Two-phase Flow Pattern Identification Research Articles

Identification of gas-liquid two-phase flow patterns based on flexible ultrasound array and machine learning

Ultrasound technology has been recognized as the mainstream approach for the identification of gas-liquid two-phase flow patterns, which holds great value in engineering domain. However, commercial rigid probes are bulky, limiting their adaptability to curved surfaces. Here, we propose a strategy for autonomous identification of flow patterns based on flexible ultrasound array and machine learning. The array features high-performance 1–3 piezoelectric composite material, stretchable serpentine wires, soft Eco-flex layers and a polydimethylsiloxane (PDMS) adhesive layer. The resulting ultrasound array exhibits excellent electromechanical characteristics and offers a large stretchability for an intimate interfacial contact to curved surface without the need of ultrasound coupling agents. We demonstrated that the flexible ultrasound array combined with machine learning can accurately identify gas-liquid two-phase flow patterns, in a circular pipeline. This work presents an effective tool for recognizing gas-liquid two-phase flow patterns, offering engineering opportunities in petroleum extraction and natural gas transportation.

Comprehensive analysis and identification of energy performance and unsteady two-phase flow patterns based on experiments and comparison between two distinct multiphase flow models

ABSTRACT In the petroleum sector, electrical submersible pumps (ESP) face numerous challenges when handling multiphase flow of gas and liquid. The primary challenge is from the build-up of the gas-bubbles within the impellers of ESPs, leading from moderate to severe deterioration in pump efficiency. Consequently, a comprehensive investigation is conducted, involving a combination of experimental and numerical analyses, to thoroughly explore the effects of gas entrainment on the performance and intricate internal flow mechanisms of a five-stage mixed-flow ESP. All these analysis was carried out under both design (Qd ) and off-design conditions (0.8Qd , and 1.2Qd ). For unsteady numerical calculations, two widely used multiphase-models: The volume of fluid (VOF) and The Euler-Euler model are employed and compared to investigate the versatility/validity of these models in predicting multiphase performance of ESP. The comparison between tests, Eulerian, and VOF models demonstrated that the Eulerian model aligns well with the experimental results. and shows higher-adaptability pattern with the test results. This reveals its best capability/versatility in predicting two-phase performance and internal-flow results for ESP. Moreover, it can more accurately capture the phase slippage inside ESP under two-phase flow. Additionally, this work provides a vision to the broad-applicability of Eulerian-model for complicated machinery such as ESP.

Nonlinear Analysis for Identification of Vertical Upward Oil–Water Two-Phase Flow Patterns

Nonlinear Analysis for Identification of Vertical Upward Oil–Water Two-Phase Flow Patterns

Two-Phase Flow Pattern Identification in Vertical Pipes Using Transformer Neural Networks

The oil and gas industry consistently embraces innovative technologies due to the significant expenses associated with hydrocarbon transportation, pipeline corrosion issues, and the necessity to gain a deeper understanding of two-phase flow characteristics. These solutions involve the implementation of predictive models utilizing neural networks. In this research paper, a comprehensive database comprising 4864 data points, encompassing information pertaining to oil–water two-phase flow properties within vertical pipelines, was meticulously curated. Subsequently, an encoder-only type transformer neural network (TNN) was employed to identify two-phase flow patterns. Various configurations for the TNN model were proposed, involving parameter adjustments such as the number of attention heads, activation function, dropout rate, and learning rate, with the aim of selecting the configuration yielding optimal outcomes. Following the training of the network, predictions were generated using a reserved dataset, thus facilitating the creation of flow maps depicting the patterns anticipated by the model. The developed TNN model successfully predicted 9 out of the 10 flow patterns present in the database, achieving a peak accuracy of 53.07%. Furthermore, the various predicted flow patterns exhibited an average precision of 63.21% and an average accuracy of 86.51%.

Identification of two-phase flow patterns in Z-shaped offshore pipelines based on deep learning technologies

Flow pattern identification is of great significance for the flow assurance of multiphase flow in offshore pipelines. For this purpose, a deep learning algorithm integrating convolutional neural networks (CNNs) and long short-term memory (LSTM) networks is developed in this study to identify different flow patterns of two-phase flow in a Z-shaped pipeline. To verify the applicability and performance of the proposed algorithm, a series of experiments are conducted in the multiphase flow loop. Then, the flow pattern pictures, mass flow rates, and vibration signals are collected by the NI data acquisition system and subjected to denoising using the density-based spatial clustering of applications with noise (DBSCAN) algorithm combined with LSTM. Meanwhile, the Kendall rank correlation is employed for feature processing. Finally, the proposed deep learning algorithm achieves accuracies of 0.98077 with engineering parameters and 0.955 with time-domain parameters. In addition, a comparison analysis is conducted with four other common deep learning algorithms in terms of computation time, learning rate, accuracy, and different sample quantities. The results demonstrate that the proposed algorithm can not only improve the identification accuracy but can also achieve faster convergence speed and higher computational efficiency.

