Simulation of Gas Injection Processes in Gas-Condensate Reservoirs Using a Binary Pseudo-Component Representation

Abstract

Abstract A method is presented for simulating gas injection processes in gas-condensate reservoirs below the processes in gas-condensate reservoirs below the dewpoint using a binary pseudo-component representation. The fluid phase behaviour (including mixing with injection gas) Is first modelled with an equation of state using a large number of components. A lumping scheme is then found that allows a projection into a binary pseudo-component subspace which minimises the pseudo-component subspace which minimises the variation in the mass fraction of each pseudo-component in the vapor phase during liquid pseudo-component in the vapor phase during liquid re-vapourization. These two pseudo-components are then used in the PVT representation. The manner in which inverse lumping is performed to obtain separator products is described. The third SPE comparative solution project is used to illustrate the application of the method with a conventional black oil simulator, however the procedure is equally applicable to limited compositional or two component K-value simulators. Introduction A substantial amount of work has been carried out on the development of multicomponent compositional simulators for use in modelling reservoir processes such as miscible and multiple contact miscible gas flooding of oil reservoirs and cycling of gas-condensate reservoirs below the dewpoint. In these cases it is necessary to take account of the variation of fluid properties not only with pressure, but also with composition. Nowadays the pressure, but also with composition. Nowadays the thermodynamic constraints which describe the phase equilibria, and which must be satisfied simultaneously with the solution of the component transport equations, are usually formulated with cubic equations of state (EOS). Although it is possible to obtain a good description of the phase possible to obtain a good description of the phase behaviour of a fluid by matching experimental data with a multicomponent characterisation, this does not necessarily translate to an accurate model of reservoir compositional processes. There are two main reasons for this. Firstly, it is necessary to reduce the number of components in the characterization by forming pseudo-components (lumping components together). This arises because of the computational demands created by solving for a large number of transport equations coupled with flash calculations in each grid block. Secondly, compositional simulators are prone to high levels of numerical dispersion. This problem is exacerbated when a larger number of components are used. Methods have been developed to reduce numerical dispersion but these are not commonly implemented in commercial three dimensional simulators. For field scale simulations of injection processes it is therefore usual to use some form of limited compositional, or extended black oil, simulator. However, unless the formulation has been based on decoupling the PVT properties of surface separator products from those of the reservoir fluids, it is products from those of the reservoir fluids, it is not possible to maintain a correct material balance in the simulation. This applies to all forms of black oil modelling where the two pseudo-components (or separator oil and gas) contain physical components in common. To try and minimise the material balance error authors have used partial densities, inversion of the 'flash' equations, and additional parameters to account for mass transfer between separator products. The first physically consistent binary pseudo-component PVT representation for use in (extended) conventional black oil simulators without code modifications was presented by Drohm and Goldthorp in 1987. Since presented by Drohm and Goldthorp in 1987. Since then Coats et al have used a two component K-value approach for decoupling reservoir and surface fluid properties, but this scheme still uses oil and gas properties, but this scheme still uses oil and gas as pseudo-components. P. 45