![]() ![]() |
![]() |
UFDC Home | UF Institutional Repository | UF Theses & Dissertations | Internet Archive | | Help |
Material Information
Record Information
|
Table of Contents |
Title Page
Page i Acknowledgement Page ii Table of Contents Page iii Page iv Page v List of Tables Page vi List of Figures Page vii Page viii Page ix Page x Nomenclature Page xi Page xii Page xiii Abstract Page xiv Page xv Chapter 1. Introduction Page 1 Page 2 Page 3 Page 4 Chapter 2. Literature review and background Page 5 Page 6 Page 7 Page 8 Page 9 Page 10 Page 11 Page 12 Page 13 Page 14 Page 15 Page 16 Page 17 Page 18 Page 19 Page 20 Page 21 Page 22 Page 23 Chapter 3. Experiment system Page 24 Page 25 Page 26 Page 27 Page 28 Page 29 Page 30 Page 31 Page 32 Page 33 Page 34 Page 35 Page 36 Page 37 Page 38 Page 39 Page 40 Page 41 Page 42 Page 43 Page 44 Page 45 Page 46 Page 47 Page 48 Page 49 Page 50 Page 51 Chapter 4. Single bubble boiling experiment Page 52 Page 53 Page 54 Page 55 Page 56 Page 57 Page 58 Page 59 Page 60 Page 61 Page 62 Page 63 Page 64 Page 65 Page 66 Page 67 Page 68 Page 69 Page 70 Page 71 Page 72 Page 73 Page 74 Page 75 Page 76 Page 77 Page 78 Page 79 Page 80 Page 81 Page 82 Page 83 Page 84 Page 85 Chapter 5. Dual bubble coalescence Page 86 Page 87 Page 88 Page 89 Page 90 Page 91 Page 92 Page 93 Page 94 Page 95 Chapter 6. Heat transfer effects of coalescence of bubbles from various site distributions Page 96 Page 97 Page 98 Page 99 Page 100 Page 101 Page 102 Page 103 Page 104 Page 105 Page 106 Page 107 Page 108 Page 109 Page 110 Page 111 Page 112 Page 113 Page 114 Page 115 Page 116 Page 117 Page 118 Page 119 Page 120 Page 121 Page 122 Page 123 Page 124 Page 125 Chapter 7. Mechanistic model for bubble departure and bubble coalescence Page 126 Page 127 Page 128 Page 129 Page 130 Page 131 Chapter 8. Conclusions and future work Page 132 Page 133 Page 134 Appendix A. Notes of programming codes for data acquisition Page 135 Page 136 Page 137 Page 138 Page 139 Page 140 Page 141 Page 142 Page 143 Page 144 Page 145 References Page 146 Page 147 Page 148 Page 149 Biographical sketch Page 150 Page 151 Page 152 Page 153 |
Full Text |
EFFECTS OF BUBBLE COALESCENCE AND HEATER LENGTH SCALE ON NUCLEATE POOL BOILING I By TAILIAN CHEN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2002 ACKNOWLEDGMENTS I would like to express my sincere appreciation to my advisor, Dr. Jacob N. Chung, for his invaluable support and encouragement. Without his direction and support, this work would not have been possible. Many thanks go to Dr. Jungho Kim for his help build up the experimental setup and instructions in this research. I also greatly acknowledge Dr. James F. Klausner for his invaluable suggestions in this research. I have also had the privilege to work with Drs. William E. Lear, Jr., Zhuomin Zhang and Ulrich H. Kurzweg as other members of my committee. Their suggestions and encouragement have shaped this work considerably. Finally, I feel indebted to my wife, Lu Miao, and my son, Matthew Chen. Without their continuous support, this work would not have been possible. TABLE OF CONTENTS page ACKN OW LEDGM ENTS................................................................................................... ii LIST OF TABLES ............................................................................................................. vi LIST OF FIGURES........................................................................................................... vii NOM ENCLATURE........................................................................................................... xi ABSTRACT .................................................................................................................... xiv CHAPTER 1 INTRODUCTION ............................................................................................................ 1 1.1 Statem ent of the Problem .......................................................................................... 1 1.2 Research Objectives.................................................................................................. 2 1.3 Significance and Justification ................................................................................... 3 2 LITERATURE REVIEW AND BACKGROUND.......................................................... 5 2.1 Bubble Dynam ics and Nucleate Boiling................................................................... 5 2.2 Coalescence of Bubbles .......................................................................................... 15 2.3 Critical Heat Flux.................................................................................................... 21 3 EXPERIM ENT SYSTEM .............................................................................................. 24 3.1 M icroheaters and Heater Array............................................................................... 24 3.1.1 Heater Construction ......................................................................................... 24 3.1.2 Heater Specifications ....................................................................................... 25 3.2 Constant Temperature Control and Data Acquisition System................................ 27 3.2.1 Feedback Electronics Loop.............................................................................. 27 3.2.2 gProcessor Control Board and D/A Board....................................................... 28 3.2.3 A/D Data Acquisition Boards .......................................................................... 29 3.2.4 Heater Interface Board (Decoder Board) ......................................................... 29 3.2.5 Software ........................................................................................................... 29 3.3 Boiling Conditioin and Apparatus .......................................................................... 30 3.3.1 Boiling Condition............................................................................................. 30 3.3.2 Boiling Apparatus ............................................................................................ 30 iii 3.4 Experiment Procedure............................................................................................. 30 3.4.1 Heater Calibration............................................................................................ 30 3.4.2 Data Acquisition and Visualization ................................................................. 36 3.5 Heat Transfer Analysis and Data Reduction........................................................... 37 3.5.1 Qualitative Heat Transfer Analysis ................................................................. 37 3.5.2 Data Reduction Procedure................................................................................ 38 3.5.3 Determination of Natural Convection on the Microheaters............................. 40 3.5.4 Uncertainty Analysis........................................................................................ 44 4 SINGLE BUBBLE BOILING EXPERIMENT ............................................................ 52 4.1 Introduction............................................................................................................. 52 4.2 Experiment Results ................................................................................................. 54 4.2.1 Time-averaged Boiling Curve.......................................................................... 54 4.2.2 Time-resolved Heat Flux.................................................................................. 55 4.2.3 Time-resolved Heat Flux vs. Superheat AT..................................................... 58 4.2.4 Visualization Results and Bubble Growth Rate............................................... 63 4.3 Comparison and Discussion.................................................................................... 67 4.3.1 Bubble Departure Diameter............................................................................. 67 4.3.2 Size Effects on Boiling Curve and Peak Heat Flux ......................................... 68 4.3.3 Bubble Incipient Temperature.......................................................................... 72 4.3.4 Peak Heat Flux on the Microheater.................................................................. 72 4.4 Deviations from Steady Single Bubble Formation ................................................. 73 4.4.1 Discontinued Bubble Formation ...................................................................... 73 4.4.2 Bubble Jetting................................................................................................... 79 5 DUAL BUBBLE COALESCENCE.............................................................................. 86 5.1 Synchronized Bubble Coalescence......................................................................... 86 5.2 Dual Bubble Coalescence and Analysis.................................................................. 87 5.3 Heat Transfer Enhancement due to Coalescence.................................................... 93 5.4 Bubble Departure Frequency .................................................................................. 95 6 HEAT TRANSFER EFFECTS OF COALESCENCE OF BUBBLES FROM VARIOUS SITE DISTRIBUTIONS .............................................................. 96 6.1 Coalescence of Dual Bubbles with a Moderate Separate Distance A Typical Case ......................................................................................................................... 9 6 6.2 Dual Bubble Coalescence from Heaters #11 and #14 at 100C Larger Separation Distance Case................................................................................................ 103 6.3 Dual Bubble Coalescence from Heaters #11 and #14 at 130C Larger separation Distance and Higher Heater Temperature Case............................................ 107 6.4 History of Time-resolved Heat Flux for Different Heater Separations................. 109 6.5 Time Period of a Bubbling Cycle.......................................................................... 109 6.6 Average Heat Flux of a Heater Pair Effects of Separation Distance.................. 113 6.7 Time-averaged Heat Flux from Heater #1 ............................................................ 113 6.8 Coalescence of Multiple Bubbles.......................................................................... 115 6.9 Heat Transfer Enhancement due to Coalescence Induced Rewetting................... 122 7 MECHANISTIC MODEL FOR BUBBLE DEPARTURE AND BUBBLE COA LESCEN CE ......................................................................................... 126 7.1 Rew getting M odel................................................................................................... 127 7.2 Results from the Rewetting Model ....................................................................... 128 8 CONCLUSIONS AND FUTURE WORK ................................................................. 132 8.1 Summary and Conclusions.................................................................................... 132 8.2 Future W ork .......................................................................................................... 133 APPENDIX NOTES OF PROGRAMMING CODES FOR DATA ACQUISITION.. 135 REFEREN CES................................................................................................................ 146 BIOGRAPHICAL SKETCH........................................................................................... 150 LIST OF TABLES Tables Page 3.1 The specifications of the D/A cards ........................................................................... 29 3.2 The uncertainty sources from calibration.................................................................... 46 4.1 Single bubble growth time and departure diameter ................................................... 68 4.2 Properties of FC-72 at 56C........................................................................................ 68 7.1 Effective thermal conductivity (w/m.K) used in the rewetting model...................... 129 LIST OF FIGURES Figures Page 2.1 The numerical model and results given by Mei et al. (1995)........................................ 7 2.2 The numerical model and four growth domains by Robinson and Judd (2001)........... 8 2.3 Research on bubble coalescence by Li (1996)............................................................ 17 2.4 The coalescence research performed by Bonjour et al. (2000)................................... 20 3.1 Heaters and heater array ............................................................................................. 26 3.2 Wheatstone bridge with feedback loop....................................................................... 28 3.3 Boiling apparatus ........................................................................................................ 31 3.4 Schematic of calibration apparatus and temperature control loop.............................. 32 3.5 Part of the calibration results....................................................................................... 34 3.6 Comparison of calibrated resistances of heater #1 with the calculated resistances from property relation........................................................................................... 35 3.7 The schematic showing the heat dissipation from a heater......................................... 39 3.8 Heat flux comparison from experimental and calculated results ................................ 41 3.9 Derived natural convection heat fluxes for different heater configurations................ 44 3.10 The circuit schematic for temperature control and voltage division......................... 47 3.11 The uncertainty at different temperatures................................................................. 50 4.1 The boiling curve of the single bubble boiling........................................................... 56 4.2 The heat flux variation during one bubble cycle......................................................... 57 4.3 Time-resolved heat flux variation at different heater superheats................................ 59 4.4 The trend of maximum and minimum heat fluxes during one bubble cycle with various heater superheats ...................................................................................... 61 4.5 A hypothetical model for bubble departure from a high temperature heater.............. 62 4.6 Bubble images of a typical bubble cycle taken from the bottom................................ 64 4.7 Measured bubble diameters at different time.............................................................. 66 4.8 Bubble growth rate at different time ........................................................................... 66 4.9 The visualization result of bubble departure-nucleation process for heater #1 at 54C ............................................................................................................................... 69 4.10 The relationship between bubble departure diameter and growth time.................... 70 4.11 Comparison of boiling curves ................................................................................... 71 4.12 Comparison of peak heat fluxes................................................................................ 71 4.13 The process of vapor explosion together with onset of boiling ................................ 75 4.14 The ruler measuring the distance in figure 4.13 after vapor explosion..................... 75 4.15 The bottom images for vapor explosion and boiling onset process.......................... 76 4.16 The heat flux variation corresponding to the vapor-explosion and boiling onset process................................................................................................................... 78 4.17 The chaotic bubble jetting process for heater #1 at 110C....................................... 81 4.18 The bottom images of chaotic bubble jetting process for heater #1 at 110C.......... 82 4.19 The heat flux traces corresponding to the bubble jetting process............................. 83 4.20 The heat flux traces for bubble jetting process from heater #1 at 105C ................. 84 4.21 The heat flux traces for bubble jetting process from heater #1 at 120C.................. 85 5.1 The heat flux variation for pair #1 with #11 and pair #1 and #12............................... 