Search

As a critical closure equation to the two-fluid model and an important tool to characterize the two-phase-flow interfacial transport, the interfacial area transport equation (IATE) was formulated by taking various physical mechanisms causing interfacial area change into account. To fulfill the dynamic prediction advantage of IATE and further replace the flow regime–based constitutive relations, the IATE model should be validated by transition data to ensure model reliability and robustness. Air-water experiments are performed in bubbly-to-slug transition flows in a 200 × 10-mm narrow rectangular duct. Four-sensor conductivity probes are used to measure the local void fraction, interfacial area concentration (IAC), and bubble velocity at three axial locations. The void fraction distribution changes significantly with the flow developing. Flow conditions with a similar area-averaged void fraction but different superficial mixture velocities are compared, and it is found that the superficial mixture velocity significantly affects the IAC. In addition, the two-group IATE model for narrow rectangular channel is evaluated using the collected data. The average relative error for the total IAC prediction is 11.4%, but the group II IAC is overestimated for most flow conditions. [ABSTRACT FROM AUTHOR]

Highlights • Two-group IATE are evaluated using the small pipe data for the first time. • Recent progress on two-group IATE modeling are summarized and evaluated. • New models for bubble coalescence using log-normal size distribution are developed. Abstract Accurate prediction of interfacial area concentration (IAC) is of great importance in the two-fluid model. As a promising IAC prediction method, the interfacial area transport equation (IATE) was developed to dynamically predict the IAC and to eliminate the shortcomings of the flow regime-based IAC correlations, which is currently used in most thermal–hydraulic system codes. To extend the applicability of the IATE from bubbly flow to slug and churn-turbulent flow, the two-group IATE was developed by representing the IAC of sphere/distorted bubbles and cap/slug bubbles separately. Several studies have been performed in the past focusing on the constitutive modeling and experimental benchmark. However, few data are collected in small size pipes ( D H ⩽ 2 D c ), hence the applicability of the current two-group IATE models on small size pipes has not been well evaluated. In this study, a comprehensive database including bubbly, slug, churn-turbulent flow is established in a round pipe with a diameter of 25.4 mm (D H ≈ 2 D c), and the applicability of two-group IATE models for round pipe has been checked against the data. The measured IAC can be predicted by the two-group IATE with the constitutive models developed for moderate size pipe with errors less than 25% except at the bubbly to slug transition flow conditions. At the transition flow, drastic intergroup void fraction and IAC transport are observed along the flow direction. The model cannot predict this intergroup transfer and overestimates the IAC. Recent improvements on the original models focusing on the one-group to two-group transition flow in a moderate size pipe are also evaluated and the results are still not satisfying. The reasons for this discrepancy are analyzed in detail and improvements including re-deriving the coalescence terms using more accurate log-normal bubble size distribution and experimental coefficients optimizing have been performed. Using the newly derived models with optimized coefficients, the two-group IATE can predict the IAC axial development for all flow conditions with an error less than 25%. The averaged prediction error for the bubbly to slug transition flow conditions decreases from 47.19% to 8.03%. [ABSTRACT FROM AUTHOR]

