ONSET OF NUCLEATE BOILING AND CRITICAL HEAT FLUX WITH BOILING WATER IN MICROCHANNELS

Transcription

1 International Journal of Microscale and Nanoscale Termal ISSN: Volume 4, Number 1 Nova Science Publisers, Inc. ONSET OF NUCLEATE BOILING AND CRITICAL HEAT FLUX WITH BOILING WATER IN MICROCHANNELS R. R. Bide 1, S. G. Sing 2, Vijay S. Duryodan 1, Arunkumar Sridaran 1, and Amit Agrawal 1, 1 Indian Institute of Tecnology Bombay, Powai, Mumbai, India 2 Indian Institute of Tecnology, Hyderabad, India ABSTRACT Tis paper focuses on experimental determination of onset of nucleate boiling (ONB) and critical eat flux (CHF) at te microscales, and comparison of tese wit available correlations. Te working fluid is deionised water and microcannel of four different ydraulic diameters: 65, 70, 107 and 125 m, ave been tested. Effect of ydraulic diameter ( m), mass flux ( kg/m 2 s) and eat flux (0-910 kw/m 2 ) on ONB and CHF as been studied in detail. Te eat flux for onset of nucleate boiling increases wit ydraulic diameter and mass flux. Te critical eat flux tends to increase wit a decrease in ydraulic diameter and wit increasing mass flux. Te effect of surface rougness on CHF as also been tested to a limited extent; no clear cange in te CHF value was observed upon canging te surface rougness by an order of magnitude. Te empirical correlations tested in tis study predict te experimental data to varying extent. Tese results may elp better determine te lower and upper limits of eat flux wile designing eat sink for electronic cooling. Keywords: Two pase flow, ONB, CHF, Boiling incipience NOMENCLATURE Symbol used Description Units A Area m 2 Cp Specific eat of te fluid J/kg-K d Hydraulic Diameter m H Heigt of cannel m Corresponding autor: Dr. Amit Agrawal, Department of Mecanical Engineering, Indian Institute of Tecnology Bombay, Powai Mumbai , INDIA, amit.agrawal@iitb.ac.in, Pone: , Fax:

2 26 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al. fg Latent eat of evaporation of fluid J/kg G Mass flux kg/m 2 -s L Lengt in te microcannel m m Mass flow rate kg/s P Power W ΔP Pressure drop mbar q Heat flux W/m 2 Q Volumetric flow rate m 3 /s T Temperature ºC u Velocity m/s W Widt of te microcannel m x Quality - ε Surface rougness m ρ Density kg/m 3 ρ H Homogenous Density kg/m 3 υ Specific Volume m 3 /kg μ Viscosity cp σ Surface Tension N/m Re Reynolds Number - Fr H Froude Number - We H Weber Number - Subscript w sat in exit cross air supp res can pred expt f g tp sp i,f a,f avg fo go Description Wall Saturation Inlet Exit Cross-sectional Heated Ambient Supplied Reservoir Cannel Predicted Experiment Fluid Vapour Two pase Single pase Frictional Pressure drop Acceleration Pressure drop Average Fluid only Vapour only

3 Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water INTRODUCTION In te recent past, tere as been an increasing trend to use miniaturised systems. Te development of fabrication tecniques and te advances in semiconductor tecnology ave led to te development of miniaturised electronic components. Tese devices need to be maintained below a certain temperature for efficient working. Due to increased storage capacity, faster performance, te eat to be dissipated from electronic components as increased considerably. It as been found tat dissipating eat fluxes greater tan 100 W/cm 2 using conventional air cooling would be difficult. Te use of microfluidic devices to remove ig eat fluxes is tus being widely researced as a viable alternative. Apart from cooling of electronic components, te micro-termal-mecanical systems can be used in portable computer cips, radar and aerospace avionics components, and in microcemical reactors [1]. A large amount of work as been devoted to te study of fluid flow and eat transfer mecanisms in microcannels. Specifically, researcers ave concentrated on te prediction of flow patterns [2-6], eat transfer caracteristics [4-9] and instability [5, 10-12] in two pase flow in microcannel. Sing et al. [13] found a strong dependence of two-pase pressure drop on aspect ratio in rectangular microcannel, wit a minimum at an aspect ratio of about 1.6. Sing et al. [5] ave reported te pressure drop in a 109 m ydraulic diameter trapezoidal cross-section microcannel over a wide range of inlet mass flow rate and eat flux values. Te pressure drop was found to exibit a maximum wit a reduction in mass flow rate for a constant eat flux. Wit a subsequent reduction in mass flow rate, te pressure drop rises rapidly. Teir compreensive data is furter analyzed in tis paper. Tis work was extended by Bide et al. [12] to study mean and r.m.s. of pressure drop in sub-undred sized microcannels. Tey found a reduction in pressure instabilities wit a reduction in ydraulic diameter and increase in wall rougness. See Agrawal and Sing [36] and Sing et al. [37] for a recent review on flow boiling in microcannel. Onset of nucleate boiling (ONB) and critical eat flux (CHF) are important issues in te study of two pase flow, especially in te design of microcannel eat sinks. Onset of nucleate boiling marks te beginning of te region of improved eat transfer. Several researcers [25-27] ave developed correlations to predict ONB for flow boiling in conventional scale flow passages. Basu et al. [33] performed subcooled flow boiling experiments on conventional scale passages fabricated using copper plate and nine zircalloy rod bundle. Tey developed a correlation for eat flux and wall supereat required for bubble inception, and suggested tat tese parameters are dependent upon mass flow rate, liquid subcooling, and contact angle. Also, tey developed a correlation for nucleation site density wic is primarily dependent upon te contact angle. Qu and Mudawar [34] performed experiments on parallel microcannels to measure te incipient boiling eat flux. Tey developed a model to predict te incipient boiling eat flux, accounting for te complexities of bubble formation along te flat and corner regions of a rectangular flow microcannel. Tey ave also accounted for te likeliood of bubbles growing sufficiently large to engulf te entire flow area of a microcannel. Liu et al. [35] experimentally investigated te onset of nucleate boiling and developed an analytical model to predict important parameters suc as incipience eat flux, bubble size, etc. during ONB. CHF is te maximum eat flux tat can be applied to te eater surface witout causing permanent damage to te device and terefore represents te upper limit for eat transfer in