Identification of two-phase flow patterns based on capacitance data of electrical capacitance tomography with semi-supervised generative adversarial network.

Currently, the flow pattern identification algorithms based on ECT (electrical capacitance tomography) technology have low identification accuracy for complex flow patterns and require a large amount of label data for learning. A novel flow pattern identification method based on a semi-supervised generative adversarial network (SGAN) with capacitance data of ECT is proposed. First, the principles of the ECT technique and general GAN are briefly described, and the model parameters, loss function, and training process of the SGAN are explained in detail. Second, a capacitance data sample set of 11 400 random flow patterns is constructed by co-simulations of COMSOL and MATLAB, and then, the SGAN and BP (back propagation) and SVM (support vector machine) network models are trained and validated by the training set. Finally, static experiments are conducted on the self-developed ECT system, and the identification results of different algorithms are compared and analyzed by modifying the label sample size of the training set. The experimental results show that SGAN maintains a higher average identification accuracy under the training condition where the number of label samples of SGAN is ten times smaller than that of the other two algorithms.

One-dimensional structure reparameterized convolutional neural network for two-phase image reconstruction based on ERT

Electrical resistance tomography (ERT) can be applied to two-phase flow pattern identification which is a key research direction for improving the operational safety of different industrial equipment systems with complex flow fields. Aiming at the existing problem that the traditional algorithm for defining flow patterns cannot accurately establish the mapping relationship between the measured voltage from ERT system and the two-phase flow conductivity distribution, a novel one-dimensional structure reparameterized convolutional neural network (1D-SRPCNN) algorithm for two-phase flow pattern image reconstruction based on ERT is proposed. First, finite element method and deep learning software framework are used to build dataset and train the neural network model respectively. Second, a deep residual network (ResNet) is used as the main network structure in the algorithm, and the one-dimensional multiscale feature extraction block (1DMSFE-Block) is improved by structural reparameterization. Then, multiscale convolution is introduced to 1DMSFE-Block for extracting features of different receptive field sizes and performing linear fusion, and the predicted two-phase flow conductivity pixel vector is obtained by the feature map passing with three fully connected layers. The results show that 1D-SRPCNN has high reconstruction performance, the average relative image error is 5.15%, the average correlation coefficient is 97.2%, and it has high anti-noise performance and generalization performance. Different experimental data also show that 1D-SRPCNN has high image reconstruction accuracy and efficiency. The research will provide important theoretical support for accurately identifying two-phase flow patterns in different fields.

Identification of Gas-Water Two-Phase Flow Patterns in Horizontal Wells of Shale Gas Reservoirs Based on Production Logging Data

In order to clarify the gas-water two-phase flow law in horizontal wells and study the gas-water two-phase flow characteristics in horizontal wells, firstly, the gas-water two-phase in a horizontal well is numerically simulated and analyzed, and the flow pattern distribution under different well inclination angles and different phase separation flow rates is obtained. Secondly, a series of production logging instruments including CAT instrument was used to conduct experimental research on gas-water two-phase flow under different flow conditions, and the measured values of each CAT probe were extracted to reflect the local holdup under different flow patterns. Finally, SSA-BP neural network algorithm is used to identify a gas-water two-phase flow pattern in a wellbore by using experimental parameters such as center holdup, well inclination angle, spinner revolution, and CAT probe measurements. The recognition accuracy of the neural network was improved from 83.75% to 91.66%, and the operation speed was accelerated. It provides a research idea to explore the flow characteristics of gas-water two-phase flow in horizontal wells.

