Search

The electrical submersible pump (ESP), an efficient artificial lift method, was developed to increase production rates from wellbores (Bates et al. 2004). As the number of ESP installations increases annually, there is a greater awareness of their environmental impact and a growing responsibility to reduce the associated carbon footprint because frequent workovers to replace failed ESPs are primary sources of carbon dioxide emissions from the oil and gas industry. Because of this, companies are beginning to pursue cost reductions and look for methods to mitigate the consequences of production. Systems for monitoring ESP performance in real time are currently being developed on the basis of data analysis to detect potential problems in advance. This paper presents a digital solution for tracking ESP performance that includes an automated data processing pipeline and the use of statistical metrics to analyze the dynamics of failure and run life at the system node level. This comprehensive analysis helps diagnose problematic equipment nodes using developed web applications. The recommendation system determines the most-reliable ESP configurations under the necessary operating conditions.

The objective of this work is to develop a predictive model for multiphase wellbore flows using the machine learning approach. The artificial neural network is developed and then trained on the dataset generated using the numerical simulator of the full-scale transient wellbore flows. After the training is completed, the neural network is used to predict one of the key parameters of the wellbore flow, namely, the bottomhole pressure. The novelty of this work is related to the application of the neural network to analyze highly transient processes taking place in wellbores. In such processes, most of the parameters of interest can be represented by interdependent time series of variables linked through complex physical phenomena pertinent to the nature of multiphase flows. The proposed neural network with two hidden layers demonstrated the capability to predict the bottomhole pressure within 5% of the normalized root mean squared error for many complex wellbore configurations and flows. It is also shown that relatively higher prediction errors are mainly observed in the case of slug flows where the transient nature of flows is pronounced the most. Finally, the developed model is tested on data affected by noise. It is demonstrated that although the error of prediction slightly increases in contrast to the data without noise, the model captures essential features of the studied transient process. Description of the developed models, analysis of various test use cases, and possible future research directions are outlined.

We present an approach of passive acoustic logging data interpretation to estimate wellbore fluid holdup variation along the wellbore due to the fluid inflow. The algorithm uses machine learning methods for the analysis of acoustic fields generated, in particular, by flow noise in the reservoir near wellbore zone. The method is designed using acoustic fields generated by numerical simulations. The study of simulation results shows the significant influence of wellbore resonances on acoustic field spectrograms and on intensity distributions along the wellbore. The interpretation results demonstrate that the suggested machine learning model predicts water holdup in a zone after the water inflow with high accuracy. The predictions of water holdup before the water inflow interval are less accurate because resonance characteristics are less sensitive to them. We also studied the influence of passive acoustic logging data distortion by contaminating noise on the model learning and on prediction accuracy for the developed interpretation algorithm. As expected, the estimation of water holdup before the water inflow interval is more sensitive to signal interference. The novelty of the suggested approach to passive acoustic logging data interpretation is in using resonance structures of the acoustic noise spatial frequency characteristics to locate the inflow interval and to estimate the oil and water volume fractions. The resonances contain a clear fingerprint of the fluid holdup variation in wellbore fluid, as shown by our study, and the corresponding information can be interpreted by the machine learning algorithms.

The paper considers the problem of increasing the approximation order of transparent boundary conditions for the wave equation while using finite difference schemes up to the sixth order of accuracy in space. As an example, the problem of wave propagation in a semi-infinite rectangular waveguide is formulated. Computationally efficient and highly accurate formulas for discretizing operator of transparent boundary conditions are proposed. Numerical results confirm the accuracy and stability of the obtained difference algorithms.

We propose highly accurate finite-difference schemes for simulating wave propagation problems described by linear second-order hyperbolic equations. The schemes are based on the summation by parts (SBP) approach modified for applications with violation of input data smoothness. In particular, we derive and implement stable schemes for solving elastodynamic anisotropic problems described by the Navier wave equation in complex geometry. To enhance potential of the method, we use a general type of coordinate transformation and multiblock grids. We also show that the conventional spectral element method (SEM) can be treated as the multiblock finite-difference method whose blocks are the SEM cells with SBP operators on GLL grid., Comment: 39 pages