A general regression artificial neural network for two-phase flow regime identification

Supplementing the collection of artificial neural network methodologies devised for monitoring energy producing installations, a general regression artificial neural network is proposed for the identification of the two-phase flow that occurs in the coolant channels of boiling water reactors. The utilization of a limited number of image features derived from radiography images affords the proposed approach with efficiency and non-invasiveness. Additionally, the application of counter-clustering to the input patterns prior to training accomplishes an 80% reduction in network size as well as in training and test time. Cross-validation tests confirm accurate on-line flow regime identification.     

Non-invasive on-line two-phase flow regime identification employing artificial neural networks

A novel non-invasive approach to the on-line identification of BWR two-phase flow regimes is investigated. The proposed approach receives neutron radiography images of coolant flow recordings as its input and performs feature extraction on each image via simple and directly computable statistical operators. The extracted features are subsequently used as inputs to an ensemble of self-organizing maps whose outputs demonstrate swift and accurate classification of each image into its corresponding flow regime. The novelty of the approach lies in the use of the self-organizing map which generates the different classes by itself, according to feature similarity of the corresponding images; this contrasts traditional artificial neural networks where the user has to define both the number of distinct classes as well as to supply separate training vectors for each class.     

Flow regime identification using chaotic characteristics of two-phase flow

Chaotic nature of two-phase flow is investigated for its dependency upon the flow regime by constructing the pseudo phase space with the time sequential impedance signals of the void fraction. In order to construct the pseudo phase space, the autocorrelation function (ACF) and the average mutual information (AMI) for the delayed time, and the false nearest neighborhood (FNN) for the dimensions as the embedding parameters were employed here. It was found that the delayed time and embedded dimension are highly dependent upon the flow regime. To visualize the trajectory of signals in the phase space, a density map is produced by the projection of the trajectory into the two-dimensional plane which is correspondent with the two-dimensional probability distribution functions (2D-PDF). Since the density map of 2D-PDF showed clear distinction of flow patterns, we developed a method to identify flow regime by applying simple classification rules to the density map. The proposed method successfully identifies the flow regimes of the experimental data of impedance signals for the void fraction produced flow regime map for the vertical channel with diameter of 25.4 and 50.8 mm.     

Flow regime identification methodology with neural networks and two-phase flow models

Vertical two-phase flows often need to be categorized into flow regimes. In each flow regime, flow conditions share similar geometric and hydrodynamic characteristics. Previously, flow regime identification was carried out by flow visualization or instrumental indicators. In this research, to avoid any instrumentation errors and any subjective judgments involved, vertical flow regime identification was performed based on theoretical two-phase flow simulation with supervised and self-organizing neural network systems. Statistics of the two-phase flow impedance were used as input to these systems. They were trained with results from an idealized simulation that was mainly based on Mishima and Ishii's flow regime map, the drift flux model, and the newly developed model of slug flow. These trained systems were verified with impedance signals measured by an impedance void-meter. The results conclusively demonstrate that the neural network systems are appropriate classifiers of vertical flow regimes. The theoretical models and experimental databases used in the simulation are shown to be reliable.     

Vertical two-phase flow identification using advanced instrumentation and neural networks

Most two-phase flow measurements, including void fraction measurements, depend on correct flow regime identification. There are two steps taken towards the successful identification of flow regimes: first, develop a non-intrusive instrument to demonstrate area-averaged void fluctuations and second, develop a non-linear mapping approach to perform objective identification of flow regimes. In this paper, an advanced non-intrusive impedance void-meter provides input signals to neural networks which are used to identify flow regimes. After training, both supervised and self-organizing neural network learning paradigms performed flow regime identification successfully. The methodology presented holds considerable promise for multiphase flow diagnostic and measurement applications.     

Local two-phase flow measurements using sensor techniques

Radial profiles of various local parameters in bubbly two-phase flow were obtained. Measurements of the local void fraction, the local interfacial area concentration, the bubble interfacial velocity and Sauter mean diameter were made using the double sensor probe method. At the same locations, local liquid velocity and turbulence intensity measurements were made using a hotfilm anemometer. Data was taken at three different axial locations (L/D=2, L/D=32 and L/D=62) along a 3.24 m test section with an inner diameter of 0.0508 m. In comparison to previous data sets, the following data is more complete in the sense that both interfacial area measurements are combined with one of the local driving forces for interfacial transfer, namely the liquid turbulent diffusion. There have been few, if any, studies done combining local liquid turbulence and the local interfacial area concentration. The data taken will eventually be applied to the closure relations required by the one-dimensional, time-averaged interfacial area transport equation.     

