Experimental Study on Liquid Film Dryout Under

th

Proceedings of the 17 International Conference on Nuclear Engineering ICONE17 July 12-16, 2009, Brussels, Belgium

ICONE17-75400 EXPERIMENTAL STUDY ON LIQUID FILM DRYOUT UNDER OSCILLATORY FLOW CONDITIONS Yosuke Yamagoe Department of Mechanical Engineering Osaka University 2-1, Suita-shi, Osaka 565-0871, Japan

Taisuke Goto Department of Mechanical Engineering Osaka University 2-1, Suita-shi, Osaka 565-0871, Japan

Tomio Okawa Department of Mechanical Engineering Osaka University 2-1, Suita-shi, Osaka 565-0871, Japan ABSTRACT The use of high power density core is one of the promising ways to improve economic efficiency of advanced boiling water reactors. It is however known that in boiling two-phase flows, an increase in power density commonly reduces the margin to the onset of unanticipated flow instability. Hence, in the development of a boiling water reactor of high power density core, ability to predict the occurrence of boiling transition is considered indispensable even when the coolant flow rate is not in the steady state. In the present work, sinusoidal oscillation was applied to the inlet mass flux and the experimental measurement of the critical heat flux was carried out under flow oscillation conditions. It was shown that the critical heat flux decreases monotonically with increased values of oscillation amplitude and oscillation period. These results are consistent with experimental data reported by previous investigators. A simple theory was then proposed to estimate the critical heat flux in oscillatory flow condition. Considering the application to the advanced boiling water reactors, the triggering mechanism of the critical heat flux condition is supposed to be the liquid film dryout in annular two-phase flow regime of high vapor quality. Under the flow oscillation condition, it is expected that long waves are formed on a liquid film due to the time variation of inlet mass flux. Assuming that the wave evolution within a boiling channel is influential in the occurrence of the local dryout of a liquid film, an available nonlinear wave theory was applied to the estimation of critical heat flux under the flow oscillation condition. It was demonstrated that the critical heat fluxes

measured under the oscillatory conditions agree with the proposed theory fairly well.

NOMENCLATURE c fi fw G J L P q t T u uf0 z0

wave velocity (m/s) interfacial friction factor wall friction factor mass flux (kg/m2s) superficial velocity (m/s) heated length (m) pressure (Pa) critical heat flux (W/m2) time (s) time period (s) or fluid temperature (K) velocity (m/s) characteristic film velocity (m/s) characteristic distance (m)

Greek symbols G

amplitude of flow oscillation (kg/m2s) density (kg/m3) wavelength (m) volume fraction

Superscript *

1

dimensionless

Copyright © 2009 by ASME

Subscripts AVE EX f IN MAX MIN OSC v

average outlet liquid film inlet maximum minimum flow oscillation vapor phase

1. INTRODUCTION Further improvements of safety and economic efficiency are required in advanced nuclear reactors. As one of the promising ways to satisfy these requirements, a high power density core is frequently selected in recent designs of advanced boiling water reactors (Aoyama et al., 1997; Chaki et al., 2006; Andersen et al., 2006; Arai et al., 2006). It is however known that significant temporal fluctuation of coolant flow rate could be induced when the power density is increased in boiling two-phase flow systems. Hence, accurately predicting the critical heat flux in forced-convective boiling even when the flow instability is initiated is considered one of the crucially important issues to develop sufficiently reliable advanced boiling water reactors. As described in the extensive review by Ozawa et al. (2002), many experimental and analytical studies have been conducted in order to clarify relationships between the flow instability and the critical heat flux in a boiling channel. It was found that the critical heat flux decreases noticeably if the flow instability is incepted. In particular, Ozawa et al. (1993) and Umekawa et al. (1999) measured the critical heat fluxes in boiling channels, imposing forced oscillation on the inlet mass flux. These experiments were effective to clarify the influences of individual oscillation parameters such as the oscillation amplitude and the oscillation frequency. It was clearly shown that an increase in the amplitude and a decrease in the frequency cause a reduction of the critical heat flux. There are two main mechanisms for the onset of critical heat flux condition in flow boiling: the liquid film dryout in an annular two-phase flow regime and the departure from nucleate boiling (Collier and Thome, 1994). However, the liquid film dryout is considered as the main triggering mechanism of critical heat flux condition in boiling water reactors since the vapor quality at the exit of heated section is rather high. In the present work, sinusoidal oscillation was applied to the inlet mass flux and the experimental measurement of the critical heat flux was carried out under flow oscillation conditions. The two dimensionless parameters are then introduced to interpret the dependences of the critical heat flux on the oscillation amplitude and the oscillation frequency. One is the dimensionless critical heat flux that is

