(54l) Real-Time Performance Monitoring and Diagnostics of Rod Pump System for Oil Elevation Using Streaming Dynamometer Card Data (poster version)

Rod pump systems are the most commonly used mechanism for oil lift in the world. The ability to monitor the rod pump operation and diagnose any impending problems with the pump and/or the well is therefore of primary interest to oil field operation. Downhole dynamometer cards (obtained by plotting the measurements of the rod displacement vs. the computed load on the pump, originally developed by Walton E. Gilbert) are viewed as the main information source for performance monitoring and diagnostics of the rod pump and the well today.
Methods to deduce the operating condition of the well and the pump from downhole dynamometer card data are reported both by academia and industry. The use of downhole dynamometer cards are at the heart of many commercial software/hardware products that target performance monitoring and diagnostics of rod pump systems in oil fields around the world.

In this paper, we present an algorithm for automated performance monitoring and diagnostics of the well and pump operating conditions where the streaming card data is processed in real-time. The proposed algorithm uses a hybrid classification approach where both labeled and unlabeled card data would be consumed to generate insight for the operators. The key objective here is to enable the detection of an anomaly in the operation by maximally utilizing prior knowledge while enabling automatic capture of the unique characteristics of a well/rod pump unit solely based on the live operation data.