Oscillatory subcooled flow boiling heat transfer of R-134a and associated bubble characteristics in a narrow annular duct due to flow rate oscillation

Oscillatory subcooled flow boiling heat transfer of R-134a and associated

bubble characteristics in a narrow annular duct due to flow rate

oscillation

S.L. Wang, C.A. Chen, T.F. Lin

Department of Mechanical Engineering, National Chiao Tung University, 1001 University Road, Hsinchu City 30010, Taiwan

a r t i c l e

i n f o

Article history:

Received 26 August 2012

Received in revised form 2 April 2013 Accepted 7 April 2013

Available online 30 April 2013 Keywords:

Oscillatory subcooled flow boiling R-134a boiling heat transfer Bubble characteristics Flow rate oscillation

a b s t r a c t

An experiment is conducted here to investigate how an imposed time periodic flow rate oscillation in the form of a triangular wave affects the long time subcooled flow boiling heat transfer and associated bubble characteristics of refrigerant 134a in a horizontal narrow annular duct. In the experiment the mean R-134a mass flux G varies from 200 to 500 kg/m2_{s, imposed heat flux ranges from 0 to 45 kW/m}2_{, and the}

amplitude of the mass flux oscillation changes from 0 to 30% of G with the period of the mass flux oscil-lation varied from 20 to 120 s for the inlet liquid subcooling ranging from 0 to 6 °C. The duct gap is fixed at 2.0 mm. The results indicate that the inlet liquid subcooling significantly affects the oscillatory flow boiling heat transfer characteristics. Besides, when the imposed heat flux is close to that for the onset of stable flow boiling, intermittent flow boiling appears. The intermittent boiling prevails in a very differ-ent range of the Boiling number for a change in the inlet subcooling. Moreover, in the subcooled boiling the heated wall temperature, bubble departure diameter and frequency, and active nucleation site den-sity also oscillate periodically in time. Furthermore, in the persistent boiling at high imposed heat flux the resulting Twoscillation is stronger for a higher inlet liquid subcooling and for a longer period and a larger

amplitude of the mass flux oscillation. And for a larger amplitude of the mass flux oscillation, stronger temporal oscillations in dp, f and nacare noted. Finally, a flow regime map is provided to delineate the

boundaries separating different boiling regimes for the oscillatory R-134a subcooled flow boiling in the annular duct.

Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Energy efficiency improvement in various engineering systems is receiving ever increasing attention worldwide intending to re-duce the release of CO2(g) to the atmosphere during the combus-tion of fossil fuels. Recently, the use of variable frequency, instead of ON/OFF, compressors in air conditioning and refrigera-tion systems to meet the temporally changing thermal load has been found to significantly augment their energy efficiencies. In these systems the refrigerant flow rate varies with time to accom-modate the changing thermal load. How the time varying refriger-ant flow rate and heat flux affect the characteristics of boiling and condensation processes in the refrigeration cycles employed in these systems remains largely unexplored.

Flow boiling of refrigerants at constant flow rate in small ducts subject to time independent heating has received some attention. Boiling heat transfer of various fluids in a small duct can be dom-inated by nucleation or convection depending on the levels of

im-posed heat flux, wall superheats and vapor quality[1–5]. Besides, at lower liquid subcooling bubble nucleation was found to be more important[6]. Moreover, boiling heat transfer is better for smaller duct and liquid subcooling and bubbles in the flow are suppressed to become smaller by raising the flow rate and subcooling[7].

For several decades, different forms of dynamic instabilities in the flow boiling of various liquids in a long heated channel have been recognized[8]. Significant temporal oscillations in pressure, temperature, mass flux and boiling onset occur at certain operating conditions. In a study for R-11 in a long horizontal tube, Ding et al.

[9]examined the dependence of the oscillation amplitude and per-iod on the system parameters and located the boundaries among various types of the oscillations. A similar experimental study was carried out by Comakli et al.[10]and the effect of the channel length on the boiling flow dynamic instabilities was examined.

The dynamic behavior of boiling flow in a horizontal channel connected with a surge tank for liquid supply has received some attention[11]. The boiling onset in an upward flow of subcooled water in a vertical tube of 7.8-m long could cause substantial flow pressure and density-wave oscillations [12]. These boiling-onset oscillations were attributed to a sudden increase of pressure-drop

0017-9310/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijheatmasstransfer.2013.04.006

⇑ Corresponding author.

E-mail address:tfl[email protected](T.F. Lin).

Contents lists available atSciVerse ScienceDirect

International Journal of Heat and Mass Transfer

across the channel and a large fluctuation in the water flow rate at the onset of nucleate boiling. The effects of the inlet flow condi-tions on the boiling instabilities were found to be relatively signif-icant[13]. A similar study for subcooled flow boiling of deionized water was conducted by Shuai et al.[14]and the pressure-drop

oscillation was also noted. Reviews of two-phase flow dynamic instabilities in tube boiling have been conducted recently by Kakac and Bon[15]and Tadrist[16].

In view of its importance on thermal hydraulic safety of nuclear reactors, how an inlet flow oscillation affects the critical heat flux Nomenclature

As outside surface area of the heated inner pipe (m2) ATw amplitude of the heated wall temperature oscillation

(°C)

Bo Boiling number, Bo ¼ q

Gifg

(dimensionless) Di, Do inner and outer diameters of duct (m) Dh hydraulic diameter, Dh= (Do Di), (m) dP bubble departure diameter (mm) f bubble departure frequency (1/s)

G, G instantaneous and time-average mass fluxes (kg/m2_s) hr boiling heat transfer coefficient (W/m2°C)

ifg enthalpy of vaporization (J/kg) nac active nucleation site density (n/m2)

P, Pin mean system pressure and instantaneous inlet pressure (kPa)

q imposed heat flux (W/m2₎

qb heat flux due to bubble nucleation (latent heat transfer) (W/m2₎

qONB, qONB steady-state and time-average heat flux for the onset of nucleate boiling (W/m2₎