4 28 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al. two pase flow. CHF is te outcome of events tat cause a sudden appreciable decrease in te eat transfer coefficient of a surface on wic boiling is occurring. However, te value of CHF is seldom reported in literature because it is rater difficult to obtain more so at te microscales. Qu and Mudawar [14] found tat CHF increases wit mass flux but does not depend on te inlet temperature. Tey attributed tis to te vapour back flow due to te parallel cannel instability wic negated te effect of liquid subcooling. Tey developed a correlation from te data for water in rectangular microcannels and te data for R-113 in circular mini/ microcannel eat sinks available from teir earlier study. Te developed a correlation as a function of vapour and liquid densities, cannel dimensions, and Weber number. Jiang et al. [7] carried out experiments for calculation of te CHF in microcannels of 40 and 80 μm ydraulic diameters. Altoug tey did not carry out flow visualisation study, from te temperature data collected tey postulated tat te boiling mecanism may be different for smaller microcannels. Tey observed different trends in te streamwise temperature profile for microcannels of 40 μm and 80 μm ydraulic diameters at CHF conditions. For te case of 40 μm cannel, tey postulated tat te bubble nucleation activity is suppressed and boiling takes place troug forced convection vaporization. Tey concluded tat te bubble dynamics mecanism is very different and tat te mecanism of bubble growt may be completely suppressed in smaller microcannels. Some studies wit working fluid oter tan water are also available; for example, Ribatski et al. [15] worked wit refrigerants (R-134a and R-245fa) wile Lee and Mudawar [16] used a dielectric fluid (HFE 7100). Following important points are noted from te CHF data reported in te literature: (i) Te exit quality at wic CHF occurs decreases by about 1.7 times for te two-fold increase in mass flux, and (ii) te correlations of Katto, Sudo and Mudawar suggest tat CHF in microcannels is a weak function of inlet subcooling. However, te correlation of Misima sows substantial dependence on inlet subcooling. CHF studies for refrigerant-123 are available in Refs. [17-19]. Kosar [20] used te annular flow model of Qu and Mudawar and mass deposition coefficient of Patankar and Puranik [21] to predict te value of CHF for bot water and refrigerant. Te model predicts te experimental data for water wit a mean absolute error of 28.9%, and in most cases te model predicted a larger value of CHF tan experimentally determined. Te discrepancy was attributed to eiter presence of parallel cannel instability wic leads to pre-mature CHF, or large tickness of te microcannel wall wic tends to make te wall temperature uniform tereby delaying CHF. Kuan and Kandlikar [18] investigated te effect of flow instabilities in six parallel rectangular microcannels, eac aving a cross-sectional area of µm 2. Tey postulated tat te ratio of evaporation momentum to surface tension force is an important parameter. Tis formed te basis of teoretical analysis of flow boiling penomena and teoretical CHF model is proposed using tese underlying forces to predict CHF in microcannels. Te proposed correlation agrees wit te experimental data wit a mean average error of 8.2% for water. Recently, Roday and Jensen [19] compared te data for water and R-123 wit existing micro/microcannel correlations. Tey found tat existing large-sized correlations do not predict CHF in microcannels. Terefore, tey developed a new correlation in low-flow subcooled boiling situation from teir experimental data. Candraker et al. [22] pointed out tat te mecanism of CHF even at te conventional scales is not well understood and needs careful investigation. It is clear from te literature survey tat very few experimental studies ave reported te value of CHF. Tis study intends to partly fill tis gap by providing data-points for four

