Improving Performance of the Phase Equilibrium Calculations in the Surface Network Portion of an Integrated Reservoir Simulator/Surface Network Compositional Model

Abstract

A network model can be extremely sophisticated for some miscible gas injection simulations. Sometimes, miscible gas injection, gas cap reinjection, and gas lift within the network are required to be solved simultaneously with a reservoir model. The phase equilibrium calculation for a network can consume a significant proportion of the total central processing unit (CPU) time. Equation of state (EOS) -based phase equilibrium can be divided into single-phase and two-phase calculations. Most research focuses on reducing the number and computational cost of EOS equilibrium calculations for reservoir modeling. Improving the performance of the phase equilibrium calculation for network flow is rarely discussed in the literature. This paper describes methods applied to speed up the network calculations. The entire network is represented as a series of nodes and connections between nodes. It can be divided into several segments. If there is no introduction of other fluids from connecting segments, the overall composition is constant within a particular segment. Based on information from the previous phase status of each node or current phase status of up/downstream nodes, the nodes with a phase status change can be identified by regular single-phase stability analysis and flash calculation methods. Therefore, the segments can be divided into several single gas/oil and two-phase regions. For the nodes within two-phase regions, some correlation models can be applied to calculate equilibrium constant K-values for each component. Then, the flash calculation can be performed using the Rachford-Rice procedure. For single-phase regions, a phase envelope generation approach is used. A saturation pressure calculation is performed at a low value of pressure and temperature. Then, several points on the rest of the curve are extrapolated sequentially in a multistage extrapolation. If the temperature of a node is between two temperatures where an approximate value of the saturation pressure was identified, interpolation can be used to estimate the saturation pressure for this node. A benchmark of example problems demonstrates that this approach can significantly speed up both single-phase stability tests and two-phase flash calculation with little accuracy loss. Phase equilibrium calculation is usually intensive in terms of CPU time. This paper designs a multistep procedure to identify the phase status along pipes. The improvement of computational efficiency in the phase equilibrium calculations within the surface network flow simulation can reduce the overall simulation CPU time and make it practical to run a full-field miscible gas injection model.

A simple approximate density-based phase envelope construction method

“…Gu et al [34]), using a very simple scaling law for equilibrium constants. Some limitations of the approximate method are mentioned, such as mixtures with several critical points, mixtures with no critical point and some complex mixtures for which the approximation of equilibrium constants is inadequate.…”