Data Assimilation-Based Real-Time Estimation of Downhole Gas Influx Rate and Void Fraction Distribution in a Drilling Riser

Abstract

Abstract The ability to perform an accurate estimation of the downhole influx rate and the distribution of the invaded formation gas is of great importance for the modeling of a riser gas event, particularly when the gas kick is to be circulated out through the riser using Managed Pressure Drilling (MPD) systems. Due to the lack of knowledge of the formation pore pressure and the great complexity of wellbore two-phase flow, the influx rate of gas entering the riser is difficult to estimate or direct measure, bringing great uncertainties in the current methods of riser gas modeling. An Extended Kalman Filter (EKF) is used as a Data Assimilation (DA) method for real-time estimation of riser gas influx rate and the enhanced accuracy of the existing physics-based flow models. Data from a set of full-scale experiments is introduced in this study. A riser gas event in a Water-Based Mud (WBM) system is replicated in the experiments by injecting gas from the bottom of an experimental wellbore. The real-time measurement data, including the surface backpressure, the pressure at the subsea blowout preventer (SSBOP) level, and the liquid outflow rate, are used to estimate the rate of gas entering the riser using the EKF. An online calibrated Drift Flux Model (DFM) implemented with DA is used to estimate the distributed flow parameters, including the profiles of gas void fractions. The gas injection rate measured by a Daniel flowmeter and the downhole Distributed Acoustic Sensing (DAS) data from the full-scale experiments were used to validate the estimation results of the proposed method. Results show satisfying agreement between the estimation and the measurement data. The benefits of this work are seen in the estimation of the parameters that are difficult to directly measure or control and in the maximized usage of measurement data of different types, realizing the more accurate prediction of gas influx behaviors in the riser. The inclusions of the model online calibration with coefficients adaptively tuned help to further improve the performance of the existing two-phase flow models and help the design of riser gas management strategies.