Identification of Oil and Gas Two-Phase Flow Patterns in Aero-Engine Bearing Chambers Based on Kriging Method

The identification of the flow pattern within the bearing chamber's oil and gas two-phase flow is crucial for its lubrication design. Aiming at the lack of accuracy and universality of the current bearing chamber oil and gas two-phase flow pattern recognition, this paper introduces a new method for recognising oil and gas two-phase flow patterns based on the Kriging model. In this paper, the oil-gas two-phase flow field in the bearing chamber is modelled in Comsol simulation software using the level set-k- ε model. Subsequently, the finite element analysis results were used as samples to fit the Kriging model, and the effects of different variables on the flow pattern were analysed and compared with the actual results. The results demonstrate that the flow pattern discrimination results fitted by the Kriging model exhibit higher accuracy and greater generalisability. This holds significant implications for future lubrication and structural design of bearing chambers.

Two-Phase Flow Pattern Identification by Embedding Double Attention Mechanisms into a Convolutional Neural Network

There are inevitable multiphase flow problems in the process of subsea oil-gas acquisition and transportation, of which the two-phase flow involving gas and liquid is given much attention. The performance of pipelines and equipment in subsea systems is greatly affected by various flow patterns. As a result, correctly and efficiently identifying the flow pattern in a pipeline is critical for the oil and gas industry. In this study, two attention modules, the convolutional block attention module (CBAM) and efficient channel attention (ECA), are introduced into a convolutional neural network (ResNet50) to develop a gas–liquid two-phase flow pattern identification model, which is named CBAM-ECA-ResNet50. To verify the accuracy and efficiency of the proposed model, a collection of gas–liquid two-phase flow pattern images in a vertical pipeline is selected as the dataset, and data augmentation is employed on the training set data to enhance the generalization capability and comprehensive performance of the model. Then, comparison models similar to the proposed model are obtained by adjusting the order and number of the two attention modules in the two positions and by inserting other different attention modules. Afterward, ResNet50 and all proposed models are applied to classify and identify gas–liquid two-phase flow pattern images. As a result, the identification accuracy of the proposed CBAM-ECA-ResNet50 is observed to be the highest (99.62%). In addition, the robustness and complexity of the proposed CBAM-ECA-ResNet50 are satisfactory.

Gas/Liquid Two-Phase Flow Pattern Identification Method Using Gramian Angular Field and Densely Connected Network

Accurate identification of flow patterns is of great importance for industrial production. A flow pattern identification method based on Gramian angular field (GAF) and densely connected network (DenseNet) was proposed. Experimental data of gas–liquid two-phase flow in a vertical upward pipeline were collected using a resistance sensor array, and the multivariate measurement data were downscaled into univariate time series for feature reduction. Then, the data were encoded into 2-D images using the Gramian angular summation field (GASF) and Gramian angular difference field (GADF) to highlight the characteristic differences between flow patterns, and the evolutionary behavior of flow patterns was further analyzed based on the images. A DenseNet model was established, and the GAF images were used as model inputs for training and testing to achieve the flow pattern identification. The results show that the GAF images can effectively reflect the characteristics of different flow patterns; in particular, the combination of GADF and DenseNet has the best recognition effect, and the average flow pattern identification accuracy reaches 98.3%. This method provides a new perspective for distinguishing complex gas–liquid two-phase flow patterns.

Identifying Flow Patterns in a Narrow Channel via Feature Extraction of Conductivity Measurements with a Support Vector Machine.

In this work, a visualization experiment for rectangular channels was carried out to explore gas-liquid two-phase flow characteristics. Typical flow patterns, including bubble, elastic and mixed flows, were captured by direct imaging technology and the corresponding measurements with fluctuation characteristics were recorded by using an electrical conductivity sensor. Time-domain and frequency-domain characteristics of the corresponding electrical conductivity measurements of each flow pattern were analyzed with a probability density function and a power spectral density curve. The results showed that the feature vectors can be constructed to reflect the time-frequency characteristics of conductivity measurements successfully by introducing the quantized characteristic parameters, including the maximum power of the frequency, the standard deviation of the power spectral density, and the range of the power distribution. Furthermore, the overall recognition rate of the four flow patterns measured by the method was 93.33% based on the support vector machine, and the intelligent two-phase flow-pattern identification method can provide a new technical support for the online recognition of gas-liquid two-phase flow patterns in rectangular channels. It may thus be concluded that this method should be of great significance to ensure the safe and efficient operation of relevant industrial production systems.