5 Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water 29 different size microcannels and at different eat flux values. Te obtained values are compared wit available correlations. CHF is detected in te experiments by a sudden rise in temperature of te eater surface due to a decrease in te eat transfer coefficient. Literature available on ONB suggest tat te bubble dynamics in microcannel differs from tat in conventional scale passages, and terefore as an even greater impact on performance of microcannel based eat sinks. Also, it as been observed tat tere is lack of available data on effect of microcannel surface condition on ONB. Hence, data for te onset of boiling is also obtained and compared wit existing empirical correlations. 2. EXPERIMENTAL SETUP AND DATA REDUCTION 2.1. Fabrication of Microcannels Te fabrication of microcannels is done in-ouse at IIT Bombay. Te microcannels are fabricated on a 2-inc, 275 ± 25 μm-tick, p-type, <100> double-side-polised silicon wafer. Trapezoidal microcannels of dimension 77 μm 270 μm (at te top) 20mm (L) (yielding a widt of 158 μm at te bottom and a ydraulic diameter of 107 μm) are fabricated by a sequence of process steps. Te size of reservoir at te two ends of te microcannel is 6 mm 6 mm. Note tat oter ydraulic diameter trapezoidal microcannels suc as 65 μm, 70 μm and 125 μm ave also been employed in tis work. Te surface rougness as determined using profilometer was found to be less tan 0.1 μm for all microcannels (oter tan for 70 μm cannel). Te sealing of te microcannels wit a quartz plate is a crucial step in fabrication and special care was taken to avoid leakage. A detailed description of microcannel fabrication is provided elsewere [5, 6, 12, 13]. A multi-film stack of Ti-Pt is used to fabricate te microeater for controlled eat flux generation. Fabrication and caracterization details are again mentioned in Ref. [13]. Te resistance of te fabricated micro eater is found to be 445 om at room temperature. Te amount of subcooling at te inlet varies between 15 to 45 0 C, wile te exit pressure is atmosperic (1 bar). Te eat loss was calculated using te standard tecnique of supplying power to te test section witout any flow of water. All te eat supplied in tis case would be lost to te atmospere. Te surface temperature is monitored at te steady-state condition using four termocouples, wic probe different locations of te cip. Knowing te average surface temperature and te eat flux supplied, te average eat transfer coefficient for eat loss is obtained. Tis eat transfer coefficient in combination wit te measured surface temperature for a given experimental run is used for subtracting te eat loss Experimental Setup Te scematic of te experimental setup is sown in Figure 1. Te test section consists of te microcannel wit a serpentine eater integrated on te back side of te silicon wafer. Heat is supplied by a DC power source (Keitley Sourcemeter, 2400 series). Deionised water is used as te working fluid; te water is vigorously boiled and ten cooled to room temperature to remove dissolved air. Te water is pumped troug a peristaltic pump

6 30 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al. (Masterflex Easyflow II EW ) at a predetermined flow rate wic is maintained constant for a given data point. Te flow rate troug te system is given by te pump itself and as a range of ml/min. Te flow rate was independently cecked by measuring te mass of te water collected over 10 minutes duration and weiged on a microbalance. (a) Figure 1. (a) Scematic of te experimental setup along wit top and bottom view of te microcannel test section, (b) 3D view of microcannel geometry sowing te measured geometrical parameters. Te microcannel as four ports, two at te inlet and two at te outlet. Two of tese serve as entry and exit ports for te working fluid, and te remaining two are for inserting termocouples (bead size of 0.5 mm, measurement accuracy ±1 K, response time of 10 ms). Tree equally-spaced K-type termocouples (bead size 25 m, measurement accuracy of ±1 K, and response time of 1 ms) probe te surface temperature along te lengt of te microcannel. All te termocouples used are connected to a data logger (Graptec GL450) wic collects data at 10 Hz frequency. A pre-calibrated digital pressure gauge (Keller, Leo 1 wit a range of -1 to 3 bar, resolution of 0.05% of full scale, response time of 1 s) is connected at te inlet of te microcannel and provides te overall pressure drop (note tat te pressure at te exit of te microcannel is atmosperic). Te microcannel is oriented orizontally. Te uncertainty in te different quantities, bot measured and derived, is (b)