Qn net power input (W)

t time instant (sec)

tc time constant (sec)

to time instant at beginning of a periodic cycle (sec) tp period of mass flux oscillation (sec)

DTsat, Tsat instantaneous and time-average saturated tempera-ture of refrigerant (°C)

Tw temperature of heated wall (°C)

z coordinate (downstream coordinate for annular duct flow) (mm)

Greek symbols

DG amplitude of mass flux oscillation (kg/m2s) DTsat wall superheat, (Tw Tsat) (°C)

DTsub time-average inlet liquid subcooling (°C)

d duct gap (mm) BY-PASS FLOW WATER FLOWMETER THERMOSTAT WATER PUMP MASS FLOWMETER PREHEATER N2 VALVE

WATER LOOP

REFRIGERANT LOOP

BY - PASS FLOW RELEASE VALVE

PUMP SUBCOOLER ACCUMULATOR FILTER/DRYER RECEIVER COOLING TOWER

WATER-GLYCOL LOOP

ANNULAR DUCT UNIT PROGRAMMABLE

of water in boiling channels has been intensively investigated at low mass flux and pressure conditions[17–24]. Significant reduc-tion in the critical heat flux by the flow oscillareduc-tion was noted. How-ever, to the knowledge of the authors the effects of an inlet flow oscillation on the nucleate boiling heat transfer have not been re-ported in the open literature.

Considerable amount of works was carried out in the past to examine bubble characteristics associated with stable subcooled flow boiling of various liquids in a duct. It was noted that the bub-ble departure frequency was suppressed by raising the mass flux and subcooling of R-134a, and only the liquid subcooling signifi-cantly affected the bubble size[25]. But the bubble departure fre-quency increases with the heat flux and the bubble growth rate dropped sharply after the bubble lift-off[26]. Besides, the waiting

time between two bubble cycles decreases significantly at increas-ing mass flux [27]. Moreover, the size of coalesced bubbles de-creases for an increase in the water mass flux and the mass flux only exhibits a strong effect on the bubble size[28].

The above literature review clearly indicates that the unsteady flow boiling heat transfer of HFC refrigerants in small diameter channels resulting from time varying mass flux and/or heat flux re-mains largely unexplored. In a recent study[29], we experimen-tally investigated the time periodic saturated flow boiling heat transfer and associated bubble characteristics for R-134a in a hor-izontal narrow annular duct due to flow rate oscillation. In the present study we move further to investigate how the liquid sud-cooling affects the R-134a flow boiling in the same duct subject to the same periodic flow rate oscillation. The effects of the

ampli-160

Flow Inlet

93

38

38

Flow Outlet

85

0

Y

Z

Side

View

Window

Test Section

Developing Channel

Outlet Section

Unit : mm

+

Smooth Tube Surface Thermocouple

Refrigerant Flow

Joined-Bolt

O-Ring

Epoxy PlugRefrigerant Flow MgO

Cartridge Heater

Copper Tube

Copper Tube Stainless Steel Flange

A

A

C

C

80 mm

Pyrex Glass Pipe

50 mm

110 mm

tude and period of the flow rate oscillation on the time dependent subcooled boiling heat transfer characteristics will be examined in detail. Particularly, visualization of the boiling flow will be con-ducted to improve our understanding of the time varying flow boil-ing processes in the narrow channel.

2. Experimental apparatus and procedures

The experimental system used in the previous study[29]for the oscillating saturated flow boiling is also employed here to investi-gate the subcooled flow boiling of R-134a due to flow rate oscilla-tion in the same narrow annular duct. It is schematically depicted inFig. 1. The experimental apparatus consists of three main loops, namely, a refrigerant loop, a water-glycol loop, and a hot-water loop. Refrigerant R-134a is circulated in the refrigerant loop. In or-der to control various test conditions of the refrigerant in the test section, we need to control the temperature and flow rate in the other two loops. The description of the system is detailed in the previous study and is not repeated here.

As schematically shown inFig. 2, the test section of the appara-tus is a horizontal annular duct with the outer pipe made of Pyrex glass to permit the visualization of boiling processes in the refrig-erant flow. The glass pipe is 160-mm long with an inside diameter of 20.0 mm. Its wall is 4.0-mm thick. Both ends of the pipe are con-nected with copper tubes of the same size by means of flanges and are sealed by O-rings. The inner copper pipe has 16.0 mm nominal outside diameter with its wall being 1.5 mm thick and is 0.41-m long. Thus the gap of the annular duct is 2.0 mm (Dh= 4.0 mm). Note that the outside surface of the inner pipe is polished by fine sandpaper. In order to insure the gap between the inner and outer

pipes being uniform, we first measure the outside diameter of the inner pipe and the inside diameter of the glass pipe by digital cal-ipers whose resolutions are 0.001 mm with the measurement accuracy of ±0.01 mm. Then we photo the top and side view pic-tures of the annular duct and measure the average radial distance between the inside surface of the glass pipe to the outside surface of the inner tube. From the above procedures the duct gap is ascer-tained and its uncertainty is estimated to be 0.02 mm. An electric cartridge heater of 160 mm in length and 13.0 mm in diameter with a maximum power output of 800 W is inserted into the inner pipe. Furthermore, the pipe has an inactive heating zone of 10-mm long at each end and is insulated with Teflon blocks and thermally nonconducting epoxy to minimize heat loss from it. Thermal con-tact between the heater and the inner pipe is improved by coating a thin layer of heat-sink compound on the heater surface before installing the heater. Then, 8 T-type calibrated thermocouples are positioned at three axial stations along the inner pipe. The outside surface temperature of the inner pipe Twis then derived from the measured inside surface temperature by taking the radial heat con-duction through the pipe wall into account.

