forward a new method for predicting oil ring volume of secondary condensate gas reservoir with high CO2 based on gas cap gas component fitting. The following research results are obtained. 1) The fluid phase behavior and volume change of oil ring under CO2 injection can be divided into four stages. At the beginning of CO2 injection, CO2 dissolves in the oil and the volume of oil ring expands. In the early stage, CO2 dissolves, replaces and extracts the light components in the oil, so that the volume of the oil ring decreases rapidly. In the middle stage, CO2 mainly extracts the light and medium components of oil, so the volume of the oil ring decreases slowly. At the later stage, the mass transfer of CO2-oil almost stops, so the proportion of oil ring volume further decreases due to the compression effect. 2) The strong fluidity of light components makes the gas composition of gas cap uniform, and the gravity differentiation makes the oil ring composition gradient change longitudinally. 3) On this basis, the strong correlation between gas cap gas composition, gas-oil ratio and oil ring volume are revealed. 4) The oil ring volumes ratio predicted by the new gas cap gas fitting method and field exploration at different well depths are 19.21% and 22.30%, respectively, which are in good agreement with the oil ring volume ratio (20.60%) obtained from the CO2 injecting visualization experiment.

Key words: gas reservoir, mass transfer between CO2 and oil, fluid phase behavior, oil ring volume prediction, component gradient distribution, visual extraction experiment