Measurement of both gas and liquid velocity profiles for bubble-induced turbulent flow H. Takiguchi 1*, M. Furuya 1, T. Arai 1, T. Kanai 1 1: Central research institute of electric power industry (CRIEPI) * Correspondent author: t-hiro@criepi.denken.or.jp Keywords: Laser Doppler velocimetry, Wire mesh sensor, Bubbly flow, Velocity, Gas density dependence ABSTRACT To understand the forces acting at the interfaces of a bubble under high pressure and temperature in a nuclear reactor, detailed velocity profiles of both gas and liquid phases are important. However, direct numerical simulation (DNS) requires vast time and memory space to resolve them in phasic interfaces, so the experimental approach is still crucial in multi-dimensional two-phase flow simulation. In this study, a novel technique to measure gas and liquid velocity profiles using laser-doppler velocimetry (LDV) and wire mesh sensor (WMS) has been developed. In this paper, we focused on the effect of gas density dependence on a two-phase velocity field, since gas density of water vapor at high pressure and temperature is widely different from that in the atmosphere. To make a detailed correlating equation using the drift-flux model, taking into account the effect of gas density, the gas density dependent data of void fraction, gas velocity, liquid velocity and its fluctuation in one-dimensional vertical direction were measured with helium, nitrogen, air and argon. Each gas, with differences of up to approximately 10 times, was selected to simulate change of gas density caused by change of temperature and pressure. A vertical rectangular test loop 200 mm wide 30 mm deep 800 mm high was fabricated to allow laser beams to access the test section, and to simulate the 2-dimensional case in which bubble motions in the depth direction are inhibited to simplify complicated turbulent behavior. The experiment targeted bubbly flow with uniform and trans-mutative bubbles (apparent diameter: 3-5 mm). From the measured data, the negative trend that gas drift velocity shifted closer to liquid phase velocity with increasing gas density was indicated. Besides, it was confirmed that this mutual trend was the same in all conditions of total volumetric flux. Finally, the void fraction dependence of gas drift velocity was compared with the drift-flux model. While this model indicates that gas density has little effect on gas drift velocity, the experimental data suggested the existence of gas density effect. It can be considered that a change of relative frictional forces is generated in phasic interactions due to bubble deformation undefined in the drift-flux model. For this reason, it was revealed that gas density gives non-negligible effect on velocity field in bubbly flow involving bubble deformation. 1. Introduction Direct Numerical Simulation (DNS), which can be regarded as a numerical experiment, has been yielding a solid numerical database for single-phase flow (of gas or liquid).
Applicability of DNS for gas-liquid flow is limited to a simple flow structure such as a stratified flow or a single bubble in a liquid column, since DNS requires vast time and memory space to resolve phasic interphases for three dimensional thermal-hydraulic analysis of a boiling water reactor (BWR)[1,2]. Therefore, experimental methods are yet crucial to explore the interfacial structure [3-7]. For instance, Prasser et al. [8] have developed the Wire-Mesh Sensor (WMS) to acquire the cross-sectional profile of void fraction and one-dimensional bubble velocity. In order to evaluate computational multi-fluid dynamics (CMFD) for two-phase turbulent flow from the experimental approach, three-dimensional liquid-phase velocity vector and turbulent property must be determined for not only the gas phase, since the liquid-phase momentum is much larger than that of the gas-phase. Laser-Doppler Velocimetry (LDV), which is a non-intrusive measurement method, has been able to quantify the liquid-phase velocity vector and turbulent property at high temporal and spatial resolutions. The drawback of LDV is the scattering of light on the gas-liquid interface due to the difference of the reflective index of light. The goal of this experimental study was to investigate combined methods to balance the pros and cons of WMS and LDV to measure the velocity profile of both phases. In this paper, we focused not only on developing measuring technique, but also on the gas density effect of a twophase velocity field at gas-liquid interfaces in order to clarify forces acting on bubbles with change of pressure and temperature. In a nuclear reactor under high pressure, this force has been calculated by a correlating formula using water and air [9]. However, since the density of vapor differs extremely from that of air under high pressure, understanding the gas density dependence of forces acting around bubbles is important. This experiment demonstrated gas density effect on a two-velocity field of two-phase flow from measurement of gas velocity, void fraction, liquid velocity and its fluctuation in a 1D-vertical direction. 2. Experimental techniques The WMS is able to measure void fraction, three-dimensional gas-phase velocity vector and bubble size distribution by detection of a bubble with an electric potential of wire [10]. LDV is applicable to measurement of liquid-phase velocity by frequency gap sensing of scattered light received from sub-micron tracer particles with approximately the-same density as water, such as nylon and silver-coated midair glass beads, which are passed through fringe spacing patterns induced by two laser beams of each equal wavelength and intensity.
