Enhanced Flow Boiling Heat Transfer in Microchannels with Structured Surfaces at Varied Mass Flow Rates by David Bian SUBMITTED TO THE DEPARTMENT OF MECHANICAL ENGINEERING IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF BACHELOR OF SCIENCE IN MECHANICAL ENGINEERING ARCHWEB AT THE MASSACHUSETTS INS'TIT(UTE OF TECHNOLOLGY MASSACHUSETTS INSTITUTE OF TECHNOLOGY JUN 2 4 2015 June 2015 LIBRARIES @ 2015 David Bian. All rights reserved. The author hereby grants MIT permission to reproduce and to distribute publicly paper and electronic copies of this thesis document in whole or in part in any medium now known or hereafter created. Signature of Author:_ Signature redacted ____ Department of Mechanical Engineering May 22, 2015 Certified by: Signature redacted 61 10 Accepted by: V velyn N. Wang Professor of Mechanical Engineering Thesis Supervisor Signature redacted Anette Hosoi Professor of Mechanical Engineering Undergraduate Officer ......... . ..... ..... Enhanced Flow Boiling Heat Transfer in Microchannels with Structured Surfaces at Varied Mass Flow Rates by David Bian Submitted to the Department of Mechanical Engineering on May 22, 2015 in partial fulfillment of the requirements for the degree of Bachelor of Science in Mechanical Engineering Abstract This thesis investigates the role of mass flux on flow boiling heat transfer in microchannels with surface micropillar arrays. The motivation for this investigation was to determine the general trends of the optimal micropillar array geometry in terms of its heat transfer capabilities. The experiment was conducted with three microchannels: a flat surface microchannel, a sample called the 5-15 (height h = 25 pm, diameter d = 5 pm, and pitch 1 = 15 jm) and a sample called the 10-40 (height h = 25 jm, diameter d = 10 jm, and pitch 1 = 40 jm). The structured surface microchannels, due to their capillary pressure-induced wicking capabilities, exhibited less temperature rise and pressure drop fluctuations at high heat fluxes. Furthermore, it was verified that the critical heat flux value of all microchannels increased with mass flux. In addition, it was concluded that at lower mass fluxes, the relative percentage heat transfer enhancement of the structured surface microchannels over the flat surface microchannel was greater. The trend observed suggests that denser samples are better at lower mass fluxes. However, if a sample is too dense, there may be too much viscous drag. Thus, an optimal balance between capillary force and viscous drag must be found in order to determine the optimum micropillar array geometry and density for maximizing the critical heat flux value. Finally, for a given mass flux, the pressure drop across every microchannel was approximately equal at all heat fluxes. This implies that no additional power consumption is required to pump a particular mass flux through a structured surface microchannel than a flat surface microchannel, though there is certainly additional power required to increase the mass flux. This work provides insights into the roles of both the micropillar array surface structures and the mass fluxes on the heat transfer performance of flow boiling in microchannels. The results and observations of this experiment may prove helpful in guiding future work in an attempt to optimize microchannels for heat transfer applications in electronics. Thesis Supervisor: Evelyn N. Wang Title: Professor of Mechanical Engineering Acknowledgements I would like to extend my gratitude to Professor Evelyn Wang, who graciously allowed me into the Device Research Laboratory to pursue a senior thesis project, and for the general friendliness and understanding of the rest of the laboratory members. My most enthusiastic thank you of all to Yangying Zhu and Dr. Dion Savio Antao for providing me with such a wonderful thesis topic, their patience and support throughout the project, their constant willingness to guide me through the physical concepts and experimental procedure, their belief in my ability to conduct experiments, analyze data, and draw conclusions on my own, and finally, for their forgiveness for breaking their precious samples. Contents 1 2 Introduction 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 10 1.2 1.3 11 12 Experiment Motivation and Summary . . . . . . . . . . . . . . . . . . . . . . Structure of Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Details 2.1 Microchannel Samples 2.1.1 2.1.2 2.1.3 2.2 2.3 2.4 2.5 3 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 13 Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . RTD Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 16 18 Experimental Setup . . . . . . . . . Experimental Preparation . . . . . Experimental Procedure . . . . . . Measurements and Data Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 22 24 24 Results and Discussion 26 3.1 Flow Instability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.2 3.3 Heat Transfer Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pressure Drop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 34 Conclusion 37 5.1 5.2 Works Cited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 39 41 5.2.1 MATLAB Script for Processing Experimental Data . . . . . . . . . . 41 5.2.2 LabVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 5 6 List of Figures 2.1 2.2 Schematic of a generic operating microchannel with micropillar structures on the bottom surface, an inlet, an outlet, and a heater indicated by heat generation are shown. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graph of how capillarity varies the with structural pillar geometry. Adapted from Xiao et al. [13] 2.3 2.4 2.5 2.6 3.1 3.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fabrication process of the microchannel. (a) Micropillars of 25 im height were etched in Si using Deep Reactive Ion Etching (DRIE). (b) A Si wafer was etched through using DRIE to define the channel. (c) The two wafers from the first two steps were bonded together using Si-Si fusion bonding. (d) Inlet and outlet ports were laser-drilled on a Pyrex glass wafer. (e) The Si layers were bonded using direct Si-Si bonding. A silicon dioxide (Si02) layer was thermally grown on the Si surface. The Pyrex layer was bonded to the top Si using anodic bonding. (f) A Pt layer was deposited on the backside of the microchannel using E-beam evaporation and patterned to form heater and temperature sensors. From Zhu et al. [9] . . . . . . . . . . . . . . . . . . Images of a representative fabricated microchannel with micropillar arrays. Optical images of the (a) front and (b) back side of a sample. (c) A crosssectional image as well as scanning electron micrograph of the micropillars on the channel bottom surface. From Zhu et al. [9] . . . . . . . . . . . . . . . . Measured data and the linear fit of the resistance of the RTDs with temperature to determine calibration constants oz and 3 for the linear fit equation 14 15 17 18 T =aR+..................................... 20 Schematic of the Experimental Setup. "P", "T", and "M" indicate pressure transducers, thermocouples, and a flow meter, respectively. The blue lines indicate fluid flow, red lines indicate measurements, green lines indicate data transfer and recording, and gray lines indicate physical connections. . . . . . 22 Temporal data comparison of temperature and pressure drop for each microchannel sample with a 150 kg/m 2 s mass flux at a range of representative heat fluxes. The uncertainties of the temperature and pressure drop measurements were approximately 6.5'C and 0.45 kPa, respectively. . . . . . . . Temporal data of temperature and pressure drop for each microchannel sample with a 300 kg/m 2 s mass flux at a range of representative heat fluxes. The uncertainties of the temperature and pressure drop measurements were approximately 3.5'C and 0.22 kPa, respectively. . . . . . . . . . . . . . . 