7 Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water 31 provided in Table 1. Te experimental setup as been carefully validated as discussed elsewere [5, 12]. Table 1. Uncertainty for various parameters measured in te experiment Parameter Maximum error Q 0.01 ml/min L 100 µm W 0.5 µm H 0.5 µm P 0.02 W T 0.5 ºC ΔP 2 mbar A C 1.49 % d 1.87 % q'' 5.54 % P can 5.42 % Te following procedure is adopted for performing te experiments. Te water reservoir is filled wit de-gassed, deionized water and te micro-pump is set for te desired flow rate. Te pressure drop, temperature of inlet and outlet, and flow rate values are measured simultaneously using te data logger. Te pressure drop across te entire microcannel is measured, wic includes entry and exit losses. Te pressure drop due to expansion and contraction (at te entry and exit) is owever estimated to be negligibly small (0.3%) as compared to te overall pressure drop. Te microeater dc power supply was set for a predetermined eat flux value. Te experiment is ten repeated for different flow rates and eat fluxes. Te experiments ave been carried out on a single microcannel of different ydraulic diameters (65, 70, 107, and 125 m) and 2 cm lengt. Te flow rate as been varied from ml/min corresponding to mass flux values of kg/m 2 s. Te power supplied is in te range of 0-7 W (q = 0-91 W/cm 2 ). Te maximum exit quality in tese experiments is Data Reduction Data like input power, inlet/outlet temperature, and mass flow rate obtained troug measurement as been processed using following approac. Water enters te inlet reservoir at atmosperic temperature and gets eated. Te power used to eat water in reservoir is: P mc T T reservoir p in amb. (1) Te eat gained by water at te inlet reservoir needs to be subtracted from te supplied power. Tis supplied power is directly obtained from te source meter. Terefore, power supplied to eat water in microcannel is given as,

8 32 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al. P P (1 ) P cannel supplied reservoir (2) were, λ is te percentage of eat loss to te atmospere. Microeater fabricated on te back side of te microcannel is uniformly distributed along te microcannel lengt. Terefore, Power per unit lengt of microcannel is given as, q ' Pcannel l. (3) Te eat input to te microcannel is utilised for bot sensible and latent eating of water. Tus, energy balance of te system gives te termodynamic quality as, x ' qlmcptsat Tin fg m (4) Heat flux supplied to te microcannel is given q '' P A cannel s (5) were, A s is te sum of te areas of side walls and te bottom wall. Mass flux is given as, m G A c (6) were, A c is te cross-sectional area of te microcannel. 3. ONSET OF NUCLEATE BOILING (ONB) Apart from te pressure drop and eat transfer caracteristics, te onset of nucleate boiling (ONB) and critical eat flux (CHF) are oter two important parameters of concern to a termal engineer. Te ONB point is basically te eat flux at wic te bubble nucleation is initiated in te microcannel. In most cases subcooled boiling leads to ONB. Te penomenon of ONB is important because it marks an abrupt cange from single pase flow to two pase flow. It is terefore te lower limit for eat sinks operating in two-pase regime and te upper limit for single-pase eat sinks.

9 Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water Tecnique for Measurement of ONB In tese experiments, te onset of nucleate boiling was determined from te cange in te slope of te pressure drop versus eat flux curves. Tese curves are for a fixed value of mass flow rate. In single pase, te pressure drop decreases wit an increase in eat flux, wic is due to reduction in viscosity of water wit temperature. Upon ONB, bubbles appear along te microcannel lengt wic leads to an increase in te overall pressure drop. Tus, te point of cange of slope in te pressure drop versus eat flux curve is taken as te point of ONB (Figure 2). ONB can also be obtained from te boiling curves. Te first cange in slope of te boiling curve signifies te ONB point. Te point from boiling curve agrees wit data obtained independently from te pressure drop curves for all te cases (not sown). Figure 2. Variation of experimental pressure drop for different flow rates for 65 µm cannel. Figure 3 sows te variation of te ONB obtained experimentally wit mass flux, in tree different ydraulic diameter microcannels. For a given microcannel, te eat flux for ONB increases wit an increase in mass flux, as expected. Te eat flux required for ONB also increases wit an increase in ydraulic diameter, for te same mass flux Comparison wit ONB Correlations Te eat fluxes for ONB obtained experimentally are compared wit different correlations. Te comparison of te current dataset is limited to te correlations of Bergles

10 34 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al. and Rosenow [23] and Tom et al. [24]; tese correlations are presented in Table 2. Te eat flux required for ONB based on energy consideration, wit T sat = 100 o C is calculated as, q'' mc ( T T ) / A boil p sat in. (7) Figure 3. Variation of ONB wit mass flux for different ydraulic diameters. Note tat most of te correlations for ONB are based on te work of Hsu [25]. Te subsequent models of Sato and Matsumara [26], Davis and Anderson [27] and Kandlikar et al. [28] are of similar form but wit a cange in te empirical constants. Tese additional correlations were also tested in tis work but were found to substantially over-predict most of te data-points; ence comparison wit tese correlations is not included ere. Te comparisons wit te aforementioned correlations are presented troug Figs. 4, 5 and 6. Note tat ratio of q predicted to q experimental as a function of mass flux is presented in te figures. Te variation is largest wit respect to te correlation of Tom et al. [24] (Figure 5). Te correlation of Bergles and Rosenow [23] also tends to over-predict te data (Figure 4). A relatively better matc wit te data (wit most of te points lying witin 50%) is found wile comparing wit q boil (Figure 6). On te wole, eat flux for ONB is predicted better at iger mass fluxes. Among all te correlations tested in tis study, te simple energy balance calculation provides te best prediction. From Figs. 4-6 we note tat te data for 125 m cannel is consistently over predicted by Tom et al. correlation as well as energy balance; but te energy balance metod seems to fare better. For te 65 and 70 m cannel, Tom et al. correlation as well as energy balance under predicts te data. Wile simple energy balance sow better prediction of ONB wit increasing mass fluxes, Tom s correlation sows a

11 decreasing trend for Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water 35 q predicted q Expt Bergles and Rosenow also sows a decreasing trend for fluxes and te ratio approaces unity for iger mass fluxes. wit increasing mass flux. Interestingly, te correlation of q predicted q Expt wit increased mass Figure 4. Comparison of ONB wit te correlation of Bergles and Rosenow [23]. Figure 5. Comparison of ONB wit te correlation of Tom et al. [24].

