Two-phase Flow Regime Identification Research Articles

Two-phase flow regime identification using multi-method feature extraction and explainable kernel Fisher discriminant analysis

PurposeIdentifying the flow regime is a prerequisite for accurately modeling two-phase flow. This paper aims to introduce a comprehensive data-driven workflow for flow regime identification.Design/methodology/approachA numerical two-phase flow model was validated against experimental data and was used to generate dynamic pressure signals for three different flow regimes. First, four distinct methods were used for feature extraction: discrete wavelet transform (DWT), empirical mode decomposition, power spectral density and the time series analysis method. Kernel Fisher discriminant analysis (KFDA) was used to simultaneously perform dimensionality reduction and machine learning (ML) classification for each set of features. Finally, the Shapley additive explanations (SHAP) method was applied to make the workflow explainable.FindingsThe results highlighted that the DWT + KFDA method exhibited the highest testing and training accuracy at 95.2% and 88.8%, respectively. Results also include a virtual flow regime map to facilitate the visualization of features in two dimension. Finally, SHAP analysis showed that minimum and maximum values extracted at the fourth and second signal decomposition levels of DWT are the best flow-distinguishing features.Practical implicationsThis workflow can be applied to opaque pipes fitted with pressure sensors to achieve flow assurance and automatic monitoring of two-phase flow occurring in many process industries.Originality/valueThis paper presents a novel flow regime identification method by fusing dynamic pressure measurements with ML techniques. The authors’ novel DWT + KFDA method demonstrates superior performance for flow regime identification with explainability.

Flow regime identification using fuzzy and neuro-fuzzy inference applied on differential pressure sensor

Intelligent identification of two-phase flow regime is indispensable for the safe design of industrial systems. This work presents the development and examination of intelligent prediction methods based on fuzzy logic designed to operate on a differential pressure sensor signal for two-phase flow regime identification. The effectiveness of intelligent paradigms for flow regime identification is established using experiments carried out in a 0.9 mm circular glass tube kept horizontally. Air and water are the two-phase fluids. Three flow regimes, namely bubbly, slug and annular flows are observed for various combinations of superficial fluid velocities. The pressure drop in the flow system is measured and recorded online using a differential pressure sensor connected to a data acquisition system. Signal features such as peak current, the difference in current and signal frequency are extracted after extensive analysis. Fuzzy rules are outlined and flow regime output is analyzed. The bottleneck observed in these techniques is qualitatively designated in terms of accuracy and required high human effort. A solution to this bottleneck is found out using adaptive neuro-fuzzy inference system (ANFIS). ANFIS applied on pressure sensor signal is found to provide an accurate characterization of flow regime.

Self-supervised learning-based two-phase flow regime identification using ultrasonic sensors in an S-shape riser

Two-phase flow regime identification is an essential transdisciplinary topic that spans digital signal processing, artificial intelligence, chemical engineering, and energy. Multiphase flow systems significantly impact pipeline safety, heat transfer, and pressure drop; therefore, precisely identifying the governing flow regime is crucial for effective modeling and design. However, it is challenging due to the geometrical complexity of flow regimes in multiphase flow. With the advances in sensor measurement and machine learning, applying non-destructive tests and self-supervised learning to practical industrial problems has become technically feasible and cost-effective. This study applies a weak-supervised learning-based two-phase flow regime identification solution using a non-destructive tests ultrasonic sensor in an S-shape riser experimental bed by proposing a self-supervised feature extraction algorithm. The proposed self-supervised feature extraction algorithm reduces time/labor consumption and human error in data annotation using SSL, which provides full supervision without manual annotation. The self-supervised feature extraction algorithm uses a bottlenecked neural network and encoder-decoder structure to extract compact features. The self-supervised feature extraction algorithm performance is evaluated using an established convolutional neural network-based classifier. The source data was collected from a 10 × 50 m riser experimental rig. The dataset is made available to the community as part of this study. The performance of the approach is comparable with state-of-the-art methods and is also the first successful attempt to apply self-supervised learning to multiphase flow regime ultrasonic signal identification. This study achieved 98.84%, 0.000663, 0.00312, and 7.71 × 105 in accuracy, root mean square error, categorical cross-entropy, and model complexity, respectively. The practical experiment justifies the robustness, fairness, and practicability in the practical application environment. The proposed self-supervised feature extraction brings new approaches and inspirations for the feature extraction step in identifying a two-phase flow regime, and it will be beneficial to generalize this study in different riser shapes in the future.