7 27 28 ....... ... ... .. ... .. ..46 3.3 3.4 3.5 3.6 5.1 Temporal data of temperature and pressure drop for each microchannel sample with a 500 kg/m 2 s mass flux at a range of representative heat fluxes. The uncertainties of the temperature and pressure drop measurements were approximately 0.3'C and 0.28 kPa, respectively. . . . . . . . . . . . . . . Boiling curves for all the flow rates for each microchannel. The effects of varying mass flux can be clearly observed for each microchannel. The initial slope indicates heating single phase flow, whereas the sudden jump back indicates the onset of nucleate boiling. The following upwards slope is heating of the two phase flow, and finally, the critical heat flux is indicated by when the slope changes significantly from the two phase flow slope. Note: the 10-40 sample broke during the experiment at 500 kg/m 2 s, and since there was no slope change observed, the critical heat flux was not reached and is therefore unknown. The error bars were approximately 1% for the heat flux q" and ranged from 2.0'C to 6.7'C for the temperature rise. . . . . . . . . . . . Boiling curves for all the microchannels at each mass flux. Differences between the microchannels can be clearly observed at each mass flux. The initial slope indicates heating single phase flow, whereas the sudden jump back indicates the onset of nucleate boiling. The following upwards slope is heating of the two phase flow, and finally, the critical heat flux is indicated by when the slope changes significantly from the two phase flow slope. Note: the 10-40 sample broke during the experiment at 500 kg/m2 s, and since there was no slope change observed, the critical heat flux was not reached and is therefore unknown. The error bars were approximately 1% for the heat flux q" and ranged from 2.0'C to t6.7'C for the temperature rise. . . . . . . . . . . . Boiling curves for each flow rate as well as each microchannel. Differences between the microchannels can be clearly observed at each mass flux. The error bars were approximately +1% for the heat flux q" and ranged from 0.36 kPa to 0.60 kPa for the pressure drop. . . . . . . . . . . . . . . . . . Screenshot of the front panel of the LabVIEW file used to monitor the entire . .. . . . . . . . . .. .. . system loop. 8 29 31 32 35 List of Tables 2.1 2.2 2.3 3.1 Geometric parameters of tested structured surface microchannels . . . . . . . At a given temperature, the measured temperature and resistance of each RTD was recorded and averaged in order to calibrate the flat surface microchannel sample. The same process was used to calibrate the other microchannels as 14 well. 19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Calibration constants of the microchannel samples that linearly relate the resistance of the RTDs to their temperatures, where ais the slope with units 'C/Qand Pis the intercept with units 'C. (a) Flat Sample. (b) 5-15 Sample. (c) 10-40 Sam ple. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Critical heat flux values of each microchannel at each flow rate as well as their critical heat flux enhancement percentage relative to the flat surface microchannel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 9 Chapter 1 Introduction Background 1.1 Moore's Law projects that the density of transistors in an integrated circuit will double every two years. The limits of heat dissipation have recently been unable to match the microprocessor power trends enabled by transistor density, partially due to the lack of affordable, high performance thermal management options [1]. As a result, multi-core processors have been developed to produce the required power because conventional thermal solutions consisting of a fin set and fan combination are unable to dissipate heat fluxes approaching 100 W/cm2 [2]. Furthermore, power electronic devices, concentrated photovoltaics, and laser diodes require much more power and generate heat fluxes on the order of and greater than 1000 W/cm 2 [1]. The high density power electronic devices of today have increased the demand for high density heat transfer technology. Two-phase thermal management systems such as heat pipes and vapor chambers first emerged because they utilized the latent heat of vaporization of fluids to successfully transport and dissipate heat fluxes up to 100 W/cm 2 [3]. As the heat fluxes increase with the density of transistors, the heat dissipation capabilities of devices must improve. Conventional thermal solutions such as a fin set and fan combination utilize conduction and forced convection. Other alternative and better methods utilize two-phase systems, which include pool boiling, spray cooling, and flow boiling using refrigerants or Pool boiling with water has demonstrated critical heat flux (CHF) values of 100 water. W/cm 2 on smooth surfaces [4] and 250 W/cm 2 with the micro and nanoscale structured surfaces [5]. The method with the most potential is known as jet impingement, which has a predicted CHF of up to 100,000 W/cm 2 [6]. However, this method has been known to cause major instabilities and large variations in temperature, which is unacceptable for electronics 10 cooling applications. On the other hand, flow boiling with water has the potential to achieve CHF values up to 1000 W/cm 2 [7,8]. In fact, Zhu et al. opted to explore the solution of using water in flow boiling in two-phase microchannel heat sinks due to their extremely compact form factor [9]. The design and use of two-phase microchannels present two main challenges. This includes maximizing the critical heat flux as well as minimizing the flow instabilities. Vapor bubbles form and expand along the channel because the cross-sectional length scale is smaller than the capillary length of water. These bubbles cause a large pressure build up until the pressure gradient forces the bubble away. However, as the vapor bubble expands, the channel begins to dry, a process known as dryout. Once dry, the heat dissipation performance of the microchannel plummets and the surface temperature of the channel spikes. Previous studies have shown that microstructured surfaces have been able to delay dryout through capillary wicking. For example, Li et al. [10] and Yang et al. [11,12] both used silicon nanowire-coated channel surfaces to improve heat transfer performance. The critical heat flux enhancement is hypothesized to be the result of the wicking capabilities of the hydrophilic nanowires which assist in preventing dryout. Additionally, the structures and fluid flow may serve a similar purpose to a micro-scale fin set and fan combination, with the structures spreading the heat flux across a larger surface area to be taken away by the fluid flow in the channel. 1.2 Experiment Motivation and Summary Microchannels with structured surfaces have been shown to enhance flow boiling heat transfer capabilities. However, the exact effects of the structured surface geometry on the flow instability and heat transfer performance are not well-studied. Previous research, such as that done by Zhu et al. [9], has examined structured surfaces in the form of micropillar arrays. Their work has found a near-optimal geometry for the micropillar arrays at a particular mass flux rate, 300 kg/m 2 s through the microchannel. While Zhu et al. examined how heat transfer performance varied with geometry at a constant mass flow rate [9], this thesis examines the additional dimension of varying the mass flow rate. The experiment considers the performance of three particular microchannels, one flat surface and two structured surfaces at different mass flow rates, 150 kg/m 2 s and 500 kg/m 2 s. The author emphasizes that the microchannels were provided by Zhu et al. [9] for the purpose of conducting these experiments. The goal of doing so is to observe how the performance of each microchannel varies with 11 mass flow rate, as well as the performance of each microchannel relative to one another. This information may reveal a relationship between optimum heat transfer performance, geometry, and mass flow rate in the microchannels that will be crucial to the future of characterizing of flowing boiling heat transfer at the micro-scale. 