Void fraction and flow regime in adiabatic upward two-phase flow in large diameter vertical pipes

In pipes with very large diameters, slug bubbles cannot exist. For this reason, the characteristics of two-phase flow in large pipes are much different than those in small pipes. Knowledge of these characteristics is essential for the prediction of the flow in new nuclear reactor designs which include a large chimney to promote natural circulation. Two of the key parameters in the prediction of the flow are the void fraction and flow regime. Void fraction measurements were made in a vertical tube with diameter of 0.15 m and length of 4.4 m. Superficial gas and liquid velocities ranged from 0.1 to 5.1 m/s and from 0.01 to 2.0 m/s, respectively. The measured void fractions ranged from 0.02 to 0.83. Electrical impedance void meters at four axial locations were used to measure the void fraction. This data was verified through comparison with previous data sets and models. The temporal variation in the void fraction signal was used to characterize the flow regime through use of the Cumulative Probability Density Function (CPDF). The CPDF of the signal was used with a Kohonen Self-Organized Map (SOM) to classify the flow regimes at each measurement port. The three flow regimes used were termed bubbly, cap-bubbly, and churn flow. The resulting flow regime maps matched well with the maps developed previously through other methods. Further, the flow regime maps matched well with the criteria which were proposed based on Mishima and Ishii's (1984) criteria.     

Flow regime transition criteria for upward two-phase flow in vertical narrow rectangular channels

In relation to the cooling system of high performance microelectronics, a high power research reactor with plate type fuels and plasma facing components of a fusion reactor, study of two-phase flow in a narrow rectangular channel has been paid considerable attention, recently. For the two-fluid model, direct geometrical parameters such as the void fraction should be used in flow-regime criteria. From this point of view, flow-regime transition criteria for vertical upward flows in narrow rectangular channels have been developed considering the mechanisms of flow-regime transitions. The basic concept of the present modeling followed the Mishima and Ishii model for vertical upward two-phase flows in round tubes. Newly developed criteria have been compared with the existing experimental data for air–water flows in narrow rectangular channels with the gaps of 0.3–17 mm. The present criteria showed satisfactory agreements with those data. Further comparisons with data for steam–water in a rectangular channel at relatively high system pressures have been made. The results confirmed that the present flow-regime transition criteria could be applied over wide ranges of parameters as well as to boiling flow.     

A study of two-phase flow characteristics using reactor noise techniques

Recent work at the University of Washington has demonstrated the potential for studying two-phase flow characteristics using neutron noise techniques. This method offers the advantage of measuring global characteristics of two phase phenomenon without disturbing the flow or requiring large numbers of sensors. Not only can the presence of the injected air into the flow loops be seen in the neutron spectrum but also the neutron spectrum reflects the type of flow by producing a characteristic signature.     

Properties of disturbance waves in vertical annular two-phase flow

Disturbance waves play an important role in interfacial transfer of mass, momentum and energy in annular two-phase flow. In spite of their importance, majority of the experimental data available in literature on disturbance wave properties such as velocity, frequency, wavelength and amplitude are limited to near atmospheric conditions (Azzopardi, B.J., 1997. Drops in annular two-phase flow. International Journal of Multiphase Flow, 23, 1-53). In view of this, air-water annular flow experiments have been conducted at three pressure conditions (1.2, 4.0 and 5.8 bar) in a tubular test section having an inside diameter 9.4 mm. At each pressure condition liquid and gas phase flow rates are varied over a large range so that the effects of density ratio, liquid flow rate and gas flow rate on disturbance wave properties can be studied systematically. A liquid film thickness is measured by two flush mounted ring shaped conductance probes located 38.1 mm apart. Disturbance wave velocity, frequency, amplitude and wavelength are estimated from the liquid film thickness measurements by following the statistical analysis methods. Parametric trends in variations of disturbance wave properties are analyzed using the non-dimensional numbers; liquid phase Reynolds number (Re_f), gas phase Reynolds number (Re_g), Weber number (We) and Strouhal number (Sr). Finally, the existing correlations available for the prediction of disturbance wave velocity and frequency are analyzed and a new, improved correlation is proposed for the prediction of disturbance wave frequency. The new correlation satisfactorily predicted the current data and the data available in literature.     