Highlights • Effect of pipe size on drift-flux analysis for horizontal flow is insignificant. • Gas phase moves slower than the liquid phase in horizontal bubbly flow. • Horizontal plug and slug flows are relatively more homogeneous than bubbly flow. • Differences are observed for C 0 and 〈〈 V gj 〉〉 between current work and these in TRACE for vertical flow. • Correlations modified from vertical flow cannot predict void fraction well for horizontal flow. Abstract The current work seeks to perform the one-dimensional drift-flux analysis for horizontal gas-dispersed flow, to provide closure models for void fraction and to investigate the relative motion between gas and liquid phases. A reliable experimental database for void fraction and bubble velocity is first established over a wide range of flow conditions for bubbly, plug and slug flows in different pipe sizes, due to the limitations of existing database. It is found that the effects of flow regime and pipe size on the drift-flux analysis are insignificant. The drift velocity is found to be negative in horizontal bubbly flow. This is consistent with the conclusion by the 〈 α 〉–〈 β 〉 analysis, where the slope is found to be greater than one. This indicates that the gas phase moves slower than the liquid phase in horizontal bubbly flow. The analysis also indicates that the horizontal plug and slug flows are relatively more homogeneous than horizontal bubbly flow. As such, the homogeneous flow model can predict the void fraction better for plug and slug flows. The current drift-flux analysis provides closure relations for C 0 and 〈〈 V gj 〉〉 that can be employed in TRACE, while existing closure relations are developed for vertical flow and cannot be used for horizontal flow prediction. In addition, various void fraction models in literature are evaluated. It is found that the correlations modified from vertical flow generally cannot predict the void fraction well for horizontal flow. The correlations for plug and slug flows by Franca and Lahey, and Lamari underestimate the void fraction, which may be because these correlations are developed for relatively low j f conditions, while the current work focuses on conditions at j f greater than 2.0 m/s. [ABSTRACT FROM AUTHOR]

System level analysis codes have been proved effective tools for accident analysis and other transient analysis in nuclear reactors. With increasing interests in the natural circulation flow driven systems, flow instabilities occurring under low power and low pressure conditions have been receiving more research attention. This work aims to demonstrate the performance of one of the most widely used system codes, i.e., RELAP5, in simulating startup transients, as well as its capability to predict flow instabilities. The experimental database was acquired from a well-scaled natural circulation test facility, which was designed to investigate the startup flow instabilities for a BWR-type Novel Modular Reactor (NMR-50). The startup simulations by RELAP5/MOD3.2 code were compared with various thermal-hydraulics measurements including system pressure, temperature, natural circulation flow rate, and void fraction, etc. In addition, two power ramp rates were applied in the startup simulations to study heating rate effects on flow instabilities. In general, the RELAP5 code can consistently simulate the two startup transients and predict two flow instabilities, i.e., flashing instability and density wave oscillations, occurred in the natural circulation test facility at low pressures. The predicted natural circulation flow velocities had a good agreement with the measurements. However, the RELAP5 could underpredict void fractions in the heated section when the system is under thermal non-equilibrium conditions at low power and low pressure. [ABSTRACT FROM AUTHOR]

Seismic vibration can induce additional interphase forces acting on bubbles and change flow characteristics in two-phase flow system. Local measurements of air-water two-phase flow parameters, namely void fraction, interfacial area concentration and Sauter mean diameter, were conducted in a vibrating annulus. The vibration acceleration ranged from 0.05 g to 0.24 g. Nine inlet flow conditions were selected, which covered bubbly and slug flows. No significant difference was observed in the time-averaged flow parameters between reference and vibration experiments. During vibration cycle, the radial two-phase flow distribution varied continuously with vibration phase. The change in non-dimensional line-averaged void fraction decreased with the increase of void fraction. As the vibration amplitude increased from 3.04 mm to 10.08 mm, the vibration effects on variations of local flow parameters were greatly enhanced with the mean relative variation (MRV) of void faction reached up to 55%. The study on MRV of local parameters showed that the vibration effects were strongly related to specific flow conditions. With the increase of superficial gas velocity and void fraction, the vibration effects were reduced and the MRV of void fraction decreased from 55% to 18%. [ABSTRACT FROM AUTHOR]

