Computational Fluid Dynamics (CFD) Model for Phase Separation at Branching Tee Junctions

Abstract

Abstract A 2-D CFD model has been developed to predict the phase separation in branching junctions for several air-water and steam-water experimental data sets. The inlet conditions of the flow are assumed to be homogeneous and dispersed. The model is based on individual phase mass and momentum balances and uses empirical correlations for the wall and interfacial drag. The only adjustable parameter is the diameter of the dispersed liquid droplets, a quantity that can be estimated from empirical correlations (but here was used as a fitting parameter). Overall, the CFD code does a reasonable job of matching experimental data, and fits the high pressure steam-water data better than the low pressure air-water systems. The code is limited by the 2-D implementation and its inability to predict the inlet flow regime; both limitations could be vastly improved with a 3-D implementation. Introduction The success of a thermal recovery process depends on efficient delivery of heat to the reservoir. In steamflooding, the majority of the energy is delivered through the latent heat of the two-phase steam, thus requiring that high quality steam be delivered to the reservoir sand face. The steam is created at the surface, typically exiting the generator in the 70–80% quality range, and is delivered to the individual wells through a distribution system. At every junction in the distribution system, there exists the possibility of phase separation; that is, the steam quality exiting the outlets is different from the entering steam quality. The separation is driven by the different inertia of the phases and the inherent ease in which either phase can accomplish a change in direction. Obviously, if this separation is severe enough, and repeated often enough, a very asymmetric distribution of heat may result, possibly degrading the overall performance of the project (some, however, claim that OVERALL, the efficiency is not compromised). The efficient partitioning of a multiphase system is important in several industries; in fact, an E&P analog to steam distribution is gas transportation where the associated liquid condensate may be unevenly distributed. Furthermore, the efficient distribution of steam is certainly not unique to the upstream E&P business. However, in most thermal projects, due to the competing costs of water purification and the desire to impart the highest energy steam to the reservoir, the steam generators are limited to producing in the 70–80% quality range. Many other industries, notably the power and especially the nuclear industry, typically deal with either low quality steam or superheated steam. In this sense, the distribution of high quality steam is a problem most frequently encountered in thermal E&P applications. The creation and distribution of steam is generally the highest expense associated with a steamflood. Phase separation can reduce the efficiency of the distribution. The ultimate goal of this study was not just to understand phase separation at junctions, but to develop techniques and products that minimize this phenomenon. Several such devices have been developed and there have been recent review articles dealing with their efficiency. However, if the problem can be understood (and predicted) at a fundamental level, perhaps even more comprehensive solutions can be developed, that may generalize to a whole host of steam problems (e.g., downhole steam distribution in vertical and horizontal injection wells). While most attempts to model phase separation have used either empirical or phenomenological approaches, we have attempted, with this Computational Fluid Dynamics (CFD) code, to develop a predictive capability from essentially first principles. Background In the absence of more sophisticated hardware for steam distribution, surface piping networks employ two primary types of junctions: branching tees and impacting tees. Figure 1 shows a schematic of these two devices and the nomenclature associated with each. Initially, branching tees have been more prevalent in steam distribution networks, but impacting tees show less tendency toward phase separation. P. 211