The photographic apparatus established here to record the bub-ble characteristics in the time periodic flow boiling in the annular duct consists of an IDT X-Stream™ VISION XS-4 high speed CMOS digital camera, a Mitutoyo micro lens set, a 3D positioning mecha-nism, and a personal computer. The data for some bubble charac-teristics are collected in a small region around the middle axial location (z = 80 mm). After the experimental system reaches a sta-tistical state, we start recording the boiling activity. The digital camera can store the images which are later downloaded to the personal computer. Then, the time variations of the space-average bubble departure diameter dpand frequency f and active

ation site density nacin a periodic cycle are calculated by viewing more than 1000 frames at each time instant. Specifically, the images of the bubbles during the boiling processes are directly visualized and measured by showing them on the computer screen in a much slower rate. By considering the pixel depth, size, resolu-tion, center-to-center spacing, sensor image area, minimum inter-frame rate, and integration time of the digital camera and com-puter screen, the uncertainties of the measured dp, f and nacare estimated to be ±10%, ±4.5% and ±5.5%, respectively.

Before a test is started, the temperature of refrigerant R-134a in the test section is compared with its saturation temperature corre-sponding to the measured pressure and the allowable difference is kept in the range of 0.2–0.3 K. Otherwise, the system is re-evacu-ated and then re-charged to remove the air existing in the refriger-ant loop. A vacuum pump is used to evacuate noncondensable gases in the loop. Note that in the test the refrigerant at the inlet of the test section is maintained at the required subcooled liquid state by adjusting the water–glycol temperature and flow rate. In addition, we adjust the thermostat temperature in the water loop to stabilize the inlet R-134a temperature. Then, we regulate the time-average refrigerant pressure at the test section inlet by adjusting the opening of the gate valve locating right after the exit of the test section. Meanwhile, by controlling the inverter current of the AC motor connecting to the refrigerant pump, the refrigerant flow rate can be varied to procure the preset mean level and the se-lected period and amplitude of the mass flux oscillation. The im-posed heat flux from the heater to the refrigerant is adjusted by varying the electric current delivered from the programmable DC power supply. By measuring the current delivered to and voltage drop across the heater and by photographing the bubble activity,

we can calculate the heat transfer rate to the refrigerant and obtain the bubble characteristics. All tests are run at statistical state con-ditions. The whole system is considered to be at a statistical state when the variations of the time-average system pressure and im-posed heat flux are respectively within ±1% and ±4%, and the vari-ations of the time-average heated wall temperature are less than ±0.2 °C for a period of 100 min. Then all the data channels are scanned every 0.05 s for a period of 360 s.

3. Data reduction and verification of experimental system The imposed heat flux q to the refrigerant flow in the narrow annular duct is calculated on the basis of the net power input Qn and the total outside surface area of the inner pipe of the annular duct Asas q = Qn/As. The total power input Qtis obtained from the product of the measured voltage drop across the cartridge heater and electric current passing through it. Hence the net power input to the test section is equal to (Qt Qloss).

The total heat loss from the test section Qlossis evaluated from the correlation for natural convection around a circular cylinder by Churchill and Chu[30]. To reduce the heat loss from the test section, it is covered with a polyethylene insulation layer. The re-sults from the present heat loss test indicate that the total heat loss from the test section is generally less than 1% of Qtno matter when single-phase flow or two-phase boiling flow is in the duct. The flow boiling heat transfer coefficient at a given axial location at a given time instant is defined as

hr¼ q ðTw TsatÞ

ð1Þ

Fig. 4. Time variations of imposed mass flux, inlet pressure, and heat transfer coefficient at z = 80 mm in persistent boiling for various q, DGG and tpat DTsub¼ 3C at

Uncertainties of the imposed heat flux, measured heat transfer coef-ficients and other parameters are estimated according to the proce-dures proposed by Kline and McClintock[31]for the propagation of errors in physical measurement. The results from this estimate show that the uncertainties of the dimension, temperature, pres-sure, mean mass flux, period of oscillation, amplitude of oscillation, imposed heat flux, and boiling heat transfer coefficient

measure-ments are less than ±1%, ±0.2 °C, ±0.2 kPa, ±2%, ±0.25 s, ±4.8%, ±4.5% and ±14.0%, respectively.

To check the suitability of the experimental system for measur-ing the flow boilmeasur-ing heat transfer coefficients, the steady state sin-gle-phase liquid R-134a heat transfer coefficients for a constant liquid Reynolds number ranging from 3,648 to 11,420 are mea-sured first and compared with the well-known traditional forced

Fig. 5. Variations of amplitudes of heated wall temperature with imposed heat flux for various refrigerant mass fluxes (a), saturated temperatures (b), inlet subcoolings (c), amplitudes of mass flux oscillation (d) and periods of mass flux oscillation (e).

convection correlation proposed by Gnielinski [32]. The results manifest that the present steady heat transfer data can be well cor-related by his correlation with a mean absolute error of 3.9%. Be-cause of the lack of the unsteady turbulent forced convection heat transfer data in the open literature, direct validation of the present time periodic liquid heat transfer data is not possible.

4. Results and discussion

In the present experiment the mean refrigerant mass flux G var-ies from 200 to 500 kg/m2_{s, imposed heat flux q from 0 to 45 kW/} m2_{, average inlet liquid subcoolingDT}

subfrom 0 to 6 °C, and mean system pressure P set at 414.6 kPa and 488.6 kPa (corresponding to the mean R-134a saturation temperature Tsat= 10 °C and 15 °C) for the duct gap d = 2.0 mm. Besides, the amplitude of the refrigerant mass flux oscillationDG is set at 0, 10, 20 and 30% of G. Moreover, the period of the mass flux oscillation tpis fixed at 20, 30, 60 and 120 s. The ranges of the parameters chosen above are in accor-dance with some air-conditioning applications. In the following, attention will be mainly paid to examining the effects of the inlet liquid subcooling on the oscillatory R-134a subcooled flow boiling heat transfer performance. Note that for the limiting case of DGG = 0% we have subcooled flow boiling of R-134a at a constant

refrigerant mass flow rate in the test section, which is designated as stable subcooled flow boiling and has been investigated by Lie and Lin[7]. In the following the thermal characteristics of the oscil-latory flow boiling is illustrated by the time variations of the instantaneous heated pipe wall temperature and boiling heat transfer coefficient. Moreover, selected flow photos and data de-duced from the images of the boiling processes are presented to show the time dependent bubble characteristics in the boiling flow.