An experimental study on air-water two-phase flow under vibration condition has been conducted using double-sensor conductivity probe. The test section is an annular geometry with hydraulic diameter of 19.1 mm. The vibration frequency ranges from 0.47 Hz to 2.47 Hz. Local measurements of void fraction, interfacial area concentration (IAC), and Sauter mean diameter have been performed along one radius in the vibration direction. The result shows that local parameters fluctuate continuously around the base values in the vibration cycle. Additional bubble force due to inertia is used to explain lateral bubble motions. The fluctuation amplitudes of local void fraction and IAC increase significantly with vibration frequency. The radial distribution of local parameters at the maximum vibration displacement is specifically analyzed. In the void fraction and IAC profiles, the peak near the inner wall is weakened or even disappearing and a strong peak skewed to outer wall is gradually observed with the increase of vibration frequency. The nondimensional peak void fraction can reach a maximum of 49% and the mean relative variation of local void fraction can increase to more than 29% as the vibration frequency increases to 2.47 Hz. But the increase of vibration frequency does not bring significant change to bubble diameter. [ABSTRACT FROM AUTHOR]

• The widely-accepted view that flow maldistribution degrades heat exchanger (HX) performance is insufficiently accurate. • What determines HX performance is not the maldistribution itself, but the degree of maldistribution mismatch. • Flow maldistribution needs not necessarily be avoided, but instead matched between two fluid streams, to maximize the HX performance. • The new HX design paradigm could result in CSP capital cost savings as large as 16.6%. Compact heat exchangers have gained increased attention in recent years, particularly in demanding applications where high temperatures, high pressures, and/or high power densities are required. For decades, the heat exchanger (HX) community believes that flow maldistribution is a key factor for HX effectiveness, that is, reducing the degree of flow maldistribution (MALD) can help increase the HX effectiveness. Therefore, significant efforts have been devoted in the past to optimizing the header geometry to minimize flow maldistribution. This work was initially motivated by this, and the original goal was to figure out a HX header design with the lowest maldistribution. However, by systematically constructing a comprehensive maldistribution matrix, the analysis revealed that the HX effectiveness is not actually determined by the MALD, but instead dominated by the degree of maldistribution mismatch (MISM). This conclusion was also theoretically generalized, which indicated that matching of the local heat capacity rate is key for achieving maximum performance. The MISM provides a local means of tracking this information, while the MALD only provides a global approximation of the maldistribution itself. With this new perspective, flow maldistribution needs not necessarily be avoided, but instead matched between two fluid streams, to improve the HX performance. We demonstrated that by carefully designing the header geometry to match the velocity profiles of the two fluids in a 2 MW PCHE with molten salt and supercritical carbon dioxide (sCO2) as the heat transfer fluids, the HX could achieve a higher effectiveness even when the maldistribution increased. A technoeconomic study using a CSP system as an example revealed that the use of this new HX design paradigm could result in CSP capital cost savings as large as 16.6%. [ABSTRACT FROM AUTHOR]

• Local measurement in churn to annular flows was conducted using the novel probe. • Two-phase flow parameters of droplets, bubbles, and ligaments were measured. • Volume fraction of gas and droplets both showed center-peak distribution. • Volume fraction of ligaments and bubbles both showed near-wall-peak distribution. • The increase of gas and liquid flux caused an increase of the bubble volume fraction. The local phase distribution of adiabatic air-water two-phase flow in churn-turbulent to annular flows was experimentally investigated using the Two-Sensor Droplet-Capable Conductivity Probe (DCCP-2). The experiments were performed in a 2.54 cm inner diameter vertical round pipe at downward flows. Twelve flow conditions in churn-turbulent to annular flows were chosen for the current study. The inlet superficial gas velocity ranged from 5.68 m/s to 17.19 m/s, and the inlet superficial liquid velocity ranged from 1.21 m/s to 2.64 m/s. The DCCP-2 is able to distinguish droplets, liquid ligaments, bubbles, and continuous gas. The radial profiles of volume fraction, frequency, axial interfacial velocity, and chord length of droplets, ligaments and bubbles were measured. The experimental results showed that gas volume fraction had a center-peak distribution, and the maximum gas volume fraction in centerline could exceed 95%. The volume fraction of droplets was very low, only on the order of 0.1%. Droplet concentration also showed a center-peak distribution. Both ligament and bubble volume fraction showed a near-wall peak distribution. The ligament volume fraction in the near wall region could reach 80%, which indicated the existence of a film region near the wall and likely includes surface waves on the film. The volume fraction profile of gas, droplets, ligaments, and bubbles showed the tomography structure in churn-turbulent to annular flows. The effects of flow conditions on phase distribution were also studied. An interesting feature was that both the increase of gas flux and the increase of liquid flux tended to increase the bubble volume fraction. The statistical distribution of droplet size was found to be invariant of radial position. The droplet velocity at centerline was generally larger than the superficial gas velocity, which can be attributed to the gravitational effects at downward flows. [ABSTRACT FROM AUTHOR]