88 5.2 The heat flux variation of one typical bubble cycle for two configurations (a) and (b) for heater #1 with #11, (c) and (d) for heater #1 with #12.................................... 89 5.3 The heater dry area before and after coalescence ...................................................... 90 5.4 The boiling heat flux for the two pairs (#1 with #11) and (#1 with #12) as they are set at different temperatures to generate bubbles and coalesce .................................. 92 5.5 The heat flux increase due to coalescence.................................................................. 94 5.6 Comparison of bubble departure frequency from heater #1 for coalescence and non- coalescence cases .................................................................................................. 95 6.1 The departing and nucleation process for heaters #11 and #13 at 100C ................... 99 6.2 The coalescence process (2 cycles of oscillation) for heaters #11 and #13 at 100C. ..... .. ....... ...... ............................................ .. .. .. .. ................ ...... ............ 100 6.3 The heat flux history for dual bubble coalescence .................................................... 101 6.4 Photographs showing the interface interaction ......................................................... 102 6.5 The side-view photographs of coalescence-departure-nucleation process for heaters #11 and #14 at 100C .......................................................................................... 104 6.6 The bottom view of coalescence-departure-nucleation process for heaters #11 with # 14 at l OOC ........................................................................................................ 105 6.7 The heat flux history from heater #11 when #11 and #14 are set at 100C .............. 106 6.8 The heat flux history for heaters #11 and #14 at 130C ........................................... 108 6.9 The heat flux history from heater #1 with time......................................................... 110 6.10 The bottom images for one bubble cycle and dryout changing (time in second) ... 111 6.11 The time duration of one bubble cycle at different superheats ............................... 112 6.12 The average heat fluxes from different pairs of heaters at different superheats..... 114 6.13 The average heat fluxes from heater #1 at different superheats.............................. 114 6.14 Four heater configurations for multiple bubble coalescence experiment ............... 115 6.15 The heat fluxes from each heater for case A (figure 6.14) vs. superheat................ 117 6.16 The heat fluxes from each heater for case C (figure 6.14) vs. superheat ................ 117 6.17 The average heat fluxes for different heater configurations vs. superheat.............. 118 6.18 The coalescence sequence for heaters #3, #5, #7, and #15 at 80C ........................ 120 6.19 The coalescence sequence for heaters #1, #3, #5, #7, and #15 at 80C.................. 120 6.20 The coalesced bubble formed on heaters #3, #5, #7, and #15 at 100C ................. 121 6.21 The coalesced bubble formed on heaters #1, #3, #5, #7, and #15 at 100C ........... 121 6.22 The heat fluxes from heater #11 when the other heaters are at various superheats. 123 6.23 The heat flux history from heater #11 during 0.8 second ....................................... 123 6.24 The heat flux history from heater #11 when the bubble formed on it is pulled toward the primary bubble from heaters #1, #2, #3, and #4............................... 124 7.1 Typical heat flux spikes during bubble coalescence................................................. 126 7.2 The fluid flow induced by the bubble departure....................................................... 127 7.3 The fluid flow induced by the bubble coalescence................................................... 128 7.4 The results from the rewetting model for the bubble coalescence............................ 130 7.5 The results from the rewetting model for the bubble departure................................ 131 A. 1 The main interface to select heaters......................................................................... 135 A.2 Heaters temperature form......................................................................................... 137 A.3 The architecture for addressing the selected heaters................................................ 141 A.4 The pins layout of the D/A cards ............................................................................. 143 A.5 The data acquisition form......................................................................................... 144 NOMENCLATURE A heater surface area (cm2) C platinum constant coefficient (Q/f2C) Cp,i specific heat (kJ/kg.K) Db bubble departure diameter Db dimensionless bubble departure diameter Fo Fourier number g gravitational acceleration (m/s2) g(Pr) function of Prandtl number in Eq.(3.8) GrL Grashof number defined in Eq.(3.7) Hfg latent heat of vaporization (kJ/kg) Ja Jacob number k thermal conductivity (w/m.K) keff effective thermal conductivity (w/m.K) L heater characteristic length Lo heater characteristic length n constant Nux Nusselt number NuL average Nusselt number for a heater NUL,- corrected Nusselt number for a heater Phd hydrodynamic pressure Pv vapor pressure inside a bubble P.o bulk liquid pressure Pr\ Prandtl number q"t heat flux (w/cm2) q"condi conductive heat transfer rate per unit area from a heater to substrate with boiling (w/cm2) q"cond2 conductive heat transfer rate per unit area from a heater to substrate without boiling (w/cm2) q"natural natural convection heat transfer rate per unit area from a heater (w/cm2) q'radl radiation heat transfer rate per unit area from a heater with boiling (w/cm2) q"rad2 radiation heat transfer rate per unit area from a heater without boiling (w/cm2) q"rawi total heat transfer rate per unit area supplied to a heater with boiling (w/cm2) q"raw2 total heat transfer rate per unit area supplied to a heater without boiling (w/cm2) q"s(t) heat transfer with time in the rewetting model(w/cm2) q "top boiling heat transfer (w/cm2) R electrical resistance at temperature T Rb bubble diameter (mm) Ro resistance at ambient temperature To t time (seconds) Ti bulk liquid temperature (C) TS heater surface temperature (C) ambient temperature (C) voltage across the heater (V) offset voltage of the Opamp (V) Greek Symbols e uncertainty associated with temperature (C) w uncertainty associated with heat flux (w/cm2) pA liquid density (kg/m3) pv vapor density (kg/m3) O1 thermal diffusivity (kJ/kg.K) Teg dimensionless bubble growth time Tg bubble growth time (seconds) /9 coefficient of thermal expansion v kinematic viscosity (m/s2) 1t liquid viscosity (N-s/m2) ar surface tension (N/m) ATe heater superheat (C) Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EFFECTS OF BUBBLE COALESCENCE AND HEATER LENGTH SCALE ON NUCLEATE POOL BOILING By Tailian Chen August 2002 Chairman: Jacob N. Chung Major Department: Mechanical Engineering Nucleate boiling is one of the most efficient heat transfer mechanisms on earth. For engineering applications, nucleate boiling is the mode of choice owing to its narrow operating temperature range and high heat transfer coefficient. Though much effort has been expended in numerous investigations over several decades, controversies persist, and a complete understanding of the boiling mechanisms still remains elusive. In this research, microscale array heaters have been utilized, where each heater has a size of 270 gtm x 270 gim. The time and space-resolved heat fluxes from constant- temperature heaters were acquired through a data acquisition system with side and bottom images taken from the high speed digital visualization system. Together with the heat fluxes from the heaters, we have been able to obtain much better new results. For the single bubble boiling, we found that in the low superheat condition, microlayer evaporation is the dominant heat transfer mechanism while in the high superheat condition, conduction through a vapor film is dominant. During the bubble departure, a heat flux spike was measured in the lower superheat regime whereas a heat flux dip was found in the higher superheat regime. As the heater size is reduced in pool boiling, the boiling curve shifts towards higher fluxes with corresponding increases in superheats. Comparing with the single bubble boiling, two major heat flux spikes have been recorded in dual bubble coalescence. One is due to the bubble departure from the heater surface and the other one is due to the bubble coalescence. The overall heat transfer is increased due to the rewetting of the heater surface as a result of bubble-bubble interaction. A typical ebullition cycle includes nucleation, single bubble growth, bubble coalescence, continued bubble growth and departure. We have found that in general the coalescence enhances heat transfer as a result of creating rewetting of the heater surface by colder liquid and turbulent mixing effects. The enhancement is proportional to the ebullition cycle frequency and heater superheat. It was also measured that the longer the heater separation distance is the higher the heat transfer rate is from the heaters. For the multiple bubble coalescence, the key discovery is that the ultimate bubble that departs the heater surface is the product of a sequence of coalescence by dual bubbles. For the heat transfer enhancement, it was determined that the time and space averaged heat flux for a given set of heaters increases with the number of bubbles involved and also with the separation distances among the heaters. In particular, we found that the heat flux levels for the internal heaters are relatively lower than those of the surrounding ones. CHAPTER 1 INTRODUCTION 1.1 Statement of the Problem In nucleate boiling from a heated surface, vapor bubbles generated tend to interact with neighboring bubbles when the superheat is high enough to activate higher nucleation site density. It is believed that bubble-bubble interaction and coalescence are responsible mechanisms for achieving high heat transfer rates in heterogeneous nucleate boiling. Bubble-bubble coalescence creates strong disturbances to the fluid mechanics and heat transfer of the micro- and macro-layer beneath the bubbles. Because of the complicated nature, the detailed physics and the effects of the bubble-bubble interaction process have never been completely unveiled. During the terrestrial pool boiling, the critical heat flux (CHF), which is the upper heat transfer limit in the nucleate boiling, represents a state of balance. Because the buoyancy force strength is relatively constant on earth, for heat fluxes lower than the CHF, this force is more than that required for a complete removal of vapor bubbles formed on the heater surface. At the CHF, the buoyancy force is exactly equal to the force required for a total removal of the vapor bubbles. For heat fluxes greater than the CHF, the buoyancy force is unable to remove all the bubbles, thus resulting in the accumulation and merging of bubbles on the heater surface, which eventually leads to a total blanketing of the heater surface by a layer of superheated vapor. Heat transfer through the vapor film, so-called film boiling, is much less efficient than the nucleate boiling and produces very high heater surface temperatures. In many modem heat transfer applications, the length scale of heat source gets smaller, where traditional boiling theory may not be applicable. How does the length scale affect the boiling phenomenon? How do we maximize the heat transfer from a limited heater surface? Those are frequent questions asked by industrial engineers. There is no way to accurately answer these questions without the knowledge of microscale boiling phenomenon. 1.2 Research Objectives This research seeks to perform high-quality experiments to unmask the effects of bubble coalescence and length scale of heaters on heterogeneous nucleate boiling mechanisms. Through this research, we would be able to achieve the following: 1. To find the basic physics of bubble coalescence and its effects on fluid mechanics and heat transfer in the micro- and macro-layers and to develop a simple mechanistic model for this phenomenon. 2. To obtain a fundamental understanding of the effects of heater length scale on the boiling mechanism and boiling heat transfer. We intend to study the detailed physics of bubble formation on small heaters, bubble coalescence and bubble dynamics, and heat and mass transport during bubble coalescence. The purpose is to delineate through experiment and analysis the contributions of the key mechanisms to total heat transfer. This includes micro/macro layer evaporation on single and merged bubbles attached to a heated wall, and heat transfer enhancement during coalescence of bubbles on the heater wall. We intend to provide answers to the following: How will the bubble coalescence affect the heat transfer from the heater surface? What mechanisms are at play during bubble coalescence? In other words, how do the thermodynamic force, surface tension force, and hydrodynamic force that are associated with the merging process balance one another? What controls the bubble nucleation, growth, and departure from the heater surface in nucleate boiling when bubble coalescence is part of the process? How does the heater surface superheating level affect the bubble coalescence? How does the heater length scale affect the bubble inception and boiling heat transfer? 1.3 Significance and Justification Nucleate boiling has been recognized as one of the most efficient heat transfer mechanisms. In many engineering applications, nucleate boiling heat transfer is the mode of choice. Boiling heat transfer has the potential advantage of being able to transfer a large amount of energy over a relatively narrow temperature range with a small weight to power ratio. For example, boiling heat transfer has been widely used in microelectronics cooling. Apart from the engineering importance, there are science issues. Currently, the mystery of critical heat flux remains unsolved. As a matter of fact controversies over the basic transport mechanisms of bubble coalescence and its effects on the role of microlayer and macrolayer, liquid resupply and heater surface property continue to puzzle the heat transfer community. With the micro-array heaters, high speed data acquisition system and high speed digital camera, we would have a better chance to unlock the secrets of nucleate boiling. Boiling is also an extremely complex and elusive process. Although a very large number of investigators have worked on boiling heat transfer during the last half century, unfortunately, for a variety of reasons, far fewer efforts have focused on the physics of boiling process. Most of the reported work has been tailored to meeting the needs of engineering applications and as a result has led to 4 correlations involving several adjustable parameters. The correlations provide a quick input to design, performance and safety issues; hence they are attractive on a short-term basis. However, the usefulness of the correlations diminishes very rapidly as parameters of interest start to lie outside the range for which the correlations were developed. CHAPTER 2 LITERATURE REVIEW AND BACKGROUND 2.1 Bubble Dynamics and Nucleate Boiling Boiling is an effective heat transfer mode because a large amount of heat can be removed from a surface with a relatively small temperature difference between the surface and the bulk liquid. Boiling bubbles have been successfully applied in ink-jet printers and microbubble-powered acuators. The technologies still in the research and development stage for possible applications include TIJ printers, optical cross-connect (OXC) switch, micropumping in micro channels, fluid mixers for chemical analysis, fuel mixers in combustion, prime movers in micro stem engines, and the heat pumps for cooling of semiconductor chips in electronic devices. The boiling curve first predicted by Nukiyama (1934) has been used to describe the different regimes of saturated pool boiling. But until now, there are no theories or literature that exactly explains the underlying heat transfer mechanisms. Forster and Grief (1959) assumed that bubbles act as micropumping devices removing hot fluid from the wall, replacing it with cold liquid from the bulk. The proposed equation for calculating the boiling heat transfer is i=P T ---T,)f (2.1) Mikic and Rohsenow (1969) developed a model whereby the departing bubble scavenges away the superheated layer, initiating transient conduction into the liquid. They also proposed the heat flux calculation by the expression S.