The time constant tcassociated with the response of the heated wall temperature to the imposed time dependent refrigerant flow rate is an important physical quantity in the oscillatory flow boil-ing. It is measured directly here from the time response of Tw sub-ject to a step change in the mass flux for various q, TsatandDTsub. The measured data show that in the transient subcooled boiling flow tcis longer for lower Tsatand higherDTsub. Over the ranges of the present experimental parameters tc varies from 16.3 to 20.3 s. The average tcis about 18.6 s. The large tcvalue mainly re-sults from the inertia of the refrigerant flow rate change with time and the time needed for the heat diffusion and convection from the pipe wall to the refrigerant flow. The time constant associated with the pipe wall of the inner copper pipe alone is rather short, ranging from 0.06 to 0.12 s, due to its high thermal diffusivity. It is of

est to note that when the period of the imposed mass flux oscilla-tion is much shorter than tcthe boiling flow is not able to quickly respond to the fast oscillation and we essentially obtain steady flow boiling heat transfer data for the imposed time-average mass flow rate. On the other end as the period of the mass flux oscilla-tion is substantially longer than tcthe heat transfer in the boiling flow is affected only in the initial stage of the flow rate oscillation. After that quasi-steady flow boiling heat transfer prevails. Thus the transient variations of the boiling heat transfer characteristics are expected to become pronounced when the flow rate oscillation period does not deviate from the time constant of the boiling flow to a large degree. Hence in the present study tpis chosen to vary from 20 to 120 s, as given previously.

4.1. Time-average subcooled boiling curves and heat transfer coefficient

At first, the time-average boiling curves and heat transfer coef-ficients measured at the middle axial location (z = 80 mm) of the narrow annular duct for various amplitudes and periods of the refrigerant mass flux oscillation were obtained for variousDTsub, G, q and Tsat. The data clearly indicate that in the present time dependent subcooled flow boiling of R-134a the time-average boil-ing curves and heat transfer coefficients are not affected by the amplitude and period of the refrigerant mass flux oscillation to a significant degree. In fact, they are nearly the same as that for the stable subcooled flow boiling[7].

4.2. Time dependent subcooled flow boiling heat transfer characteristics

The oscillatory subcooled flow boiling heat transfer characteris-tics for the R-134a flow in the annular duct resulting from the peri-odic refrigerant mass flux oscillation are illustrated inFigs. 3 and 4

by presenting the measured time variations of the heated wall

temperature Twand boiling heat transfer coefficient hrat the mid-dle axial location in the statistical state for various G, Tsat,DGG, tp, and q atDTsub¼ 3C. For clear illustration, the measured data for the variations of the imposed refrigerant mass flux are also given here. The mass flux varies like a triangular wave. By and large, the measured oscillatory subcooled flow boiling heat transfer char-acteristics are qualitatively similar to the oscillatory saturated flow boiling [29]. However, some noted differences do exist. It is of interest to find from these data that in addition to the persistent boiling, intermittent boiling also appears depending on the level of the imposed heat flux, as that for the oscillatory saturated flow boiling withDTsub¼ 0C. But the temporal oscillations in Twand hr are noticeably stronger for the subcooled boiling atDTsub¼ 3C. Note that the Twand hroscillations are also periodic in time and are at the same frequency as the mass flux. And the Twand hr oscil-lations become more intense for a higherDTsub and for a higher amplitude and a longer period of the mass flux oscillation. The stronger Twoscillation at a larger tpis attributed to the accumula-tion and dispersion of thermal energy in the heated pipe wall, respectively in the first half and second half of every periodic cycle over a longer period. However, the amplitude of the Twoscillation varies nonmonotonically with the imposed heat flux. Note that in the single-phase flow the heated surface temperature oscillates stronger at a higher q at a relatively low imposed heat flux with q < qONB. Here qONBis the time-average heat flux for the onset of nucleate boiling for the cases with oscillating mass flux. But an opposite trend is noted in the persistent boiling which prevails at high heat fluxes. The data given inFigs. 3 and 4also show that the Twoscillations slightly lag the mass flux oscillation. This time lag apparently results from the large time constant of the flow boil-ing heat transfer in the duct.

A close inspection of these data further reveals that for the sin-gle-phase convection at low q and in the first half of the periodic cycle in which the mass flux decreases with time, the wall temper-ature is found to rise also with time, suggesting that the

single-(a) G =500 kg/m2_{s, ΔG/G = 10%, tp= 120sec (b) G =500 kg/m}2_{s, ΔG/G = 30%, tp= 20sec (c) G =500 kg/m}2_{s, ΔG/G = 30%, tp= 120sec} 6mm (1) t=to, G = 551.3 kg/m2s 6mm (5) t=to+4tp/8, G = 449.7 kg/m2s 6mm (1) t=to, G = 650.1 kg/m2s 6mm (5) t=to+4tp/8, G = 350.2 kg/m2s 6mm (1) t=to, G = 651.2 kg/m2_s 6mm (5) t=to+4tp/8, G = 349.9 kg/m2_s (2) t=to+tp/8, G = 524.6 kg/m2s (6) t=to+5tp/8, G = 475.4 kg/m2s (2) t=to+tp/8, G = 588.6 kg/m2s (6) t=to+5tp/8, G = 414.6 kg/m2s (2) t=to+tp/8, G = 587.4 kg/m2_{s G = 421.1 kg/m}(6) t=to+5tp/8, 2_s (3) t=to+2tp/8, G = 502.3 kg/m2s (7) t=to+6tp/8, G = 501.5 kg/m2s (3) t=to+2tp/8, G = 501.3 kg/m2s (7) t=to+6tp/8, G = 502.7 kg/m2s (3) t=to+2tp/8, G = 504.6 kg/m2s (7) t=to+6tp/8, G = 503.7 kg/m2s (4) t=to+3tp/8, G = 475.7 kg/m2_s (8) t=to+7tp/8, G = 525.1 kg/m2_s (4) t=to+3tp/8, G = 414.1 kg/m2_s (8) t=to+7tp/8, G = 588.7 kg/m2_s (4) t=to+3tp/8, G = 420.4 kg/m2s (8) t=to+7tp/8, G = 587.9 kg/m2s flow