the critical heat flux in the oscillation condition scaled by those in the steady states. The other is the dimensionless heated length. Under the flow oscillation condition, it is considered that a thick film region and a thin film region are present in a boiling channel, corresponding to high and low mass flow rates at the inlet. It is hence postulated that the reduction of critical heat flux associated with the liquid film dryout is minimized if the heated length is long enough for the thick and thin film regions to interact sufficiently. To derive the dimensionless heated length, the theory for nonlinear waves (Witham, 1999) is used, since the dryout of liquid film in annular two-phase flow regime was considered as the triggering mechanism of critical heat flux condition. The validity of the proposed theory is tested using experimental data measured under the flow oscillation conditions.

2. EXPERIMENTAL SETUP 2.1 Experimental Loop A schematic diagram of the experimental loop used in this work is displayed in Fig. 1. Filtrated and deionized tap water was used as a working fluid, and was circulated by a multistage pump. Two needle valves and a turbine flow meter were used to control the inlet mass flux GIN. The inlet flow rate was oscillated sinusoidally by changing the electric power applied to the circulation pump. A 15 kW plug heater was used to keep the water temperature at the inlet of the heated section TIN around 374 K. After exiting the pre-heating section, the water entered a vertical round tube of 12 mm in internal diameter and 0.8 mm in thickness; the tube was made of SUS 316 stainless steel. A 60 kW DC power supply was used to heat the test section round tube ohmically and generate steam-water annular flow inside it. The heated length, defined as the distance between the upper end of the lower electrode and the lower end of the upper electrode, was 1,360 mm. The copper electrodes were covered with rubber sheets for thermal insulation. Above the heated section, a glass window section was equipped for the visualization of a liquid film. The height of the glass section was 40 mm and the distance between the upper end of the heated section and the center of the window was 298 mm. The top of the vertical test section was connected with the upper tank that was open to the atmosphere in order to separate the vapor phase from liquid phase. The liquid phase was then returned to the circulation pump, while the vapor phase was condensed in the separator or released to the atmosphere. 2.2 Measurement Methods The fluid temperature was measured at the inlet and outlet of the preheating section with type-K thermocouples. Pressure transducers were used to measure the pressures at the inlet and

2


outlet of the heated section. The heat flux applied to the fluid was calculated from the electric current passing through the heated section and the electrical resistance of the tube material. Ten type-K thermocouples were spot-welded to the outer surface of the heated section to measure wall temperatures. Initiation of the boiling transition was expected to occur at the exit of heated section since the channel wall was heated uniformly. Four thermocouples were therefore placed as the BT (boiling transition) detectors near the exit of the heated section to detect the transition to critical heat flux condition without fault. The heated section was not thermally insulated, but heat loss to the ambient air was estimated as less than 1 kW/m2 from calculation using the heat transfer equation proposed by Churchill and Chu (1975). Experimental data were recorded every 1 ms using a data acquisition system. T: thermocouple P: pressure transducer F: flow meter

Separator

P Glass window

Buffer tank Electrode

T : BT detectors

3. EXPERIMENTAL RESULTS 3.1 Detection of onset of critical heat flux condition The measurement of critical heat flux was conducted both in the steady states and flow oscillation conditions. Displayed in Figs. 2a and b are the typical time transients of the wall temperatures TW and the inlet mass flux GIN measured when critical heat flux condition was reached; here, TW,BT denotes the wall temperatures measured by the BT detectors, and Figs. 2a and b indicate the results obtained under steady and oscillatory conditions, respectively. In the flow oscillation conditions, the characteristic temperature rise was measured by the BT detectors periodically as shown in Fig. 2b. In the present work, the wall heat flux at which the amplitude of the characteristic temperature rise exceeded 5 K was regarded as the critical heat flux. The critical heat flux in the steady state (Fig. 2a) was determined using the same method for consistency. Because of the finite heat capacity of the channel wall, the inner surface of the heated wall should be kept dry for a certain time period t for the transition to critical heat flux condition to be detected using the present method. From the results of numerical calculations using the heat conduction equation, t was estimated to within 10% of the oscillation period TOSC. Since t was reasonably shorter than TOSC, it was assumed that the measured critical heat flux was not significantly influenced by the heat capacity of the channel wall. 3.2 CHFs measured in the steady states

