Use of Advanced Pattern-Recognition and Knowledge-Based System in Analyzing Dynamometer Cards

Abstract

Summary A quick and accurate identification of downhole problems is essential in rod pumping to minimize operating costs and to maximize oil production. TOTAL-CFP developed a system to assist engineers in day-to-day oilfield operations in rod pumping. The system will make rod-pumping expertise available to field operations personnel. It provides a tool for improving data analysis and can be used as a training aid for engineers dealing with rod-pumping wells for the first time. This system has three subsystems: a dynamogram acquisition system, a dynamogram management system, and a knowledge-based system. This paper presents the knowledge-based system, which is used to diagnose downhole problems and to suggest corrective actions. To achieve these goals, the system applies advanced pattern-recognition techniques that consist mainly of structural pattern-recognition methods and hierarchical data structures called quadtrees. The diagnostics and recommendations are based on the analysis of surface dynamometer and downhole cards. The interface between the user and the system is very friendly. The system provides the user with graphical explanations of its diagnostics. The system prototype was installed in March 1991 and the specifications of the final version are planned for the end of this year.