Fig. 7. Photos of bubbles in time periodic subcooled flow boiling of R-134a in a small region around the middle axial location at eight time instants in a typical periodic cycle at Tsat¼ 15C, DTsub¼ 3C, q = 25 kW/m2and G = 500 kg/m2s for various DGG and tpat d = 2.0 mm.

phase convection heat transfer over the heated surface is poorer at a lower G. This trend agrees with the traditional forced convection heat transfer in single-phase flow[32]. The opposite process occurs in the second half of the cycle. The corresponding Twoscillation amplitude in the single-phase flow can be significant. It is worth noting in the oscillatory subcooled flow boiling that for the im-posed heat flux somewhat higher than qONBin the persistent boil-ing regime, the heated surface temperature can be well below that in the single-phase flow regime prevailed at much lower imposed heat flux in the entire periodic cycle. This outcome obviously re-sults from the relatively significant temperature overshoot at ONB in the oscillatory subcooled flow boiling for R-134a, as seen in the stable flow boiling[7]. We also note that intermittent flow boiling exists less often in the time periodic subcooled flow boiling, when compared with that in the saturated flow boiling[29]. The time instants for the start and termination of the nucleate boiling are marked on the Twcurves for the cases with the presence of the intermittent boiling. At this intermediate imposed heat flux the Twoscillation is very weak. Note that in the persistent boiling at higher q in the first half of the cycle in which G decreases with time, Twalso decreases with time, suggesting that the flow boiling heat transfer over the heated surface is better at a lower G, as sup-ported by the data for hr(Fig. 4). The reverse trend is noted in the second half of the periodic cycle in which G decreases with time. This unusual result of increasing (decreasing) hrat reducing (upris-ing) G can be attributed to the unique effects of the refrigerant

mass flux variation on the bubble characteristics in the boiling flow, which will be examined later. The results inFig. 4also show that the hroscillation is significantly weaker at a higher imposed heat flux in the persistent boiling.

The time variations of the refrigerant pressure at the inlet of the test section are also shown inFigs. 3 and 4for selected cases. The results indicate that the low frequency component of the inlet refrigerant pressure oscillates nearly in phase with the mass flux oscillation and also in the form of triangular waves. In fact, the in-let pressure oscillation is characterized by high frequency compo-nents superimposed on the low frequency component. A close examination of these data, however, reveals that the mass flux oscillates slightly behind the low frequency component of the pressure oscillation.

Then, we move further to present the data inFig. 5to elucidate the effects of the experimental parameters on the amplitude of the Twoscillation over a wide range of the imposed heat flux covering the single-phase, intermittent and persistent boiling flow regimes. The results for a given inlet liquid subcooling of 3°C at the high G of 500 kg/m2_{s shown inFig. 5(a) clearly indicate that in the} single-phase flow the oscillation amplitude of Twincreases significantly with the imposed heat flux for given Tsat,DGG and tp. Note that when the intermittent boiling appears the Twoscillation starts to weaken substantially with an increase in the imposed heat flux. At a certain higher q but still in the intermittent boiling regime the Twoscillation decays to a minimum point and then a further

in-Fig. 8. Photos of bubbles in the time periodic intermittent flow boiling of R-134a in a small region around the middle axial location at eight time instants in a typical time periodic cycle for Tsat¼ 3C, DTsub¼ 3C, G = 500 kg/m2s, q = 19.5 kW/m2, DGG = 30% and tp= 120 s at d = 2.0 mm.

crease in q causes Twto oscillate in a larger amplitude. This non-monotonic variation of the Twoscillation amplitude with q in the intermittent boiling can be attributed to the completely opposite trends in the Twoscillation for the single-phase and persistent boil-ing flow, as noted inFig. 3. Specifically, Twincreases as G decreases in the single-phase flow but in the boiling flow Tw decreases at decreasing G. Thus as the bubble nucleation starts to appear in the single-phase flow the Twoscillation is weakened. This trend continues until the boiling flow dominates over the single-phase flow for a rise in q to a certain level. Then, the Twoscillation gets stronger for a higher q. In the persistent boiling regime at high im-posed heat flux the Twoscillation gets gradually weaker at increas-ing q. Note that a much stronger Twoscillation results for a higher inlet liquid subcooling, as evident fromFig. 5(c). Besides, the Tw oscillation is also somewhat stronger for a higher amplitude and a longer period of the mass flux oscillation (Figs. 5(d) and (e)). However, the mean saturated temperature of the refrigerant shows a relatively insignificant effect on the amplitude of the Tw oscilla-tion (Fig. 5(b)).