12 36 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al. Figure 6. Comparison of ONB wit q boil. 4. CRITICAL HEAT FLUX Before moving on to te experimentally obtained CHF data (in Section 4.3), te procedure for determining CHF is discussed briefly in Sections 4.1 and CHF from Boiling Curves CHF is estimated by plotting te boiling curve (eat flux versus surface temperature) (Figure 7a). Wile te first cange in slope of te boiling curve indicates transition from single-pase to two-pase, te second cange in slope occurs wen te surface temperature increases rapidly owing to a drastic reduction in eat transfer coefficient. Tus, CHF corresponds to te point wit a rapid increase in surface temperature. Note tat te two pase flow is accompanied by oscillations in temperature and pressure. From temperature-time plot, maximum, minimum and average temperatures can be determined. For example, in several cases te average surface temperatures may not be ig ( o C) wile te peak temperatures may reac anywere from o C. Boiling curves wit eat flux versus te maximum, minimum and average temperature are plotted and evaluated in Figure 7b. Te figure sows tat te CHF point is clearly obtained from plot of te maximum and average surface temperatures, wile it is not so clear from te minimum temperature curve.

13 Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water 37 (a) Figure 7. Boiling curve for 65 micron cannel wit (a) different mass flux values and (b) G = 316kg/m 2 s wit maximum, minimum and average temperatures. (b)

14 38 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al CHF from Pressure Drop Curves In tese set of experiments, te eat flux is kept constant wile te mass flux is systematically reduced. Te curve sown in Figure 8a sows te typical variation of pressure drop [5]. Te flow is initially in single pase region, wic is represented by stars; wit a reduction in flow rate te flow starts to boil, depicted by squares. Te reason for tis variation is discussed in sufficient detail in Sing et al. [5]. Te eat flux corresponding to te sudden increase in pressure drop (point e in Figure 8a) can be regarded as CHF for te particular value of mass flow rate. Note tat te flow exibits instability (i.e. te pressure varies rapidly wit time). Sing et al. [5] and Bide et al. [12] ave quantified te variation in pressure wit exit quality. In te former study, te normalized pressure (r.m.s. pressure by mean pressure) was found to be up to 14%. Furter, te local maxima/ minima in normalized pressure were found to correlate very well wit transitions in flow regime. Te above result is for a single eat flux value; a family of suc curves was obtained, as sown in Figure 8b. Te data obtained by Sing et al. [5] is unique because suc detailed measurements on pressure drop, covering single-pase liquid, two-pase liquid-vapour and extending to dryout, are not available. In particular, Bergles and Kandlikar [29] ave mentioned te difficulty in performing measurements in te regime d-e (in Figure 8a), igligting tat tere is paucity of suc data at te microscales. Te CHF (estimated from point e ) is influenced by te instabilities as pointed out by Bergles and Kandlikar [29]. It was specifically pointed out tat microcannels wit vaporization are prone to excursive instability wic results in a relatively smaller value of CHF CHF Data and Comparison wit Correlations Note tat instabilities are inerent in two-pase flow and no explicit attempt to suppress tem as been made in tis work. Quantification and discussion on te observed instabilities are provided elsewere [5, 12]. Te values reported ere are terefore only indicative of te actual CHF. Te experimental CHF data points were obtained from one of te two metods discussed above (but not bot) and plotted in Figure 9. Note tat data from four different size microcannels is plotted in te figure. Te CHF value increases wit an increase in te mass flux (oter tan for d = 125 m case). For a given mass flux value, te CHF appears to increase wit a decrease in ydraulic diameter. Note tat te surface rougness of 65 and 70 m microcannels at ε/d = and , respectively, are different by approximately an order of magnitude [12]. However, te effect of surface rougness wit approximately te same ydraulic diameter microcannel is not reflected in te CHF values. Tis may be owing to te fact tat CHF occurs due to DNB (formation of vapour blanket) as indicated by low quality values. Te vapour layer may be suc tat it is ticker tan te rougness elements on te tube surface, tereby masking te effect of surface rougness. Candraker et al. [22] correlated te CHF data from te literature (at 9660 data points) to te underlying flow regime. Tey found tat te CHF generally increases wit mass flux in te curn/slug region; owever, te CHF decreases wit increase in mass flux in te annular region. Te increase in CHF is attributed to te enanced level of turbulence in te curn flow regime as te flow

15 Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water 39 rate is increased; tis delays te occurrence of CHF [22]. On te oter and, in te annular flow regime te dryout mecanism is dictated by te entrainment and deposition of droplets on te liquid film. At iger flow rate in te annular regime te rate of droplet entrainment due to searing at te liquid film-vapor interface increases, tis causes early dryout and lower CHF values. Figure 8. Experimental pressure drop versus mass flow rate for (a) eat flux of W/cm 2 and (b) family of eat fluxes [5].

