Development of Fault Diagnosis Model for Sucker Rod Pump Using Motor Power Curve

A real time diagnostic system development for detecting sucker rod pump working conditions is current demand for most oil producers because of significant expenses related to costs of labor and consequent damages when a fault is not identified timely. In this paper the main objective is to develop a fault diagnostic model that helps to solve the existing diagnostic problems in sucker rod pump installations. The main source of data which relays on the use of unreliable sensors is the main cause to the existing difficulties. However, motor power curve is reliable source and has a potential to improve the efficiency of sucker rod pump diagnosis. In this research, sucker rod pump simulating model generates motor power curves considering the influences of possible scenarios in the subsurface on the subsurface boundary condition. The generated motor power curves are used to train support vector machine (SVM) and for testing. The results of the simulation indicate satisfactory performance of SVMs in solving the diagnostic problem of sucker rod pumps.