4.3. Intermittent flow boiling

In the oscillatory subcooled boiling flow in the heated annular duct for the imposed heat flux close to that needed for ONB of the stable boiling, we also have intermittent boiling, as mentioned earlier. More specifically, in a typical periodic cycle of the refriger-ant flow rate oscillation bubble nucleation on the heated pipe wall is first seen at a certain time instant in the first half of the cycle as the imposed refrigerant mass flux decreases to a certain low level and the required incipient boiling heat flux is below the imposed heat flux. The boiling process continues for some time interval. At a certain later time instant in the second half of the cycle in which the imposed refrigerant mass flux increases and the needed incipient boiling heat flux is higher than q, bubble nucleation dis-appears and boiling stops. Single-phase flow then prevails in the

duct. The above processes of the intermittent flow boiling are repeatedly seen on the heated surface. The data inFig. 5clearly manifest that the appearance of the intermittent boiling substan-tially affects the intensity of the wall temperature oscillation for the subcooled flow boiling. To be more clear, the time instants at which boiling starts and stops are marked on the Tw curves in

Fig. 3. It is noted that at a higher imposed heat flux the onset of boiling is earlier and the termination of boiling is later. Besides, the instants for the boiling inception and termination are not af-fected by the amplitude and period of the mass flux oscillation to a noticeable degree (Fig. 3).

A flow regime map to delineate the boundaries separating dif-ferent flow boiling regimes in the oscillatory subcooled flow boil-ing in terms of the mean Boilboil-ing number versus the relative amplitude of the mass flux oscillation is given inFig. 6. The results show that the intermittent boiling prevails at a much higher Boil-ing number for an increase in the inlet subcoolBoil-ing from 3°C to 6°C. Besides, the intermittent boiling exists over a wider range of the Boiling number for a higher amplitude of the mass flux oscillation, a lower mean refrigerant mass flux and a higher mean refrigerant saturated temperature.

4.4. Bubble characteristics in time periodic subcooled flow boiling The bubble characteristics associated with the stable subcooled flow boiling of R-134a in the present narrow annular duct have been examined in our recent study[7]. These characteristics indi-cate that for the imposed heat flux exceeding that for onset of nucleate boiling the duct is dominated by the persistent boiling and a number of discrete bubbles nucleate from the cavities and slide along the heating surface. At a higher q the active bubble nucleation density increases and a lot of coalescence bubbles ap-pear. More coalescence bubbles are seen and they are confined by the duct walls to become slightly deformed for a further in-crease in q. They also reported that at a higher refrigerant mass flux

Fig. 9. Mean bubble departure diameter variations with time in a typical periodic cycle for various DGG at q = 25.0 and 30.0 kW/m2

(a), tp(b), G (c), Tsat(d), and DTsub(e) at



G = 500 kg/m2

the bubble departure frequency was higher and the bubbles were smaller and in violent agitating motion. However, the active nucle-ation site density is lower at a higher mass flux. But at a lower mass flux the bubble coalescence is more important and a number of bigger bubbles form in the duct. Then, at a lower saturation tem-perature of the refrigerant the bubbles were found to grow bigger and depart at a lower rate, and the active nucleation site density is lower. In general, the bubbles are smaller and less bubble coales-cence takes place at a higher inlet liquid subcooling due to the stronger vapor condensation, along with the reduction of the bub-ble departure frequency and active nucleation sites.

By and large, in the oscillatory subcooled flow boiling investi-gated here the effects of the imposed heat flux and mean refriger-ant mass flux, saturation temperature and inlet liquid subcooling on the bubble behavior exhibit similar trends to that in the stable subcooled flow boiling. Hence, we first illustrate here how the time periodic bubble characteristics are affected by the amplitude and period of the mass flux oscillation for the persistent subcooled flow boiling inFig. 7by presenting the photos of the boiling flow for the selected cases forDTsub¼ 3C in a small region around the middle axial location at eight selected time instants in a typical periodic cycle in the statistical state. In the figure the symbol ‘‘t = to’’ signi-fies the time instant at which the instantaneous refrigerant mass flux is at the highest level and starts to decrease. The results indi-cate that for given q, Tsat,DTsub, G,DGG and tpthe bubbles get big-ger and become more crowded with time in the duct in the first half of the cycle in which the mass flux decreases. The opposite processes take place in the second half of the cycle in which the mass flux increases with time. These changes of the bubble charac-teristics with the instantaneous mass flux are more significant at increasing amplitude of the mass flux oscillation. Besides, more large bubbles appear in the duct for the cases with a longer period of the mass flux oscillation. Then, the bubble behavior in the inter-mittent flow boiling for a selected case is shown inFig. 8. The

re-sults clearly indicate that initially in the beginning of the cycle the instantaneous mass flux decreases with time but is still well above G no bubbles nucleate from the heated surface. The flow is in single-phase state. Note that the bubbles start to nucleate from the heated surface at a certain time instant slightly before t0+ tp/8 at which the instantaneous mass flux is somewhat above G: Note that the number and size of the bubbles increase noticeably with time in the second quarter of the periodic cycle for the continuing decrease of the mass flux. Then in the third quarter of the periodic cycle the number and size of the bubbles diminish noticeably with time due to the increase of the mass flux. The bubbles cease to nucleate from the heated surface at a certain time instant slightly after t0+ 7tp/8 when the mass flux exceeds G to a certain degree and bubble nucleation stops completely. Single-phase flow again dominates. We have to wait until the middle of the first quarter of the next cycle to see the bubble nucleation appearing on the heated surface. The above processes repeat in each cycle.