16 40 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al. Figure 9. CHF versus mass flux. Bergles and Rosenow [23] Tom et al. [24] Table 2. Correlations used for comparison wit ONB q ONB 1082Psat 1.8 T P sat in bar, q ONB in MW/m 2 Tsat q" ONB 22.65exp P P sat in bar; q ONB in MW/m 2 Single pase energy balance, mc p Tsat Tin q boil q", T sat = 100 o boil C A Figure 9 sows an increasing trend in CHF wit mass flux for all microcannels (oter tan 125 m). Te flow regime is owever believed to be annular in all te cases [5, 12]. Te result terefore suggests a cange in te mecanism leading to CHF at te micro-scales. Te mecanism of CHF is yet to be fully understood and calls for furter detailed investigations / Psat " w Tsat sat 87 2 ; ;

17 Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water 41 Table 3. Correlations used for comparison wit CHF Qu and Mudawar [14] Katto [30] Misima and Isii [31] q" q" q CHF CHF " CHF 1 q" q" CHF,3 CHF,3 C=0.25 for 33.43G q" fg CHF, 0 1 g f 1.11 K fg We sub, in 0.21 L d , 0.25G q" CG 0.15 G 0.26 G fg fg fg L d g f g f CHF,2 We We 1/ L 50 ; C= 0.34 for L 50 d d fg We L d L d L d d K 1 1 ; K 2 ; K 3 CWe g We f q" q" ; q" q" K K CHF, 1 CHF,2 CHF,0 CHF,1 ; 1 1 L d L 1/ 3 q" K K CHF, 1 q" CHF,2 ; if q" CHF,2 q" CHF,3 ; q" CHF,0 q" CHF,2 ; q" K K q CHF, 2 q" CHF,3 ; if q" CHF,3 q" CHF,4 ; q" CHF,0 q" CHF,3 ; " CHF, 3 q" CHF,4 ; q" CHF,0 q" CHF,4 A G cross sub, in 1 q " CHF fg 0.11 g g( f g ) d A fg Co Co g f 2 3 Sudo et al. [32] q g( " CHF fgg g f g ) ( f g ) g

18 Table 4. Comparison of q max (expt) wit different CHF correlations for various ydraulic diameters Case I: Hydraulic diameter = 65 micron Expt Comparison wit Correlations q" max q" cf (kw/m 2 ) q" cf (kw/m 2 ) q" cf (kw/m 2 ) q" cf (kw/m 2 ) q" max Sr. No. G (kg/m 2 s) (kw/m 2 ) (Katto) (Misima) (Sudo) (Mudawar) Exit qlty /q" Mu Expt q" max (kw/m 2 ) Case II: Hydraulic diameter = 70 micron Comparison wit Correlations q" cf q" cf q" cf (kw/m 2 ) (kw/m 2 ) (kw/m 2 ) (Katto) (Misima) (Sudo) q" cf (kw/m 2 ) (Mudawar) Sr. No. G (kg/m 2 s) Exit qlty /q" Mu Expt q" max (kw/m 2 ) Case III: Hydraulic diameter = 107 micron Comparison wit Correlations q" cf q" cf q" cf (kw/m 2 ) (kw/m 2 ) (kw/m 2 ) (Katto) (Misima) (Sudo) q" cf (kw/m 2 ) (Mudawar) Sr. No. G (kg/m 2 s) Exit qlty /q" Mu q" max q" max

19 Expt q" max (kw/m 2 ) Case IV: Hydraulic Diameter = 125 micron Comparison wit Correlations q" cf q" cf q" cf q" cf (kw/m 2 ) (kw/m 2 ) (kw/m 2 ) (kw/m 2 ) (Katto) (Misima) (Sudo) (Mudawar) Exit qlty Sr. No. G (kg/m 2 s) q" max /q" Mu

