Three-Phase Numerical Solution for Jet Pumps Applied to a Large Oilfield

Robert Merrill,Vivek Shankar,Thomas Chapman
DOI: https://doi.org/10.2118/202928-ms
2020-11-09
Abstract:Abstract The Mangala oil field in Barmer, Rajasthan is one of the largest onshore fields in India. The field is under an active polymer flood with over 300 wells injecting >400 kbwpd. Jet pumps have been the main mode of artificial lift for the field since production commenced in 2009. The jet pumps are "reverse circulation" type with the power fluid (water), pumped down the annulus. Despite their low efficiency, jet pumps are ideal for Mangala because heated water is used for the circulation fluid which avoids wax deposition in well tubulars. Simple performance curves, often used in commercial packages, work well under two phase liquid only or low GOR conditions. With high gas volume fractions and higher drawdowns, the additional complexity of critical flow is introduced; this occurs when the fluid velocity reaches the speed of sound within the jet pump. These conditions are often encountered in the Mangala pumps, and many jet pump solutions become grossly inaccurate when used for these conditions. A field-wide network model is being used for production optimization; the model employs coupled power fluid and production network systems. Given the size of the power fluid network, an accurate and fast method for jet pump performance under three phase conditions is a prerequisite to the modelling of both well behavior and jet pump sizing. This paper presents a full solution of the fundamental jet pump equations, which are based on Bernoulli’s principle and the conservation of momentum. The method follows the classic solution presented by Cunningham (1970, 1995), which made simplifying assumptions of perfectly incompressible liquids and ideal gases. The proposed technique eliminates these simplifications, making the solution applicable for the situations commonly encountered in the Mangala field. Existing jet pump models for Mangala are based on spreadsheet solutions. These solutions are slow and are practical only for the sizing of individual jet pumps. The new model presented in this paper solves the fundamental equations using numerical techniques and properly accounts for high compressibility fluids, multiphase flow and critical flow. The model was written in Python, which is an object-oriented language well suited to modeling the individual components of the system. The model solves the integrals involved in the revised Cunningham equations, using Newton-Raphson techniques to solve the equations iteratively. The model is used to generate complete (and complex) VLP curves in standard formats which can then be used directly with network models. The new model executes 25-100 times faster than the previous spreadsheet model.
What problem does this paper attempt to address?