MODELING MULTIPHASE FLOW IN DOWNHOLE VALVES

Oil and gas have been produced from onshore and offshore fields for more than 100 years, and production rates are falling for the most easily accessible fields. New oil fields are found at deeper and more remote areas, and exploration costs are increasing. A new tool that can enhance the well testing process is a wireline operated downhole shut-in valve. In order to get as precise as possible results from the well testing, the two-phase pressure drop across the shut-in valve must be known. The flow path through this shut-in valve is however complex and cannot easily be compared to standard tubing parts and singularities. Frictional pressure losses in pipes are well understood now and have been studied by a number of authors since the late forties. Minor pressure losses arise from singularities like bends, contractions, expansions etc. Two-phase flow minor losses have also been studied by many authors, but only for well-defined and common shapes like bends, nozzles, sharp edged contractions etc. In a typical industrial application like the shut-in valve, the flow path is complex. More research is therefore needed in order to be able to predict the two-phase pressure drop in a complex flow path. The modeling methods developed here should hopefully be applicable to other two-phase flow systems as well. The main objective for this work is therefore to find methods for modeling two-phase flow in complex geometries with several singularities and changes of cross section. The work has included design, construction and instrumentation of a full scale shut-in valve mock-up. A series of experiments have been performed with two-phase flow of air, water and two different oil types. This provides a valuable experimental data base for two-phase flow in a typical downhole valve. Furthermore an in-house simulation tool for 1-D models was implement, verified and validated. The first achievement in this thesis is the validation of 3-dimensional computational fluid dynamics (CFD) simulations of single-phase flow in the valve. Provided that the mesh is properly designed etc. the deviation in pressure drop is only 3-6\% compared to experimental data. The next achievement is the 1-D modeling of the flow in the valve. This 1-D model serves as a necessary basis for the two-phase simulations. The main achievement is the implementation of two-phase flow in the 1-D model. Two approaches are used. First classical flow pattern independent correlations are applied, and then the state-of-the-art Unified Comprehensive Model formulation is introduced. The latter provides the best results with only some 10\% deviation in pressure drop.