Modeling and Implementation of a System for Sucker Rod Downhole Dynamometer Card Pattern Recognition

This paper presents a new method for the downhole dynamometer card pattern creation. This method use artificial intelligence concepts and real card data picked up from the oil fields acquired data. The main idea is to allow field team to generate patterns based on their own experience supported by an intelligent algorithm designed to perform graphical analysis of the card shape. The main idea is to allow pattern recognition and its classification toward the solution of rod pumping problems in oil wells. Successful results were reached when the proposed method was applied in PETROBRAS/UN-BA real databases. Tests performed with real data allowed the development of a new set of patterns used to analyze well performance in Buracica and Balsamo oil fields. These tests showed good results when compared with the ones obtained from empirical patterns only. In addition, this approach allowed to immediately increasing the pattern number from two to a set of eight, improving the sucker-rod pumped wells analysis and management.