• A 2 MW PCHE is designed for high temperature (>700 °C) molten chloride salt CSP. • Zirconium carbide/tungsten (ZrC/W) is used as the structural material for the PCHE. • ZrC/W-based PCHEs are favored over nickel alloy-based and stainless steel-based PCHEs. To increase the power cycle efficiency and lower the levelized cost of electricity (LCOE) of concentrated solar power (CSP) plants, printed circuit heat exchangers (PCHEs) capable of operating above 700 °C with molten chloride salt and a sCO 2 -based fluid are needed. In this paper, the design of a high-pressure, high-temperature, 2 MW PCHE comprised of a thermomechanically-robust, zirconium carbide/tungsten (ZrC/W) composite is conducted for CSP plants. The ZrC/W composite is a material with high thermal conductivity, high stiffness, and high failure strength at high temperatures, along with excellent resistance to thermal cycling and thermal shock. In this work, a thermomechanical design analysis was conducted to select appropriate material thicknesses of the ZrC/W plates, and to determine the geometrical dimensions and the thermal performance of the PCHE. The influences of the plate number and heat exchanger length on power density and pressure drop have also been systematically investigated. Economic analyses were conducted to compare the cost of ZrC/W-based PCHEs to those comprised of IN740H (a state-of-the-art, nickel-based superalloy) and 316 stainless steel (316SS). At a sufficiently high plate production rate, the manufacturing cost of ZrC/W-based PCHEs can be significantly lower, while achieving a much higher power density, compared with state-of-the-art, nickel alloy-based and stainless steel-based PCHEs. [ABSTRACT FROM AUTHOR]

• A novel probe is developed for local measurement in liquid dispersed flows. • Locl measurements in churn-turbulent to annular transition region are performed. • The probe's capability in measuring gas-dispersed flows is demonstrated. In churn to annular transition, the liquid could exist as a dispersed phase (droplets). But the conventional probe is not able to distinguish the dispersed liquid phase from the gas phase due to electrode design. To extend the measurement capability of the conductivity probe, the Two-Sensor Droplet-Capable Conductivity Probe (DCCP-2) is developed, which incorporates one more sensor to distinguish the dispersed liquid phase from the gas phase and the continuous liquid phase. This paper presents the design of DCCP-2, such as the circuit design, signal processing, and probe manufacturing process. Its capability in measuring liquid-dispersed flow is demonstrated by experiments in churn to annular transition conditions. Also, it still preserves the capability of measuring bubbly flow, which is demonstrated through benchmark experiments with the conventional probe. Therefore, the DCCP-2 serves as a reliable measurement instrument covering a wide range of two-phase flow regimes from bubbly flow to annular flow. [ABSTRACT FROM AUTHOR]