Fig. 1 Experimental apparatus with rectangular test section. (a) (b) 20 mm Fig. 2 Snap-shot of air and water near the measuring point in the test section; (a) j L=0.131 m/s, jg=0.028 m/s and (b) jl=0.052 m/s, jg=0.017 m/s. The experimental apparatus applicable to the above-mentioned measuring techniques will be introduced. The test section in Fig. 1, made of acrylic resin to visualize and measure twophase flow, is a rectangular channel 0.8 m long, 200 mm wide and 30 mm deep, in order to allow laser beams accessing into the test loop and assumes the 2D case in which bubble motion in the depth (y) direction is inhibited to simplify complicated turbulent behavior. WMS and LDV
measurement and visualization with high-speed digital video camera (HSC) of two-phase flow were employed at 0.7m above the mixing area. The tetragonal snap shot of bubbly flow in the inlet superficial liquid velocity j L=0.131 m/s, gas velocity j G=0.028 m/s and j L=0.052 m/s, gas velocity jg=0.017 m/s are illustrated in Fig. 2. It was confirmed that 2D-assumed and approximately-4 mm-uniform bubbles were transformed and rose as shown in these snap shots. Liquid and gas phases in the test loop are water and air at room temperature and atmospheric pressure. Liquid was supplied from the pump and flowed in an upward direction, gas was fed into single-phase flow through a horizontally long air injector, which was selected to make bubbles flow uniformly in a cross-wise direction. Wire electrodes for WMS -measurement made of 250 µm-stainless steel were utilized as transmitter and receiver electrodes, and the WMS had 576 points (9 64 points) for local measurement regions to be able to measure velocity vector profiles. The axial distance between transmitter and receiver wires was 2.5 mm, and the pitch of both transmitter and receiver wires was 3 mm. Additionally, another WMS measuring section was installed with a 30 mm gap to measure gas-phase velocity vector by detecting time delay of the gas-phase behavior between the pair of WMS with cross-correlation. Our target in the experiments was the quantitative-comparison of 2D-9 64 matrix distribution data with CMFD using experimental data with void fraction, gas-phase velocity vector, bubble size, liquid-phase velocity vector and turbulent properties. In this experiment, as a part of our aim, in order to demonstrate the verification of this combined-measurement method and to elucidate gas density effect of a liquid and gas velocity field, we measured time-averaged liquid-phase velocity and root mean square (RMS), gas-phase velocity and void fraction in 1Dvertical direction at the voluntary measuring point while changing gas density from 0.1785 kg/m to 1.784 kg/m. 3 3 3. Results and discussion In this experiment, we targeted the bubbly-flow, which is moved upward twodimensionally with approximately uniform 4 mm bubbles, the void fractions measured by WMS were less than those of the homogeneous model (HEM) due to gas velocity drift as shown in Fig. 3. Then, since each inlet superficial gas-phase volumetric flux with four types of gases used in this experiment was regulated to be roughly the same by adjusting from density and specific heat of each gas to investigate only the effect of gas density on the velocity field, void fractions acquired by WMS were also indicated to be approximately equal as shown in 20 mm 20 mm
tetragonal snap shots of Figs. 2 and 4. Inlet flow rate (total volumetric flux) was set from 0.068 m/s to 0.158 m/s. Fig. 3 Flow regime: bubbly flow, comparison of time-averaged void fraction between measurement values of WMS with homogeneous model (HEM). (a) (b) 20 mm Fig. 4 Snap-shot of argon near the measuring point in the test section; (a) jl=0.131 m/s, jg=0.028 m/s and (b) jl=0.052 m/s, jg=0.017 m/s. Table 1 Gas density (unit: kg/m 3 ) [11]. Air Helium Nitrogen Argon 1.293 0.1785 1.250 1.784 In order to indicate the measurement verification of the two combined methods, we compared the time-averaged void fraction, gas velocity, liquid velocity and its fluctuation in 1D-
vertical direction with changing gas density in helium, nitrogen, air and argon in Table 1. The relationship of liquid velocity with gas drift velocity regarding gas density dependence at the Fig. 5 Relationship of time-averaged data between gas-phase velocity and liquid-phase velocity with helium, nitrogen, air and argon. Fig. 6 Time-averaged liquid-phase velocity of helium, nitrogen, air and argon.