T T T

DC power supply

T T T Electrode P

The critical heat fluxes measured in the steady states are compared with those calculated using the generalized correlation by Katto and Ohno (1984) and the empirical correlation by Bowring (1972) in Fig. 3. It can be seen that the present data are in fairly good agreements with these correlations. To investigate the relationship between critical heat fluxes under steady states and those under oscillatory conditions, the critical heat fluxes obtained in the steady states were approximated using a secondorder polynomial. The resulting fitting curve is also indicated by the solid line in Fig. 3. 3.3 Results under flow oscillation conditions

T Bypass valve Flow control valve Pump

Pre-heater T F

Fig. 1: Schematic diagram of experimental apparatus

The critical heat fluxes measured under flow oscillation conditions qOSC are plotted against G/GAVE using TOSC as a parameter in Figs. 4a-c, where GAVE and G are the timeaveraged value and the oscillation amplitude of inlet mass flux and TOSC is the oscillation period. It can be seen that the critical heat flux measured under flow oscillation condition qOSC decreases with increased values of G/GAVE and TOSC. These dependences of critical heat flux on oscillation parameters are consistent with the experimental results reported by previous investigators (Ozawa et al., 1993; Umekawa et al., 1999).

3


GIN (kg/m2s)

100

50

50

0

140

140 TW.IN (℃)

0

120

120

100

100

140

140 TW (℃)

TW (℃)

TW,BT (℃)

GIN (kg/m2s)

100

120

100 0

100

TIN 10

20

120

TIN 0

30

10

20

30

T (s)

T (s) (a) Results obtained under steady state

(b) Results obtained under flow oscillation condition

Fig. 2: Typical time transients of inlet mass flux and wall temperatures when the critical heat flux condition is reached; results obtained under (a) steady state (PEX = 118 kPa, TIN = 372 K, GIN = 48 kg/m2s, qW = 215 kW/m2), and (b) flow oscillation condition (PEX = 115 kPa, TIN = 372 K, GAVE = 48 kg/m2s, G/GAVE = 0.48, TOSC = 5 s, qW = 119 kW/m2).

4


800

Experiment

800

Fitting Katto and Ohno

TOSC=5s

TOSC=10s

TOSC=20s

(b) intermediate mass flux

TOSC=2s

TOSC=5s

TOSC=10s

TOSC=20s

TOSC=2s

TOSC=5s

TOSC=10s

TOSC=20s

400

2

CHF (kJ/m s)

Bowring

TOSC=2s

(GAVE =48–50 kg/m s)

2

2

CHF (kW/m )

(a) low mass flux

0 800

400 2

CHF (kW/m )

2

(GAVE =94–98 kg/m s)

0 0

100

400

200

0

2

GIN (kg/m s) 2

CHF (kW/m )

Fig. 3: Critical heat fluxes measured in steady states, and those calculated by the generalized correlation by Katto and Ohno (1984) and the empirical correlation by Bowring (1972).

800

400 (c) high mass flux 2

(GAVE =186–229 kg/m s)

3.4 Dimensionless Critical Heat Flux A large number of experiments have been carried out to measure the critical heat flux in the steady states. Therefore, the critical heat flux under flow oscillation condition is scaled using the data obtained in the steady states. If the wall heat flux is increased gradually, it is considered that instantaneous dryout of liquid film is first measured when the axial location corresponding to the minimum inlet flow rate GMIN = GAVE – G arrives at the exit of the boiling channel. A notable difference caused by the flow oscillation would be the presence of a larger amount of liquid film around the axial location where the film flow rate is lowest. If the surrounding liquid has no influence on the minimum film region, the critical heat flux under the flow oscillation condition qOSC may be equal to qMIN, that is the critical heat flux in a steady state when the inlet mass flux is set to GMIN. On the other hand, if the surrounding liquid and the minimum liquid region interact sufficiently, the liquid film would be completely mixed in the axial direction. In this case, qOSC may be equal to qAVE, that is the critical heat flux in a steady state when the inlet mass flux is equal to GAVE.

0 0

0.5

1

ΔG/GAVE Fig. 4: Critical heat flux data measured under flow oscillation conditions; (a) low mass fluxes (GAVE = 48–50 kg/m2s), (b) intermediate mass fluxes (GAVE = 94–98 kg/m2s), and (c) high mass fluxes (GAVE = 186–229 kg/m2s)

Based on the above discussion,it would be appropriate that the critical heat flux under the flow oscillation condition is scaled using those in the steady states as q*

qOSC q AVE

qMIN qMIN

(1)

where q* is the dimensionless critical heat flux under an oscillatory flow condition. It can be confirmed that q* = 0 when qOSC = qMIN and q* = 1 when qOSC = qAVE.