• A method to predict velocity profile in rod bundle has been proposed. • Gas velocity distribution can be simplified to turbulent velocity distribution. • Liquid velocity distribution can be derived by interfacial shear stress correlation. • The predictions show reasonable agreement with data. Annular flow in rod bundle occurs frequently in boiling water reactor (BWR) under normal operation and in pressurized water reactor (PWR) under accident conditions. When annular flow occurs, rod film thickness is one of the most important parameters for safety concern. In order to calculate rod film thickness distribution inside rod bundle, both gas and liquid velocity distribution inside rod bundle is necessary. Also, flow distribution inside rod bundle is of importance for calculation of interfacial shear stress and other parameters in each sub-channel. However, entrained droplets in gas core can sharpen the gas velocity profile while existing rods can flatten the gas velocity profile, making it different from that of single phase flow. The liquid velocity distribution inside rod bundle is hard to obtain from experiment because of the complex structure of its geometry. Due to these reasons, this paper deducts gas and liquid velocity profiles from existing adiabatic rod film thickness data. Even the gas velocity profiles for different flow conditions are different, there exists a gas velocity profile that can be applied to all current flow conditions with minimum uncertainties. Based on that gas velocity profile, liquid velocity profile can be calculated. From these flow velocity profiles, film thickness on each rod and interfacial shear stress in each sub-channel can be calculated. [ABSTRACT FROM AUTHOR]

A new conductivity probe design has been developed in order to measure dispersed liquid particles in churn-turbulent and annular flows which cannot be detected by the conventional conductivity probes. The probe incorporates a common sensor near the measurement point to detect the local conductance signals between two sensor tips that are 150 µm apart. Once fully developed, the probe is capable of measuring local two-phase parameters of various fields including small and large bubbles, large liquid droplets and a continuous liquid field. Preliminary benchmarking studies of the probe have been performed with a specially designed droplet dispensing setup. A high speed imaging system is used to provide benchmarking data such as droplet diameter, velocity, and chord length, for individual droplets, and the time-averaged data such as droplet volume fraction, volumetric flux and interfacial area concentration. Reasonable agreement has been obtained by comparing individual droplets with a diameter range of 1.4–4.0 mm, and a velocity range of 1.6–4.8 m/s. For time- and area-averaged parameters, the results from two test runs show that the maximum absolute relative errors are 6.42%, 7.43% and 5.72% for droplet volume fraction, interfacial area concentration and volumetric flux, respectively. The error is within 5% compared with the global droplet flow rate measurement. [ABSTRACT FROM AUTHOR]

Highlights • A study is performed on the effect of the electrode distance to impedance meter. • A sensitivity study of the electrode distance is done by numerical simulation. • Void fraction measurement is done using impedance meters with different geometries. • Gamma densitometer and conductivity probes are used in the experiment as comparisons. • The performance of flow regime identification is analyzed with the impedance signals. Abstract Experimental study on the effect of electrode distance on the performance of impedance void meter is presented. The study includes multiple approaches. In the numerical simulation approach, based on the mathematical similarity between electric and temperature fields, the electric fields with different electrode distances are simulated using ANSYS Fluent to generate temperature fields. In the experimental study, an adiabatic air-water experiment is performed to compare two impedance meters with different electrode distances. The test conditions cover the flow regimes from bubbly flow to annular flow. Gamma densitometer and multi-single sensor conductivity probe are utilized for the void fraction comparison. The effect of the electrode distance on the flow regime identification using the artificial neural network is investigated, the signals of the different impedance meters are applied as the inputs for the artificial neural network. It is found that impedance meter with a properly large electrode distance has better performance in void fraction measurement, especially at high superficial velocity. [ABSTRACT FROM AUTHOR]

• Obtain the general rules and methods of RELAP5/MOD 3.4 to simulate CMT. • The RELAP5/MOD 3.4 can accurately simulate most of the CMT operating conditions. However, large uncertainty is expected in the simulation of delayed time. • The CMT delayed drainage experiment cannot verify the accuracy of the RELAP5/MOD 3.4 direct contact condensation model. The RELAP5 code has been developed for best-estimate transient simulation of light water reactor coolant systems during accident and has been widely used in the field of nuclear thermal-hydraulics and safety analysis. The core make-up tank (CMT) is a crucial device of the passive core cooling system in the AP series pressurized water reactor nuclear power plants. Since the CMT not only provides full-pressure injection but also triggers auto depressurization system by its level signal during small-break loss of coolant accident. Therefore, it is necessary to carry out CMT separate effect tests to verify the feasibility and accuracy of RELAP5/MOD 3.4 to simulate CMT. The previous studies found that the lower version of the RELAP5 program has problems in simulating CMT, and the applicability of the latest RELAP5/MOD3.4 has not been evaluated. In this paper, a CMT separate effect test facility were presented and then RELAP5/MOD 3.4 was used to establish a CMT test bench model to evaluate the applicability, and obtain the general rules and methods of RELAP5/MOD 3.4 to simulate CMT. It's found that RELAP5/MOD 3.4 can accurately simulate most of the CMT operating conditions. However, large uncertainty is expected in the simulation of delayed time and the flowrate of delayed drainage stage is not affected by the heat transfer coefficient of steam direct contact condensation. The simulation of delay time depends on experimental data and the CMT delayed drainage experiment cannot verify the accuracy of the RELAP5/MOD 3.4 direct contact condensation model. [ABSTRACT FROM AUTHOR]

