Few, if any, black oil-type correlations are available for describing the PVT behaviour of Gas Condensates and Volatile Oils. Such near-critical fluids are generally modelled using multi-component Equations of State (EoS). However, EoS require detailed laboratory analyses to be performed on actual reservoir fluids. For modelling production wells, this data is often of poor quality or not available at all.
In a production system, there is usually a regular measurement of the post- separator condensate and gas gravities (stock tank oil and gas) and the total Condensate-Gas-Ratio (CGR). In this study, this data was used to generate a four pseudo-component description that was then used with a conventional EoS to make predictions of fluid behaviour.
The basis for the selection of the four pseudo-components is discussed as is the calculation of the component mole fractions using the two stock tank gravities, the CGR and the additional constraint of the mole fractions summing to unity.
The EoS model was tuned to data obtained from a North Sea Gas Condensate well. The tuned model was then used with an appropriate flow correlation to successfully predict top hole conditions, having specified measured bottom hole conditions.
Considerable effort has been devoted to the prediction of the flow of oil and gas mixtures in wellbores. A summary of this research can be found in Beggs. An essential element required for the accurate modelling of these systems is the prediction of the fluid behaviour, especially over the wide range of pressures and temperatures encountered.
Traditionally, gas and oil mixtures have been modelled with correlations, which consider the fluid as a two-component or blackoil model: the components, or more strictly pseudo-components, are stock tank gas and stock tank oil. Gas Condensates and Volatile Oils, being near-critical hydrocarbon mixtures, are not ameanable to analysis by blackoil-type correlations and are usually modelled by multi-component Equations of State (EoS).
A variety of EoS are used, of which by far the most common are the so- called cubic van-derWaals EoS, such as those due to Soave-Redlich-Kwong (SRK) and Peng-Robinson (PR). Generally, these models cannot be used in a purely predictive way, even when modified by the volume translation approach of Peneloux et al., and therefore require that representative samples of the fluid be taken and subjected to a suite of laboratory analyses. Sampling is a costly process that is prone to systematic error.