\1/2 = 4kplcp, [TwTf2 (2.2) Mei et al. (1994) investigated the bubble formation and growth by considering simultaneous energy transfer among the vapor bubbles, liquid microlayer, and the heater. They presented the sketch for a growing bubble, microlayer and heating solid which is shown in figure 2.1(a), where the bubble dome has the shape of a sphere of radius of Rb(t), the microlayer has a wedge shape with a radius Rb(t), wedge angle 0 centered at r =0. They found that four dimensionless parameters governing the bubble growth rate are Jacob number, Fourier number, thermal conductivity and diffusivity ratios of liquid and solid. And the Jacob number is the most important one affecting c and cl, two empirically determined constants that depict the bubble shape and microlayer wedge angle. Using this model, they examined the effects of varying Ja, Fo, k, ac separately. The results are shown in figure 2.1(b), and they claim that the effects of Ja, Fo, k, 04 on the normalized growth rate Rb(t) are the following: (1) increasing Ja and ai will result in an increasing Rb(t), and (2) increasing Fo and k will result in a decreasing Rb(t). Recently, Robinson and Judd (2001) developed a theory to manifest the complicated thermal and hydrodynamic interactions among the vapor, liquid and solid for a single isolated bubble growing on a heated plane surface from inception. They use the model shown in figure 2.2(a) to investigate the growth characteristics of a single isolated hemispherical bubble growing on a plane heated surface with a negligible effect of an evaporating microlayer. They demarcate the bubble growth into four regions, surface tension controlled growth (ST), transition domain (T), inertial controlled growth (IC), and W ). 3~Q a ct~ S" "i&a boo WUPPW todr W" ~or bdM i-aimaw S() (a) (b 1.2 1|- 1.B | ..- I 1.0 -0.01 -0.0 - 1-, 0.01 0.-a 0, 0.001- S0.6 0.001- 0.6 0.0005 0.4 0.4 0.2 0.106 0.2 I I I 1.0 :. I I 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.6 02 0.4 0.6 04 1.0 1.2 T (0) (4) Parmetric depmndec of the nomnalized growth rate R(t). (a) Ja 1, 10,100 ad 1000 at (Fo. K. a) -(, 000., 0.005); (b) Fo 0.01, 1, 10. 100, 1000 and 10000 at (Ja. a) (0. 0.005. 0.005); (c) S- 0.0005. 0.001.0.01, 0.05 at (1a, PO. 2) (10. 1. 0.005) and (d) a 0.0005.0.001.0.01 and 005 at (A.,. oa,. C)- (10, 1. 0.005). (b) Figure 2.1 The numerical model and results given by Mei et al. (1995). (a) The numerical model; (b) The numerical results. 8 Z Liquid, P., T. /Vapour "" (re. A) \ lu Heatd Su&Nfe rv (a) 200 0.10 140 2.I 0.05 Plan "td Surfiooo 180 (a S60 -0.05 400 / 1 100 180 .20 P \ / 0 -10 3.__ 2a | -20 T T C t 640 0 0.000001 0.0001 0.01 1 100 Time (mIs) (b) Figure 2.2 The numerical model and four growth domains by Robinson and Judd (2001). (a) The numerical model-Hemispherical bubble growing on a plane heated surface; (b) The four growth domains with contiuent pressures at each domain. heat transfer controlled growth (HT), which are shown in figure 2.2(b). During the surface-controlled domain, they claim that energy is continuously transferred into the bubble by conduction through liquid. But the average heat flux, thus the growth rate dRb/dt, is so small that the contribution of the hydrodynamic pressure in balancing the equation of motion is insignificant so that it essentially reduces to a static force balance, Pv P. = 2o/Rb, as shown in figure 2.2(b). The bubble growth in this domain is accelerated due to a positive feedback effect in which the increase in the radius, Rb, is related to a decreasing interfacial liquid temperature. This corresponds to an increase in q", through the increase in the magnitude of the local temperature gradient, which feeds back by a proportional increase in the bubble growth rate aRb/t. In the earliest stage of the surface tension domain, this feedback is not significant. However, in the latter stage, it becomes appreciable as indicated by a noticeable increase in Rb away from Rb, a significant decrease in both Tv and Pv and a sharp increase in q". At the transition domain, the hydrodynamic force Phd rises sharply due to the significant liquid motion outside of the bubble interface. Though q" and aRdat increase at the beginning of this stage, they are shown to decrease in the latter stage and reach the maximum value. One of the reasons for this decrease in spite of a positive feedback of surface-controlled growth can be additional resistance associated with forcing the bulk liquid out radially. The other reasons can be due to the conduction and advection occurring in the liquid adjacent to the interface. Each of these heat transfer mechanisms acts in such a way as to diminish the temperature gradients in the immediate vicinity of the vapor-liquid interface and thus has a detrimental influence on the rate at which q" and aRblat increase. The inertial controlled growth refers to the interval of bubble growth in which the rate of bubble expansion is considered to be limited by the rate at which the growing interface can push back the surrounding liquid (Carey, 1992). In this domain, the average heat flux into the bubble is very high, so heat transfer to the interface is not the limiting mechanism of the growth. The pressure difference, Pv P.o, is now balanced by the hydrodynamic pressure at the interface. The hydrodynamic pressure comprises two "inertial" terms, i.e., the acceleration term, plRb(d2Rb/dt2), and the velocity term, 3/2p1(dRb/dt)2. The two terms are of differing signs and thus tend to have an opposite influence on the total liquid pressure, and thus the force of the liquid on the bubble interface. This inertial controlled growth domain is characterized by a decreasing average heat flux and a decelerating interface. This signifies that the positive influence that the decreasing vapor temperature tends to have on the local temperature gradient is not sufficient to compensate for the rate at which advection and conduction serve to decrease the temperature gradient at the interface. The heat transfer controlled growth domain refers to the interval of bubble expansion and is considered to be limited by the rate at which liquid is evaporated into the bubble, which is dictated by the rate of heat transfer by conduction through the liquid (Carey, 1992). In this latter stage of bubble growth, the interface velocity has slowed enough so that the hydrodynamic pressure, Phd, becomes insignificant compared with the surface tension term, 2a/Rb, in balancing the pressure difference, P, Po.. This is shown in figure 2.2(b). Because the liquid temperature at the interface is now constant, the positive feedback effect, responsible for the rapid acceleration of the vapor-liquid interface in the surface tension controlled region, does not occur in this domain of growth. Conversely, the "shrinking" and "stretching" of the thermal layer in the liquid due to conduction and advection are responsible for the continuous deceleration of the interface due to the diminishing interfacial temperature gradients. The steady cyclic growth and release of vapor bubbles at an active nucleation site are termed an ebullition cycle. This bubble growth process begins immediately after the departure of a bubble. Bulk fluid replaces the vapor bubble initially. A period of time, called waiting period, then elapses during which transient conduction into the liquid occurs but no bubble growth takes place. After this, the bubble begins to grow as the thermal energy needed to vaporize the liquid at the interface arrives. This energy comes from the liquid region adjacent to the bubble that is superheated during the waiting period and from the heated surface. As the bubble emerges from the site, the liquid adjacent to the interface is highly superheated and the transfer of heat is not a limiting factor. However, the resulting rapid growth of the bubble is resisted by the inertia of the liquid and, therefore, the bubble growth is considered to be inertia controlled. During the inertia-controlled growth, the bubble generally grows radially in a hemispherical shape. A thin microlayer (evaporation microlayer) is formed between the lower portion of the bubble interface at the heated wall. Heat is said to transfer across the thin film from the wall to the interface by directly vaporizing liquid at the interface. The liquid region adjacent to the bubble interface (relaxation microlayer) is gradually depleted of its superheat as the bubble grows. Therefore, as growth continues, the heat transfer to the interface may become the limiting factor and the bubble is said to be heat transfer controlled. It is easy to see that vapor bubbles grow in two distinct stages. Fast growth emerges initially, followed by a slow growth period until detachment. In general, very rapid, inertia controlled growth is more likely under the following conditions which typically produce a build-up of high superheat levels during the waiting period and/or cause rapid volumetric growth (Carey, 1992). High wall superheat High imposed heat flux Highly polished surface having only very small cavities Very low contact angle (highly wetting liquid) Low latent heat of vaporization Low system pressure (resulting in low vapor density) Heat transfer controlled growth of a bubble is more likely for conditions which result in slower bubble growth or result in a stronger dependence of bubble growth rate on heat transfer to the interface. They include: Low wall superheat Low imposed heat flux A rough surface having many large and moderately sized cavities Moderate contact angle and moderately wetting liquid High latent heat of vaporization Moderate to high system pressures The forces acting on a bubble can be a very complicated issue. It is highly dependent on the bubble stages during its cyclic process. Among all forces, dominant force can vary depending on the bubble stages, surface superheat, boiling condition, and so forth. In general, the following major forces can be involved during boiling: surface tension, buoyancy force, inertia of induced liquid flow, drag force, internal pressure, adhesion force from the substrate, etc. In a flow boiling field, due to possible bubble slide before the bubble lifts off, boiling becomes even more complicated. Klausner et al. (1993) have studied bubble departure on an upward facing horizontal surface using R-113 as the test fluid. By balancing forces due to surface tension, quasi steady drag, liquid inertia force due to bubble growth, buoyancy, shear lift force, hydrodynamic pressure and contact pressure along and normal to the heater surface, they found that the bubble will slide before lift-off. The predicted bubble diameter at the beginning of sliding motion was found to agree well with their data. It was noted that liquid inertia force resulting from the bubble growth played a more important role in holding the bubble adjacent to the heater surface than the surface tension. However, such a conclusion is highly dependent on the base diameter that is used in calculating the force due to surface tension. In a subsequent study, Zeng et al. (1993) correlated both bubble diameter at departure and the lift off diameter by assuming that the major axis of the bubble became normal to the surface after the bubble began to slide. By balancing the components of buoyancy, liquid inertia, and shear lift force in the direction of flow, they were able to predict bubble diameter at departure. The bubble lift-off diameter was determined by balancing forces due to buoyancy and liquid inertia associated with bubble growth. For the evaluation of liquid inertia, the proportionality constant and the exponent in the dependence of bubble diameter on time were obtained from the experiments. Mei et al. (1999) have included the surface tension force, liquid inertia, shear lift force and buoyancy to determine the bubble lift-off diameter. For the model the bubble lift off diameter was shown to decrease with flow velocity and slightly increase with wall superheat. Maity and Dhir (2001) experimentally investigated the bubble dynamics of a single bubble formed on a fabricated micro cavity at different orientations of the heater surface with respect to the horizontal surface. They found that under flow boiling conditions bubbles always slide before lift off and the bubble departure diameter depends solely on the flow velocity and is independent of the inclination of the surface. Bubble departure diameter decreases with the flow velocity. Bubble lift-off diameter depends on both the flow velocity and the angular position. Lift-off diameter increases with the angular position but decreases with the flow velocity. On the other hand, commercial success of bubble jet printers (Nielsen, 1985) has inspired many researchers to apply bubble formation mechanisms as the operation principle in microsystems. For this purpose, the bubble from a microheater should be designed to present a stable, controllable behavior. Therefore, it is important to understand the bubble formation mechanisms in microheaters before they may be optimally designed and operated. Some previous work has been done in investigating bubble formation mechanism. lida et al. (1994) used a 0.1mm x 0.25mm x 0.25tm film heater subjected to a rapid heating (maximum 93 x 106 K/s). They measured the temperature of heaters by measuring the electrical resistance. The temperature measured at bubble nucleation suggested homogeneous bubble nucleation in their experiment. But the heater they used does not ensure a uniformly heated surface. Lin et al. (1998) used a line resistive heater 50 x 2 x 0.53 tm3 to produce microbubbles in Fluorinert fluids. By a computational model and experimental measurements, they concluded that homogeneous nucleation occurs on these micro line heaters. They also reported that strong Marangoni effects prevent thermal bubbles from departing. Avedisian et al. (1999) performed experiments on a heater (64.5ptm x 64.5tm x 0.2jm) used on the commercial thermal inkjet printer (TIJ) by applying voltage pulses with short duration. They claimed that homogeneous nucleation at a surface is the mechanism for bubble formation with an extremely high heating rate (2.5 x 108 K/s), and this nucleation temperature increases as the heating rate increases. Zhao et al. (2000) used a thin-film microheater of size of 100m x l lO0m to investigate the vapor explosion phenomenon. They placed the microheater underside of a layer of water (about 6gm), and the surface temperature of the heater was rapidly raised electronically well above the boiling point of water. By measuring the acoustic emission from an expanding volume, the dynamic growth of the vapor microlayer is reconstructed where a linear expansion velocity up to 17 m/s was reached. Using the Rayleigh-Plesset equation, an absolute pressure inside the vapor volume of 7 bars was calculated from the data of the acoustic pressure measurement. In a heat transfer experiment, Hijikata et al. (1997) investigated the thermal characteristics using two heaters of 50gm x 50gim and 100m x lO100m, respectively. They found that 70-80% of heat generated was released through a phase change process and heat is initially conducted in the glass substrate, then, it is transferred to the liquid layer above the heater and finally released through the evaporating process. Most of these works are based on applying a constant current or voltage so that the heat flux on the heater is maintained constant. 