Effect of flow-induced vibration on local flow parameters of two-phase flow

A preliminary study was conducted experimentally in order to investigate the effect of flow-induced vibration on flow structure in two-phase flow. Two kinds of experiments were performed, namely ‘reference’ (no vibration) and ‘vibration’ experiments. In the reference experiment, an experimental loop was fixed tightly by three structural supports, whereas the supports were loosen a little in the vibration experiment. In the vibration experiment vibration was induced by flowing two-phase mixture in the loop. For relatively low superficial liquid velocity, flow-induced vibration promoted the bubble coalescence but liquid turbulence energy enhanced by the vibration might not be enough to break up the bubble. This leaded to the marked increase of Sauter mean diameter, and the marked decrease of interfacial area concentration. Accordingly, flow-induced vibration changed the void fraction profile from ‘wall peak’ to ‘core peak’ or ‘transition’, which increased distribution parameter in the drift-flux model. For high superficial liquid velocity, shear-induced liquid turbulence generated by two-phase flow itself might be dominant for liquid turbulence enhanced by flow-induced vibration. Therefore, the effect of flow-induced vibration on local flow parameters was not marked as compared with that for low superficial liquid velocity. Since it is anticipated that flow structure change due to flow-induced vibration would affect the interfacial area concentration, namely interfacial transfer term, further study may be needed under the condition of controlled flow-induced vibration.     

Droplet entrainment correlation in vertical upward co-current annular two-phase flow

Upward annular two-phase flow in a vertical tube is characterized by the presence of liquid film on the tube wall and entrained droplet laden gas phase flowing through the tube core. Entrainment fraction in annular flow is defined as a fraction of the total liquid flow flowing in the form of droplets through the central gas core. Its prediction is important for the estimation of pressure drop and dryout in annular flow. In the following study, measurements of entrainment fraction have been obtained in vertical upward co-current air–water annular flow covering wide ranges of pressure and flow conditions. Comparison of the experimental data with the existing entrainment fraction prediction correlations revealed their inadequacies in simulating the trends observed under high flow and high pressure conditions. Furthermore, several correlations available in the literature are implicit and require iterative calculations.Analysis of the experimental data showed that the non-dimensional numbers, Weber number (We = ρ_gj_g²D/σ(Δρ/ρ_g)^1/4) and liquid phase Reynolds number (Re_f = ρ_fj_fD/μ_f), successfully collapse the data. In view of this, simple, explicit correlation was developed based on these non-dimensional numbers for the prediction of entrainment fraction. The new correlation successfully predicted the trends under the high flow and high pressure conditions observed in the current experimental data and the data available in open literature. However, in order to use the proposed correlation it is necessary to predict the maximum possible entrainment fraction (or limiting entrainment fraction). In the current analysis, an experimental data based correlation was used for this purpose. However, a better model or correlation is necessary for the maximum possible entrainment fraction. A theoretical discussion on the mechanism and modeling of the maximum possible entrainment fraction condition is presented.     

垂直上升矩形流道内气液两相流流型图的数值模拟

两相流流型在分析换热、流动不稳定性以及临界热流密度方面具有基础性作用.本文基于VOF(Volume of Fluid)多相流模型,对垂直上升矩形流道内气液两相流动进行数值模拟,表观气速0.1～110 m/s,表观液速0.1～3.2 m/s.得到了流道内气液两相流的主要流型:泡状流、弹状流、搅混流和环状流,分析了流道内截面含气率分布与流型的对应关系,以及截面含气率与气液两相流容积含气率的关系;分析了各种流型下的压降分布特性,并绘制了基于气液表观动能通量的不同流量下气液两相流的流型图,直观的表示出各种流型的分布区域及各流型间的流型转换边界,与已发表文献的实验结果对比符合较好.     

A self-standing two-fluid CFD model for vertical upward two-phase annular flow

In this paper, a new two-fluid CFD (computational fluid dynamics) model is proposed to simulate the vertical upward two-phase annular flow. This model solves the basic mass and momentum equations for the gas core region flow and the liquid film flow, where the basic governing equations are accounted for by the commercial CFD package Fluent6.3.26^®. The liquid droplet flow and the interfacial inter-phase effects are accounted for by the programmable interface of Fluent, UDF (user defined function). Unlike previous models, the present model includes the effect of liquid roll waves directly determined from the CFD code. It is able to provide more detailed and, the most important, self-standing information for both the gas core flow and the film flow as well as the inner tube wall situations.     