Numerical Gas–Liquid Two-Phase Flow Regime Identification in a Horizontal Pipe Using Dynamic Pressure Data

Gas–liquid two-phase flow is very common in industrial pipelines. Flow regime identification is the first step to design, analyze, and operate the gas–liquid system successfully. The purpose of this study is to develop a methodology for identification of a two-phase flow regime using post signal processing techniques, namely Fast Fourier Transform (FFT) and Probabilistic Density Function (PDF). Three different flow regimes were simulated in a 6 m horizontal pipe with a 0.050 m inner diameter. A Level-Set (LS) method coupled with Volume of Fluid (VOF) method is used to model the air–water interface. After validation of the numerical method, dynamic pressure readings were collected with the intent to identify the associated flow regimes by post-processing of these signals. It was concluded that dynamic pressure signals of different flow regimes show different characteristics (like dominant frequency, FFT amplitude, PDF location and PDF magnitude) in the time and frequency domains. These characteristics can be potentially used as differentiating factors to distinguish different flow regimes. This research is limited to stratified, slug, and annular flow in the horizontal pipe. This paper uses a new approach to identify the flow regime in a horizontal pipe by Fast Fourier Transform and Probability Density Function of dynamic pressure readings obtained by using numerical simulation.

Novel Infrared Approach for the Evaluation of Thermofluidic Interactions in a Metallic Flat-Plate Pulsating Heat Pipe

A novel and advanced analysis tool, based on the resolution of the inverse heat conduction problem, is used to evaluate wall-to-fluid heat fluxes in a metallic flat-plate pulsating heat pipe. The device under analysis is made of copper and formed by 16 channels having a squared section of 3 × 3 mm2 and filled with a water–ethanol mixture (20 wt.% of ethanol) with a volumetric filling ratio of 50%. One flat side of the device is externally coated with a highly emissive paint to perform temperature measurements by means of a medium-wave infrared camera. The acquired infrared maps are first processed by a three-dimensional Gaussian filter and then used as inputs for the inverse approach for the evaluation of heat fluxes locally exchanged between the fluid and the thin walls of each channel. The suggested procedure is successfully validated by means of synthetic data. The resulting space–time heat flux distributions are therefore statistically investigated in terms of amplitude and space–time variations, providing quantitative references for the identification of two-phase flow regimes. These unique data give an evaluation of the local heat transfer behavior, which is essential to provide empirical values for the numerical models of pulsating heat pipes.

Identification of Horizontal Gas-Liquid Two-Phase Flow Regime using Deep Learning

Two-phase flow is of great importance in various industrial processes. A characteristic feature of two-phase flow is that it can acquire various spatial distribution of phases to form different flow patterns/regimes. The knowledge of flow regime is very important for quantifying the pressure drop, the stability and safety of two-phase flow systems and it holds great significance in petrochemical and thermonuclear industries today. The objective of this study is to develop a methodology for identification of flow regime using dynamic pressure signals and deep learning techniques. Stratified, slug and annular flow regimes were simulated using a Level-Set (LS) method coupled with Volume of Fluid (VOF) method in a 6 m horizontal pipe with 0.050 m inner diameter. Dynamic pressure signals were collected at a strategic location. These signals were converted to scalograms and used as inputs in deep learning architectures like ResNet-50 and ShuffleNet. Both architectures were effective in classifying different flow regime and recorded testing accuracies of 85.7% and 82.9% respectively. According to our knowledge no similar research has been reported in literature, where various Convolutional Neural Networks are used along with dynamic pressure signals to identify flow regime in horizontal pipe. This research provides a benchmark for future research to use dynamic pressure for identification of two-phase flow regimes. This research provides a benchmark for future research to use dynamic pressure for identification of two-phase flow regimes. This study can be extended by collecting data over broader range of flow parameters and different geometries.