2.2 Coalescence of Bubbles The suggested theory about boiling mechanism is that as the temperature of a heater surface increases from the onset of nucleate boiling, more bubbles are nucleated and coalesce simultaneously on the heater surface, which makes the heat flux higher and higher until the critical heat flux point (CHF). It has been considered for a long time that bubble-bubble coalescence plays an important, if not dominant, role in the high heat flux nucleate boiling regime and during the CHF condition as well. Because of the microscopic nature and complicated flow and heat transfer mechanisms, the research on coalescence has not progressed very fast in both experimental and theoretical fronts. Coalescence of bubbles on a surface is a highly complicated process that is involved with a balance among surface tension, viscous force and inertia. The phenomenon is intrinsically a fast transient event. For the above reasons, the research of coalescence between bubbles has been rather limited. Li (1996) claims when two small bubbles approach each other, a dimpled thin liquid film is formed between them, as shown in figure 2.3(a). He developed a model for the dynamics of the thinning film with mobile interfaces, in which the effects of mass transfer and physical properties upon the drainage and rupture of the dimpled liquid film are investigated. The model predicts the coalescence time, which is the time required for the thinning and rupture of the liquid film, given only the radii of the bubbles and the required physical properties of the liquid and the surface such as surface tension, London- van der Waals constants, bulk and surface diffusion coefficients. The comparison of the predicted time of coalescence and experimental results is shown in figure 2.3(b), which shows that predicted coalescence is less than experimental results, and much less than the results predicted when the immobile interface is considered. In his research, no heat transfer is considered to affect the bubble coalescence. Yang et al. (2000) performed a numerical study to investigate the characteristics of bubble growth, detachment and coalescence on vertical, horizontal, and inclined downward-facing surfaces. The FlowLab code, which is based on a lattice-Boltzmann model of two-phase flows, was employed. Macroscopic properties, such as surface tension and contact angle, were implemented through the fluid-fluid and fluid-solid interaction potentials. The model predicted a linear relationship between the macroscopic properties of surface tension and contact angle, and microscopic parameters. Hydrodynamic aspects of bubble coalescence are investigated by simulating the growth and detachment behavior of multiple bubbles generated on horizontal, vertical, and 17 z. Phase I (Phase C Szh; (r'.lt) h f(r hi't)-t! h Z',t (rt) = Phase 2 (a) - immobile interface (0) pe- s ent study a e000 experimmntol / .400 7 , 4oo- m-- - 200 s o. o- i iii ii .1 (b) 600, s BOW .400- / 2 4oo. ,/. '-yI 200 -s oJ , OW. 600- .5400- *^ w 200 - 0.00 0.02 0.04 0.06 R; (cm) (b) Figure 2.3 Research on bubble coalescence by Li (1996). (a) The thin liquid film between bubbles; (b) Comparison of the predicted results with experimental results. inclined downward-facing surfaces. For the case of horizontal surface, three distinct regimes of bubble coalescence were represented in the lattice-Boltzmann simulation: lateral coalescence of bubbles situated on the surface; vertical coalescence of bubbles detached in a sequence from a site; and lateral coalescence of bubbles, detached from the surface. Multiple coalescence was predicted on the vertical surface as the bubble detached from a lower elevation merges with the bubble forming on a higher site. The bubble behavior on the inclined downward-facing surface was represented quite similarly to that in the nucleate boiling regime on a downward facing surface. Bonjour et al. (2000) performed an experimental study of the coalescence phenomenon during nucleate pool boiling. Their work deals with the study of the coalescence phenomenon (merging of two or more bubbles into a single larger one) during pool boiling on a duraluminium vertical heated wall. Various boiling curves characterizing boiling (with or without coalescence) from three artificial nucleation sites with variable distance apart are presented. The heat flux ranges from 100 to 900 w/cm2 and the wall superheat from 5 to 35 K. They pointed out that the coalescence of bubbles growing on three sites results in higher heat transfer coefficients than single-site boiling, which is attributed to the supplementary microlayer evaporation shown in figure 2.4(a). However, the highest heat transfer coefficients are obtained for an optimal distance between the sites for which coalescence does not occur. They used different inter-site distances (a = 0.26mm, 0.64mm, 1.05mm, 1.50mm, 1.82mm) to generate bubbles and stated that for low and high intersite distances, the heat flux deviation is limited by the seeding phenomenon and the intersite distance, respectively, whereas for moderate intersite distances, the heat flux deviation is maximum because of the vicinity of the sites and absence of seeding. They also showed that due to bubble coalescence there is a noticeable change of the slope of the coiling curve. They claim that such a change is attributed to coalescence and not to the progressive activation of the sites. They also showed that the coalescence of two bubbles has a much lower effect on the slope of the boiling curve, because the supplementary microlayer evaporation with two bubbles is lower than with three bubbles and consequently has a lower effect on the heat transfer coefficient. They used the influence area shown in figure 2.4(b) to depict the locations of the influence areas for various intersite distances. For a = 0.26mm, the influence area is overlapped for the three bubbles, thus a small overall influence area, which for the heater of larger area results in a large heat transfer coefficient. They also proposed a map for activation and coalescence that is shown in figure 2.4(c). In their experiment, it is also shown that coalescence results in a decrease in the bubble frequency. Haddad and Cheung (2000) found that the coalescence of bubbles is one of the phases in a cyclic process during nucleate boiling on a downward-facing hemispherical surface. Bubble coalescence follows the phase of bubble nucleation and growth but precedes the large vapor mass ejection phase. A mechanistic model based on the bubble coalescence in the wall bubble layer was proposed by Kwon and Chang (1999) to predict the critical heat flux over a wide range of operating conditions for the subcooled and low quality flow boiling. Comparison between the predictions by their model and the experimental CHF data shows good agreement over a wide range of parameters. The model correctly accounts for the effects of flow variables such as pressure, mass flux and inlet subcooling in addition to geometry parameters. Ohnishi et al. (1999) investigated the mechanism of secondary bubble creation induced by bubble coalescence in a drop tower experiment. They also performed a two-dimensional numerical simulation study. They )O. Q (a) \,' '''t6 i'-~~ j y .1. ( %1 (b) 0 0 OS i i l .11 t- p m.-..o (c) Figure 2.4 The coalescence research performed by Bonjour et al. (2000). (a) The additional microlayer formed; (b) Schematic representation of influence-area; (c) Coalescence and activation map. reported that the simulation results agree well with the experimental data and indicate that the size ratio and the non-dimensional surface tension play the most important role in the phenomenon. 2.3 Critical Heat Flux The boiling curve was first identified by Nukiyama (1934) more than sixty years ago. Since then the critical heat flux (CHF) has been the focus of boiling heat transfer research. A plethora of empirical correlations for the CHF are now available in the literature, although each is applicable to somewhat narrow ranges of experimental conditions and fluids. Recently a series of review articles (Lienhard 1988a,b, Dhir, 1990, Katto, 1994, Sadasivan et al., 1995) have been devoted to the discussion of progress made in the CHF research. The consensus is that a satisfactory overall mechanistic description for the CHF in terrestrial gravity still remains elusive. The following are some perspectives on issues in CHF modeling elucidated in a recent authoritative review by Sadasivan et al. (1995). 1. CHF is the limiting point of nucleate boiling and must be viewed as linked to high-heat-flux end of nucleate boiling region and not an independent pure hydrodynamic phenomenon. The experiments dealing with CHF will be meaningful if only measurements of the high heat-flux nucleate boiling region leading up to CHF are made together with the CHF measurement. This would help resolve the issue of the role of dry area formation and the second transition region on CHF. Simultaneous surface temperature measurements are necessary. 2. Measurements of only averaged surface temperature in space or in time actually mask the dynamics of the phenomena. An improved mechanistic explanation of CHF also requires that experimental efforts be directed towards making high resolution and high-frequency measurements of the heater surface temperature. Experiments designed to make transient local point measurements of surface temperature (temperature map) and near surface vapor content will help in developing a clearer picture of the characteristics of the macrolayer and elucidating the role of liquid supply to the heater surface. Microsensor technology appears to be one area that shows promise in this respect. 3. Identifying the heater surface physical characteristics such as active nucleation site distribution, static versus dynamic contact angles, and advancing versus receding angles. These would help understand the heater surface rewetting behavior. Sakashita and Kumada (1993) proposed that the CHF is caused by the dryout of a liquid layer formed on a heating surface. They also suggested that a liquid macrolayer is formed due to the coalescence of bubbles for most boiling systems, and that the dryout of the macrolayer is controlled by the hydrodynamic behavior of coalesced bubbles on the macrolayer. Based on these considerations, a new CHF model is proposed for saturated pool boiling at higher pressures. In the model, they suggest that a liquid macrolayer is formed due to coalescence of the secondary bubbles formed from the primary bubbles. The detachment of the tertiary bubbles formed from the secondary bubbles determines the frequency of the liquid macrolayer formation. The CHF occurs when the macrolayer is dried out before the departure of the tertiary bubbles from the heating surface. One of the formulations of the model gives the well-known Kutateladze or Zuber correlation for CHF in saturated pool boiling. The vast majority of experimental work performed to date utilized the heat flux- controlled heater surface to generate bubbles. Rule and Kim (1999) were the first to utilize the micro heaters to obtain a constant temperature surface and produced spatially and temporally resolved boiling heat transfer results. Bae et al. (1999) used identical micro heaters as those used by Rule and Kim (1999) to study single bubbles during nucleate boiling. They performed heat transfer measurement and visualization of bubble dynamics. In particular, it was found that a large amount of heat transfer was associated with bubble nucleation, shrinking of dry spot before departure, and merging of bubbles. In this research, identical micro heaters to those of Rule and Kim (1999) were used to investigate the boiling microscopic mechanisms through the study of single bubbles boiling and the coalescence of bubbles. For each experiment, the temperature of the heaters was kept constant while the time-resolved and space-resolved heat fluxes and bubble images were recorded. CHAPTER 3 EXPERIMENT SYSTEM 3.1 Microheaters and Heater Array 3.1.1 Heater Construction This section describes briefly the heater array construction. The details can be referred to T. Rule, Design, Construction, and Qualification of a Microscale Heater Array for Use in Boiling Heat Transfer, Master of Science thesis in mechanical engineering at the Washington State University, 1997. The heater array is constructed, as shown in figure 3.1(a), by depositing and etching away layers of conductive and insulating material on a quartz substrate to form conductive paths on the surface which will dissipate heat when electrical current is passed through them. The basic element of the microscale heater array is the serpentine platinum heater. The heater element, as shown in figure 3.1(b), is constructed by depositing platinum onto the substrate surface, masking off the heater lines, and etching platinum away from the unmasked areas. The terminal ends of each platinum heater are connected to the edge of the chip with aluminum leads deposited on the chip. Substrate conduction was reduced by using a quartz (k = 1.5 w/mK) substrate instead of a silicon (k = 135 w/mK) substrate, since silicon is 90 times more thermally conductive than quartz. A Kapton (k = 0.2 w/mK) would further reduce substrate conduction, but it has not yet been tested as a substrate material. Quartz is an electrical insulator, so the substrate cannot be used as an electrical ground, as it was when silicon was used. An aluminum layer was deposited over the platinum heaters to serve as a common heater ground. Something must be used to separate the platinum heaters from the aluminum layer. But still the problem with the ground potential variation exists. The simplest way to eliminate the problem with ground potential variations is to provide an individual ground lead for each heater that connects to a ground bus bar. This bus bar must be large enough to provide less than 100 tV voltage difference between individual heater ground connections during all operating conditions. The construction steps are as follows: 1. A thin layer of titanium (Ti, 22) is first sputtered onto the quartz to enable the platinum (Pt, 78) to adhere to surface. 2. A 2000 A layer of platinum is deposited on top of the titanium layer. 3. The platinum and titanium are etched away to leave the serpentine platinum heaters and the power leads. 4. A layer of aluminum is then deposited and etched away to leave aluminum overlapping the platinum for the power leads and the wire bonding pads. 5. Finally, a layer of silicon dioxide is deposited over the heater array to provide a uniform energy surface across the heater. The area where wire-bond connections will later be made is masked off to maintain a bare aluminum surface. 3.1.2 Heater Specification The finished heater array measures approximately 2.7 mm square. It has 96 heaters on it, as shown in figure 3.1(c). Each individual heater is about 0.27 mm square. The lines of the serpentine pattern are 5 jgm wide, with 5 gim spaces in between the lines. The total length of the platinum lines in one heater is about 6000 pm, and the heater lines about 2000 A thick. The nominal resistance of each heater's resistance is 750 ohm. Aluminum Ground Lead Silicon Dioxide I Aluminum Power Lead Platinum Heater I Quartz Substrate U l7 3l64 63E 1 60 11591158 3 IE 1 I Ji 141 56Ir l~l 13 !55*I |f42 J llo[I l 11 !5311 144 45[46 48 49 5o0515 IB (c) Figure 3.1 Heaters and heater array. (a) Heater construction; (b) Single serpentine platinum heater; (c) Heater array with 96 heaters. 3.2 Constant Temperature Control and Data Acquisition System The system is consisted of feedback electronics circuits, interface print board connecting heaters array to the feedback electronics circuits, a D/A board used to set the heater temperature, two A/D boards for data acquisition, and software made from Visual Basics 6.0 in windows 95. 3.2.1 Feedback Electronics Loop This experiment makes use of the relationship between platinum electrical resistance and its temperature. We know that resistance of the platinum almost varies linearly with its temperature by the following relationship: (R-R IR=C(T-To) (3.1) Where R is the electrical resistance at temperature T, Ro is the resistance at a reference temperature To and C is the constant coefficient. For platinum, the value of C is 0.002 Q/QC. The key part of the loop is the Wheatstone bridge with a feedback loop where Rh is the platinum heater. Each heater has a nominal resistance 750Q. For a temperature change of 1C, the heater's resistance would change by 1.5Q, while R2, R3 and R4 are regular metal film resistors which values are not sensitive to temperature. The resistance of the digital potentiometer, Rc, can be set by the computer. Each heater has an electronic loop to regulate and control the power across it. Wheatstone bridge shown in figure 3.2 is used to carry out the constant temperature control. The bridge is said to be balanced when V1 = V2. This occurs when the ratio between R4 and Rh is the same as that between R2 and (Re+R3). The feedback loop maintains the heater at a constant temperature by detecting imbalance and regulating the current through the bridge in order to bring it back into balance. The amplifier will increase or decrease the electrical current to the circuit until the heater reaches the resistance necessary for the bridge to maintain balance. Therefore, the exact value of Rc corresponds to the temperature of the heater Rh. Figure 3.2 Wheatstone bridge with feedback loop. 3.2.2 piProcessor Control Board and D/A Board Each heater on the heater array can be individually controlled. g^Processor control card programs each of the 96 feedback control circuits with correct control voltage. Rather than having 96 separate wire connections from the computer control board, a multiplexing scheme is used, where a single wire carries a train of voltage pulses to all the boards and an address bus directs the voltage singles to the correct feedback circuit. D/A board is used to connect the computer for the software to send the addressing signals. The details can be referred to T. Rule, "Design, Construction, and Qualification of a Microscale Heater Array for Use in Boiling Heat Transfer", Master of Science thesis in mechanical engineering at Washing State University. 