The Purdue NMR (Novel Modular Reactor) represents a BWR-type small modular reactor with a significantly reduced reactor pressure vessel (RPV). Specifically, the NMR is one third the height and area of a conventional BWR RPV with an electrical output of 50 MWe. Experiments are performed in a well-scaled test facility to investigate the thermal hydraulic flow instabilities during the startup transients for the NMR. The scaling analysis for the design of natural circulation test facility uses a three-level scaling methodology. Scaling criteria are derived from non-dimensional field and constitutive equations. Important thermal hydraulic parameters, e.g. system pressure, inlet coolant flow velocity and local void fraction, are analyzed for slow and fast normal startup transients. Flashing instability and density wave oscillation are the main flow instabilities observed when system pressure is below 0.5 MPa. And the flashing instability and density wave oscillation show different type of oscillations in void fraction profile. Finally, the pressurized startup procedure is recommended and tested in current research to effectively eliminate the flow instabilities during the NMR startup transients. [ABSTRACT FROM AUTHOR]

Passive safety systems are widely used in the advanced nuclear reactors, such as AP1000 and CAP1400. As an important component in the passive safety system, the core make-up tank (CMT) plays a key role for the safety injection and the core decay heat removal. Although the structure of the CMT system is simple, it involves many thermal-hydraulic phenomena, most of which significantly impact the peak temperature of the fuel cladding and the liquid level of the core. Since the CMT not only provides full-pressure injection but also triggers the auto depressurization system by its liquid level signal during small-break loss of coolant accident (SBLOCA), its thermal hydraulic behaviors significantly affect the performance of the passive core cooling system (PXS). Therefore, separate effect tests are inevitable to verify the feasibility of CMT applied to passive safety and injection systems and to study its operating characteristics. In this work, a separate-effect CMT test facility was designed and constructed. Experiments were performed to investigate the CMT's thermal hydraulic characteristics under the effects of system pressure, steam condensation, and flow friction. The study found that the delay drainage phenomenon is the result of a pressure balance between the pipe line and the gravity head. Whether delayed drainage occurs depends on the degree of subcooling of the fluid at the top of the CMT. The flowrate is not affected by the heat transfer coefficient of steam direct contact condensation. The flow friction of the drainage pipeline has no effect on the delayed drainage process, but has a great influence on the gravity drainage process. [ABSTRACT FROM AUTHOR]

The core make-up tank (CMT) is a crucial device of the passive core cooling system in the Westinghouse AP-series pressurized water reactor nuclear power plants. In the prototype plant, the CMT is a vertical cylindrical vessel with hemispherical top and bottom end sockets. Although the structure of the CMT system is simple, it involves many thermal-hydraulic phenomena, most of which significantly impact the peak temperature of the fuel cladding and the liquid level of the core. Since the CMT not only provides full-pressure injection but also triggers the auto depressurization system by its liquid level signal during small-break loss of coolant accident (SBLOCA), its thermal hydraulic behaviors affect the performance of the passive core cooling system (PXS). Therefore, it is necessary to carry out separate effect tests to verify the feasibility of CMT applied to passive safety and injection systems and to study the operating characteristics of CMT. This article first presents the design of a CMT separate effect test facility and analysis of the operating characteristics of the CMT with the experimental data. The influence of pressure, non-condensable gas, wall condensation, and thermal resistance on the CMT behavior were investigated. The study found that the non-condensable gas would stratify in the CMT, which had negligible effect on the wall condensation, but could greatly reduce the gas-liquid interface condensation. [ABSTRACT FROM AUTHOR]

