Abstract

Gas condensates are very complex reservoir fluids because of their high gas-oil ratio. Hence, a large number of components (about 13–18) are generally needed to properly characterize the condensation behavior. A procedure has been developed to reduce the number of components needed for reservoir simulation study. In this approach, a proportional adjusting factor is used between the molar average and the weight average weighting factor to match the laboratory PVT data, specifically the liquid drop out. Using this approach we were able to reduce to seven the number of pseudo-components needed to fully represent six gas condensate fluids for reservoir simulation. These seven pseudo-component models matched both the amount of laboratory measured condensate and the original fully compositional (12–13 components in this case) model in the reservoir studies. This paper describes the procedure and compares the results to fully compositional models.

Introduction

Compositional models are generally used for gas condensate reservoir studies to simulate depletion and cycling processes. Gas condensate is considered a very complex reservoir fluid because of its high gas-oil ratio(1). Hence a large number of components (13 to 18) are typically needed to properly characterize the condensation behavior. A procedure was developed to reduce the number of components needed for reservoir simulation study. This reduces significantly the amount of simulation time needed to run a compositional model for a gas condensate system.

In this study, the Peng-Robinson equation of state (PREOS) was used to characterize the gas condensate phase behavior. The characterization procedure consisted of two steps; first, the condensate was characterized with a fully compositional (FC) model made up of 11 to 12 pseudocomponents and second, the FC model was grouped into a lesser number of pseudo-components (six to seven). To validate the grouping procedure, data from a gas condensate field study was used. This field has six different productive reservoirs and thus six sets of measured gas condensate PVT data were available. To further check the grouping accuracy, the SPE's third comparative study on compositional reservoir models(2) was used. The actual reservoir simulation results are reported separately.

The measured PVT data, including gas compositions and residual oil composition during constant volume depletion tests and the amount of liquid drop-out, are used to verify the FC model results. In comparing the FC and the reduced pseudocomponent models, the reservoir pressure, produced gas-oil ratio, produced oil rate and cumulative oil produced are used to compare results between models.

EOS CHARACTERIZATION

The available PVT data consisted mostly of reservoir fluid analyses to C7+. Equation of state (EOS) characterization was done in two steps. First, the gas condensate was characterized using the measured compositions up to C6 and by splitting the heavier ends (C77+) into one to three components. This resulted in 11 to 12 pseudo-components in the FC models. The second step involved further grouping of the pseudo-components in the FC models into a lesser number of components (typically seven) and these models were called pseudo-component (P-C) models.

This content is only available via PDF.
You can access this article if you purchase or spend a download.