3.2.3 A/D Data Acquisition Boards Each of the A/D data acquisition boards used in this experiment has 48 channels. For 96 heaters, to acquire data simultaneously, we installed two A/D cards. But we just need one of them because maximum eight heaters were selected for the purpose of this research. The major parameters of the A/D cards have been shown in Table 3.1. Table 3.1 The specifications of the D/A cards # phanl 48 single ended, 24 differential or modified differential Resolution 12 bits, 4095 divisions of full scale Accuracy 0.01% of reading +/-1 bit Type successive approximation Speed micro-seconds 3.2.4 Heater Interface Board (Docoder Board) Heater interface board is used to interface the heater array with the feedback control system. Since the heaters are independently grounded on the heater array, there are 192 wires extended from the heater array to the interface board, which are accessed by the feedback electronics loops. 3.2.5 Software The software used in this experiment functions as following: 1. Address heaters so that heaters can be selected. 2. Send signals from the computer to the computer control board and D/A card to set the heaters temperature. 3. Automatic and manual heaters calibration. 4. Data acquisition. The software is developed in the Microsoft Visual Basic 6.0 environment under windows 95 running in PC. 3.3 Boiling Condition and Apparatus 3.3.1 Boiling Condition In this experiment, we choose FC-72 to be the boiling fluid. The reason for choosing FC-72 is that it is dielectric, which makes it possible for each heater to be individually controlled. The bulk fluid is at the room condition (1 atm, 25C), where its saturation temperature is 56C, thus it is subcooled pool boiling. 3.3.2 Boiling Apparatus Initially the boiling experiments were performed in an aluminum chamber. To improve the visualization results, a transparent boiling chamber was built. Another advantage of the new chamber is its flat glass walls that effectively prevent the image distortion taken by the fast speed camera described in section 3.4.2. Figure 3.3 shows the boiling experimental setup. In addition to heater array and electronics feedback system, the computer is used to select the heaters and set their temperature through D/A card and acquire data through A/D cards. 3.4 Experiment Procedure 3.4.1 Heater Calibration 3.4.1.1 Calibration apparatus The calibration apparatus includes the constant temperature oil tank with oil- circulating pump and temperature control system, as shown in figure 3.4. The constant oil tank functions to impinge the constant oil into the heater array surface. The temperature on I- Dfl--- Decodei Digital camera for bottom view Figure 3.3 Boiling apparatus. control system is used to keep the circulating oil at a constant temperature. Calibration is the beginning of the experiment, and it is also a very important step since the following boiling experiment will be based on the calibration data. Therefore, much more care should be exercised to ensure the accuracy. 3.4.1.2 Calibration procedure The calibration procedure is as follows: 1. Set the temperature controller at a certain temperature. 2. Circulating the fluids in the calibration tank. Power the heating components. After this, several minutes or more are needed to maintain the temperature of the circulating fluid at the stable temperature. 3. Start the calibration routine in PC and calibrate. The computer automatically saves the calibration result. Lubricate Oil Insulation (b) Figure 3.4 Schematic of calibration apparatus and temperature control loop. (a) The calibration system; (b) The electrical loop to maintain the temperature of calibration oil. 4. Set the heater array at another temperature. Follow the first and second steps till all calibration is completed. 5. Two calibration methods are included, automatic method and manual method. Automatic method is for regular calibration use. We can select heaters to calibrate, though, normally we calibrate all 96 heaters at one time. To find the corresponding control voltage for the set temperature, progressive increment method was used. Increment the control voltage gradually, and compare the voltage output across the heater with last time value. If the voltage across the heater is different from the last time value, we can say, the controlling system starts to regulate the circuit. Then the control voltage is the corresponding value for the set temperature. It is worth mentioning that how to set the difference to compare the voltage is important. Initially, the difference was set as 0.01. The result proves the control voltage is a little bit higher than the actual value. If the difference is set as 0.003, then the result is good. How to prove the result is the correct value for the set temperature, manual calibration method is used to check it out. Manual method is an alternative to the regular automatic method. It is designed to check the accuracy of the automatic method. By referring to the figure 3.5(a), we can see by increasing the control voltage, the heater's voltage starts to increase at about 91 axial value. Below 91, the heater's temperature is at the set temperature; the voltage across the heater is very small and at a relatively constant level. Starting at 91, increasing the control voltage will increase the heater's voltage so that heater's temperature increases accordingly. The starting value of 91 is the control voltage corresponding to the set temperature. This method is much more viewable than automatic method. Its disadvantage is slow, only one heater can be done at one time. But we can rely on its result to verify the accuracy of automatic method. 3.4.1.3 Calibration results Based on the relations between the heater resistance and temperature indicated in Eq. (3.1). For each heater, its resistance almost linearly changes with temperature. Thus, ideally, the resistance-temperature figure is an approximately straight line. The slope of the line is determined by the material property. Thus the lines representing the resistance- temperature relations have almost the same slope. On the other hand, the temperature change also depends on heater's initial resistance values. This regard is shown on figure 3.5(b) for calibration results of heaters #1 through #9, where the slope of the lines for different heaters. s 2.5 I2 | 1.5 & -- -- Voltage acquired from the heater vs. DQ values DQ value (digital pot) 500 450 400 8.350 S300 * 250 200 o 150 100 50 Calibration results for heaters (1-9) from50Cto 120C #4 /- . . #5 #7 S- #8 ----- #9 50 60 70 80 90 100 110 Impinging liquid temperature (C) 120 Figure 3.5 Part of the calibration results. (a) Result of the manual calibration method; (b) The calibration results (only heaters #1 through #9 were shown). The calibration results obtained for each heater will be used in boiling experiment to set heater temperatures. For the present research, the heaters will be always set at the same temperature in any experiment performed, thus nominally, there will not be any temperature gradient among heaters. 3.4.1.4 Comparison of calibrated resistances with the calculated resistances To validate the calibration results, i.e. the temperature of the heaters obtained by calibration, we have measured the resistances of each heater at different temperatures and then compare them with the resistances from calculation using the platinum property relation given by Eq. (3.1), the result is shown in figure 3.6. From the above comparison, we conclude that measured resistances match well with calculated resistance. It implies: (1) The heater temperature is reliable. (2) We can use calculated resistance to evaluate heat dissipation with only a small uncertainty introduced. 1030 1000 U Calculated R 0 Measured R S970 S940 S910 880 850 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 Temperature *C Figure 3.6 Comparison of calibrated resistances of heater #1 with the calculated resistances from property relation. 3.4.2 Data Acquisition and Visualization To investigate the bubble coalescence, we need to generate bubbles with appropriate separation distances. Since each of the 96 heaters on the heater array is individually controlled by the electronic feedback control system, we can set active one or more of the heaters by powering them individually so that they reach a certain temperature while leaving all other heaters unheated. To obtain different cases of bubble coalescence, all we need to do is to select heaters. In reference to figure 3.1, by powering two heaters such as heaters #1 and #11, we obtain dual bubble coalescence. Also, if we choose #1 and #3, we obtain dual bubble coalescence with a shorter separation distance. However, by powering two heaters that are too far apart such as #1 and #25, two single bubbles will be generated, but the bubble departure sizes are not large enough for them to touch and merge before they depart. Therefore, for coalescence to take place, the active heaters have to be close enough within a certain range. The power consumed by the heater was acquired by the computer data acquisition system (figure 3.3). Data were acquired at a sampling rate of 40,000 Hz for each channel of the A/D system, after allowing the heater to remain at a set temperature for 15 minutes. The data were found to be repeatable under these conditions. The acquired data were converted to heat flux from the heater according to the following basic relationship: q" = (V2 /R) I/A (3.2) For each heater configuration, the heaters were always set at the same temperature for the bubble coalescence experiment, and this temperature is varied to investigate the effect of heater superheat on the bubble coalescence. For all cases, the dissipation from heaters was acquired in sequence. Because steady state has been reached in all these experiments, sequential data acquisition does not affect the data accuracy. The bubble visualization includes both bottom and side views; though bottom views are more suitable for multiple bubble coalescence. The semi-transparent nature of the heater substrate made it possible to take images from below the heater. The setup of experiment has been shown in figure 3.3. A high-speed digital camera (MotionScope PCI 8000S) was used to take images at 2000 fps with a resolution of 240 x 210, with maximum 8 seconds of recording time. The bubble visualization was performed using the shadowgraph technique. In this technique, the bubbles were illuminated from one side while the images were taken from the other side. 3.5 Heat Transfer Analysis and Data Reduction 3.5.1 Qualitative Heat Transfer Analysis Each heater has a dimension of 0.27mm x 0.27mm. For such a small heater, the heat transfer behavior is hardly similar to that for large heaters. Specifically, the edge effects greatly affect the heat transfer. Since the heaters used in our experiment are always kept at a constant temperature, the data reduction turns out to be much simpler. Qualitatively, the heat dissipated from the heater at a certain temperature, by reference to figure 3.7(a), is composed of the following components: 1. Boiling heat transfer from the heater in boiling experiment, when the heater is superheated high enough. 2. Conduction to the substrate on which the heater is fabricated. This is due to the temperature gradient between the heater and the ambient through the substrate. 3. Radiation heat transfer due to the temperature difference between the heater and ambient. 4. Natural convection between the heater and air and FC-72 vapor mixture when the heater array is positioned vertically to be separated from the liquid for data reduction experiment. This natural convection is replaced by the boiling heat transfer from the heater when the boiling occurs on the heater surface. 3.5.2 Data Reduction Procedure To obtain the heat transfer rates due to boiling only, we conduct the experiments by the following procedure: 1. Measure the total heat flux supplied to the heater during the boiling process at different heater temperatures. The total heat flux with boiling: q"rawl = q"top + q"condl + q"radl. (3.3) 2. Tilt the boiling chamber 90, so that the heater is exposed to the air and FC-72 vapor, while separated from FC-72 liquid, and measure the total heat flux without boiling at corresponding temperatures. Therefore, the total heat flux without boiling: q'raw2 = q"natural + q"cond2 + q"rad2 (3.4) The q"condl in Eq.(3.3) and q"cond2 in Eq.(3.4) are the conduction to the substrate and ambient through the substrate. Because the heater is held at a constant temperature, they are independent of the state of fluid above the heater, thus we assume q"condji=q"cond2. The same reasoning also goes with radiation. Thus, from Eq.(3.3) and Eq.(3.4) the heat dissipated above the heater during boiling can be derived as the following: q"top = q"rawl q"raw2 + q"natural (3.5) where q"naturai is the contribution from natural convection with the mixture of air and FC- 72 vapor. To get a good estimate of the natural convection component in Eq.(3.5), we need to pay special attention to the small size of the heater we used. The details for this estimation are given in the following. Bulk liquid Ambient room (a) q a SMixture of air and vapor ^^^^^Bk ^(flnaftur Ambient room 'aui q'&nd2 4. q'rd2 (b) Figure 3.7 The schematic showing the heat dissipation from a heater. (a) The heat transfer paths from the microheater during boiling experiment; (b) The heat transfer paths from the heater during data reduction experiment. 3.5.3 Determination of Natural Convection on the Microheaters We have done a significant amount of literature research in order to rationally determine the natural convection on the heaters we were using. We first use the empirical correlations to evaluate the heat dissipated from the heater, in which we use the results of Ostrach (1953) to calculate this natural convection. Then the calculated heat dissipation is used to compare with experimental results. Conduction. We assume one dimensional conduction from the heater to the ambient through the quartz substrate. The ambient has a constant temperature of 25C0(2. Conductivity of the quartz substrate is ksub = 1.5 w/mK. For different heater temperatures To, using Fourier law of conduction, we calculate the conduction heat flux. For simplicity, we neglected the epoxy thickness that is used to seal the heater at the bottom. Since we calculate the conduction based on quartz substrate only, neglecting the heat transfer resistance of epoxy, the calculated conduction heat transfer should be larger than that if the epoxy layer is accounted for. Convection. When the liquid was separated from the heater, there is natural convection heat transfer to the mixture of air and vapor from the heater. For this calculation, the mixture of FC-72 vapor and air is approximated as ideal gas of air. Radiation. With the approximation of black body, the radiation heat flux is calculated as follows: q"= ( T,4- To4) (3.6) where cr is the Stefan-Boltzmann constant. The calculation results for the above heat transfer modes have been shown in figure (3.8). From this figure, obviously, the sum of convection, conduction and ~60 ---- Experimental heat flux from a heater -0- Calculated heat flux from a heater 0 1 . ., , I . [ . . | I . 