Identification of gas-liquid two-phase flow regime in pipelines with low liquid holdup based on ResNet1D-34

Flow pattern identification is an important topic in multiphase flow research. To overcome the subjectivity of manual identification, intelligent identification of flow patterns has attracted much attention in recent years. Both traditional machine learning methods and deep learning methods have been utilized in this field. However, traditional machine learning methods lack accuracy, and existing deep learning methods mostly rely on artificial feature extraction or complex preprocessing. In this paper, we propose a new method with high accuracy and low preprocessing dependency to solve these issues. We modify ResNet, which has proven high performance in computer vision, to fit the data collected by the wire-mesh sensor system (WMS). Due to its outstanding feature extraction ability, the new model can reach high accuracy with simple normalization as the preprocessing step. Additionally, the model can directly process data at various scales without retraining or rebuilding, which gives it high usability and economic value. The experimental results show that the accuracy of this method can reach 99.58% on our dataset.

Identification of high-pressure two-phase flow regime transition using image processing and deep learning

The commercial production process of methane gas from methane hydrate has been developed by the MH21-S R&D consortium (MH21-S) in recent years. In order to establish the gas production method, an accurate gas-liquid two-phase flow regime identification under high-pressure conditions is indispensable. In the present study, a 28 m long test section with an inner diameter of 52.7 mm was newly constructed, and a nitrogen-water two-phase flow experiment was conducted at 3.0 MPa G. Two-phase flow images were successfully acquired using the high-speed camera from the visualization port. The sequential image of upward gas-liquid two-phase flow was merged into a single image by the time strip method. Then, flow regime identification using the convolutional neural network (CNN) trained under sequential images was carried out. The proposed method has shown an excellent capability to identify flow regime images obtained at the current high-pressure conditions. In addition, for the slug flow regime, it was found that the Taylor bubble length highly influences the identification performance, and the model proposed by Street and Tek showed good agreement for the present dataset. Furthermore, the bubbly-slug transition region was successfully evaluated and quantified through the present model. The current methods have shown a good correlation with the existing model. These results indicate that the gas void fraction at the transition region to be larger than the theoretical value for the elevated pressure system.

Identification of two-phase flow regime in the energy industry based on modified convolutional neural network

The identification of the two-phase flow regime is the most fundamental and crucial target of fluid mechanics in the nuclear thermal-hydraulic analysis since the accuracy of fluid and heat transfer models depends on the reasonable prediction of flow regimes. The trend is quantitative identification based on the measurement of quantifiable parameters or the quantitative information of the two-phase flow images. Machine learning is currently the mainstream methodology for flow regime identification. The state-of-the-art method is using flow regime images for identification based on convolutional neural networks (CNNs). To achieve a higher classification accuracy and overcome the disadvantage of the traditional CNN (time-consuming owing to a large number of images required for training), this paper proposes a more efficient methodology for two-phase flow regime identification, which combines the advantages of CNN for image feature information extraction and the support vector machine (SVM) for classification - the CNN-SVM method. Furthermore, the traditional CNN needs a large number of images for training. Considering the limited flow regime dataset, the CNN-SVM method in this paper uses the idea of transfer learning. A small-scale flow regime dataset is used for validating the effectiveness of the proposed method. The accuracies for SVM, CNN and CNN-SVM were 93.9%, 97.2% and 98.8% respectively, which show that the proposed integrated method can realize better classification results since it combines the capability of strong feature extraction of CNN and classification techniques of SVM. The improvement is mainly reflected in the higher classification accuracy of slug flow and churn flow. Additionally, according to the results, the proposed method is also more robust.

