Statistical control applied to events detection in oil production with artificial lift system ESP

The growing need not only to understand, control and optimize oil production systems but also for detecting anomalies in a timely manner to mitigate losses, favored the need to do some research in order to provide support elements for the characterization of failure events. This research is aimed at evaluating different statistical control techniques - Univariate and Multivariate - in order to achieve efficient management of information in real time as well as identifying patterns associated with operational failures in artificial lift systems with electric submersible pumping systems (ESP) in oil production. Statistical control techniques analyzed were exponentially weighted moving average EWMA, CUSUM cumulative sums, Hotelling´s T2, Frequency Analysis and combination of these. From this study it was determined that the phenomena of gasification and solid precipitation in artificial lift systems (ESP), generate specific behavior patterns in variables and multivariable system, which once identified can help generate strategies or methodologies to mitigate losses due to operational problems.