20 44 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al. Figure 10. Comparison of CHF wit te correlation of Sudo et al. [32] and Qu and Mudawar [14] for 107 µm cannel. Te CHF value obtained experimentally is compared wit te correlations of Katto [30], Misima and Isii [31], Sudo et al. [32] and Qu and Mudawar [14]. Tese correlations are tabulated in Table 3. Figure 10 and Table 4 provide a comparison of predicted and experimental values. Of te correlations tested, te correlation of Qu and Mudawar is for microscales wile te oter correlations are for conventional (macro) scale. It is seen from Table 4 tat te CHF values obtained from te correlation of Qu and Mudawar compares wit te experimental data better tan te oter tested correlations. In general, te correlation of Katto over-predicts te CHF data, wile te correlation of Misima under-predicts. A similar observation as been reported by Qu and Mudawar [14]. Te correlation of Sudo et al. predicts te CHF data reasonably well; tis correlation was owever developed for flow in vertical rectangular cannels. DISCUSSION AND CONCLUSION Experimental eat flux values for onset of nucleate boiling and critical eat flux are reported in tis paper, as a function of mass flux and ydraulic diameter. Te primary goal of tis work is to add data points to te existing literature on microcannels, so tat a more appropriate correlation for CHF at te microscales can eventually emerge. Te obtained range of ONB and CHF are kw/m 2 and kw/m 2, respectively, wit ydraulic diameter in te range of m and mass flux in te range of kg/m 2 -s. Te CHF

21 Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water 45 is obtained from te boiling curve were a distinct increase in surface temperature occurs for a comparatively small increase in te applied eat flux. Alternatively, te CHF is obtained from te pressure drop curves in te two pase region, were backflow was observed. Te CHF points are accompanied by fluctuations in pressure and temperature, and are responsible for premature CHF in microcannels. In several cases te average surface temperatures may not be ig ( o C) wile te peak temperatures may reac anywere from o C. Suc ig temperatures toug observed intermittently, can cause serious damage to te device. Besides providing ONB and CHF data in extremely small cannels, tis paper also provides comparison wit different correlations. In particular, te experimental CHF values are found to compare witin 25% wit te correlation of Qu and Mudawar [14]. Tis correlation as been developed specially from microcannel database for CHF obtained from experiments done on water and R-113. It is needless to empasize tat te CHF data points are difficult to obtain and are rarely reported in te literature especially in te context of microcannels. Te generation of CHF data is important as te occurrence of CHF is an important design consideration in two pase microcannel eat sinks. Tese results terefore provide useful lower and upper limits on eat flux to maintain two pase flow regime in microcannels. ACKNOWLEDGMENTS We are grateful to Mr. Ansul Jain for some initial calculations. REFERENCES [1] J.R. Tome, Boiling in microcannels: a review of experiment and teory, Int. J. Heat and Fluid Flow 25 (2004) [2] M Lee, Y. Y. Wong, M. Wong, Y. Zoar, Size and sape effects on two pase flow patterns in microcannel forced convection boiling, Journal of Micromecanics and Microengineering, 13 (2003) [3] W. Qu, I. Mudawar, Flow boiling eat transfer in two-pase micro-cannel eat sinks I. Experimental investigation and assessment of correlation metods, Int. J. Heat Mass Transfer 46 (2003) [4] L. Zang, J.M. Koo, L. Jiang, M. Asegi, K.E. Goodson, J.G. Santiago, T. W. Kenny, Measurements and modelling of two pase flow in microcannels wit nearly constant eat flux boundary conditions, J. Microelectromecanical Systems, 11 (2002) [5] S.G. Sing, R. Bide, B. Puranik, S.P. Duttagupta, A. Agrawal, Two-pase flow pressure drop caracteristics in trapezoidal silicon microcannels, IEEE Transactions of Components and Packaging Tecnology, 32 (2009) [6] S. G. Sing, A. Jain, A. Sridaran, S. P. Duttagupta, A. Agrawal, Flow map and measurement of void fraction and eat transfer coefficient using image analysis tecnique for flow boiling of water in silicon microcannel, J. Micromec. Microeng., 19 (2009)