Gas-liquid flow regimes identification using non-intrusive Doppler ultrasonic sensor and convolutional recurrent neural networks in an s-shaped riser

The problem of gas-liquid (two-phase) flow regime identification in an S-shaped riser using an ultrasonic sensor and convolutional recurrent neural networks (CRNN) is addressed. This research systematically evaluates three different schemes with four CRNN-based classifiers over fourteen experiments. Four metrics are used as the evaluation criteria: categorical accuracy, categorical cross-entropy, mean square error (MSE), and computation graph complexity. Compared with existing results, a compatible performance is achieved while considerably reducing the model complexity. The testing and validation accuracies were 98.13% and 98.06%, while the complexity decreased by 98.4% (only 117,702 parameters). The proposed approach is i) accurate, low complexity, and non-intrusive and hence suitable for industry, and ii) could provide a benchmark for flow regime identification.

機械学習によるクラシフィケーションと気液二相流への応用

Gas-liquid two-phase flow is seen in various engineering disciplines and understanding of its interfacial structure is a great importance for proper model development. Flow regime maps have been developed by various researchers in the past. However, identification of flow regime involves subjectivity and technical difficulty as the nature of flow regime transition mechanism and criteria are still yet unknown. In the present article, various two-phase flow regime identification methods utilizing machine learning (ML) approach will be reviewed. Two-phase flow features such as two-phase mixture impedance, dynamic force signal, and high-resolution images from high-speed camera were selected for the ML model training and testing. For the machine learning models, artificial neural network (ANN), convolutional neural network (CNN), and convolutional long short-term memory (ConvLSTM) were tested, depending on the feature types. As a result, the ML approach can identify two-phase flow regime with high accuracy. There still remains issues related to ML usage, including hyperparameter selection, and ability to explain its decisions. However, adaptation of ML tools for multiphase flow research fields may provide utmost benefits for efficient signal and image classification approach.

Two-phase flow regime identification based on the liquid-phase velocity information and machine learning

Two-phase flow regime identification in a horizontal pipe was realized based on the liquid phase velocity information and the machine learning method. Ultrasound Doppler velocimetry was employed to measure the liquid velocity. Statistical features extracted from the velocity time series data, such as mean, root mean square, and power spectral density, were used to realize real-time flow regime identification. In addition, two novel parameters—maximum velocity ratio and maximum velocity difference ratio—were proposed to identify plug and decaying slug flow. Different classification algorithms were employed to achieve a high identification accuracy. Moreover, transient flow regime identification with a fast response was realized based on two classification algorithms—long–short term memory and convolutional neural network. The results show that the accuracy of real-time flow regime identification based on a flow regime map can reach up to 93.1% using support vector machine, the maximum velocity ratio and maximum velocity difference ratio are effective in identifying plug and decaying slug flow, and transient flow regime identification under slug flow condition can be realized with an accuracy of 94% based on a convolutional neural network (CNN). Decaying slugs with long lengths confuse the CNN and are responsible for the error in identification. The results presented herein are expected to expand the available knowledge on two-phase flow regime identification.

Experimental characterization of boiling two-phase flow structures under BWR core operating conditions

Local two-phase flow parameters have been measured at the Westinghouse FRIGG facility in a test fuel bundle under Boiling Water Reactor (BWR) core operating conditions, over a wide range of flow and power (up to critical power). The measurements were performed at end of heated length using sapphire optical probes and Pitot tubes placed in various radial positions of the bundle. By using both instruments simultaneously, the void fraction, mixture velocity, dispersed phase (bubble or drop) characteristic lengths and volumetric interfacial area can be measured. In addition, the identification of two-phase flow regimes can be attempted. The measurements are of particular interest since there is very limited available data in the literature at high pressure relevant to BWR operation.Experimental measurements are reported in this paper covering a range of inlet flow and bundle power relevant to BWR operation, at a system pressure of 70 bar. The dynamic void fluctuations are used to establish two-phase flow regime maps, one based on global parameters (for 1-D applications) and the second based on local parameters (for CFD applications). However, it is discovered that transition flow regimes typically observed at low pressure (cap, slug, etc) are not similarly observed at high pressure. In addition, the measured characteristic lengths of the dispersed phase are reported in bubbly and annular two-phase flow regimes. In view of the results, recommendations are provided for further experimental work and modeling improvements.

