Testing of shale oil wells equipped with electric submersible pumps (ESPs) is particularly challenging because both rates and pressures decline rapidly and continuously, which is aggravated by rising gas-liquid ratios (GLR) and intermittent production, which means that wells are often in transient condition. High-frequency measurement is therefore necessary to obtain a representative profile of rate and pressure versus time, which requires a dedicated multiphase flowmeter or test separator on each well. This paper investigates a novel method for obtaining very high frequency measurements using ESP data from both the downhole permanent gauge and the surface VSD (Variable Speed Drive), thus leveraging the standard ESP installation to obtain critical reservoir information.

The algorithms use real-time data, which provide the necessary measurement frequency, resolution, and repeatability to capture well performance transients. The liquid flow rate calculation is based on the principle that the power absorbed by the pump is equal to that generated by the motor, which can be resolved for rate. Water cut is calculated by modeling the production tubing as a gradiometer. The tubing pressure drop provides a measurement of the average fluid density, which is then converted to a water cut. Analytical equations are used throughout the process ensuring that the physics are respected at all times, which yields greater reliability than analogous methods based on correlations and artificial intelligence.

The calculated rates were compared to measured rates on a shale oil well equipped with an ESP and a dedicated test separator to provide a reference for validating the calculations. An excellent match was obtained when averaging the high frequency data to match the separator's frequency of daily measurements. In addition, a single calibration over a period of 5 months was all that was needed confirming that the models respected the underlying physics at all times and provided an accurate trend, despite an increasing gas void fraction and numerous stops and starts. One of the reasons that a single calibration was sufficient for a long time period was that the liquid rate algorithm is independent of specific gravity and therefore the computation can handle varying water cut and GLR. The high-frequency flow rate data enabled modeling of fracture efficiency because reservoir flow regimes can be identified. A method was also identified for providing an approximate prediction of flowing pressure as a function of flow rate [i.e., a pseudo inflow performance relationship (IPR) curve for shale oil wells], which is essential to well performance optimization and predicting the timing of the switch to beam pumping.

In addition to providing a cost-effective method for well testing when transients dominate well production, such virtual flowmeters also have value in conventional oil well testing. They can reduce the cost of testing operations while concurrently improving back allocation by virtue of the high frequency, resolution, and repeatability of the rate calculations.