Increasing the Computational Speed of Flash Calculations With Applications for Compositional, Transient Simulations

Summary

Approaches are presented for reducing the computation time spent on flash calculations in compositional, transient simulations. In a conventional flash calculation, the majority of the simulation time is spent on stability analysis, even for systems far into the single-phase region. A criterion has been implemented for deciding when it is justified to bypass the stability analysis. With the implementation of the developed time-saving initiatives, it has been shown for a number of compositional, transient pipeline simulations that a reduction of the computation time spent on flash calculations by approximately 85 to 90% can be achieved.

Introduction

Modeling of oil production requires simulations of transient flow. Examples are miscible-gas displacement in petroleum reservoirs and multiphase flow in pipelines, including slug formation and startup scenarios. In a typical transient-flow simulation, the system is discretized into a number of cells or sections. Phase amounts and phase properties are needed for each cell or section to solve the conservation equations in the model, and specifically for transient pipeline simulations to calculate such properties as heat loss to the surroundings, liquid holdup, and pressure drop. If the overall composition is constant during the simulation, the phase properties can be stored in precalculated tables listing the needed properties as a function of pressure and temperature (Bendiksen et al. 1991; Xu and Shea 2001). This is referred to in the following sections as a noncompositional, table-based simulation.

When simulating miscible-gas displacement in reservoirs, the assumption of a constant overall composition is not adequate because the injection gas will dissolve in the reservoir fluid and vice versa. Similarly, in many typical multiphase pipelines, the fluid composition will vary because of velocity differences between phases, interfacial mass transfer, and merging networks. In these situations, a compositional approach is useful. A compositional model has the drawback that the computation time is much higher than that of a noncompositional table-based approach. The phase amounts and properties must be evaluated in each cell or section in each timestep. Furthermore, phase compositions are required to calculate the interfacial mass transfer. Nevertheless, the increased accuracy in the fluid description makes the compositional approach attractive to represent multiphase pipelines and reservoirs with large compositional variations.

Flash Equilibrium Calculations in Compositional, Transient Simulations

The physical properties of a fluid in a cell or section depend on whether the fluid is present as a single phase or splits into several equilibrium phases. A flash calculation is therefore required in each timestep to determine the number of equilibrium phases and their amounts and compositions. Even with relatively few mixture components, the computation time of a compositional, transient simulation far exceeds that of a corresponding noncompositional, table-based simulation; furthermore, the computation time increases with an increased number of components.