1.3 Structure of Thesis This thesis is structured as follows: Chapter 1 provides an overarching background and structure of this thesis. Chapter 2 describes all experimental details regarding the microchannel samples with structured surfaces, the experimental setup and preparation, and the experimental procedure. Chapter 3 discusses the qualitative and quantitative results to compare the heat transfer performance and flow stability for each microchannel sample at each mass flux. Chapter 4 summarizes the results obtained from the experiments, including how the mass flux affects flow stability, heat transfer characteristics, and pressure drop across the microchannel. It concludes by proposing potential directions for future research. Chapter 5 is a bibliography of all works referenced in the writing of this thesis followed by an appendix of MATLAB scripts and a LabVIEW diagram utilized in the experimental setup and data analysis. 12 Chapter 2 Experimental Details 2.1 2.1.1 Microchannel Samples Design The microchannels used in this experiment came from Zhu et al.'s experiments [9] and are 10 mm long with cross-sectional dimensions of 500 im x 500 4m. On the bottom surface of each structured surface microchannel is an array of micropillars. In addition, there is a thin-film metal heater 9 mm wide x 350 4m long integrated directly below the microchannel that provides Joule heating. There are also four resistance temperature detectors (RTDs) along the length of the microchannel; RTD1 and RTD2 are positioned at the inlet of the channel, RTD3 is positioned directly at the midpoint of the microchannel as well as the heater, and RTD4 is positioned at the outlet of the channel. Figure 2.1 depicts a fully labeled operating microchannel from a cross-sectional side view. 13 outlet inlet heat generation bottom surface, Figure 2.1: Schematic of a generic operating microchannel with micropillar structures on the shown. are an inlet, an outlet, and a heater indicated by heat generation The three specific microchannels utilized for this thesis were: a flat sample (smooth surface, = 5 no pillar array), a sample henceforth called the 5-15 (height h = 25 ym, diameter d im, yim, and pitch 1 = 15 ym) and a sample henceforth called the 10-40 (height h = 25 2.1. Table in viewed diameter d = 10 ym, and pitch 1 = 40 ym), whose geometries can be These cover the relative cases of smooth surface, dense pillar array, and sparse pillar array microchannels, respectively. 25 5 15 5 0.33 Table 2.1: Geometric parameters of tested structured surface microchannels The flat surface microchannel was used as a control, but the length scales for the microscale pillars of 5-25 pm were originally chosen for Zhu et al.'s experiments for the following reasons: (1) Fabricating structures of those sizes on silicon are standard, well-controlled etching processes that present no complications. (2) Capillary pressure that can be generated the at this length scale with moderate mass fluxes is on the same order of magnitude as wicking. pressure drop across the microchannel, and can therefore effectively promote liquid as liquid (3) The pillars' structural integrity are robust and will not deform appreciably the evaporates. The specific selection of the geometric parameters was based on balancing 14 maximization of capillary driving pressure and minimizing viscous drag resistance to flow. A model created by Xiao et al., depicted in Figure 2.2, shows how capillarity of the pillar array varies with its geometric parameters, specifically the spacing 1. 0.1 A large viscous drag 0.12 4A 0 0 0.1 - - .. 0.08 -i/d=4 Wd=6 40 0.06 -o -+Wd=10 / -d 0~ -*Wd=infinity 0.04 Cr 0.2Q p in dequate apillarity I ___I .1 0.15 0.2 0.25 i 0.35 0.3 d/1 Ratio I I I I 0.4 0.45 0.5 0.55 Figure 2.2: Graph of how capillarity varies the with structural pillar geometry. Adapted from Xiao et al. [13] As the labels indicate, a low d/1 ratio translates to a very low capillary pressure induced by the pillars. On the other hand, too large a d/1 ratio induces large viscous drag forces, thereby inhibiting the flow. Therefore, the structured surface microchannels fabricated primarily fell into the middle range: d/l = 0.25-0.5. Flow boiling using water in microchannel heat sinks generally operate in the annular flow regime not only because of the capillary length of water, which is greater than 2.5 mm, but also because of the high vapor quality of the fluid at high heat fluxes. The uniform micropillar arrays at the bottom surfaces of the microchannels are designed for the purpose of enhancing and sustaining stable liquid film evaporation through the mechanism of capillary flow. This was done because only the bottom surface is heated and therefore the most likely 15 to experience dryout. While the bottom surface is structured, the side walls of a microchannel have a roughness of 1-20 im in order to facilitate nucleation. By avoiding nucleation from the heated bottom surface, the possibility of dryout on the structured surface is mitigated. This decoupling of film evaporation from the heated surface and the nucleation from the side walls enables the microchannels to dissipate high heat fluxes while maintaining stable heat transfer performance. 2.1.2 Fabrication Standard silicon microelectromechanical systems (MEMS) fabrication processes were used to create the microchannel samples and are summarized in Figure 2.3. The pillar arrays, all with heights of 25 tm, were etched out of a 500 rm thick silicon wafer using deep reactive ion etching (DRIE) to the channel's bottom surface (Fig. 2.3a). Another silicon wafer of the same thickness was etched through using DRIE to create the channel's side walls (Fig. 2.3b). The two wafers from the first two steps were bonded together using Si-Si fusion bonding (Fig. 2.3c). The inlet and outlet ports were laser drilled through a Pyrex wafer (Fig. 2.3d). Si-Si bonding was then used to bond the two silicon wafers together. A 1 tm hydrophilic, silicon dioxide coating was thermally grown on the channel walls, as well as on the back side as an electrical insulation layer. The Pyrex wafer was then bonded to the silicon wafers using anodic bonding to enclose the microchannel and facilitate observation of the flow (Fig. 2.3e). Finally, a ~170 nm thick layer of platinum was deposited on the back side of the channel with E-beam evaporation and patterned by the lift-off technique to serve as the heater and RTDs (Fig. 2.3f). The fabrication process was the same for the flat samples with the exception that step (a) requires no etching. 16 (ab1(e) (b) (c) ff (d) Figure 2.3: Fabrication process of the microchannel. (a) Micropillars of 25 pm height were etched in Si using Deep Reactive Ion Etching (DRIE). (b) A Si wafer was etched through using DRIE to define the channel. (c) The two wafers from the first two steps were bonded together using Si-Si fusion bonding. (d) Inlet and outlet ports were laser-drilled on a Pyrex glass wafer. (e) The Si layers were bonded using direct Si-Si bonding. A silicon dioxide (Si02) layer was thermally grown on the Si surface. The Pyrex layer was bonded to the top Si using anodic bonding. (f) A Pt layer was deposited on the backside of the microchannel using E-beam evaporation and patterned to form heater and temperature sensors. From Zhu et al. [9] Figure 2.4 displays a comprehensive set of images of the microchannel. Fig. 2.4a and 2.4b show the full front and back sides of a fabricated microchannel sample. The trapezoidal air chambers next to the center microchannel in Fig. 2.4a were included for thermal insulation to reduce heat loss through the sides of the microchannel. A cross-sectional image as well as scanning electron micrograph of the micropillars on the channel bottom surface are shown in Fig. 2.4c. 17 0 3 30mm RTDs Pt heater (C) Thermal insulation chamber Figure 2.4: Images of a representative fabricated microchannel with micropillar arrays. Optical images of the (a) front and (b) back side of a sample. (c) A cross-sectional image as well as scanning electron micrograph of the micropillars on the channel bottom surface. From Zhu et al. [9] 2.1.3 RTD Calibration The temperature of the microchannels are obtained by measuring the resistance of each RTD, which is a function of the temperature. The relationship between the temperature and the resistance of each RTD acquired by calibration, which is important because they are not all perfectly identical, despite having the same fabrication process. The microchannel sample was secured in a fixture created for the samples. The purpose of the fixture is to provide electrical contact to the RTDs on the sample and is connected to an external electrical circuit, from which the RTD voltages were measured and their resistances calculated. The entire fixture and sample were then placed in a controlled thermal chamber along with a separate thermocouple to measure the internal temperature of the chamber. The temperature of the 0 0 thermal chamber was set to 25'C, 50 C, 75 C, and 100'C. Two 30-second sets of raw voltage data from the thermocouple and the four RTDs on the sample were recorded and afterwards converted into temperature and four resistance values, respectively. These values were then averaged, and a sample of the averaged data for the flat sample calibration is shown below in Table 2.2. 18 102.27 310.10 302.69 304.35 300.25 Table 2.2: At a given temperature, the measured temperature and resistance of each RTD was recorded and averaged in order to calibrate the flat surface microchannel sample. The same process was used to calibrate the other microchannels as well. The data was then plotted, and the trend appeared linear, which was expected because the RTDs are made of platinum, whose resistance varies linearly with temperature at relatively low temperatures. The plot is shown in Figure 2.5. 19 120 Expe i--ntal Data RR12 Expe ri nental Data A Expe ri Tental Data R3 x Experi Tiental Data R41 Linear (Experimental lData RI) - Linear (Experimental 'Oata R2) Linear (Experimental D~ata R3) Linear (Experimental Oata R4) 100 0~ + * * y = 1.7224x 419.13 y = 1.7 )73x - 417.4 y 15867x-374. C y 1.6766x - 417 71 801 Q = 60 a. EGD I- 40 20 1 4 0 250 260 270 280 290 300 310 320 Resistance (D) Figure 2.5: Measured data and the linear fit of the resistance of the RTDs with temperature to determine calibration constants a and 0 for the linear fit equation T = aR + 0 Therefore, the governing equation decided as the best fit for the data was a line equation depicted by Equation 2.1. T = aR+ (2.1) As Equation 2.1 implies, the slope of each best fit line is the a value for the RTD, while the temperature, or y-axis, intercept, is the 0 value. The list of the values for each microchannel sample and their RTDs are shown in Table 2.3 20 (a) (b) (C) - (Q) R2 (0) R3 (0) R4 (0) 31.93 -351.95 -349.15 -352.93 R 1 (0) R2 (Q) R3 (0) R4 (n) -361.98 -363.79 R 10-40 0 33.8-363.18 Table 2.3: Calibration constants of the microchannel samples that linearly relate the resistance of the RTDs to their temperatures, where otis the slope with units 'C/Qand Pis the intercept with units 'C. (a) Flat Sample. (b) 5-15 Sample. (c) 10-40 Sample. 2.2 Experimental Setup The flow boiling test setup is a closed loop system, a schematic of which is shown in Figure 2.6. The water reservoir was heated, maintaining degassed water at saturation temperature and pressure. The saturated water leaves the reservoir and flows through tubing that is fed into a peristaltic pump, continues through a flow meter, and through a heated segment (called the pre-heater) of the loop which heats the water closer to saturation temperature. The water then passes a needle valve and enters another heated portion (called the postheater) of the loop which heats the water to near boiling -90-95'C. The flow then enters the microchannel, which is being heated through its built-in resistive element. The channel can be observed through an inverted microscope as well as an attached high speed camera, which can capture flow boiling videos. The tubing at the outlet of the microchannel leads back into the reservoir to complete the closed loop. 21 DAQ Card T P Power Supply P T Test Fixture with Microchannel Sample Post-Heater Needle High Speed Inverted Valve Camera Microscope thDgsd with DegassedT ReservoirT Liquid Peristaltic Pulve Pre-Heater M transducers, thermoFigure 2.6: Schematic of the Experimental Setup. "P", "T", and "M" indicate pressure measurements, indicate lines red flow, fluid indicate lines blue The couples, and a flow meter, respectively. connections. physical indicate lines gray and recording, and transfer data indicate green lines transAdditionally, sensors were placed throughout the loop: 7 thermocouples, 3 pressure on ducers, and a flow meter. As mentioned in Chapter 2, there are the four thermocouples the fixture of the microchannel sample. Additionally, there is one thermocouple and pressure transducer each in the reservoir, at the inlet of the microchannel, and at the outlet. The single flow meter lies just downstream of the peristaltic pump. The sensors are all connected to a data acquisition (DAQ) card. The sensors and heaters also each have their own variable power supply. 2.3 Experimental Preparation The reservoir was made ~90% full of deionized water by adding deionized water through cona valve using a syringe. Deionized water was used to minimize particulates and other before taminants that could affect the results of the experiment. The valve had to be closed detaching the syringe, in that order, to prevent outside contaminants from being sucked into the reservoir, as the reservoir pressure is generally less than atmospheric pressure at room all temperature. To prevent unintentional contamination or flow, the reservoir valves should above be closed before continuing. The water level was maintained throughout experiments 80% capacity. were Using the LabVIEW program "flow boiling DP test," the reservoir pressure readings 22 monitored to ensure safety. The reservoir's PID controller was turned on and set to 102'C, and its heater was turned on to 45 V in order to ensure adequate power to heat the reservoir. While the reservoir was heating up, the sample was cleaned, again to remove any contamination amassed through previous experiments or obtained during storage between experiments. Clean tweezers were used to hold the edge of the sample, and then, while ensuring the back side did not get wet, acetone, isopropyl alcohol, and water were flowed through the inlet to the outlet of the channel in that order for about a minute each to dissolve any possible organic contaminants. Then the air gun was used to blow dry the sample for a minute, making sure to aim at the inlet so that all the liquids were flushed from channel through the outlet. To further clean the sample, it was laid face up inside a plasma cleaner so the inlet and outlet ports would be exposed to the plasma gas. The plasma cleaner valve was then closed, the pump was turned on, and the oxygen valve was opened until the pressure reading stabilized. The power switch was turned to "HIGH" and the purple plasma gas was seen inside the chamber. After 10 minutes of plasma cleaning, everything was turned off, and the valve was slowly opened to prevent outside contamination from being sucked into the chamber. After the sample was removed from the plasma cleaner, it was placed in the plastic fixture that secures the sample in the loop. The heater pads were lined up with the pogo pins on the fixture cover; incorrect orientation of the sample could result in burning it. The sample was then secured in the fixture by tightening the bolts such that they could be loosened with the slightest effort. Once the reservoir's measured temperature