5


3.5 Dimensionless Heated Length It is considered that a finite distance is necessary for the surrounding fluid to interact with the minimum film thickness region. To derive the characteristic length scale, we start from the simplified mass conservation equations for a liquid film: f

t

x

(

f

uf )

(2)

0

where t is the time, x is the axial coordinate, is the volume fraction, u is the velocity and the subscript f denotes the liquid film. Supposing that the wall friction force is equal to the interfacial shear force acting on a liquid film, the following equation is derived: 1 fw 2

1 fi 2

u 2f

f

v

J v2

cf

cf0

c f sin

2 x

(10)

where cf0 is the mean value of cf, cf is the amplitude and is the wave length. The time evolution of Eqs. (8)–(10) leads to the formation of shock in the distributions of f and cf (Whitham, 1999). The shock reaches a maximum strength at ts = /4 cf and then decays. In this study, the following propagation distance of wave z0 is used as the characteristic length scale:

(3)

where fw is the wall friction factor, is the density, fi is the interfacial friction factor, J is the superficial velocity and the subscript v denotes the vapor phase. The following simple expressions are used for fw and fi (Wallis, 1969): fw

Equation (9) indicates that cf and f are linearly related in the present formulation if the compressibility of gas and liquid phases and the axial variation of vapor flow rate are neglected as the first approximation. As a particular case, the following sinusoidal initial distribution is assumed to cf:

z0

cf 0

c f 0 ts

(11)

4 cf

The dimensionless heated length L* is therefore defined by L*

4 cf L

L z0

(12)

cf 0

(4)

0.005

Using Eq. (7), the mass flux of liquid film Gf is expressed by fi

0.005 1 300

(5)

D

Gf

f

f

uf

f

(1 37.5

f

)

f

Gv

(13)

v

where is the film thickness and D is the tube diameter. From Eqs. (3)–(5), uf is expressed by uf

Jv

v

1 300

f

D

Hence,

(1 37.5

f

v

)

Jv

f

cf

cf

(1 75

f

x

f )

(8)

0

v f

Jv

Gf

v

Gv

f

(14)

(15)

To estimate the values of cf, cf0 and in Eq. (12), it is further assumed that the initial distribution does not decay significantly within L and the droplet entrainment is negligible. In this case, Gv and Gf may be approximated by

Using Eq. (7), Eq. (2) is transformed to f

G*

1 150G * 75

(7)

f

t

is expressed by 1

(6)

Supposing a sufficiently thin liquid film, uf is approximated by uf

f

Gv

GMIN

Gf

G Gv

G AVE

(16)

G

G GAVE

G

(17)

(9) From Eqs. (9) and (14)–(17), cf = (cf,max – cf,min)/2 and cf0 are calculated by

6


GMIN

cf

2

1 300

v

f

GMIN

cf 0

v

1 150 f

G GMIN

v

1

2 1 Single: GAVE =49 (kg/m s) Double: GAVE =96 (kg/m2s)

(18)

f

G

v

GMIN

f

Stripe:

(19)

2

GAVE =200 (kg/m s)

G/GAVE =0.26 G/GAVE =0.51

is estimated by

q

*

The wave length

(20)

c f 0 tOSC

0.5

G/GAVE =0.76

Substituting Eqs. (18)–(20) into Eq. (12), one finally obtains L*

G*

2L

v

1 300 G * 1 1 150 G *

f

GMIN tOSC

(21)

0

G

v

GMIN

f

10–2

(22)

10–1 L

It is expected that the initial distributions of f and cf decay and consequently q* defined by Eq. (1) increases from 0 to 1 with an increase in L*. However, substantial simplifications were applied in deriving the functional form of L* given in Eqs. (21) and (22). The relations between q* and L* are hence investigated using the experimental data in the following section. *

3.6 Relationships between q and L

*

The relationships between the experimental values of q* and L are displayed in Fig. 5. It can be seen that q* increases with rising value of L* as an overall trend. This would indicate that L* and q* are a promising set of scaling parameters to characterize the critical heat flux under the flow oscillation condition. However, although the dependences on the oscillation amplitude and the oscillation frequency are collapsed fairly well, the scattering due to the time-averaged mass flux GAVE is rather significant. In particular, if the value of L* is in the same range, q* takes greater values in higher mass flux experiments. It is known that the existence of disturbance waves becomes significant when the liquid flow rate is high (Sekoguchi et al., 1985). It is therefore considered that the liquid transport associated with the disturbance wave might be one of the main reasons of scattering found in Fig. 5. Further accumulation of experimental data and the improvement of theory are still necessary to achieve accurate prediction of the critical heat flux in oscillatory flow conditions. *