To be more quantitative, we estimate the time variations of the space-average bubble departure diameter and frequency and the number density of the active nucleation sites on the heated surface in a typical periodic cycle at the middle axial location for the per-sistent subcooled flow boiling from the images of the boiling flow stored in the video tapes. Selected results from this estimation are manifested inFigs. 9–11by presenting the effects of the experi-mental parameters on the time dependent bubble characteristics. In these plots the time lag in the flow is ignored. The data inFigs. 9– 11indicate that as the refrigerant mass flux oscillates time period-ically, the bubble departure diameter and frequency and active nucleation site density also vary time periodically and to some de-gree like a triangular wave as the mass flux oscillation. More spe-cifically, the size of the departing bubbles and the active nucleation site density on the heated surface increase but the bubble depar-ture frequency decreases significantly in the first half of the peri-odic cycle in which the mass flux decreases with time. While in

Fig. 10. Mean bubble departure frequency variations with time in a typical periodic cycle for various DGG at q = 25.0 and 30.0 kW/m2

(a), tp(b), G (c), Tsat(d), and DTsub(e) at



G = 500 kg/m2

the second half of the cycle an opposite process is noted when the mass flux increases with time. Besides, the results in Figs.9(a),

10(a) and11(a) show that at a larger amplitude of the mass flux oscillation the bubble departure diameter, frequency and the ac-tive nucleation site density oscillate somewhat stronger. Moreover, the results in Figs.9(b),10(b) and11(b) indicate that f, dpand nac are not affected by the period of the mass flux oscillation. Further-more, increases in the mean refrigerant mass flux and saturation temperature and a decrease in the imposed heat flux cause the departing bubbles to become smaller but do not change the wave form of the dpvariation with time (Fig. 9(a), (c) and (d)). At higher q, G and Tsat the bubble departure frequency is higher (Fig. 10(a), (c) and (d)). Note that nacis significantly larger at a lower G and higher q and Tsat(Fig. 11(a), (c) and (d)). It is also noted that dp, f and nacare somewhat smaller for a higher inlet liquid subcooling (Figs.9(e),10(e) and11(e)). It is worth mentioning that in the pres-ent study even the size of the largest departing bubble is below 0.25 mm which is much smaller than the diameter of the outer glass pipe in the test section (Do= 20.0 mm). Thus the observation of the bubble size through the curved surface of the glass pipe is not expected to produce significant error.

Based on the data presented inFigs. 9–11for the oscillatory subcooled flow boiling of R-134a in the duct at d = 2.0 mm, the dependence of the quantitative bubble characteristics on the R-134a mass flux oscillation can be approximately expressed as dp -/ Ga_{, f / G}b

and nac/ Gc, when the short time lag in the Tw oscil-lation is neglected. Here the exponents a, b and c range respectively from 0.5 to 0.54, 0.79 to 0.84, and 0.44 to 0.52. Note that the latent heat transfer resulting from bubble nucleation in the persistent boiling qbis proportional to d

3

p, f and nac, as reported in the litera-ture[7,33]. Thus, qb/ Gd, here d varies from 1.15 to 1.3. This re-sult clearly indicates that the flow boiling heat transfer gets better at decreasing refrigerant mass flux since qbprevails in the boiling heat transfer. This in turn causes the heated wall

tempera-ture to decrease during decreasing refrigerant mass flux and vice versa in the oscillatory subcooled flow boiling, as already seen fromFig. 3and discussed earlier in Section4.2. Besides, the domi-nant effect of the mass flux oscillation on the boiling heat transfer comes from the very strong influences of the mass flux on the bub-ble departure size.

5. Concluding remarks

The measured heat transfer data for the oscillatory subcooled flow boiling of R-134a resulting from the refrigerant mass flux oscillation in the narrow annular duct have been presented here along with the bubble behavior. Effects of the mean level and oscil-lation amplitude and period of the refrigerant mass flux on the oscillatory subcooled R-134a flow boiling have been investigated. The major results obtained here can be summarized in the following.

(1) The time-average boiling curves and heat transfer coeffi-cients for the oscillatory subcooled flow boiling of R-134a are not affected to a noticeable degree by the amplitude and period of the refrigerant mass flux oscillation.

(2) The heated pipe wall temperature and bubble characteristics also oscillate periodically in time and at the same frequency as the mass flux oscillation. Experiments also show that the resulting oscillation amplitudes of the wall temperature get larger for a longer period and a larger amplitude of the mass flux oscillation and for a higher inlet liquid subcooling. Besides for a larger amplitude of the mass flux oscillation, stronger oscillations in the bubble characteristics, such as dp, f and nac, are noted.

(3) Increases in the bubble departure size and active nucleation site density at decreasing refrigerant mass flux overwhelm the bubble departure frequency in the oscillatory subcooled

Fig. 11. Mean active nucleation site density variations with time in a typical periodic cycle for various DGG at q = 25.0 and 30.0 kW/m2_{(a), t}

p(b), G (c), Tsat(d), and DTsub(e) at



flow boiling. This in turn causes an increase in the latent heat transfer and a drop in the heated wall temperature at decreasing G, opposite to that in the single-phase flow. (4) The intermittent boiling exists in narrower ranges of the

experimental parameters in the subcooled flow boiling. A flow regime map is given to delineate the boundaries sepa-rating different boiling flow regimes in the annular duct.

Acknowledgments

The financial support of this study by the engineering division of National Science Council of Taiwan, R.O.C. through the contract NSC 96-2221-E-009-133-MY3 is greatly appreciated.

References

[1]Z.Y. Bao, D.F. Fletcher, B.S. Haynes, Flow boiling heat transfer of Freon R11 and HCFC123 in narrow passages, Int. J. Heat Mass Transfer 43 (2000) 3347–3358. [2]T.N. Tran, M.W. Wambsganss, D.M. France, Small circular- and rectangular-channel boiling with two refrigerants, Int. J. Multiph. Flow 22 (1996) 485–498. [3]B. Agostini, A. Bontemps, Vertical flow boiling of refrigerant R134a in small

channels, Int. J. Heat Fluid Flow 26 (2005) 296–306.

[4]S.G. Kandlikar, M.E. Steinke, Flow boiling heat transfer coefficient in minichannels – correlation and trends, Proceedings of the Twelfth International Heat Transfer Conference 3 (2002) 785–790.

[5]B. Watel, Review of saturated flow boiling in small passages of compact heat exchangers, Int. J. Therm. Sci. 42 (2003) 107–140.

[6]B.S. Haynes, D.F. Fletcher, Subcooled flow boiling heat transfer in narrow passages, Int. J. Heat Mass Transfer 46 (2003) 3673–3682.