22 46 R. R. Bide, S. G. Sing, Vijay S. Duryodan et al. [7] L. Jiang, M. Wong, Y. Zoar, Pase cange in microcannel eat sinks wit integrated temperature sensors, J. of Microelectromecanical systems, 8 (1999) [8] M.E. Steinke, S.G. Kandlikar, An Experimental Investigation of Flow Boiling Caracteristics of Water in Parallel Microcannels, Transactions of ASME, 126 (2004), [9] G. Hestroni, A. Mosyak, E. Pogrebnyak, Z. Segal, Periodic boiling in parallel microcannels at low vapour quality, Int. J. Multipase flow, 32 (2006) [10] H.Y. Wu and P. Ceng, Boiling instability in parallel silicon microcannels at different eat flux, Int. J. of Heat and Mass Transfer, 47 (2004), [11] G. Hetsroni, A. Mosyak, E. Pogrebnyak, Z. Segal, Explosive boiling of water in parallel micro-cannels, Int. J. Multipase Flow, 31 (2005) [12] R.R. Bide, S.G. Sing, A. Sridaran, S.P. Duttagupta, A. Agrawal, Pressure drop and eat transfer caracteristics of boiling water in sub-undred micron cannel, Experimental Termal and Fluid Science, 33 (2009) [13] S.G. Sing, A. Kulkarni, S.P. Duttagupta, B.P. Puranik, A. Agrawal, Impact of aspect ratio on flow boiling of water in rectangular microcannels, Experimental Termal and Fluid Science 33 (2008) [14] W. Qu, I Mudawar, Measurement and correlation of critical eat flux in two-pase micro-cannel eat sinks, International Journal of Heat and Mass Transfer 47 (2004) [15] L. Wojtan, R. Revellin, J.R. Tome, Investigation of saturated critical eat flux in a single, uniformly eat microcannel, Experimental Termal and Fluid Science, 30 (2006) [16] J. Lee, I. Mudawar, Critical eat flux for subcooled flow boiling in micro-cannel eat sinks, International Journal of Heat and Mass Transfer, 52 (2009) [17] Kosar, and Y. Peles, Critical Heat Flux of R-123 in Silicon-Based Microcannels, Journal of Heat Transfer, 129 (2007) [18] W.K. Kuan, and S.G. Kandlikar, Experimental Study and Model on Critical Heat Flux of Refrigerant-123 and Water in Microcannels, Journal of Heat Transfer, 130 (2008) [19] A.P. Roday, and M.K. Jensen, Study of te Critical Heat Flux Condition wit Water and R-123 During Flow Boiling in Microtubes. Part I: Experimental Results and Discussion of Parametric Effects, Int. J. Heat Mass Trans., 52 (2009) [20] Kosar, A model to predict saturated critical eat flux in minicannels and microcannels. Int. J. of Termal Sciences 48 (2009) [21] P.U. Patankar, B.P. Puranik, Modifications and extensions to te annular flow model, in Proceedings of te Fourt International Conference on Nancannels, Microcannels, and Minicannels, Limerick, Ireland, ICNMM (2006). [22] D.K. Candraker, P.K. Vijayan, D. Saa and R.K. Sina, Investigation on te caracteristic of CHF in various flow pattern regimes based on look-up table data, Nuclear Engineering Design, 238 (2008), [23] A.E. Bergles, W.M. Rosenow, Te determination of forced-convection surface-boiling eat transfer, J. Heat Transfer 86 (1964) [24] J.R.S. Tom, W.M. Walker, T.A. Fallon, and G.F.S. Reising, Boiling in subcooled water during flow up eated tubes or annuli. Symposium on Boiling Heat Transfer in

23 Onset of Nucleate Boiling and Critical Heat Flux wit Boiling Water 47 Steam Generating Units and Heat Excangers, Mancester, September IMecE, London (1965). [25] Y.Y. Hsu, On te size range of active nucleation cavities on a eating surface, J. Heat Transfer 84 (1962) [26] T. Sato, H. Matsumura, On te conditions of incipient subcooled-boiling wit forced convection, Bull. JSME 7 (1963) [27] E.J. Davis, G.H. Anderson, Te incipience of nucleate boiling in forced convection flow, AICE J. 12 (1966) [28] S.G. Kandlikar, V. Mizo, M. Cartwrigt, E. Ikenze, Bubble nucleation and growt caracteristics in subcooled flow boiling of water. National Heat Transfer Conference, HTD-342, ASME, 1997, [29] E. Bergles, S.G. Kandlikar, On te Nature of Critical Heat Flux in microcannels, J. Heat Transfer 127 (2005) [30] Y. Katto, General features of CHF of forced convection boiling in uniformly eated rectangular cannels, Int. J. Heat Mass Transfer 24 (1981) [31] K. Misima, M. Isii, Critical eat flux experiments under low flow conditions in a vertical annulus, ANL-82-6, NUREG/CR-2647, [32] Y. Sudo, K. Miyata, H. Ikawa, M. Kaminaga, M. Okawara, Experimental study of differences in DNB eat flux between upflow and downflow in vertical rectangular cannel, J. Nucl. Sci. Tecnol. 22 (1985) [33] N. Basu, G. R. Warrier, V. K. Dir, Onset of nucleate boiling and active nucleation site density during subcooled flow boiling, Journal of Heat Transfer 124 (2002) [34] W. Qu, I. Mudawar, Prediction and measurement of incipient boiling eat flux in micro-cannel eat sinks, International Journal of Heat and Mass Transfer, 45 (2002), [35] D. Liu, P. Lee, S. V. Garimella, Prediction of te onset of nucleate boiling in microcannel flow, International Journal of Heat and Mass Transfer, 48 (2005), [36] Agrawal, S.G. Sing, A review of state-of-te-art on flow boiling of water in microcannel, International Journal of Microscale and Nanoscale Termal and Fluid Transport Penomena, 2 (2011) [37] S.G. Sing, R.R. Bide, S.P. Duttagupta, A. Sridaran, A. Agrawal, Experimental study of water boiling in microcannel, Advances in Multipase Flow and Heat Transfer, Eds: L. Ceng and D. Mewes, Vol. 3, pp , Capter 6, 2010 (Bentam Science Publisers). Received 03 October 2011, received in revised form 12 August 2012; accepted 18 August 2012.