approached 100 C, the heater was turned down to ~35 V. Once the reservoir water temperature reached 101'C, the vent valve at the top of the reservoir was opened and kept open to degas the reservoir and prevent the pressure inside from getting too high. The resistance temperature detector (RTD) power supply was turned on to 2 V and the microchannel heater power supply was turned on. Then a different LabVIEW program "Flow boiling input power control new 2," which displays all the measurements for the experiment (temperatures, pressures, mass flow rate, heat flux, and resistance), was opened and run. Monitoring the values here were crucial to ensuring that the system did not overheat or overpressurize. The calibration constants, the alpha and beta values, for each RTD were entered to the corresponding value for the sample in the loop in order to obtain accurate temperature data. In addition, the heater resistance was set to 300 Qand the maximum heater voltage to 2 V. Then, the preheater and postheater were set to the appropriate voltages such that at steady state, which takes 30-60 minutes to achieve depending on the flow rate, the inlet water temperature is between ~90-95'C. The needle valve was also set such that there was approximately a 10 kPa pressure rise after the pump. 23 2.4 Experimental Procedure Once the inlet water temperature was stable between ~90-95'C, the experiment could begin. The LabVIEW program was stopped and rerun before saving a new data file; otherwise, the data would be saved to the older file. For each data set, the program was run, the desired heat flux specified, and the maximum heater voltage increased as necessary such that the actual heater voltage was slightly below the maximum heater voltage and the actual heat flux matched the desired value. Once the measurements, such as temperature, reached steady state, data was recorded for at least 2 minutes, the recording was stopped, and finally the program was stopped. Notes of interesting and benchmark phenomena, such as the onset of two-phase flow, were recorded by hand in a laboratory notebook. During the experiment, high speed videos were taken in order to document interesting and benchmark phenomena in the flow boiling process throughout the experiment. Video was only taken while data was being recorded, so as to match up the data with the video for future reference. To do so accurately, the camera's clock had to be synchronized with the computer's clock. Then video would be taken for a few seconds, after which only a selected, representative fraction of that video would be saved. Once the system appeared to have achieved its critical heat flux or if the measured temperature of the microchannel consistently rose above ~200'C the experiment was stopped in order to protect the samples. At this point, the desired heat flux and the maximum heater voltage settings were gradually decreased to zero both to ensure that the actual heat flux becomes zero and to prevent the sample from cooling too quickly and causing a thermal fracture, damaging the sample. Then, the preheater, postheater, and reservoir heater were turned to 0 V and turned off. fllr wr The pump was turned off, and then all the valves of the osed. Once it was verified that the temperatures in the microchannel were dropping, then the heater power could be turned off and the LabVIEW program could be stopped and closed. 2.5 Measurements and Data Processing By following the procedure above, the resistances of the RTDs were measured. The temperatures throughout the loop were measured, excluding those in the test fixture. The pressure sensors at the inlet and outlet of the microchannel measured the inlet and outlet pressures. Finally, the mass flux was measured by the flow meter after the pump. Instrument error included the temperature measurements from the RTDs (2'C), pressure 24 ___ - - - 1.. -11 _11 measurements from the pressure transducers ( 0.3 kPa), instantaneous mass flux measurements from the flow meter ( 1 mL/min, or approximately 70 kg/m 2 s), and voltage and current measurements from the power supply ( 0.4 mA and 0.06 V, respectively). The quantities of interest were the input heat flux from the heater, the temperature rise of the fluid, and the pressure drop across the channel. The equations for calculating each of the values are given below in Equations 2.2, 2.3, and 2.5, respectively. q"= I1AAV (2.2) Aheater where I and AV are the input current and voltage from the power supply. AT = T3 - Tsat (2.3) where T 3 is the temperature of RTD3, calculated using the resistance of RTD3 in Equation 2.1 along with its calibration constants a and 3. Tat is a function shown below in Equation 2.4 that varies with the average of the inlet and outlet pressures, defined here as Psat. Tsat = 0.00000502877Psat - 0.00266178sat + 0.661681943Psat + 55.02536145 (2.4) This relationship is important to include because saturation temperature changes with the pressure of the environment and it is upon this value that the temperature rise is based. APb Pinlet - Pouthet The above equations produced the results discussed in the following section. 25 (2.5) Chapter 3 Results and Discussion 3.1 Flow Instability During the heated flow through the microchannels, flow instabilities in the form of temperature and pressure fluctuations could be observed in time. The heater surface temperature and the pressure drop across the entire channel were measured. Every sample (flat, 5-15, and 10-40) exhibited increasingly large oscillations with increasing heat fluxes for the lower mass fluxes 150 kg/m 2 s and 300 kg/m 2 s. For the 150 and 300 kg/m 2 s mass fluxes, the flat sample exhibited much greater temperature and pressure fluctuations than the structured samples due to more prolonged periodic dryout. These temporal fluctuations can be observed in the figures below, in which the heater surface temperature and the pressure drop across the channel were plotted for each sample at each mass flux for a representative range of heat fluxes. The following results are in reference to experiments performed at the mass flux of 150 kg/m 2 s, during which a representative range of heat fluxes was examined after two phase flow began, 15 W/cm 2 to 431 W/cm 2 , and these temporal results can be observed in Figure 3.1. At the lower heat fluxes, 150 W/cm 2 and 264 W/cm 2 , before the critical heat flux of the flat sample was reached, the flat sample experienced temperature fluctuations of t10'C and pressure drop fluctuations of t0.7 kPa to temperature fluctuations of 6'C to 1.0 kPa. The 5-15 sample experienced 8'C and pressure drop fluctuations of 0.6 kPa to 0.8 kPa. The 10-40 sample experienced temperature fluctuations of t0.3'C to and pressure drop fluctuations of 0.3 kPa. 7'C to 2.5'C 2 At the higher heat fluxes, 357 W/cm and 431 W/cm 2 , after the critical heat flux of the flat sample was reached, the 5-15 sample experienced temperature fluctuations of 0.6 kPa to 10'C to 130 C and pressure drop fluctuations of 0.8 kPa. The 10-40 sample experienced temperature fluctuations of 26 5'C to 8'C and pressure drop fluctuations of 150 kg/m^2*s, 0.6 kPa. 153 Wlcm^2 15) k/m^2-i 153 W/cme2 150 kg/n^2 264 W/cm^2 Time (s) 150 kgje 2* -264 Wckm2 TimeW 150 kgirn Pme i5o kgim"2~i 1I0 kg/me2*, , -357 Wfcm2 -357 W/cm^2 me (S) 150 kg/m"2 431 WAw^2 (S) -431 W/UnI2 kNAn Figure 3.1: Temporal data comparison of temperature and pressure drop for each microchannel sample with a 150 kg/m 2 s mass flux at a range of representative heat fluxes. The uncertainties of the temperature and pressure drop measurements were approximately 6.5'C and 0.45 kPa, respectively. 