Artificial Lift Methods Optimising and Selecting Based on Big Data Analysis Technology

Shi, Junfeng (RIPED, CNPC) | Chen, Shiwen (RIPED, CNPC) | Zhang, Xishun (RIPED, CNPC) | Zhao, Ruidong (RIPED, CNPC) | Liu, Zhaoyu (Drilling & Production Engineering Department of Jilin Oilfield Company, CNPC) | Liu, Meng (RIPED, CNPC) | Zhang, Na (RIPED, CNPC) | Sun, Dakui (Drilling & Production Engineering Department of Jilin Oilfield Company, CNPC)

OnePetro 

Abstract

The best artificial lifting method for a well is to lifting more with less during the whole life cycle, however, it is difficult to select the best method because there are many factors affecting the choice of artificial lifting methods and most factors cannot be described with mathematical models. At present, the selection mainly depends on experts’ experience, which results in many incorrect decisions and lead to huge economic loss. In China, there are a large number of artificial lifting wells, and the large amount of data generated by the wells is of great value, which can be used to select the best artificial lifting method for different type of oil wells. This paper selects 11 parameters, including well fluid characteristics, fluid production capacity, well trajectory, pump depth, pump efficiency, pump inspection period, and maintenance cost per year as influence factors. About 40,000 artificial lifting wells of CNPC were selected as the big data analysis sample initially. As not all the wells are good samples, an effect evaluation function is established by taking pump efficiency, power consumption with lifting per ton liquid 100m, annual operating maintenance cost into account, then the samples are selected further according to the value of effect evaluation function, which ensured all the samples participated in machine learning are good samples. A deep recurrent neural network was established which can select the best artificial lifting method through deep learning. Experimental results showed that the neural network model had fast convergence and high prediction accuracy. With the application of this model, artificial lifting method selecting and effect analysis have been conducted for more than 5,000 oil wells of CNPC with different reservoir characteristics, different wellbore structures and different fluid characteristics. The coincidence rate between calculation results of this model and actual production situation is 90.56%. Big data analysis provides a reliable, practical and intelligent method for optimizing and selecting artificial lift.