60 90 120 150 Heat temperature (C) Figure 3.8 Heat flux comparison from experimental and calculation results. radiation is far less than the total heat flux q"raw2, when the heater is separated from the liquid. This is sufficient to prove that since the heater size is much smaller than the regular heater size, classical horizontal isothermal correlation is not applicable for this heater. We need to find another approach to evaluate the natural convection occurring from the heater for the data reduction. Baker (1972) has investigated the size effects of heat source on natural convection. He argued that as the heat source area decreases, the ratio of source perimeter to surface area increases and since substrate conduction is proportional to the source perimeter, the portion of the heat transferred by conduction into the substrate increases as the surface area decreases. In our experiment, since the heater is kept at the same constant temperature both in boiling and no-boiling conditions, the conduction to the substrate should be the same regardless of the condition on the surface of the heater. However, this small size effect does affect the natural convection calculation in our data reduction. Baker (1972) used the experimental setup similar to that used in this research to study the forced and natural convection of small size heat source of 2.00 cm2, 0.104 cm2 and 0.0106 cm2, respectively, smallest of which is over 10 times larger than the heater used in this experiment. Therefore, we can not use Baker's data in our experiment. Also according to Park and Bergles (1987) and, Kuhn and Oosthuizen (1988), due to the small size, the edge effects are important. The approach to determining q"natur in this study for a heater with height L and width W is that we first used the results of Ostrach (1953) for two-dimensional laminar boundary layer over a vertical flat plate with a height of L. Then the two-dimensional Nusselt number is corrected for the transverse width of W by the correlation developed by Park and Bergles (1988). The following provides the details of the procedure. The Grashof number, GrtL, at the trailing edge of a vertical plate with height L is defined as Gr gf(-T)L3 V 2 (3.7) where fP is the coefficient of thermal expansion, Ts is the heater temperature, T, is the ambient fluid temperature, L is the heater length, and v is the kinematic viscosity of the fluid. Based on Ostrach (1953), the local Nusselt number at the distance L from the leading edge of the heater surface is given by: NuL = Y g(Pr) (3.8) where g(Pr) is a function of the Prandtl number, Pr, of the fluid and is given by LeVevre (1956) as below: g(Pr) -= 0.75Pr (3.9) (0.609 +1.221Pr + 1.238Pr)Y The average Nusselt number NuL of the heater surface with a height of L can be obtained by: 4 NuL =-4NuL (3.10) 3 This two-dimension Nusselt number needs to be corrected to take into account the finite width effects. This correction is based on the relation given by Kuan and Oosthuizen (1988), which is repeated here: 0,4752 NuL =NuLx1+ 362.5 (3.11) NuLxdge NL JQ0-73 lRaw j In the above, Raw is the Rayleigh number based on the width W of the heater and Raw is equal to Grw x Pr. This corrected Nusselt number NuL.,edge has been used to evaluate the natural convection heat transfer from the microheaters. The derived natural convection heat fluxes for different heater configurations are given in figure 3.9 and they were used in equation (3.3) to find the boiling heat transfer flux, q"top. The order of magnitude of the derived natural convection heat fluxes using the above approach has been found to be consistent with the results given by Kuan and Oosthuizen (1988), where they investigated numerically the natural convection heat transfer of a small heat source on a vertical adiabatic surface positioned in an enclosure which is very similar to our experimental condition. Natural convection heat flux derived for 10 heaters in this experiment 9 -A-- Single heater, L=0.027cm -V-- Two heaters, L=0.038cm 8 ---- Three heaters, L=0.047cm . -0-- Four heaters, L=0.054cm S 7 -u---- Five heaters, L=0.061 cm E ^O50 60 70 80 g0 100 110 120 130 140 150 160 Heater temperature (AT) Figure 3.9 Derived natural convection heat fluxes for different heater configurations. 3.5.4 Uncertainty Analysis Uncertainty analysis is the analysis of data obtained in experiment to determine the errors, precision and general validity of experimental measurements. For single sampled experiments, the methods introduced by Kline and McClintock (1953) have been popularly used to determine the uncertainty. The theory of the method is introduced in the following. Assume R is a given function of the independent variables xs, x2, x3, ... x,, that is, R = R(xl, x2, X3,.... Xn). Let wR be the uncertainty in the result of R, and wi, w2,4w3, ... wn, be the uncertainties in the independent variables. If the uncertainties in the independent variables are all given with the same odds, then the uncertainty in the result with these odds is given as: 6 X .4 3- 2 40 50 60 70 80 90 100 110 120 130 140 150 160 Heater temperature (AT) Figure 3.9 Derived natural convection heat fluxes for different heater configurations. 3.5.4 Uncertainty Analysis Uncertainty analysis is the analysis of data obtained in experiment to determine the errors, precision and general validity of experimental measurements. For single sampled experiments, the methods introduced by Kline and McClintock (1953) have been popularly used to determine the uncertainty. The theory of the method is introduced in the following. Assume R is a given function of the independent variables x1, x2, X3, ... x,,, that is, R = R(xj, x2, x3, ... x.). Let WR be the uncertainty in the result of R, and w1, W2, W3, ---.. be the uncertainties in the independent variables. If the uncertainties in the independent variables are all given with the same odds, then the uncertainty in the result with these odds is given as: 2 W) \(R 2 /1+ W)+.+ RW )2]1/2 [(aRY (QR Yf/ liY fa/ 1" =WR -[ 2 a3 +n (3.12) w axR+ -2 w -wax. Because of the square propagation of the separate uncertainties, it is the larger ones that predominate the final uncertainty. Thus any improvement in the overall experimental result must be achieved by improving the instrumentation or technique connected with these relatively large uncertainties. It should be noted that it is equally as unfortunate to overestimate uncertainty as to underestimate it. An underestimate gives false security, while an overestimate may make one discard important results, miss a real effect or buy too much expensive instruments. The purpose of this exercise is to analyze the possible sources of uncertainties and give a reasonable estimate of each uncertainty, and finally obtain the overall uncertainty. In each experiment, the heaters are always set at the same temperature, and this temperature was increased to investigate how the boiling heat transfer changes with heater temperature. The boiling data are always associated with the pre-set temperature. Therefore, the uncertainties will include uncertainty for heater temperature and uncertainty for boiling heat transfer. The uncertainty sources come from calibration, boiling and data reduction experiments. They have been summarized in the following. The uncertainty sources from calibration. The uncertainty sources from calibration basically include the oil fluctuation in the calibration chamber, heat loss due to oil impinging on the heaters, discrete increment of control resistance and slew rate of the opamp which is used to balance the wheat-stone bridge. Each uncertainty for these has been estimated and given in table 3.2. Table 3.2 The uncertainty sources from calibration The uncertainty sources from boiling experiment. To have a better idea of uncertainty sources from boiling experiment, figure 3.10 needs to be referred. They include: (1) Wiring resistance from heater to feedback system and from feedback system to A/D card. Estimated total wiring resistance is about 50 ohm. Since heater's nominal resistance is 1000 ohm. Thus uncertainty introduced is 5%. (2) A/D card resolution, 2.44mV: 12 bits 4095 divisions of full scale (10V). For present experiments, the average voltage is about 10V. Thus the uncertainty is 2.44mV/10V, which is less than 0.03%. (3) Voltage division device. Mainly comes from the deviation of two division resistors. They are metal film resistors (1% accuracy). Thus, the voltage division uncertainty can be 101%/99% = 1.02%. Summary of uncertainty sources in this experiment. The uncertainty sources analyzed above can be divided into uncertainty sources for heater temperature and for heat flux. They are summarized as follows. (1) Uncertainty sources for temperature Oil temperature fluctuation in calibration, eTI = +0.5C. Sources of uncertainty Estimated uncertainty value Oil temperature fluctuation eT, = +0.5C Heat loss due to oil impinge en = +0.050C Discrete increment of control resistance en = +0.1 C Slew rate of Opamp Negligible +24V Qll 2N2222 OAuF From feedback card Output to DQ- RI CLK_ > .LOOK 4J Figure 3.10 The circuit schematic for temperature control and voltage division. Discrete increment of control resistance, EC2 = +0.05C. Drift of Voff of the opamp, ET3 = +0.1 C. Temperature fluctuation during boiling, especially during the vapor-liquid exchange process, ET4 = +0.05C. (2) Uncertainty sources for boiling heat transfer Wiring resistance, Evi = 5%. Resolution of A/D card, Ev2 =0.03%. Voltage division device, Ev3 = 1.02%. Natural convection, eHl = 15%. Determination of overall temperature uncertainty. Since the individual uncertainties given above are absolute values relative to one variable (temperature), the overall temperature uncertainty can be obtained by simply summing up each of them, that is: ET = ETI + ET2+ eT3 + eT4 = 0.5C + 0.1 C + 0.05 C + 0.05C = 0.7C. Determination of overall heat transfer uncertainty. From the uncertainty sources for boiling heat transfer analyzed above, the first three parts come from the voltage measurements. Since they come from the single variable, the total uncertainty for voltage can be obtained: ev = evi + ev2 + ev3 = 5% + 0.03% + 1.02% = 6.05% Now we are ready to calculate the overall uncertainty for boiling heat transfer. V2 V2 rawl raw2 (1 qtop ~ p natural (3.13) RA RA From Eq. (3.7), the total boiling heat transfer is a function of voltages and natural convection, neglecting the uncertainties contributed by R (resistance of heaters) and A (area of the heaters). From the uncertainty theory, the overall uncertainty can be written as w-. _q w +( 9q'wv +( f ---w 11 .' = a- # W '. ) + W\- -- W ) I + --(- W ) ] q #V I av V vq that is, 7 V (2 1/'2 In this equation, w, =e,xVr,w, =e,xV2 ,w. =e xq' In this experiment, the temperatures of the heaters are set in a temperature range, thus R is changing. To calculate uncertainty, we let R = 1000 ohm. Also the areas of the heaters are given as A = 0.000731cm2. The averaged values for voltages during boiling and data reduction are given at Vrawi = O10V, V,,2 = 4V. With these data, we can calculate the overall uncertainty at different temperatures. As an example, at 100C single bubble boiling: q, ,I = 40w/cm2. The overall uncertainty can be obtained as wq. = 9.2w/cm2. Also, at 100C, the boiling heat transfer can be read from the boiling curve: q'= 65w/cm2. Thus, the percent uncertainty for boiling heat transfer at 100C can be obtained readily as: Wov,, = 9.2/65 = 13.8%. For other temperatures, the uncertainty can be calculated following the same procedure. For the single bubble boiling, the boiling heat transfer uncertainties at different temperatures have been calculated and shown in the figure 3.11 (a). For dual bubble coalescence, we also calculated the overall uncertainty levels for boiling heat transfer for temperatures from 100C to 140C, which are shown in the figure 3.11(b) with the overall uncertainty values for single bubble boiling shown as well for comparison. As we have observed that in the dual-bubble coalescence boiling case, the overall uncertainty is smaller than that for the single bubble boiling at corresponding temperatures. The reason for this can come from two factors: (1) due to bubble 100 105 110 115 120 125 130 135 140 145 150 155 160 165 170 (a) 17% 16%. *% Uncertainty for single bubble S% Uncertainty for dual-bubble #1 with #11. 15%- 14% 13% 12% 11%1 10% 1 " 100 105 110 115 120 125 130 135 140 145 150 155 160 165 170 (b) Figure 3.11 The uncertainty at different temperatures. (a) The uncertainty for single bubble boiling at different temperatures; (b) The uncertainty for dual-bubble coalescence together with single bubble boiling. coalescence, the overall boiling heat transfer has been increased. (2) due to heater interaction, the natural convection is smaller for the single heater case. In summary, for temperatures, the oil temperature fluctuation during calibration contributes to the main uncertainty, though opamp offset also has some contributions. Natural convection is the main contributor to the boiling heat transfer uncertainty. The overall uncertainty for heater temperature is estimated about 0.7C, and the overall uncertainty of boiling heat flux is about 15% between 100C ~ 170C, and this uncertainty is temperature dependent. The uncertainty of dual bubble boiling is smaller than that of the single bubble boiling due to coalescence-enhanced heat transfer. CHAPTER 4 SINGLE BUBBLE BOILING EXPERIMENT 4.1 Introduction Applications of microtechnology must utilize components or systems with microscale fluid flow, heat and mass transfer. As the size of individual component shrinks and the length scale decreases drastically, the transport mechanisms involved go beyond those covered by the traditional theories and understanding. The development of the new theories and the fostering of up-to-date physical understanding have fallen behind the progress of micro machining and manufacturing. Extensive survey papers (Duncan and Peterson, 1994, Ho and Tai, 1998) of microscale single-phase heat transfer and fluid mechanics noted that an investigation of the flow characteristics of small channels has shown significant departure from the thermo-fluid correlations used for conventional macroscale flows. For example, for turbulent flow of gases in microtubes (with diameters of 3 mm to 81 mm), neither the Colbumrn analogy nor the Petukhov analogy between momentum and energy transport (Duncan and Peterson, 1994) is supported by the data. More recently Gad-el-Hak (1999) gave a complete review on the fluid mechanics of microdevices. He concluded that the technology is progressing at a rate that far exceeds our understanding of the transport physics in micro-devices. Therefore the study of micro-scale transport has become an integral part of not only understanding the performance and operation of miniaturized systems, but also designing and optimizing new devices. The current chapter presents an experimental study and analysis to provide a fundamental basis for boiling on a microheater and to investigate the small size effect on boiling mechanisms. Microheaters have been found in many applications, for example, inkjet printerhead, and actuators and pumps in microfluidic systems. Recently Yang et al. (2000) proposed a new model of characteristic length scale and time scale to describe the dynamic growth and departure process of bubbles. A correlation between bubble departure diameter and bubble growth time is established and a predication formula for bubble departure diameter is suggested by considering the analogue between nucleate boiling and forced convection. The predictions by the model agree well with experimental results that were obtained with basically macro-scale pool boiling conditions. Rainey and You (2001) reported an experimental study of pool boiling behavior using flat, microporous-enhanced square heater surfaces immersed in saturated FC-72. Flush-mounted 2cm x 2cm and 5cm x 5cm copper surfaces were tested and compared to a 1cm x 1cm copper surface that was previously investigated. Heater surface orientation and size effects on pool boiling performance were investigated under increasing and decreasing heat-flux conditions for two different surface finishes: plain and microporous material coated. Results of the plain surface testing showed that the nucleate boiling performance is dependent on heater orientation. The nucleate boiling curves of the microporous coated surfaces were found to collapse to one curve showing insensitivity to heater orientation. The effects of heater size and orientation angle on CHF were found to be significant for both the plain and microporous coated surfaces. Hijikata et al. (1997) investigated boiling on small heaters to find the optimum thickness of the surface deposited layer to enhance the heat removal from the heater in order to obtain the best cooling effect for a semiconductor. The square heaters they used are 50 gm and 100 pm and they claimed that the deposited layer conduction dominates the heat transfer due to the small sizes of the heater area. Also they presented the nucleate boiling curves for the two heater sizes and different deposited layer thickness. Rule and Kim (1999) used a meso-scale heater (2.7mm x 2.7mm) which consists of an array of 96 microheaters. Each of the microheaters was individually controlled to maintain at a constant temperature that enabled the mapping of the heat flux distributions during the saturated pool boiling of FC-72 fluid. They presented space and time resolved data for nucleate boiling, critical heat flux and transition boiling. Specifically, the outside edge heaters were found to have higher heat fluxes than those of the inner heaters. For the materials in this chapter, only one single microheater (marked as #1 in figure 3.1) is heated to produce bubbles. The experiment starts with setting the microheater at a low temperature of 50C where only natural convection occurs at this temperature. Then the temperature of the heater is incremented by 5C at a time, and for each increment, the heat dissipation by the heater is obtained by the data acquisition system until the superheat reaches 114C. 4.2 Experiment Results For each series of the boiling experiment, the heater temperature was set at 50C initially. After that the temperature was increased with 5C increments until it reached the superheat of 114C. For each temperature setting, the voltage across the heater was sampled at a rate of 4500 times per second. The time-resolved heat flux was obtained based on the heater area and its electrical resistance. 