2 For the experiments performed at the mass flux of 300 kg/m s, a representative range of 2 2 heat fluxes was examined after two phase flow began, 152 W/cm to 640 W/cm , and these temporal results can be observed in Figure 3.2. The flat sample experienced temperature fluctuations of 5'C to 20'C and pressure drop fluctuations of 0.3 kPa to 1.0 kPa. The 5-15 sample experienced temperature fluctuations of 5'C to 7'C and pressure drop fluctuations of 0.6. The 10-40 sample experienced temperature fluctuations of 4'C to 6'C and pressure drop fluctuations of 0.2 kPa to +1.0 kPa. 27 -A MM 300 kg/^2i. W/Cm^2 ~152 300 kg/m^2s 12 W/CMr2 30Okg/m2*s, 376 W/cm^2 300 kg/rn~2~, ii23 WIcm'Z Time is) -'376 W/cm-'2 ~00 k~Im~2 Time fs} 300 kg/m^P2s. 523 W/Cm'2 w~ 300 kg/m 2s 640 Time W/m 300 kg/m^2% -640 W/cm2 2 Is 2 with a 300 kg/m s Figure 3.2: Temporal data of temperature and pressure drop for each microchannel sample and pressure drop mass flux at a range of representative heat fluxes. The uncertainties of the temperature respectively. kPa, 0.22 measurements were approximately 3.5*C and 2 in order to The experiments performed at the mass flux of 500 kg/m s were more averaged system. The filter the non-physical measurements such as noise or slight vibrations in the 153 W/cm 2 representative range of heat fluxes examined after two phase flow began was with the to 859 W/cm 2 , and these temporal results can be observed in Figure 3.3. Even is suspected additional averaging, the 5-15 sample exhibited more pressure fluctuation. This from a to be due the relatively dense spacing of 5-15. The vapor shear stress resulting drop across combination of the high mass flux and dense pillar spacing causes the pressure the channel to increase dramatically. 28 /cm^2Is, 600 kg/mc2~, "153 W/cm12 !Oh?_L A -A 163 V1/cm~2 - A .A -V - 600 Time 500 kg/m 2 '524 W/cm 2 S00 kg/m^2*, (s) 524 W/cc 2 Tmie fs 5,00 -'c~2 746 WAITnc2 20* 41 SN/b 09 OmAA OO# -4W1 500 kgm2*,-5 W/c' T fme1p 2 Figure 3.3: Temporal data of temperature and pressure drop for each microchannel sample with a 500 kg/m s drop mass flux at a range of representative heat fluxes. The uncertainties of the temperature and pressure 0 measurements were approximately 0.3 C and 0.28 kPa, respectively. 3.2 Heat Transfer Performance In addition to flow instabilities, the time-averaged heat transfer performance of the microchannels were also considered. In general, it was observed that as the heat flux q" increases, the temperature rise AT initially increases relatively linearly, then suddenly drops, begins to rise relatively linearly again, and then begins to increase enormously compared to the heat flux. The initial increase is due to heating single phase flow, during which the water simply increases in temperature with the heat flux towards the boiling point. The 29 sudden drop indicates the onset of nucleate boiling, when the heat transfer coefficient increases substantially due to the slight mixing that occurs when bubbles separate from their nucleation sites. The subsequent rise is the continual heating of two phase flow. Finally, massive temperature rises with little heat flux increase indicates that the critical heat flux has been surpassed. The critical heat flux is the point just before the heat transfer of the microchannel suddenly becomes inefficient, meaning that a small increase in heat flux will cause the temperature to rise dramatically. This transition occurs due to a dynamic phenomenon (not captured because of the time-averaged nature of the data) known as dryout. Dryout is a condition in which the liquid on the microchannel surface has evaporated away, thus causing the heat transfer coefficient to drop, which explains the inefficiency. Boiling curves were obtained by plotting the heat flux q" against the time-averaged temperature rise of the heater (RTD3) AT for each flow rate and microchannel investigated. The boiling curves for all the flow rates for each microchannel can be observed in Figure 3.4 and all the microchannels at each mass flux can be observed in Figure 3.5. 30 -- -I---' - - - -- - - - - - Boiling Curve, Flat 1200 2N - 150, FIat - - 300, Flat 500, Flat - - 1000 800 600 X< 400 20 r 0 10 0 60 50 40 30 20 70 AT (*C) Boiling Curve, 5-15 1200 -e-300, S-15 -- 150, 5-15 -0- 500, 5.15 1000 800 600 200 II0 0 s0 40 30 20 10 60 70 80 60 70 80 AT (*C) Boiling Curve, 10-40 1200 -- 150, A 10 40 300, 10-40 - -500, 10 40 1000 800 600 X 400 200 I 0 0 10 20 40 30 50 AT (*C) Figure 3.4: Boiling curves for all the flow rates for each microchannel. The effects of varying mass flux can be clearly observed for each microchannel. The initial slope indicates heating single phase flow, whereas the sudden jump back indicates the onset of nucleate boiling. The following upwards slope is heating of the two phase flow, and finally, the critical heat flux is indicated by when the slope changes significantly from the 2 two phase flow slope. Note: the 10-40 sample broke during the experiment at 500 kg/m s, and since there was no slope change observed, the critical heat flux was not reached and is therefore unknown. The error bars were approximately 1% for the heat flux q" and ranged from 2.0'C to 6.7'C for the temperature rise. 31 -- . " qj - - --- --; -1 - Boiling Curve, 150 kg/mA2*s 600 S-150,Flat -0-150,1-40 -4-150,5-15 50c) U 400 300 10 0 50 40 30 20 60 70 AT (*C) (U Boiling Curve, 300 kg/m^2*s 1200 -A 300, fat -i- -h-- 300, 1-4 300, 5-15 1000 00 2-0 E 10 0 U 50 40 30 20 60 70 80 60 70 80 AT (*C) Boiling Curve, 500 kg/mA2*s 1200 - 500, lat -U- 500, S-15 500, 10-40 - 1000 800 600 X 400 200 0 0 10 20 40 30 50 AT (*C) Figure 3.5: Boiling curves for all the microchannels at each mass flux. Differences between the microchannels can be clearly observed at each mass flux. The initial slope indicates heating single phase flow, whereas the sudden jump back indicates the onset of nucleate boiling. The following upwards slope is heating of the two from the phase flow, and finally, the critical heat flux is indicated by when the slope changes significantly 2 two phase flow slope. Note: the 10-40 sample broke during the experiment at 500 kg/m s, and since there was no slope change observed, the critical heat flux was not reached and is therefore unknown. The error 0 0 bars were approximately 1% for the heat flux q" and ranged from t2.0 C to t6.7 C for the temperature rise. 32 For a given microchannel, Figure 3.4 clearly shows that critical heat flux increased with the mass flux, as expected. The structured surface microchannels exhibited an increased heat transfer coefficient due to the fact that evaporation is dominant in the annular flow regime and the structures facilitate a stable liquid thin film. For each mass flux, it seems that the 10-40 structured surface microchannel exhibited the greatest critical heat flux value, followed by the 5-15, and then the flat surface. Unfortunately however, the 10-40 sample broke during the 500 kg/M2s experiment, so the critical heat flux value at that flow rate could not be determined. Table 3.1 displays the critical heat flux values of each microchannel at each flow rate as well as their critical heat flux enhancement percentage relative to the flat surface microchannel. Flo 1 50 Rte\amle Flat (W/Cm,2 , %) 5-15 (W/cm,2, %) 10-40 (W/cm,2, %) kg/mS 300 kg/i-s 762 612 24% 969 58% Table 3.1: Critical heat flux values of each microchannel at each flow rate as well as their critical heat flux enhancement percentage relative to the flat surface microchannel. The critical heat flux enhancement due to structured surface microchannels relative to the flat surface microchannel decreased with increasing mass flux. This effect can be explained by the fact that capillary pressure, given by Equation 3.1, assists in prolonging dryout, is independent of mass flux, and is dependent on the spacing between pillars. r Pcap (3.1) where o is the surface tension of the fluid, and r is the radius of curvature of the liquid-vapor interface whose lower bound is the pillar spacing, which in turn limits the magnitude of the capillary pressure Pap. Therefore, because the pressure driving the flow increases with mass flux, it eventually dominates the effect of capillary flow. At the lower mass fluxes, the 10-40 microchannel achieved greater critical heat flux values 2 than the 5-15. However, at the lowest mass flux of 150 kg/m s, the critical heat flux values for the 5-15 and the 10-40 are nearly equal. The trend suggests that perhaps at even lower mass fluxes, the critical heat flux achieved by the 5-15 may surpass that of the 10-40. This is because the 5-15 has a denser micropillar array due to its smaller spacing than the 10-40. 