• The new wall shear stress model is simple with good prediction accuracy. • Wave velocity correlation's prediction improved after modifying air density. • Two two-phase multiplier correlations have good prediction accuracy. • Two pure empirical correlations cause deviations at pressure above 1 atm. Annular flow has been widely applied to industrials such as oil extraction, refrigerators, nuclear reactors and so on. Wall shear stress is one of the most important parameters for studying two-phase annular flow. When calculating wall shear stress from interfacial shear stress, the relationship between the wall shear stress and the interfacial shear stress is complicated and needs some parameters such as film thickness which requires high-precision measurement technique to acquire accurately. Besides, some wall shear stress correlations developed for annular flow are applicable to flow conditions at atmospheric pressure only. Because of these reasons, four databases for vertical upward co-current adiabatic air-water two-phase annular flow have been collected with wide range of flow conditions and different pressure conditions. Using these data, a simplified model between wall shear stress and interfacial shear stress is proposed. The new model is derived from the force balance analysis of the liquid film and gas core. As void fraction of annular flow is close to 1, the relationship between wall shear stress and interfacial shear stress can be simplified to be linearly. Based on this model and existing interfacial shear stress correlations, the wall shear stress can be calculated directly from flow conditions and fluid properties that can be readily obtained from experiments. Also, a new wall shear stress correlation calculated from wave velocities has been suggested. It is developed from frictional pressure drop correlation. Correlations for two-phase multiplier and wave velocity as well as pure empirical correlations also have been compared with existing databases. From data comparison, the newly proposed correlations in this study agree well with experimental data and other correlations with good prediction accuracy have been pointed out. [ABSTRACT FROM AUTHOR]

• Benchmark simulation of bubbly flow in an annular channel under stationary condition. • A numerical modeling method for bubbly flow under structure vibration condition. • Detailed comparison of numerical results and experimental data. Earthquake vibrations have great impacts on two-phase flow in the reactor core. In this paper, numerical simulations were conducted to identify the air-water two-phase distribution in a vertical upward annular channel under horizontal vibration condition. The inner and outer diameter of the annular channel are 19.1 mm and 38.1 mm. The simulations were performed based on the two-fluid Euler-Euler modeling framework. Benchmark simulations for bubbly flow under stationary condition were firstly conducted to validate the up-to-date interfacial force models. Applicability of the interfacial force models were evaluated and a set of Ishii-Zuber drag force model, Tomiyama lift force model, Frank wall lubrication force model and Burns turbulent dispersion force model was found to provide the best agreement with experimental results. Based on this work, two-phase flow distributions under vibration condition were studied. Horizontal vibration of the annular channel was simulated by introducing simple harmonic motion. The vibration frequency and vibration amplitude were set to 1.97 Hz and 9.5 mm, respectively. Additional interfacial force caused by inertial effect was imposed on disperse bubbles under vibration condition, which promoted the lateral migration of bubbles. The numerical results showed that the distributions of local interfacial parameters varied periodically along the vibration direction. A maximum increase of about 32% was observed in the peak void fraction compared to that under stationary condition for case 3. But the distribution of local flow parameters remained unchanged in the direction perpendicular to the vibration. A good consistency between the simulation results and the experimental data was found. The comparison results indicate that the present modeling method is able to predict the two-phase distribution of bubbly flow under vibration condition. But an inhomogeneous model considering bubble interactions and grouping is suggested for two-phase flow beyond bubbly regime. [ABSTRACT FROM AUTHOR]