A numerical technique for analysis of transient two-phase flow in a vertical tube using the Drift Flux Model

This paper presents a numerical solution of one-dimensional transient two-phase flow in a vertical channel using the Drift Flux Model (DFM). The DFM treats the two phases as a mixture, but allows slippage between the gas and the liquid phase. The DFM was used for the calculation of velocity and fraction of each phase, combined with the most relevant closure relationships models for condensation, wall evaporation, and phasic velocities. The solution of the three conservation equations for the mixture and a continuity equation for the gas phases is obtained by a semi-implicit numerical method. A finite volume method is used to discretize the governing equations on a staggered grid in the computational domain. Satisfactory agreement is shown between predicted void fraction, RELAP5 code and available experimental data under both transient and steady state conditions. Numerical solution was also obtained for a wide two-phase flow conditions: system pressure, surface heat flux, mass flow rate and inlet sub-cooling to check the model ability to predict void fraction accurately. It is concluded, therefore, that the DFM is able to predict void fraction in subcooled flow boiling with sufficient accuracy. For pressures lower than 30 bars, the DFM overestimated the void fraction in comparison with the experimental data by about 15%. The model requires less computational power to simulate than other approaches and has no limitations on the nodalization process for numerical stability. It is therefore expected that development of presented model will be useful for the assessment of experimental data, as well as performing pre-test numerical experimentation.     

Disturbance wave frequencies, velocities and spacing in vertical annular two-phase flow

A systematic study of disturbance wave properties in annular flow is reported. Frequencies and velocities were deduced from correlational analysis of film thickness records. The data is shown to agree with the more reliable earlier studies, though these were not so extensive. The spacing between waves has also been deduced. The effects of flow rates on this parameter are explained and its relationship to wave height is examined.     

Multidimensional two-phase flow regime distribution in a PWR downcomer during an LBLOCA refill phase

The multidimensional countercurrent two-phase flow regimes that occur in a pressurized-water reactor (PWR) vessel downcomer during the refill phase of a large-break loss-of-coolant accident are studied using a transparent 1/10 scale model of a PWR vessel. The various flow regimes and their distribution in the downcomer have been identified and mapped for a range of air-water flooding experiments. The two-phase flow patterns that are identified in the downcomer included various types of film flows, droplet flows, countercurrent churn flows and cocurrent flows depending on the flooding condition. Through observation of the two-phase flow dynamics it was deduced that the physical mechanisms associated with the flooding processes could be separated into a liquid entrainment process and a film flow reversal process. In addition to the above exercise, the effect of non-uniform injection of water into the downcomer via different combinations of cold leg was studied similarly by determining flooding curves and flow pattern maps. It was found that differences in the flooding characteristic were noticeable for various water inlet configurations when compared with the uniform injection case. The differences could be explained qualitatively in terms of the flooding mechanisms identified previously by examining the flow patterns in the downcomer for the non-uniform injection tests.     

Experimental investigation of local two-phase flow parameters of a subcooled boiling flow in an annulus

As a series of subcooling boiling flow tests, local two-phase flow parameters were obtained at SUBO (subcooled boiling) test facility under steam–water flow conditions. The test section is a vertical annulus of which the axial length is 4.165 m with a heater rod at the center of a channel. The inner and outer diameters of the test section and the heater rod are 35.5 mm and 9.98 mm, respectively. The test was performed by a two-stage approach. Stage-I for the measurement of local bubble parameters has been already done (Yun et al., 2009). The present work focused on the stage-II test for the measurement of local liquid parameters such as a local liquid velocity and a liquid temperature for a given flow condition of stage-I. A total of six test cases were chosen by following the test matrix of stage-I. The flow conditions are in the range of the heat flux of 370–563 kW/m², mass flux of 1110–2100 kg/(m² s) and inlet subcooling of 19–31 °C at pressure condition of 0.15–0.2 MPa. From the test, local liquid parameters were measured at 6 elevations along the test section and 11 radial locations of each elevation in addition to the previously obtained local void fraction, interfacial area concentration, Sauter mean diameter and bubble velocity. The present subcooled boiling (SUBO) data completes a data set for use as a benchmark, validation and model development of the Computational Fluid Dynamics (CFD) codes or existing safety analysis codes.