4.2.1 Time-averaged Boiling Curve Figure 4.1 shows the measured boiling curve in logarithmic and linear scales where the superheat of the heater covers a range from -6C to 114C. As the degree of superheat was increased to 54C, single bubbles were seen to nucleate. The onset of nucleate boiling (ONB) for the single bubble experiment was found at the superheat ATe of 54C to 59C. After the ONB, the degree of superheat for boiling dropped to 44C as the minimum temperature for stable boiling on this heater. It is noted that the trend of figure 4.1 remains very similar to that of the classical pool boiling curve predicted by Nukiyama (1934) and this includes the ONB phenomenon. In figure 4.1, similar to the classical macro-scale boiling, the entire boiling curve can be divided into three sections. Regime I is due to natural convection. Regimes II and III are separated by the peak heat flux (critical heat flux) which takes place at a heater superheat ATe of 90C. Also shown in figure 4.1 are the boiling data from Rule and Kim (1999) for a meso-scale heater (2.7mm x 2.7mm) which is composed of a 10 x 10 array of 96 microheaters. It is noted that the boiling heat transfer rates for a micro heater (0.27 mm) in the current work are more than twice higher than those for a meso-scale heater (2.7 mm) but the general trends are similar for both heaters. Also, the peak heat flux for a micro heater takes place at a higher heater temperature. These results are consistent with those of Baker (1972) for forced convection and natural convection that as the heater size is decreased the heat transfer increases. 4.2.2 Time-resolved Heat Flux Figure 4.2 shows the heat flux history when the microheater was set at a degree of superheat of 44C. In figure 4.2, we note that the heat flux is closely associated with the bubble life cycle during the ebullition process. [A] corresponds to a large spike that takes place during the bubble departure. When the preceding bubble departs, the heater is rewetted by the cooler bulk fluid. The establishment of the microlayer for the succeeding 45 45 T peak heat flux 40 [II] * 35 X Before ONB_. ] DRun I |30. ARun 2 *Run 3 5 0 Meso heater (Rule and Kim, 19) a % Xx U 20-' 00 [II 0o 00 1 5 0 100 10 ---- 0X0, - -'--- 1- -10 10 30 50 70 90 110 130 Superheat AT (a) 100- 90 X Before ONB 80 ORun 1 70 ARun2 60 Run 3 5-. OMeso heater (Rule and Kim, 1999) E5 .-k heat flux I40 f S30 = g 20f 0 X 10-- IX 0 1 10 100 1000 Superheat AT (b) Figure 4.1 The boiling curve of the single bubble boiling. (a) The boiling curve in linear scale; (b) The boiling curve in logarithmic scale. 120 Heat flux trace for a typical bubble cycle from #1 at AT=44*C 100 [A]| [F] ~80 50 80[C] [E] 040 20 1.5 2 2.5 3 3.5 Time (seconds) Figure 4.2 The heat flux variation during one bubble cycle. new bubble on the heater surface and the turbulent micro-convection induced by this vapor-liquid exchange lead to this large heat flux spike. [B] represents the moment when the succeeding bubble starts to grow after the vapor-liquid exchange. As the new bubble grows, the contact line that is the three-phase division expands outward. The bubble growth results in a larger dry area on the heater surface, thus the heat flux is decreasing. The low heat flux period indicated by [C], [D] and [El corresponds to the slow growth stage of a bubble. As the bubble size reaches to a certain level, the buoyancy force starts to become more important than the forces which hold the bubble to the surface, but it is still not large enough to lift the bubble from the heater surface, causing the bubble to neck. During the necking process, the contact line starts to shrink, then the dryout area is starting to decrease, thus we have observed that the heat flux is starting to increase slightly with some oscillation of a small-amplitude. Finally, the buoyancy force is large enough to detach the bubble from the heater surface, and then another bubble ebullition cycle begins. 4.2.3 Time-resolved Heat Flux vs. Superheat AT Figure 4.3 shows the time-resolved heat flux at different heater superheats (44C to 114C with increments of 10C) for a total of six-second data acquisition. Figure 4.4 shows the trends of two characteristic heat fluxes during a bubble life cycle (Point A - Point E in figure 4.2) at various heater superheats. The curve with triangles represents the peak heat flux of a spike or the minimum heat flux of a dip during the bubble departure (point A). Based on figure 4.3, the bubble departure was recorded to produce a spike for the heater temperature up to 84C, after that a heat flux dip was observed. In figure 4.4, The curve with diamonds shows the heat flux level during bubble slow growth (point E). It is clear that the two curves hold opposite trends, which is due to different controlling mechanisms as explained next. As we examine figures 4.1, 4.3 and 4.4 closely, all three figures consistently indicate that Regimes II and III are dominated by two different transport mechanisms. In figure 4.1, the heat flux increases with increasing heater superheat in Regime II while the trend reverses in Regime III. In figure 4.3, we notice that a heat flux spike is associated with the bubble departure in Regime II (heater superheat up to 84C) and a heat flux dip is seen to accompany the bubble departure in Regime III. In figure 4.4, we found that the two curves cross each other at the heater superheat of 90C which is the separating point between Regimes II and III. We believe that in Regime II bubble growth is mainly sustained by the heat transfer mechanism of microlayer evaporation that follows the rewetting of the heater surface. This scenario is supported by the presence of a heat flux spike recorded during the bubble departure. The spike is produced when the heater surface is rewetted by the liquid with micro turbulent motion which in turn causes the microlayer to form and the evaporation of the microlayer facilitates the bubble at AT=44C Max: 105.6 w/cm2, I 110 100 90 "S80 70 |60 s50 X 40 30 20 110 100 S90 80 S70 60 50 T40 30 20 I 110 100 90 .g80 S70 S60 150 40 30 20 110 100 80 3, 70 S60 B50 Z 40 30 20 Time (second) #1 at AT=64C Max: 84.6w/cm2, M in=48.9w/cm2 #1 atAT=74C Max: 74.8w/cm2, Min=51.4w/cm2 El .* . . . . i . . . . . S 0.5 1 1.5 2 2.5 3 3.5 Time (second) 4 4.5 5 5.5 6 Figure 4.3 Time-resolved heat flux variation at different heater superheats. #1 atAT=54C Max: 94.8w/cm2, Min=47.1 wlcm2 #1 atAT=84C Max: 65.6w/cm2, Min=53.2w/cm2 .. .. . . . . . . 1.5 2 2.5 3 Time (second) 3.5 4 4.5 5 5.5 6 #1 at AT=94C Max: 47.8.6w/cm2, Min=57.1w/cm2 '~ \^ ii*i- .,i~' -1 'V .x^^ ^. ii ^1 'i 1 * ^ ^ -" "** i- ^ 1-^ *--* ( " 0.5 1 1.5 2 2.5 3 Time(second) 3.5 4 4.5 5 5.5 6 #1 atAT=104C Max: 39.2w/cm2, Min=47.9w/cm2 . . .y . . . . . .'.. . . . . . . . m . . . . . . . 0.5 1 1.5 2 2.5 3 Time(second) 3.5 4 4.5 5 5.5 6 #1 at AT=114C Max: 33.7w/cm2, M in=41.5w/cm2 . ..""* !"" - -" . . " . **" . . .~^ - i . . i . |L* . . |- . . . 1 . .-- .*"**" ' 0 0.5 1 1.5 2 2.5 3 Time (second) 3.5 4 4.5 5 5.5 6 Figure 4.3--continued Time-resolved heat transfer variation at different heater superheats. 0.5 120 A Departing moment SMinimum Growth 100 Poly. ( Departing moment) Poly. ( Minimum Growth) -~80 ~60- ||40- 20 0' 40 50 60 70 80 90 100 110 120 AT Figure 4.4 The trend of maximum and minimum heat fluxes during one bubble cycle with various heater superheats. growth. While for Regime III, because of the higher heater superheat and the associated higher surface tension, the heater is covered by a layer of vapor all the time, even during the bubble departure. The heater surface is no longer wetted by liquid flow, which results in the bubble growth controlled by conduction through the vapor layer. The conduction- controlled scenario is supported by the presence of a heat flux dip recorded during the bubble departure in Regime III. The reason for this heat flux dip is mainly due to the necking process during the bubble departure. During the necking process, the top part of the bubble exerts an upward pulling force to stretch the neck. After the bubble severs from the base, the upward force disappears suddenly, which allows the lower part of the neck to spread and cover more heater surface area as depicted in figure 4.5. This larger dry area on the heater is responsible for the heat flux dip. Another support for the S Before departure After departure ) k leftover vapor stem Substrate Substrate Figure 4.5 A hypothetical model for bubble departure from a high temperature heater. conduction-controlled bubble growth in Regime III is that the heat flux level decreases slightly with increasing heater surface temperature which is caused by the decrease of FC-72 vapor thermal conductivity with increasing temperature. From figure 4.3, where each spike or dip represents the departure of a bubble, we can conclude that the bubble departure frequency increases with heater temperature. At a heater superheat of 44C, the bubble departure frequency is about 0.43 Hertz, while at 84C, it is at 0.53 Hertz. Figure 4.3 also demonstrates the repeatability of the experiment. We observed that the bubble departure size did not change much with temperature. This is consistent with the results by Yang et al. (2000). Departure criteria of the bubbles are totally determined by the forces acting on the bubbles. Buoyancy force, besides the inertia force of the bubble, is the major player to render the bubble to depart. The interfacial surface tension along the contact line invariably acts to hold the bubble in place on the heater surface. Since this interfacial adhesive tension increases with temperature, for departure to occur, it requires a larger bubble size to generate the corresponding buoyancy force to overcome the adhesive force. At higher temperatures, the thermal boundary layer that is due to transient heat conduction to the liquid is thicker which allows the bubble to grow larger. However, we observed that as the heater temperature was increased the bubble exhibited significant horizontal vibration that could be due to the higher evaporation rate at higher superheat. This vibration promotes bubble departure. On the other hand, due to the micro size of the heater, the natural convection taking place around the outside of the contact line helps bubble detach from the surface. Thus, these effects could cancel so that the bubble departure size does not change much with heater superheat. 4.2.4 Visualization Results and Bubble Growth Rate Figure 4.6 provides a sequence of bubble images during the life cycle of a typical bubble on the microheater at superheat of 54C. The images were taken from beneath the heater surface using a CCD camera of 30 fps. As shown by the images, as it grows, the bubble displays the shape of an annulus. The outer circle is the boundary of the bubble, which shows the diameter of the bubble. The inner circle represents the contact line where solid, liquid and vapor meet which measures the bubble base area on the heater. In this figure, we note that just before the departure, the micro heater is almost entirely covered by the vapor, thus resulting to a very low heat flux (the portion between point D and point E in figure 4.2), and at this point the bubble has a diameter of about 0.8 mm which is three times the size of the micro heater. The corresponding bubble size history and growth rate were estimated and plotted in figures 4.7 and 4.8, respectively. We note that the characteristics of bubble growth dynamics shown in figures 4.7 and 4.8 agree well with the theoretical prediction of Scriven (1959). 0.0333 0.0667 B~~i^M. ft ^"^ ,...s'* 9 .... . 'V ..' nil i< j~f V '^' - 0.1667 0.2000 0.3000 0.3333 0.4333 0.4667 0.5667 0.6000 .. w i 0.7000 0.1000 0.1333 0.2333 0.2667 0.3667 0.4000 0.5000 0.5333 0.6333 0.6667 0.7667 0.8000 Figure 4.6 Bubble images of a typical bubble cycle taken from the bottom. 0.8333 0.8667 0.9667 0.9000 0.9333 1.0000 1.0333 1.1000 1.1333 1.2333 1.1667 1.2000 1.2667 1.3000 1.3667 1.4000 ,4:% 1.5000 1.4333 1.5667 1.5333 1.3333 1.4667 1.6000 Figure 4.6--continued Bubble images of a typical bubble cycle taken from the bottom. 1.0667 66 Bubble diameter vs. Time Heater #1(0.07136mm2) at AT=54C 0 Diameter (mm) reading 2nd order polynomial trendline 0.9 0.8 0.7 -, 0.6 e0.5 S0.4 S0.3 0.2 0.1 0.0 0 0.8 1.0 1.2 Time (seconds) 1.4 1.6 1.8 Figure 4.7 Measured bubble diameters at different time. Growth speed of the single bubble at AT=54C 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Time (seconds) Figure 4.8 Bubble growth rate at different time. .0 0.2 0.4 0.6 The initial growth speed from 0 to 0.0333 seconds is about 11.57mm/s. o Growth Speed - Logarithrric trendlne 0v :: 0 Heater #1(0.07136mm2) at AT=54 0 o o 0 0 0 0--- __ OC A high-speed video camera has also been used to show the necking and bubble departure process. Figure 4.9 displays the images taken from the side and bottom views by the camera set at 1000 frames per second. Based on the side view images it is clear that the embryo of the next bubble is formed as a result of the necking process. It is also estimated that the bubble departs at a very high velocity of 0.16 m/s that causes a strong disturbance to the heater surface. This departure force in turn produces turbulent mixing and microlayer movement around the bubble embryo that are responsible for the spike of heat flux denoted as Point A in figure 4.2. As shown by the bottom view images, the bubble departure disturbance also produces microbubbles that stay near the bubble embryo and eventually coalesce with it. 4.3 Comparison and Discussion 4.3.1 Bubble Departure Diameter Recently Yang et al. (2000) proposed a dimensionless length scale and a corresponding time scale to correlate the bubble departure diameter and growth time. These scales are repeated in Eq. (4.1) and Eq. (4.2). D = _AD, 1 ARl(AYr (4.1) L0 B B)' where = =2pATHA B= Ja12 A B =Ja -a, V 3p, (4.2) and [= 1+1 +' 2(6Jaf 6Ja For the current study, the measured bubble departure diameters and growth times at various heater surface superheats are given in table 4.1. It is shown that the bubble Table 4.1 Single bubble growth time and departure diameter AT 49 54 59 64 69 74 79 84 89 94 99 104 Growth time 2.352 2.054 1.922 1.873 1.747 1.611 1.483 1.403 1.335 1.237 1.128 1.024 (seconds) Departure diameter 0.823 0.824 0.829 0.833 0.832 0.836 0.839 0.843 0.850 0.859 0.866 0.875 (mm) Table 4.2 Properties of FC-72 at 56C Properties Pi (kg/m3) Pv (kg/m3) jHfg (kJ/kg) Cp, (kJ/kg) C4 (10m2/s) Value 1680 11.5 87.92 1.0467 3.244 growth time is inversely proportional to the superheat while the departure size only increases very slightly with the increasing superheat as discussed above. In order to compare our data with the correlation of Yang et al. (2000), the data were converted to dimensionless forms according to Eqs. (4.1) and (4.2) using thermal properties given in table 4.2. Figure 4.10 shows the comparison. Part (a) of figure 4.10 is the reproduced correlation curve from Yang et al. (2000) and part (b) shows our data along with the correlation curve of part (a). It is seen that our data fall slightly lower than the extrapolated correlation of Yang et al. (2000). Since the data given in part (a) are from macro-scale heaters, we may conclude that on a microheater in the current study, the bubble departure size is larger and they stay on the heater longer. 