Abstract
Electrical submersible pump system (ESP) represents a very effective artificial lift method in oil production; however, the possible presence of free gas may affect the performance of the pump in several ways. Depending on the gas content at suction conditions and on the suction pressure, it can block the impeller where no pumping action takes place, a phenomenon known as gas locking. The objective of this study is to investigate the pump behavior operating near this condition by using signal processing and analysis tools, in order to understand and anticipate the phenomenon.
The tests were performed in laboratory, using a horizontal skid operating with a three-staged ESP. It was acquired time varying signals for process variables such as inlet and outlet pressures, shaft power and liquid flow rate for a constant gas mass flow rate. Then, using a mathematical software package for signal processing, the data were analyzed in both time and frequency domain to correlate these results with the pump operating nearly the gas locking point.
Thus, it was possible to observe that as the void fraction increases, the most variables tend to oscillate heavily at some particular frequencies. It suggests that the frequency analysis relates with the physical phenomenon since the same operating conditions is ensured. This preliminar analysis permits the diagnose of operation, emphasizing the behaviour of the pump before and during the gas locking point.
In literature, it is possible to find a lot of information about frequency analysis in flows through pipelines, so this work introduces a different approach of signal analysis for ESPs in order to predict the above mentioned phenomenon. Therefore, this work suggests that new kind of tests could be advantageous in future in order to evaluate the pump's behavior at different conditions, for example by varying the gas flow rate or inlet pressure.