Fig. 7 Time-averaged fluctuation (RMS) of liquid-phase velocity of helium, nitrogen, air and argon. Fig. 8 Time-averaged gas-phase velocity of helium, nitrogen, air and argon. Fig. 9 Correlating equation line with gas velocity trend data (Fig. 8) and drift-flux model regarding drift velocity versus void fraction measured by wire mesh sensor. measuring point was demonstrated through the liquid-phase velocity data by LDV and the gasphase velocity data by WMS. The time-averaged velocities (ul) and root mean square (RMS, ul-rms) were worked out with LDV for data of 50,000 liquid-phase velocities in the vertical direction; this data included only extracted velocity information fed by supplied tracer, without that caused by gas-liquid phasic interactions. Additionally, the data (WMS) of 10,000 one-dimensional gas-phase velocities and void fractions, which were gained at 1000 Hz for 10 seconds, were calculated as time-averaged velocities (ug) and void fraction (α). The reliability of LDV measurement values was confirmed
with single liquid-phase flow, then the measured values were well accorded with ±10 % from those of the readings in the range of superficial liquid-phase volumetric flux from 0.05 m/s to 0.3 m/s. In comparison with liquid-phase and gas-phase velocity under roughly the-same conditions of void fraction in Fig. 5, gas drift velocities were increased with decrease of gas density for all gases. Additionally, the dashed line indicates the HEM model, which defined that liquid velocity is equal to gas velocity. This trend indicated that gas drift velocity shifted closer to liquid-phase velocity with increasing gas density. Time-averaged data of liquid velocity, its fluctuation (RMS) and gas velocity are shown in Figs. 6, 7 and 8. These figures show that the trend of liquid velocity works in a positive direction and that of gas velocity works in a negative direction toward gas density. This result might suggest that flow velocity fluctuations were generated due to momentum transfer in bubbly flow. Furthermore, these transfer trends were demonstrated without reference to inlet superficial liquid and gas velocity from 0.068 m/s to 0.158 m/s. Finally, the gas drift velocity was compared with the drift-flux model using water and air [12] regarding void fractions for WMS, which are shown by the dashed lines in Fig. 9. The calculation with the drift-flux model was worked using Eq. (1). The void fraction in Eq. (1) employed the values measured by WMS; other thermo-physical properties were used for reference [11]. u Gd = ( α) 1 ( ρ ρ ) 4 1.5 σg L G 1 2 2 ρl (1) While the calculated lines estimated using the drift-flux model [12] did not shift relative to the changing gas density from 0.1785 kg/m to 1.784 kg/m, experimental plots were negatively 3 3 changed for gas density. Although the drift-flux model defines a bubble as a spherical form, bubbles with diameters of approximately 4 mm practically move and deform from side to side and up and down as in the snapshots in Figs. 2 and 4. Therefore, it can be considered that the deformations of bubbles generated in two-phase flow affect bubbles to mass change with change of relative frictional force, in other words, density change. As a result, in the presence of bubble deformation, the gas density effect on the velocity field in two-phase flow is not negligible. 4. Conclusions
In order to understand forces acting on bubbles in two-phase flow under high pressure and high temperature in a nuclear reactor, we developed a combined measurement method applicable to gas phase velocity, void fraction, liquid-phase velocity and its fluctuation in a turbulent field using LDV and WMS. It compared gas velocity, void fraction, liquid velocity and its fluctuation (RMS) with gas density dependence for four types gases (helium, air, nitrogen and argon) to confirm the reliability of this compensated method. The trends found through the experiment indicated that (1) gas drift velocity shifted closer to liquid-phase velocity with increasing gas density, (2) gas density affected the liquid and gas velocity field at bubble interfaces, even though the effect of buoyant force caused by gasliquid density difference was negligible, and (3) in two-phase flow with roughly 4 mm bubbles, bubbles move and deform from side to side and up and down. Therefore, it can be considered that the deformation of bubbles causes mass change of bubbles with change of relative frictional force, in other words, density change. These suggest that gas-density has a considerable effect on forces acting on bubbles under a liquid and gas velocity field. Acknowledgment The authors are grateful to Mr. Takeo Yoshioka and Mr. Tsugumasa Iiyama of Electric Power Engineering Systems Co., Ltd. for their assistance in these experiments. Nomenclature g: Gravity [m/s ] 2 jl: Inlet superficial liquid volumetric flux [m/s] jg: Inlet superficial gas volumetric flux [m/s] jt: Total volumetric flux [m/s] ul: Liquid-phase velocity [m/s] u L-RMS: ug: Liquid-phase velocity fluctuation (RMS) [m/s] Gas-phase velocity [m/s] ugd: Gas drift velocity [m/s] α: Void fraction [-] ρ: Density [kg/m ] 3 ρl: Liquid density [kg/m ] 3
ρg: Gas density [kg/m 3 ] σ: Surface tension [N/m] References [1] W. D. Pointer et al., Proc. of 16 Int. Conf. Nuclear Engineering, (2008). th [2] C. Kocar et al., Nucl. Eng. Des., 239(2009), 1839. [3] F. Wang et al., Chinese J. Sensors and Actuators, 20-3(2007), 628. [4] F. Chen et al., Adv. Mat. Res., 361-363(2012), 671. [5] H. M. Prasser et al., Flow Meas. Inst., 16(2005), 73. [6] M.E. Shawkat et al., Int. J. Multiphase Flow, 33(2007), 300. [7] S. Hosokawa et al., Int. J. Heat and Fluid Flow, 40(2013), 97. [8] H. M. Prasser et al., Flow Meas. Inst., 9(1998), 111. [9] T. Hibiki et al., Chem. Eng. Sci., 62(2007), 6547. [10] Kanai et al., Int. J. Multiphase Flow, 46(2012), 75. [11] Japan Society of Thermophysical Properties, Thermophysical Properties Handbook (2008). [12] M. Ishi, Argonne National Laboratory-77-47, (1977).