4.3.2 Size Effects on Boiling Curve and Peak Heat Flux It is also of importance to examine the heater size effect on the boiling curve. We plotted boiling curves for heater sizes ranging from 50 gtm, 100 gtm, 270 pgm, 2700 gtm, 1 2ms Figure 4.9 The visualization result of bubble departure-nucleation process for heater #1 at 54C. (a) The side views; (b) The bottom views. 0 ms 3ms (a) (b) Figure 4.10 The relationship between bubble departure diameter and growth time. (a) Experimental data and correlation curve from Yang et al. (2000); (b) Data from current study along with the correlation of Yang et al. (2000). cm to 5 cm in figure 4.11. In order to ensure a meaningful comparison, all the curves in figure 4.11 are based on fluids with similar thermal properties. The trend is very clear that as the heater size decreases, the boiling curve shifts toward higher superheats and higher heat fluxes. The peak heat fluxes for different size heaters have been plotted in figure 4.12 for comparison. It is consistent that the peak heat flux increases sharply as the heat size approaches the micro-scale. Actually, for heater sizes of 50 gtm and 100 gm (Hijikata et al., 1997), even with a very high superheat of about 200C as shown in figure 4.11 the peak heat fluxes have not been reached. 71 Size effect on nucleate boiling curves El FP E3 1000 100 10 I 100 Superheat(AT) Figure 4.11 Comparison of boiling curves. E "'30 p20 CHF for Different Size Heaters 270 urn present study 2700 urn 5 Cc 5Scm Figure 4.12 Comparison of peak heat fluxes. 13h A 50urn (Hijikata et al., 1997) 0 O1um (Hijikata etal., 1997) g 0 Current study +* X2700um (Rule and Kim, 1999) + 1cm (Chang and You, 1997) + +5cm (Rainey and You, 2001) + . I I 4.3.3 Bubble Incipient Temperature Bubble incipient temperature is relatively higher for the small size heater used in this experiment. This is consistent with the results given by Rainey and You (2001), where they observed that the incipient temperature is about 16-17C for 5 cm heater, 20- 35C for 2 cm heater, and 25-40C for 1cm heater. The incipient temperature observed in our experiment is about 45-55C. The explanation for this phenomenon by Rainey and You (2001) is that larger heaters are more likely to have more surface irregularities and therefore wider size range for bubble to nucleate from. 4.3.4 Peak Heat Flux on the Microheater As mentioned in Section 2.3, recently more research reports (Sakashita and Kumada, 1993 and Sadasivan et al., 1995) have indicated their support of the macrolayer dryout as the basic mechanism for CHF as opposed to the hydrodynamic instabilities (Liehard, 1988a). In the current research with microheaters as suggested by Sadasivan et al. (1995), the theory of macrolayer dryout has been verified further as the cause of the CHF. The support based on the results of the current single bubble experiment is summarized as follows: 1. As discussed in Section 4.2.3 Regimes II and III are separated by the CHF point. Two different mechanisms were discovered for the two regimes, respectively. It is clear that the CHF that is the starting point of the Regime III where the heat transfer is through a layer of vapor film and the heater is no longer wettable, is corresponding to the dryout of the macrolayer. 2. In the current microheater condition, there is no possibility for the formation of Kelvin-Helmholtz instability nor Taylor instability. 4.4 Deviations from Steady Single Bubble Formation For the single boiling experiment dedicated in this chapter, as the bubble incipient temperature is reached, a steady bubbling process has been observed and analyzed as above. With the help of a high speed digital camera and simultaneous data acquisition, we have noticed some deviations from the steady single bubble bubbling process. In this section, these deviations have been summarized. It must be noted that these deviations are very random according to observations in the experiments. The physics behind these phenomena requires much more experimental and analytical work. We have noticed there are three possible phenomena associated with bubble formation that could occur. Firstly, a long-waiting period is necessary before the onset of another bubble and a vapor explosion always occurs before this onset of another bubble, we call this process discontinued bubble formation. Discontinued bubble formation happens usually at lower superheats and with a highly subcooled liquid. Secondly, the direct bubble formation is the process during which another bubble forms immediately after a bubble departs. Thirdly, the bubble jetting is a chaotic phenomenon with smaller bubbles ejected from the heater surface without forming a single bubble. The direct bubble formation is the steady bubble formation process that we have analyzed previously. This section is dedicated to the other two deviations from the steady bubble formation process. 4.4.1 Discontinued Bubble Formation As we have observed, discontinued bubble formation actually occurs similarly to the initial onset of boiling on the heater. For the microheater used in out experiment, onset of nucleate boiling starts actually with vapor explosion, a physical event in which the volume of vapor phase expands at the maximum rate in a volatile liquid. During this process, the liquid vaporizes at high pressures and expands, performing mechanical work on its surroundings and emitting acoustic pressure waves. According to previous investigators, for large heaters rapid introduction of energy is necessary to initiate and sustain the vapor volume growth at the high rate. For the microheater used in this experiment, vapor explosion always occurs prior to the nucleation of small bubbles on the heater. This could be due to the limited energy the microheater can supply from the solid surface into the liquid though the superheat is well beyond the required superheat for nucleating bubbles. We recorded the vapor explosion process together with the onset of boiling, as shown in figure 4.13 for side-view images and figure 4.15 for bottom-view images. When the heater reaches about 110C, vapor explosion occurs. Following this process, small bubbles start to nucleate and coalesce almost right after. With the existence of noise due to the acoustic pressure waves, a layer of vapor is ejected with high speed up into the liquid exhibiting a shape of mushroom (about 1.0ms). The vapor enclosed in the mushroom is loose due to the rapid expansion, and then starts to shrink to form a round bubble and moves up. On top of the heater, the cool liquid replenishes the vacancy and small bubbles nucleate and coalesce to a single bubble. The high energy stored before the ONB is released suddenly to overcome all kinds of forces including the Van der Waals force from the solid surface, hydrostatic force due to the liquid depth, etc. A typical application of this process is the commercial success of thermal ink jet printers. The key to the thermal ink-jet technology is the action of exploding micro bubbles, which propel tiny ink droplets through the openings of an ink cartridge. Zhao et al. (1999) used a thin-film microheater of size of 0l0m x llOpm to investigate the vapor explosion phenomenon. They placed the microheater underside of a layer of water and the surface temperature of the heater was rapidly raised (about 6gm) Figure 4.13 The process of vapor explosion together with onset of boiling. Figure 4.14 The ruler measuring the distance in figure 4.13 after vapor explosion. o 0 0 8 07 J hLmi !- wi - MUM ** ii -1*R ITV Wpm M~iiSw ^6L l jL41 Figure 4.15 The bottom images for vapor explosion and boiling onset process. I . electronically well above the boiling point of water. By measuring the acoustic emission from an expanding volume, the dynamic growth of the vapor microlayer is reconstructed where a linear expansion velocity up to 17m/s was reached. Using the Rayleigh-Plesset equation, an absolute pressure inside the vapor volume of 7 bars was calculated from the data of the acoustic pressure measurement. In this experiment, we also measured the bubble velocity when vapor was exploded. By referring the heater size shown in figure 4.14 the distance the bubble was ejected up from the heater surface can be measured. The result is the average velocity during the first 7 milliseconds is about 0.2m/s which is well below that measured by Zhao et al. (1999). The low velocity measured in this experiment is not surprising due to the following reasons. One of the reasons is that only a layer of water was used in their experiment while pool boiling experiment was used in this experiment. Also in this experiment the boiling liquid FC-72 with a lower saturation temperature is used. Figure 4.16 illustrates the heat flux traces corresponding to the vapor explosion and boiling onset process. In figure 4.16(a) during the 6-second data acquisition, it starts at a lower level where boiling is not present with natural convection to be the heat transfer mode. All of a sudden, the vapor explosion exists and we recorded a sharp heat flux spike that is supposed to correspond to the vapor explosion process as shown in figure 4.16(b). For present data recording speed about 4,500 data points per second, we recorded the vapor explosion occurs only in a mini-second related to the heat flux variation from the heater. Following this vapor explosion, small bubbles start to nucleate and heat flux increases. 78 PrllM1 I SO ftM I no The Heat Flux History Recording Onset of Boiling Process 100 SVapor explosion and onset of boiling S00 so Natural convection nglebubble formed 40 30 1I 1 0 1 2 3 4 8 1 Seconds (a) F-061I 20p*m I The Heat Flux History During Vapor Explosion and Onset of Boiling 4. 1 ^ I I onst p () Te ht fx vOnset o1 bonlln o pces 7 b Th hea lxvaporexpiIatio d th -e b o pr 40- 30- '2.01 15 2.02 2.02. 203 2.036 2.04 S~oomwo (b) Figure 4.16 The heat flux variation corresponding to the vapor-explosion and boiling onset process. (a) The heat flux variation recording the vapor-explosion and boiling onset process; (b) The heat flux variation during the vapor-explosion and boiling onset process. 4.4.2 Bubble Jetting During the single bubble boiling process, we observed that periodically there is bubble jetting after two or three bubbling cycles. This bubble jetting lasts about 0.2 second each time before a new single bubble is formed on the heater. Images taken with the high speed camera in figure 4.17 and figure 4.18 show the bubble jetting process, and figure 4.19(a) is the heat flux traces for a six-second data acquisition which includes three bubble cycles of the departing-nucleation process. Figure 4.19(b) and 4.19(c) are the close-up views of two heat flux encircled in figure 4.19(a). Through a close observation, we find figure 4.19(b) is the heat flux variation corresponding to this bubble-jetting process shown in figure 4.17 and figure 4.18. We notice that during the bubble-jetting process, the heat flux stays much higher and drops sharply as a single bubble grows on the heater. We also observed the single bubble boiling at 105C and 120C. At these two temperatures, we also noticed the same phenomenon of bubble jetting. Figure 4.20 shows the heat flux variation from the heater when it is set at 105C, and figure 4.21 shows the heat flux variation from the heater when it is set at 120C. For both these two cases at 105C and 120C, we also observed the bubble-jetting after bubble's departure. The bubble jetting process is a characteristic of chaotic bubble nucleation, merging, and departing process with the company of acoustic waves. During the bubble jetting process, small bubbles on the heater keeps merging and departing. And all of a sudden a merged bubble stays and merges with all other small bubbles around it, then a single bubble is formed. The single bubble formation and growth also inhibits the bubble-jetting process, and at the same time dries out the microlayer around the bubble for this particular heater configuration used in this experiment. Thus the heat flux drops quickly as the single bubble grows. The possible reason for this bubble jetting phenomenon is due to the high 80 superheat of the heater, the evaporation rate is very high on top of and around the edge of the heater. The small bubbles do not have enough liquid to replenish the microlayer region for them to grow. Clearer explanation of these phenomena needs further observation and numerical approach. 1~ei -011O 1.. 1LA 69.00. -421 jPOi *3W -&* . 4 L-bI ..4. I ,a' *I a ,,a "m .* qii .6aL--"i a a, v _* -,- - * I "s, 7" -41 to-J ... -.> -J- -jif- '* .jafr** .^.< An m d W d P . Z :A :: < ..,WO wim Md* 6*" - 0-% i Figure 4.17 The chaotic bubble jetting process for heater #1 at 110C. dadgmw-mlmws AMW - W -.. 0.% -OOMMM -- a. Ak % --1- 04 4k m 56 ms 58ms 59ms 60ms 61 ms 61.5m 62 62. 5ms 63m 63.5ms 64.5 ms 67ms 70 5 ms 77 Is .At h. 8055 E I 06fma 1265 ms 128 ms 164.5ms 165ms 165.5ms 166.5ms 0* l _4__ of 167.5ms 168ms 169ms 170ms 171.5mg O- 0 of0 of 0 9 #9 .9 :9 o 173ms 173.5m 174ms 175ms 176mS Figure 4.18 The bottom images of chaotic bubble jetting process for heater #1 at 1 10C. Heat Flux - i / 1 ,t, Variation during Nucleation-Departing Process .. -- I . -,-- \ .- ... . Buotle jetting process about 0 12 seconds i Bub lede rting si ng bubble is omed 0 0.01 0.02 0.03 0.04 0.05 0.06 0 .07 0.08 0.09 0.1 C Seconds 1.11 ....0.12 0.13 0.14 0.15 0.16 0.17 0.1 0.1... 0....2 .11 0.'12 0.'13 0.'14 0.15 0.16 0.17 0.18 0.'19 0.2 FmO041 IMN r----2- 13 The heat flux variation with invisible bubble jetting 130 120 -- -- -- --- -- -- -- -- -- -- -- -- -- __ __ __ __ __ __ 110 -- S100 _ 9.0 X 70 .._ 40 3.1 3.2 3.3 3.4 Seconds (c) Figure 4.19 The heat flux traces corresponding to the bubble jetting process. (a) 6-second data acquisition; (b) Close-up view of (a) for visible bubble jetting; (c) Close-up view of (a) for direct bubble formation. F t- 1 I FPOMI I Typical heat flux variation #1 at 110 IOC 130. __I__ I I I I I I I l __ I I_ I J I I I ,,o .- 1 1 ... .IhJ! p o : n m .b u. J r = ..... .... Visible e letting Invisible bubble jetIng after depalture __ 110 .. .. . X 70 .... .. .. ... o . . .. ^ .. ^ ... . . . .. ... . : 0I t -i i i i, i iT _-' l - r - 0 1 2 3 4 5 6 Seconds I- 1 7 ft202 1 120 110 100 190 180 1 70 60 I I l3 Typical heat flux variation #1 at 105C 12D L H t I I I i t- 1 1 1 1 1 1 1 1 --i i i j_ _-..111. i i Invisible bubbW eting VIb bbble n X~ 70 40 P I I iI I I j. -' F ... i , .: .... i -+- ^ + IL __ .i, Seconds f- ," I O... I, , Typical heat flux variation #1 at 105C VI i5blebubl I el9ng M A .... ... ... ... W, 5.8 Second* Figure 4.20 The heat flux traces for bubble jetting process from heater #1 at 105C. (a) 6- second data acquisition; (b) Close-up view of (a) for visible bubble jetting; (c) Close-up view of (a) for direct bubble formation. 130 120 110 if100 9D0 )so 1 70) 60 SO 50 405. 5.9 130 120 110 100 90 1 90 170 !o 402. 50 130 120 110 100 190 19.0 X 70 S0 50 40 0 0.9 2.9 Seconds Scnd Secnd Figure 4.21 The heat flux traces for bubble jetting process from heater #1 at 120C. (a) 6- second data acquisition; (b) Close-up view of (a) for visible bubble jetting; (c) Close-up view of (a) for direct bubble formation. Typical heat flux variation #1 at 120C 1:! i 1 L1 L I I I I I I 1 Invilble bubble jetting Visible bubble Jetting 110.. z ... ........... .......- if 0 T I _ 0 1 2 3 4 5 4 Seconds Typical heat flux variation #1 at 120'C Visble bubble'jet__ .A ..... I^--"--- i , . -. . . .. . . .. . I " I I i .... -1- --i ... T i Typical heat flux variation #1 at 120C __I 1 1 1 t I .1 1_ l L i!.. Invisible bubble jelling I 1 I _ . 7 |