[7]Y.M. Lie, T.F. Lin, Subcooled flow boiling heat transfer and associated bubble characteristics of R-134a in a narrow annular duct, Int. J. Heat Mass Transfer 49 (13–14) (2006) 2077–2089.

[8]T. Otsuji, A. Kurosawa, Critical heat flux of forced convection boiling in an oscillating acceleration field: I – general trends, Nucl. Eng. Des. 71 (1982) 15– 26.

[9]Y. Ding, S. Kakac, X.J. Chen, Dynamic instabilities of boiling two-phase flow in a single horizontal channel, Exp. Thermal Fluid Sci. 11 (1995) 327–342. [10]O. Comakli, S. Karsli, M. Yilmaz, Experimental investigation of two phase flow

instabilities in a horizontal in-tube boiling system, Energy Convers. Manage. 43 (2002) 249–268.

[11]P.R. Mawasha, R.J. Gross, Periodic oscillations in a horizontal single boiling channel with thermal wall capacity, Int. J. Heat Fluid Flow 22 (2001) 643–649. [12]Q. Wang, X.J. Chen, S. Kakac, Y. Ding, Boiling onset oscillation: a new type of dynamic instability in a forced-convection upflow boiling system, Int. J. Heat Fluid Flow 17 (1996) 418–423.

[13]D. Brutin, L. Tadrist, Pressure drop and heat transfer analysis of flow boiling in a minichannel: influence of the inlet condition on two-phase flow stability, Int. J. Heat Mass Transfer 47 (2004) 2365–2377.

[14] J. Shuai, R. Kulenovic, M. Groll, Pressure Drop Oscillations and Flow Patterns for Flow Boiling of Water in Narrow Channel, in: Proceedings of International Conference on Energy and the Environment, Shanghai, China, May 22-24, 2003.

[15]S. Kakac, B. Bon, A review of two-phase flow dynamic instabilities in tube boiling systems, Int. J. Heat Mass Transfer 51 (2008) 399–433.

[16]L. Tadrist, Review on two-phase flow instabilities in narrow spaces, Int. J. Heat Fluid Flow 28 (2007) 54–62.

[17]K. Mishima, H. Nishihara, I. Michiyoshi, Boiling burnout and flow instabilities for water flowing in a round tube under atmospheric pressure, Int. J. Heat Mass Transfer 28 (1985) 1115–1129.

[18]M. Ozawa, H. Umekawa, Y. Yoshioka, A. Tomiyama, Dryout under oscillatory flow condition in vertical and horizontal tubes – experiments at low velocity and pressure conditions, Int. J. Heat Mass Transfer 36 (1993) 4076–4078. [19]Y.I. Kim, W.P. Baek, S.H. Chang, Critical heat flux under flow oscillation of

water at low-pressure, low-flow conditions, Nucl. Eng. Des. 193 (1999) 131– 143.

[20] M. Ozawa, H. Umekawa, K. Mishima, T. Hibiki, Y. Saito, CHF in oscillatory flow boiling channels, Chem. Eng. Res. Des. 79 (2001) 389–401.

[21]T. Okawa, T. Goto, J. Minamitani, Y. Yamagoe, Liquid film dryout in a boiling channel under flow oscillation conditions, Int. J. Heat Mass Transfer 52 (2009) 3665–3675.

[22]T. Okawa, T. Goto, Y. Yamagoe, Liquid film behavior in annular two-phase flow under flow oscillation conditions, Int. J. Heat Mass Transfer 53 (2010) 962– 971.

[23]M. Ozawa, M. Hirayama, H. Umekawa, Critical heat flux condition induced by flow instabilities in boiling channels, Chem. Eng. Technol. 25 (2002) 1197– 1201.

[24]D.W. Zhao, G.H. Su, Z.H. Liang, Y.J. Zhang, W.X. Tian, S.Z. Qiu, Experimental research on transient critical heat flux in vertical tube under oscillatory flow condition, Int. J. Multiph. Flow 37 (2011) 1235–1244.

[25]C.P. Yin, Y.Y. Yan, T.F. Lin, B.C. Yang, Subcooled flow boiling heat transfer of R-134a and bubble characteristics in a horizontal annular duct, Int. J. Heat Mass Transfer 43 (2000) 1885–1896.

[26]R. Situ, Y. Mi, M. Ishii, M. Mori, Photographic study of bubble behaviors in forced convection subcooled boiling, Int. J. Heat Mass Transfer 47 (2004) 3659–3667.

[27]R. Maurus, V. Ilchenko, T. Sattelmayer, Automated high-speed video analysis of the bubble dynamics in subcooled flow boiling, Int. J. Heat Fluid Flow 25 (2004) 149–158.

[28]S.H. Chang, I.C. Bang, W.P. Baek, A photographic study on the near-wall bubble behavior in subcooled flow boiling, Int. J. Therm. Sci. 41 (2002) 609–618. [29]C.A. Chen, W.R. Chang, T.F. Lin, Time periodic flow boiling heat transfer of

R-134a and associated bubble characteristics in a narrow annular duct due to flow rate oscillation, Int. J. Heat Mass Transfer 53 (2010) 3593–3606. [30] S.W. Churchill, H.H.S. Chu, Correlating equations for laminar and turbulent free

convection from a horizontal cylinder, Int. J. Heat Mass Transfer 18 (1975) 1049–1053.

[31]S.J. Kline, F.A. McClintock, Describing uncertainties in single-sample experiments, ASME Mech. Eng. 75 (1953) 3–12.

[32]V. Gnielinski, New equations for heat and mass transfer in turbulent pipe and channel flow, Int. Chem. Eng. 16 (1976) 359–368.

[33]J.R. Thome, Enhanced Boiling Heat Transfer, Hemisphere Publishing Corporation, New York, Chapter 6 (1990) 98–117.