Sucker Rod Pumping System: Challenges To Develop Diagnostic System And Role Of Dynamic Simulator

Abstract

Relevance. Production benefits and efficiency of sucker rod pump installations is highly dependent on the accuracy of monitoring and fault diagnosis system used. When sucker rod pump is operated at faulty working states, the rate of equipment failure increases and the production efficiency decreases. Moreover, since the sucker rod pump operates deep in the underground, maintenance cost is more, and production is interrupted for longer time. Hence, improving monitoring and diagnostic system for sucker rod pump operation has become very important. The information about sucker rod pump working state is embodied in the dynamometer card and motor power curve. Monitoring sucker rod pump using motor power curve is more advantageous than dynamometer card. It can be used to monitor both surface and subsurface equipment. Moreover, motor power curve is obtained using more reliable current and voltage measurements. Therefore, the motor power curve provides a better alternative evidence for development of monitoring and diagnostic systems for sucker rod pumps. The main aim of the research is to ease the challenges those impede the promotion of diagnostic models using motor power curve. Objects: electrical drive, sucker rod pumping unit, oil producing well. Methods: sucker rod pump simulation model; feature extraction method that produces a feature vector to uniquely represent each working state; diagnostic method based on support vector machine. Results. 72 labeled motor power curves representing six working states namely: normal working state, travelling valve leakage, gas affected, insufficient liquid supply, condition when the plunger hits top dead center and bottom dead center, are generated. It was seen that the feature vector constructed based on the valve working points and energy consumption represent uniquely each working state. It was also seen that the support vector machine classifier correctly classifies the samples for normal, travelling valve leakage, gas affected working states. However, some samples of insufficient liquid supply were misclassified as gas affected and normal. © 2022 Tomsk Polytechnic University, Publishing House. All rights reserved.