33 A smaller spacing would cause there to be greater capillary pressure, which is independent of mass flux. Therefore, the effect of capillary pressure would be dominant at lower mass fluxes, making denser micropillar arrays such as the 5-15 more beneficial than sparser arrays such as the 10-40 at lower mass fluxes. Based on these trends and observations, the lower the mass flux, the denser the micropillar array should be to maximize the critical heat flux value. 3.3 Pressure Drop While it has been shown that the heat transfer performance can be improved by adding structures to microchannel surfaces, the pumping costs must also be determined in order to accurately characterize the effectiveness of surface structures. Figure 3.6 shows the timeaveraged pressure drop across the microchannels as a function of heat flux. For each mass flow rate, the pressure drop across all the microchannels were nearly the same, implying that the expected additional pressure drop introduced by the surface structures was practically negligible. 34 -- --------I'll-........ --- i- Pressure Drop, 150 kg/mA2*s 6 -o- 150, Flat -0 150, 5-15 - 150, 10-40 5 4 CL 0~ 1 01T 50 0 200 150 100 250 350 300 400 450 500 Heat Flux q" (W/cmA2) Pressure Drop, 300 kg/mA2*s 18 16 -r-300, 300 Flat -- 5-1S -R-300, 10-40 14 12 CL 10 8 4 2 0 600 400 200 0 800 1000 1200 1000 1200 Heat Flux q" (W/cmA2) Pressure Drop, 500 kg/mA2*s 25 -0 500, Flat -M- 5(0, 5 15 -- 500, 1040 20 CU 1 CL 10 5 0 0 200 600 400 800 Heat Flux q" (W/cmA2) Figure 3.6: Boiling curves for each flow rate as well as each microchannel. Differences between the mi1% for the heat crochannels can be clearly observed at each mass flux. The error bars were approximately flux q" and ranged from 0.36 kPa to 0.60 kPa for the pressure drop. 35 This could have been predicted by the fact that there is a 500 im x 500 im cross-section above the structures in the structured microchannels, the same size as the entire cross-section of the flat surface microchannel. Therefore, at heat fluxes during which the structures are completely immersed in liquid, there should not be a noticeable pressure drop difference between each microchannel. Additionally, at heat fluxes during which there is insufficient liquid supply, the liquid-vapor interface forms menisci between the structures and generates capillary flow, increasing the flow rate. This capillary flow also does not require a larger pressure drop occur. Finally, comparing single microchannels at different mass fluxes, it can be observed that the pressure drop across the microchannels are comparable at 150 and 300 kg/m 2s, but that there is a noticeable increase at 500 kg/m 2 s. This could potentially be explained by the fact that at 500 kg/m 2 s, the viscous resistance has increased greatly and thus requires a much larger pressure head to propel the flow through the microchannel. 36 Chapter 4 Conclusion In an effort to find potential thermal solutions for newer and higher power density electronics, it has been demonstrated that utilizing two-phase microchannels with micropillar arrays are a viable future solution for some high power applications. These surface structures reduced the flow boiling instability because they enhanced the annular flow while the flat surface microchannel experienced more instabilities such as temperature spikes and pressure drop fluctuations due to more frequent dryout. Furthermore, because surface structures decreased the instances of dryout in microchannels, the structured surface microchannels were able to achieve greater critical heat flux values than the flat surface. In examining the heat transfer capabilities of the microchannels at various mass fluxes, it was verified that the critical heat flux value of all microchannels increased with mass flux. In addition, it was concluded that at lower mass fluxes, the relative percentage heat transfer enhancement of the structured surface microchannels over the flat surface microchannel was greater. The trend observed suggests that denser samples are better at lower mass fluxes. However, if a sample is too dense, there may be too much viscous drag and thus an optimal balance between capillary force and viscous drag must be found in order to determine the optimum micropillar array geometry and density. Finally, for a given mass flux, the pressure drop across every microchannel was approximately equal at all heat fluxes. This implies that no additional power consumption is required to pump a particular mass flux through a structured surface microchannel than a flat surface microchannel. This work provides insights into the roles of both the micropillar array surface structures and the mass fluxes on the heat transfer performance of two-phase microchannels. The results and observations of this experiment may prove helpful in guiding future work in an attempt to optimize microchannels for heat transfer applications in electronics. 37 While this work is a preliminary examination of the role of mass flux on the heat transfer properties, there are many more opportunities to further characterize flow boiling in structured surface microchannels. A few examples of future work include: (1) Examining the relationship between micropillar arrays with the same dl ratio. This study would aim to determine how and under what circumstances that microchannels with the same ratio but different diameter pillars differ. (2) Examining how the height of the micropillars given a constant cross-section affects the heat transfer and wicking properties of the microchannel. (3) Examining the performance of other common coolants that boil near the operating temperature of high power electronics in structured surface microchannels. 38 Chapter 5 5.1 Works Cited S. Krishnan, S. V. Garimella, G. M. Chrysler, and R. V. Mahajan, "Towards a Thermal Moore's Law," IEEE Trans. Adv. Packag., vol. 30, no. 3, pp. 462-474, 2007. E. Pop, "Energy dissipation and transport in nanoscale devices," Nano Res., vol. 3, no. 3, pp. 147-169, Mar. 2010. H. A. Kariya, T. B. Peters, M. Cleary, D. F. Hanks, W. L. Staats, J. G. Brisson, and E. N. Wang, "Development and Characterization of an Air-Cooled Loop Heat Pipe With a Wick in the Condenser," J. Therm. Sci. Eng. Appl., vol. 6, no. 1, pp. 011010-011010, Oct. 2013. N. Zuber, "Hydrodynamic Aspects of Boiling Heat Transfer," California. Univ., Los Angeles; and Ramo-Wooldridge Corp., Los Angeles, 1959. K.-H. Chu, Y. Soo Joung, R. Enright, C. R. Buie, and E. N. Wang, "Hierarchically structured surfaces for boiling critical heat flux enhancement," Appl. Phys. Lett., vol. 102, no. 15, pp. 151602-151602-4, Apr. 2013. Y. Qiu, Z. Liu, "Critical heat flux of steady boiling for saturated liquids jet impinging on the stagnation zone," International Journal of Heat and Mass Transfer, vol. 48, pp. 4590-4597, 2005. S. Kumagai, M. Murata, M. Izumi, and R. Shimada, "Heat Transfer Mechanism Based on Temperature Profiles and Bubble Motion in Microbubble Emission Boiling," Proc. 6 the ASME-JSME thermal Engineering Joint Conf., CD-ROM C6-219, 2003. K. Suzuki, "Microbubble Emission Boiling for Advanced Compact Heat Exchanger," Proc. 7th International Conference on Enhanced, Compact and Ultra-compact Heat Exchangers: Science, Engineering and Technology, CD-ROM, CHE2009-31, 2009. Y. Zhu, D.S. Antao, K.-H. Chu, T.J. Hendricks, E.N. Wang, "Enhanced Flow Boiling Heat Transfer in Microchannels with Structured Surfaces," International Heat Transfer Conference, Kyoto, Japan, August 10-15, 2014. D. Li, G. S. Wu, W. Wang, Y. D. Wang, D. Liu, D. C. Zhang, Y. F. Chen, G. P. Peterson, and R. Yang, "Enhancing Flow Boiling Heat Transfer in Microchannels for Thermal Management with Monolithically-Integrated Silicon Nanowires," Nano Lett., vol. 12, no. 7, pp. 3385-3390, Jul. 2012. 