Compositional simulation is widely used in the petroleum industry to predict the phase behaviour of reservoir fluids within the reservoir, flow lines and process facilities. Computational time is an important factor in compositional simulation where CPU time exponentially increases with the number of components. It has been shown that by omitting binary interaction parameters (BIPs) from the mixing rules of the equation of state (EOS), phase split calculations can be carried out much faster than when using conventional methods, when a large number of components is used to describe the reservoir fluid. However, using the rapid flash calculation method, the phase behaviour calculation may fail to converge, particularly for gas condensate systems. Omitting BIPs may also deteriorate the predictive capability of the EOS for some fluids. In this paper, the original rapid flash calculation method is modified to improve its convergence for gas condensate systems. To compensate for the omitted binary interaction parameters, the temperature dependency of the attractive term (a) in EOS has been modified to improve its phase behaviour prediction over a wide range of temperatures. A number of binary systems were used to determine the a function for supercritical compounds. This improved the EOS predictions for systems with high concentrations of supercritical compounds. The objective is to perform the phase equilibrium calculation as rapidly as possible with a large number of components in the reservoir without compromising on accuracy of predictions. Then, using the same number of components, detailed compositional analysis can be used to design efficient process facilities.