Research on Constant-Power Frequency Conversion System for Beam Pumping Units
Beam pumping unit is one of the most important mechanical equipment for oil production in on-shore oilfields.In recent years,a series of frequency conversion energy-saving technologies have been developed.However,the researches on beam pumping unit mainly focus on hardware and there are no fundamental theories,which result in low universality and inconsistency of beam pumping unit in field applications.In this paper,the mechanisms of frequency-variable measurement and control of beam pumping units were investigated,the main measurement and control components were optimized,the requirements for test data were proposed,and a frequen-cy-variable control system for beam pumping uniting was invented.A BP neural network system was constructed to identify operation anomalies,such as leakage,plunger hitting pumps and gas effects,and the frequency conver-sion calculation models were proposed for these different scenarios.Considering the negative torque and stroke limit of the pump,a constant-power frequency conversion calculation model was developed for normal wells.A case study was performed to analyze the variation laws of the pumping speed and acceleration after frequency conversion.It is found that with constant liquid production,the negative power and positive power decrease by 65%and 44%,respectively.Moreover,the frequency conversion calculation model of the anomaly of upstroke leakage was built,and the stroke frequency of upstroke was increased to form the operation mode of"fast upward and slow down-ward".The calculation shows that the system can well handle upstroke leakage and frequency-conversion energy saving by shortening the upstroke time by 2.6 s.The findings of this research provide theoretical support for the fre-quency conversion technology of suck-rod pumps and guidance for energy saving and production improvement in oil-fields.
beam pumping unitfrequency conversion controlBP neural networkconstant powermathe-matical model
NETL
NSTL
万方数据
维普
