Advanced search
Start date
Betweenand

Water cut identification in multiphase flows using centrifugal pumps

Grant number: 19/08446-4
Support type:Scholarships in Brazil - Master
Effective date (Start): August 01, 2019
Effective date (End): May 31, 2021
Field of knowledge:Engineering - Mechanical Engineering - Transport Phenomena
Cooperation agreement: Equinor (former Statoil)
Principal Investigator:Alberto Luiz Serpa
Grantee:Matheus Paris Orsi
Home Institution: Faculdade de Engenharia Mecânica (FEM). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
Associated research grant:17/15736-3 - Engineering Research Centre in Reservoir and Production Management, AP.PCPE

Abstract

The suitable operation of electrical submersible centrifugal pumps in complex flows such asthose involving different liquids (water and oil) is important in the process of artificialelevation to ensure adequate process efficiency and to identify changes in the system behaviorto plan maintenance and suitable operating conditions to obtain good levels of reliability. Thismaster's project involves the use of systems identification techniques to predict, throughpump operating parameters such as rotation, electric current, torque, pressure variation andothers, the respective flow pattern identifying the water cut or, in other words, the percentageof water present on the flow. The centrifugal pumps used in this type of artificial elevationprocess have their characteristic curves determined by the manufacturer using water asworking fluid. In the case of flows that are more complex, involving water and oil, theoperating conditions of the pump change significantly and certain operating situationsconsidered inadequate should be avoided. The main difficulty of pumping these twoimmiscible liquids is the formation of emulsions, which increase the viscosity of the mixture,causing loss of pumping efficiency and impair the life of the equipment, as well as increasingthe power needed to supply the pump. To identify and predict this condition in multiphaseflows, the use of systems identification tools based on experimental data becomes relevant. Inthis master's project, we intend to explore these techniques using classical techniques basedon least squares data fitting, techniques based on neural networks and techniques based onthe most recently used machine-learning concepts (deep learning) to predict the conditions ofoperation of the centrifugal pump and thus adjust its operating conditions to avoid undesiredoperating situations. The main expected result of this master's project is to explore somemodeling tools applied to the case of the multiphase flow problem in order to identify thewater cut. The experimental results will be obtained from test rigs, available at LABPETRO ofUNICAMP, looking for an adequate validation of the techniques employed.