Identification of two-phase flow regime using ultrasonic phased array

Flow regime is one of the key characteristics of gas-liquid two-phase pipe-flows and its identification is essential for several industrial applications. In this paper, the ultrasonic phased array technology is used to identify flow regimes of two-phase (air-water) vertical flow. The ultrasonic phased array can perform multi-point, omnidirectional detection to obtain high-resolution data suitable for image processing. The scanned images, which have distinctive features, are subjected to a series of image-treatment techniques, such as principle component analysis, to extract information necessary for flow regime identification. The K-nearest neighbors (KNN) classification algorithm is then used to identify flow regimes with high accuracy.

Two-phase flow regime identification through local temperature mapping

Two-phase flows underpin some of our most ubiquitous technologies, ranging from micro-scale liquid-liquid cooling of electronics to macro-scale liquid-vapour boiling and condensation in thermal power plants. Establishing the morphology of a two-phase flow, under a prescribed set of conditions, is considered particularly important in the design stage. As the pressure loss and heat transfer characteristics of a two-phase flow are intimately linked to the fluidic arrangement, knowledge of the prevailing flow topology enhances understanding, and can lead to the development of flow-specific correlations and/or models. This paper presents a novel experimental measurement technique for identifying the predominant two-phase flow regime in a circular tube. Specifically, the investigation presented in this paper focuses on condensing flows of steam, at typical Rankine cycle cooling conditions. However, it is proposed that the experimental arrangement and methodology can be applied to any two-phase flow scenario. The approach presented herein employs a temperature measurement platform - composed from localised instrumentation - to measure the temperature drop, associated with the presence of a liquid phase, at any point in the tube. Through analysis and interpretation of local temperature difference measurements around the inside tube circumference, and along the tube length, the predominant flow regime can be identified. In this study, measurements were taken from a 25 mm internal diameter round tube, with steam flow rates in the range of 0.42–0.94 g·s-1. The flow regime was seen to transition from an annular-type profile nearest the tube inlet to a stratified-wavy topology towards the tube exit in all instances.

A New Method of Selecting the Airlift Pump Optimum Efficiency at Low Submergence Ratios with the Use of Image Analysis

This paper presents experimental studies on the optimization of two-phase fluid flow in an airlift pump. Airlift pumps, also known as mammoth pumps, are devices applied for vertical transport of liquids with the use of gas. Their operating principle involves the existence of a density gradient. This paper reports the results of experimental studies into the hydrodynamic effects of the airlift pump. The studies involved optical imaging of two-phase gas-liquid flow in a riser pipe. The visualization was performed with high-speed visualization techniques. The studies used a transparent model of airlift pump with a rectangular cross-section of the riser. The assessment of the airlift pump operation is based on the image grey-level analysis to provide the identification of two-phase flow regimes. The scope of the study also involved the determination of void fraction and pressure drops. The tests were carried out in a channel with dimensions 35 × 20 × 2045 mm with the gas flux range 0.2–15.0 m3/h. For the assessment of the two-phase flow pattern Probability Density Function (PDF) was applied. On the basis of the obtained results, a new method for selecting the optimum operating regime of airlift pump was derived. This method provides the finding of stability and efficiency of liquid transport. It can also be applied to determine the correlation between the total lifting efficiency and the required gas flux for proper operation of the airlift pump.