100 *

Fig. 5: Relationships between L* and q* obtained in the experiments. 4. SUMMARY AND CONCLUSIONS In this study, the dryout of liquid film in annular two-phase flow regime was investigated under the flow oscillation condition. It was shown experimentally that the critical heat flux decreases with increased values of oscillation amplitude and oscillation period. These results were consistent with those reported by previous investigators. To interpret the dependences on these oscillation parameters, the critical heat flux under the flow oscillation condition was scaled using those in the steady states and the dimensionless heated length was introduced based on the available theory for nonlinear waves. The relationship between the dimensionless heated length and the dimensionless critical heat flux was investigated experimentally. It was shown that as an overall trend, the dimensionless critical heat flux increases with rising value of the dimensionless heated length. This result was consistent with the proposed theory. However, the scattering due to the difference in the time-averaged inlet mass flux was rather significant. Although the definite reason for the dependence on the mean mass flux was not identified, it was suggested that disturbance waves formed on the surface of the liquid film might enhance the axial mixing of the liquid film. Accumulation of experimental data and improvement of the model are necessary in future studies.

7


ACKNOWLEDGMENT This work was supported by KAKENHI (No. 20360419).

REFERENCES Andersen, J. G. M., Casillas, J. L., Shiralkar, B. S. (2006) Application of advanced thermal hydraulic TRACG model to preserve operating margins in BWRs at extended power uprate conditions, Proceedings of the International Congress on Advances in Nuclear Power Plants, Reno, Nevada, pp. 1066– 1077. Aoyama, M., Haikawa, K., Fennern, L. E., Yoshioka, R., Ohta, T., Anegawa, T. (1997) Optimization of core design for the next generation BWR, Proceedings of the 5th International Conference on Nuclear Engineering, Nice, France, Paper No. ICONE5-2636. Arai, K., Murase, A., Hamazaki, R., Kuroki, M. (2006) AB1600-Progress of ABWR technology toward next generation ABWR, Nuclear Engineering and Design, in press. Bowring, R. W. (1972) A simple but accurate round tube uniform heat flux dryout correlation over the pressure range 0.7-17 MN/m2, AEEW-R789 Chaki, M., Murase , M. (2006) Evaluation of the sensitivity of a two-phase flow model for steam separators analysis, Proceedings of the 14th International Conference on Nuclear Engineering, Miami, Florida, Paper No. ICONE14-89507. Churchill, S. W., Chu, H. H. S. (1975) Correlating equations for laminar and turbulent free convection from a vertical plate, International Journal of Heat and Mass Transfer 18, 1323– 1329.

Collier, J. G., Thome, J. R. (1994) Convective boiling and condensation, 3rd edition, Oxford University Press, Oxford, pp. 325–374. Katto, Y., Ohno, H. (1984) An improved version generalized correlation of critical heat flux for the convective boiling in uniformly heated vertical International Journal of Heat and Mass Transfer 27, 1648.

of the forced tubes, 1641–

Ozawa, M., Hirayama, M., Umekawa, H. (2002) Critical heat flux condition induced by flow instabilities in boiling channels, Chemical Engineering and Technology 25, 1197–1201. Ozawa, M., Umekawa, H., Yoshioka, Y., Tomiyama, A. (1993) Dryout under oscillatory flow condition in vertical and horizontal tubes-experiments at low velocity and pressure conditions, International Journal of Heat and Mass Transfer 36, 4076–4078. Sekoguchi, K., Ueno, T., Tanaka, O. (1985) An investigation of the flow characteristics in the disturbance wave region of annular flow: 2nd Report, On correlation of principal flow parameters, Transactions of the Japan Society of Mechanical Engineers, Series B 51, 1798–1806. Umekawa, H., Ozawa, M., Mitsunaga, T., Mishima, K., Hibiki, T., Saito, Y. (1999) Scaling parameters of CHF under oscillatory flow conditions," Heat Transfer-Asian Research 28, 541–550. Wallis, G. B. (1969) One-dimensional two-phase flow, McGraw-Hill, New York, pp. 315–374. Whitham, G. B. (1999) Linear and nonlinear waves, John Wiley & Sons., pp. 19–67.

8