Liqnet: A real-time monitoring network for two-phase flow patterns

Real-time identification of gas-liquid two-phase flow can help fluid systems maintain safe operating conditions. A flow pattern identification method based on a convolutional neural network (CNN) algorithm (after this referred to as liqnet) is proposed in this paper to realize automatic detection and real-time identification of two-phase flow patterns. This paper mainly focuses on solving two problems of CNN algorithm flow pattern identification (1): the experimental samples for two-phase flow classification are few, and (2): the existing methods do not fully consider the real-time nature of two-phase flow identification. Therefore, this paper constructs a two-phase flow database containing 6242 images using data enhancement, proposes a lightweight network liqnet, and compares it with six mainstream CNN models. The results show that liqnet can achieve the highest accuracy (98.65%), has the least amount of parameters (1.3708 M), and can achieve the purpose of real-time prediction (32.11FPS).

Two-phase flow pattern identification in horizontal gas–liquid swirling pipe flow by machine learning method

Gas–liquid two-phase swirling flow has been widely used in nuclear industry. Its flow pattern is fundamental to investigate the two-phase flow. Although flow patterns of non-swirling flow in a horizontal pipe have been investigated for a long history, flow patterns of swirling flow in the pipe are rarely reported. In this paper, gas–liquid two-phase flow patterns of swirling flow generated by a vane-type swirler inside a horizontal pipe were investigated by a visualization experiment. Five swirling flow patterns were observed and recorded by the backlight imaging method. Then image processing method was used to obtain void fraction, and the statistical analysis (CDF and PDF) of void fraction for each swirling flow patterns was performed. Owing to the distinguished and stable feature of PDF signals, four parameters describing the characteristics of PDF signals have been proposed as the indicator of machine learning method. Finally, five algorithms of machine learning method have been used to identify the swirling flow patterns, and RUSBoost tree algorithm performs best with an accuracy of 97.4%.

Two-phase flow pattern identification in CAES systems with dimensional analysis coupled with support vector machine

Two-phase bubbly flow can lead to challenges during the charging period for hydraulically-compensated Compressed Air Energy Storage systems. In this study, upward two-phase flow systems have been investigated in large diameter vertical pipes to better understand when the transitions to problematic flow regimes occur. A comprehensive dimensional analysis was conducted, considering vertical pipe geometry and fluid thermophysical properties. Nine dimensionless parameters and three two-phase flow maps were created. The plot of liquid Reynolds number, ReL, over the Two-phase Reynolds number, Retp,made classification between the Bubbly, Slug, and Churn flow difficult for large diameter pipes. The plot of the two-phase Capillary number,Catp, over the liquid Reynolds number, ReL, showed large overlap of the data sets extracted from the literature. This may have been due to the weakness of the capillary number in defining the larger pipe flow behaviours. The last flow map based on the Slippage number, Sltp, versus the two-phase Froude number, Frtp, showed an exponential decay relationship between the Froude and Slippage parameter. When this was converted to the semi-log scale it provided a valuable classification map of two-phase flow. Finally, a Support Vector Machine (SVM) technique applied to the 2D plot of (Sltp- Frtp), gave a useful transition boundary to separate regimes in air-water upward flows in large diameter pipes.

Two-phase flow patterns identification in porous media using feature extraction and SVM

• Support vector machine is employed to identify the flow patterns in packed beds. • Features of different flow patterns are extracted based on different pressure signals. • Feature extraction is conducted by the PDF, PSD and WES methods respectively. • Three SVM models are trained and their identification ability is compared. • Identification rate of typical flow patterns is up to 96.08%. Rapid and accurate identification of two-phase flow patterns in porous media is of great significance to the chemical industry, petroleum and nuclear engineering, etc. Based on the different pressure signals of gas-liquid two-phase flow in a porous bed, the present work proposes an intelligent recognition method to identify the two-phase flow patterns in porous media by the technologies of feature extraction and support vector machine (SVM). The analysis techniques, including time domain (PDF), frequency domain (PSD) and time-frequency domain (Wavelet), are employed to extract and summarize the corresponding characteristics of differential pressure signals of flow patterns. The intelligent recognition models are developed to identify the two-phase flow patterns in porous media by SVM. The models are trained respectively based on the characteristics of time domain + frequency domain (TF-SVM model), time domain + wavelet (TW-SVM model) and frequency domain + wavelet (FW-SVM model). The overall identification accuracy of the optimal model (TW-SVM model) reaches 96.08%.