Vertical co-current two-phase flow regime identification using fuzzy C-means clustering algorithm and ReliefF attribute weighting technique

In order to develop the flow regime identification map by objective method, air–water two-phase upward and downward flow experiments were conducted in a tubular test section with the inside diameter of 2.54cm. This study focused on (1) a new objective flow regime identification method based on the fuzzy C-means (FCM) clustering algorithm and ReliefF attribute weighting algorithm, (2) the construction of the objective flow regime map on the basis of the identification results of the proposed objective method and (3) the discussion of entrance and L/D effects for co-current upward and downward two-phase flow. The ReliefF-FCM method successfully identified flow regime and the flow regime maps obtained by this method showed the same trends as the maps of Mishima–Ishii and Usui for upward and downward flow, respectively. Furthermore, in comparing with the previous objective flow regime identification methods, the remarkable advantage of the method represented in the paper was discussed. By comparing the flow regime maps at different positions of test section, the entrance and L/D effects on the flow regime transition were also considered.

An image reconstruction algorithm based on Bayesian theorem for electrical resistance tomography

Electrical resistance tomography (ERT) is considered to be one of the most promising process tomography techniques for two-phase/multiphase flow measurement due to the advantages such as high speed, low cost and non-intrusive sensing. The iterative image reconstruction algorithm based on Bayesian theorem has been derived to solve the ERT inverse problem. It is taken into account, which is different to the existing ERT image reconstruction algorithms, including the prior probability of the permittivity distribution and the noise information in the measurement data. Both simulation and experimental data were carried out for typical two-phase flow regimes. The flow regimes are identified according to the reconstructed images of the Bayesian iteration method and conventional methods. Results obtained indicate that the Bayesian iteration method improves the reconstructed image quality with the traditional linear back projection (LBP) and reweighted least squares (RLS). Therefore, the Bayesian iteration method is suitable for identification of online two-phase flow regimes.

Study on two-phase flow regime visualization and identification using 3D electrical capacitance tomography and fuzzy-logic classification

From variety of industry-oriented imaging solutions the electrical capacitance tomography applied to the two-phase gas–liquid mixtures visualization and the phase distribution calculation is getting popular especially when flow key parameters are required. Industry demands particularly include efficient non-invasive automatic phase fraction calculation and flow structure identification in the vertical and horizontal pipelines. This can be solved by using non-deterministic fuzzy-logic based techniques for analysis of volumetric images. This paper presents a preliminary study on automated two-phase gas–liquid flow pattern identification based on a fuzzy evaluation of series of reconstructed 3D ECT volumetric images. The set of volume data is obtained by using nonlinear electrical capacitance tomography reconstruction algorithms. Finally a set of fuzzy-based features is calculated for flow substructure classification. As a result of this analysis obtained features will be used to classify given volumetric image into one of known flow regime structures.

Fast reconstruction of electrical resistance tomography (ERT) images based on the projected CG method

Electrical resistance tomography (ERT) is considered to be one of the most promising process tomography techniques for multi-phase flow measurement due to the advantages such as high speed, low cost and non-intrusive sensing. The image reconstruction for ERT is an inverse problem to find the conductivity distribution of an object by measuring the voltage between sets of electrodes placed around its periphery. The conjugate gradient (CG) method is one of the most popular methods applied for image reconstruction, although its convergence rate is low. In this paper, an advanced version of the CG method, namely the projected CG method, is proposed for image reconstruction of an ERT system. The solution space is projected into the Krylov subspace and the inverse problem is solved by the CG method in a low-dimensional specific subspace. Both simulation and experimental work were carried out for typical gas–water two-phase flow regimes. The flow regimes are identified according to the reconstructed images of the projected CG method and conventional methods. Results obtained indicate that the projected CG method improves the quality of the reconstructed images and dramatically reduces the computational time compared with the traditional sensitivity, Landweber, and CG methods. Therefore, the projected CG method is suitable for identification of online two-phase flow regimes.