A new deep neural network framework with multivariate time series for two-phase flow pattern identification

Uncovering flow dynamic behavior of different flow patterns is an important foundation of multiphase flow research. But the traditional classifier is still adopted in the flow pattern identification based on statistical features of experimental measurements, and the utility of data is not sufficient in previous works. Therefore, a novel deep neural network framework is proposed to leverage abundant details of signals. The data was input into the new model after two innovative slicing operations, which combines BiLSTM and CNN to extract the deep characteristic information of different flow patterns. In addition, attention mechanism and residual connection are introduced to improve the network performance. Meanwhile, the dynamic experiment of vertical gas-water two-phase flow is carried out, four-channel conductance signals under five typical flow patterns, namely bubble flow (BF), slug flow (SF), bubble-slug transitional flow (BSF), churn flow (CF) and slug-churn transitional flow (SCF), are collected to feed the network. Finally, in order to verify the effectiveness of the proposed model, some comparative experiments are designed and implemented. The results demonstrate that our proposed model outputs more precise flow pattern identification, which opens up a new way for investigating industrial multiphase flow.

Flow boiling heat transfer characteristics of two-phase flow in microchannels

A set of experimental platforms with widths of 0.5, 1.0, 1.5, and 2.0 mm was established to explore the mechanism of flow boiling bubble dynamics in microchannels, focusing on heat transfer characteristics, pressure loss, and two-phase flow pattern identification. Bubble flow, restricted bubble flow, and dry area were observed in all four channels. The appearance of flow pattern was related to flow rate and channel width. Under the condition of the same channel width, the initial heat flux of subcooled boiling gradually increased with increase in flow rate, and this change trend was close to the linear trend. Under the same flow rate, the initial heat flux of subcooled boiling increased with decrease in channel width. This condition was due to the faster flow rate of the working medium in the narrow channel, resulting in decrease of heating time. The increase in bubble generation frequency directly led to the increase in the wall heat transfer coefficient and the decrease in the bubble separation diameter. Mathematical analysis showed that under the condition of small flow, reduction of channel size led to reduction of the total wall heat transfer coefficient. In this condition, reduction of channel size cannot enhance heat transfer. With increasing volume flow rate, the range of hydrodynamic control area increased and the index decreased. When the flow rate was large, the total heat transfer coefficient increased greatly with the decrease in channel size. The theoretical values were in good agreement with the experimental data.

Image identification for two-phase flow patterns based on CNN algorithms

Flow patterns are essential and useful to model the interfacial structures and heat transfer in gas-liquid two-phase flow. However, the current two-phase flow patterns classification methods mostly depend on direct visual observation. This study adopted a new flow pattern classification method based on convolutional neural network (CNN) algorithms to achieve an automatic and objective identification of two-phase flow patterns. A database of 696 test conditions, including 105642 condensing flow pattern images of methane and tetrafluoromethane in a horizontal circular tube, is collected as the input of the data-driven algorithms. After 80% of image data is fed to train and fit the parameters in the algorithms, the trained models with acceptable universality are obtained to identify five flow patterns: annular flow, bubbly flow, churn flow, slug flow and stratified flow. Compared with the manual classification, the proposed method can accurately predict two-phase flow patterns with a prediction accuracy of more than 90.63% and 91.45% for the test dataset and the entire database, respectively. The average accuracy for predicting all data points in the database is more than 97.56%. The results showed that using images as input, CNN algorithms can provide objective prediction with satisfactory accuracy and universality for two-phase flow pattern identification.

Intelligent identification of two-phase flow patterns in a long pipeline-riser system

Accurate and rapid identification of multiphase flow patterns in long-distance pipelines is an important means for flow assurance. In this paper, an experiment of air-water two-phase flow has been carried out on a pipeline-S-shaped riser system with a length of 1687 m. Based on the amplitude of riser differential pressure, the working conditions are categorised into three typical flow patterns, severe slugging flow, oscillatory flow, and stable flow. Only two difference pressure signals that are most practical near the offshore platforms are used. The data of liquid phase accumulation of severe slugging is adopted as samples, and the signal features are extracted by wavelet multiresolution analysis. The parameters of six common classifiers are optimized, and the effects of different classifiers with the optimal hyperparameters on flow pattern recognition are compared and analyzed. For neural networks, decision trees, support vector machines and k-Nearest Neighbors classifier with the optimal hyperparameters, the recognition rate of severe slugging is higher than 95.8% and the average recognition rate of the three flow patterns is higher than 94.2% when the sample duration is only 6.2 s. On the premise of achieving the highest recognition rate, the number of features is substantially reduced to 15.6% of the original number by principal component analysis.