39 F. Yang, X. Dai, Y. Poles, P. Cheng, J. Khan, and C. Li, "Flow boiling phenomena in a single annular flow regime in microchannels (I): Characterization of flow boiling heat transfer," Int. J. Heat Mass Transf., vol. 68, pp. 703-715, Jan. 2014. F. Yang, X. Dai, Y. Peles, P. Cheng, J. Khan, and C. Li, "Flow boiling phenomena in a single annular flow regime in microchannels (II): Reduced pressure drop and enhanced critical heat flux," Int. J. Heat Mass Transf., vol. 68, pp. 716-724, Jan. 2014. R. Xiao, R. Enright, and E. N. Wang, "Prediction and Optimization of Liquid Propagation in Micropillar Arrays," Langmuir, vol. 26, no. 19, pp. 15070-15075, Oct. 2010. 40 5.2 5.2.1 clear clC Appendix MATLAB Script for Processing Experimental Data all (V) ) ( % Read all the raw data files i-min0 = 4; i-maxO = 29; i-minl = 4; i-maxl = 29; i-min2 = 5; i-max2 = 29; for i = i_min0:i_max0 File = [ ' TestData ' , sprintf ('%02d' , i ) , '.lvm '] 23, 0); Data{i} = dlmread(File '\t', Data0 = Data{i}; % time (s) = Data0(: ,1); t{i} % Calculated Uset f rom heater res istance = Data0(:,2); U-cal{i} % Actual Uset (V) = DataO(: ,3); U-set { i} % Heater resistance (Ohm) R-heater{i} = Data0(:,4); % power (W) DataO(: ,5); Pw{ i} C) Inlet % RTD temperature Data0(: ,6); T1{ i} C) before the heater % RTD temperature = Data0(: ,7); T2{i} C) heater cent er % RTD temperature = DataO(: ,8); T3{i} C) after the h eater % RTD temperature = DataO (:,9); T4{ i} = Data0 (:,10); % Reservoir pressur e (kPa) Pt{i} % Inlet pressure (b Pa) Data0(: ,11); Pi{i} kPa) Po{i} = Data0(: ,12); % Outlet pressure % Flow rate (mL/mir = Data0(: ,13); G{i} % Pump pressure (kF a) DataO (:,14); Pc{ i } % Microchannel inle t temperature = Data0(: ,15); T-mci{ i} % Microchannel out [et temperature = Data0(: ,16); T-to{i} end Optional step to correct temperatures if incorrec t for i = 1:29 T1{i}= (T1{i}+427.56)/0.8444*1.7308-601.78; T1{i} =T-{i}; T2{i} = (T2{i}+433.81)/0.8912*1.7998-604.68; T2{i} = T-2{i}; T-3{i} = (T3{i}+411.06)/0.8638*1.7333-572.48; T3{i} = T-3{i}; T_4{i} = (T4{i}+447.85)/0.835*1.7515-637.63; T4{i} = T_4{i}; end G-ave = 300; A-itube = pi()*0.003^2; m-ave = G-ave * A-itube; T-sc = 10; Area-heater = 500e-6 * lOe-3; T.ambient = 22; % h-itube = 1 % % % % calibration constants Average mass flux (kg/m2s) Inlet tube cross-section Area Average mass flux (kg/s) Subcooling (C) % (m^2) % (C) 41 were used % Boiling curve; HTC; P drop for i = i-min0:i-max0 Tla(i) = mean (T1{ i }); T2a(i ) = mean(T2{i }); T3a(i) = mean(T3{i}); T4a(i) = mean(T4{i}); = 0.25*T2a(i)+0.5*T3a( i)+0.25*T4a(i ); Ta(i) Pwa(i) = mean(Pw{i}); h1l(i) = -1.41033e-9*Ta(i)^3 + 9.50934e-7*Ta(i)^2 ... - 0.000122382*Ta(i) ... + 0.027040238; loss(i) = h-l(i).*(Ta(i)-T.ambient); Q{i} = (Pw{i}-loss(i))*0.95/Area-heater/le4; q(i) = (Pwa( i)-loss( i ))*0.95/ Area-heater/le4; % assuming 5% extra loss % accounting for loss as a function % Ti average % T average % Power average % % % % hloss (W/C) loss (W) heat flux (W/cm^2) Ave heat flux (W/cm^2) of temperature dP{i} = Pi{i}-Po{i}; P.mid{i} = 0.5*(Pi{i}+Po{i}); P-mida(i) = mean(P-mid{i}); dPa(i) = mean(dP{i}); T-sat(i) = 0.00000502877*P-mida(i)^3 - 0.00266178*P-mida(i)^2 + 0.661681943*P-mida(i) ... + 55.02536145; dT(i) = T3a(i)-T-sat(i); htc(i) = q(i)*1e4/dT(i)/le3; % Note this will likely underestimate the real % Pressure drop (kPa) % center pressure % average center pressure % Saturation temperature %delta T % HTC (kW/m^2 K) h Gal(i) = mean(G{i}); % ml/min Ga2(i) = Gal( i)/60/956.5/(500e-6)^2; % kg/m2 s end casenumber = 4; % Change casenumber switch casenumber case for a different plot 1 % Boiling Curve figure (1); h = axes('FontSize ' ,16); plot (h,dT(i-minO: i-maxO) ,q(i-minO: i-max),'-sb' , 'LineWidth' ,2 'Color' ,[0.8 0 0.2] ... 'MarkerFaceColor ' ,[0.8 0 0. 21 'MarkerSize ' ,6); xlabel ('{\ it \DeltaT} (^oC) ' ,' FontSize ' ,16 ,'Fontweight ','b') ylabel('Heat Flux {\itq"} (W/cm^2) ','FontSize' ,16,'Fontweight ylim ([0 ,1000]) xlim ([0 ,60]) legend (h,'10 -40','Location ','NorthWest') set (h, 'LineWidth ', 1); case 2 % Heat Transfer Coefficient figure (2); hi = axes ('FontSize ' ,l6); plot (hl ,q(i-min2 :i-max2) ,htc (imin2: 'Color ' ,[0.8 0 0.2] ... 'MarkerFaceColor ' ,[0.8 0 0. 2], ... 'MarkerSize ' 6); 42 imax2),'-sb','LineWidth ... ','b') ',2 ... at the center xlabel ('Heat Flux {\itq"} (W/cm^2)','FontSize ',16, 'Fontweight y IabeI( '{\ ith}I (kW/m^2-K)' 'FontSize ',16, 'Fontweight',b) ylim ([0 ,350]) legend (hl, ' Structured 10 -40', 'Location ', 'NorthEast') set (hl , 'LineWidth ',1); ','b') case 3 % Pressure Drop figure (3); h2 = axes ('FontSize ' ,16); ','LineWidth', 2 ... plot (h2,q(i-minl : i-maxl) ,dPa(i-minl : i-max),'-sb 'Color' ,[0.8 0 0.2],... 'MarkerFaceColor ' ,[0.8 0 0.2] ... 'MarkerSize ' 6); xlabel ('Heat Flux {\itq"} (W/cm^2) ','FontSize ' ,16 , 'Fontweight ' , 'b') ylabel ('{\ it\DeltaP} (kPa) ','FontSize ' ,16, 'Fontweight ' 'b') ylim ([0 ,14]) xlim ([0 ,900]) legend (h2, ' Structured 10 -40','Location ','NorthWest') set (h2, 'LineWidth ' , 1); case 4 % Temporal data, T3 and dP temp-data = cell (1, for i = 4:length(t) temp-data{i} length(t)-3); [ cell2mat (t(i )) cell2mat (T3(i )) end x1 x2 x3 x4 = = = = 29; 26; 27; 28; x1 = 28; x2 = xl + 1; x3 = x1 + 2; x4 = x1 + 3; time-low = 0; time-high = 200; temp-low temp-high = 0; 155; p-low = 0; p-high = 14; figure (4); subplot (4 ,2,1) plot (tf{xl},T3{x1},'-sb' , 'LineWidth ',1. 5 ,... 'Color' ,[0.8 0 0.2] ... 'MarkerEdgeColor ' ,[0.8 0 0. 2],... 'MarkerFaceColor ' ,[0.8 0 0. 2] ... 'MarkerSize ' ,2); title ([ 'q"=' sprintf('%0.lf' ,q(xl)) ' W/cm^2']) xlim ([time-low , time-high]) ylim ([temp-low ,temp-high]) subplot (4 ,2 ,3) 43 , cell2mat(dP( i)) plot(t{x2},T3{x2},'-sb','LineWidth',1.5,... 'Color' ,[0 0.65 0.32] ,... 'MarkerEdgeColor ' ,[0 0.65 0.35] 'MarkerFaceColor ' ,[0 0.65 0.35] ... 'MarkerSize ' ,2); ' W/cm^2']) title ([ 'q"=' sprintf('%0.lf ',q(x2)) xlim ([time-low , time-high]) ylim ([temp-low , temp-high]) ylabel('{Temperature \itT} (oC)','FontSize',12 ,'Fontweight subplot (4 ,2 ,5) LineWidth ' ,1.5 ,.... p = plot (t jx3}1,T3{x3}I,'- -',' 'Color ' , 'k' ,. .. 'MarkerEdgeColor' ,'k 'MarkerFaceColor' , 'k' 'MarkerSize ' ,2); title ([ 'q"=' sprintf ('%O.lf ' ,q(x3)) ' W/cm^2']) xlim ([time.low , time-high]) ylim ([ temp-low , temp.high 1) subplot (4,2,7) p = plot (t {x4},T3{x4},'-','LineWidth ' ,1.5 ,.... 'Color ', 'b' ... 'b' 'MarkerEdgeColor' 'MarkerFaceColor''b 'MarkerSize ' ,2); tit le ([ 'q"=' sprint f ('%O.lf ' ,q(x4)) ' W/cm^2'] xlim ([time-low , time.high]) ylim ([temp.low , temphigh]) xlabel ('Time {\ itt } (s ) ' , 'FontSize ' ,12, 'Fontwei ght ' , 'b') subplot (4,2 ,2) plot ( t {x1} ,dP{xl},'-sb ' , ' LineWidth ' , 1.5 ,... 'Color ' ,[0.8 0 0.2] ,... 'MarkerEdgeColor ' ,[0.8 0 0. 2], ... 'MarkerFaceColor ' ,[0.8 0 0. 2], ... 'MarkerSize ' ,2); t it le ([ 'q" =' sprintf ('%0. 1f ' ,q(xl)) ' W/cm ^2'] xlim ([time-low , time-high J) ylim ([plow , p.high]) subplot (4 ,2 ,4) plot (t {x2},dP{x2},'-sb ','LineWidth ',1.5 ,... 'Color ' ,[0 0.65 0.32] ... 'MarkerEdgeColor ' , [0 0.65 0. 3 5] 'MarkerFaceColor ' ,[0 0.65 0.35] ... 'MarkerSize ' ,2); title ([ 'q"=' sprint f ('%O.lf ' ,q(x2)) ' W/cm^2'] xlim ([time-low , time-high]) ylim ([p-low , p.high ]) ylabel ( '{\ it \DeltaP} (kPa) ','FontSize ' ,12 , 'Fontweight subplot (4,2 ,6) p = plot (t{x3},dP{x3},'-^','LineWidth ' ,1.5 ... 'Color ' , 'k' ,. 'k' 'MarkerEdgeColor' 'MarkerFaceColor', 'k' 'MarkerSize ' ,2); t itlIe ([ 'q"=' sprint f ('%0. 1f ' ,(x3)) ' W/cm ^2'] 44 ' , 'b') ' , 'b') ) xlim ([time-low , time-high]) ylim ([p-low , phigh subplot (4,2,8) p = plot (t{x4},dP{x4},'-^','LineWidth ' ,1.5 ... 'Color ', 'b' ,... 'MarkerEdgeColor' ,'b 'MarkerFaceColor' , 'b' 'MarkerSize ' 2); t it Ie ([ 'q" =' sprint f ('%0. 1f ',7q (x4)) ' W/cm ^-2'7j) xlim ([time-low , time.high]) ylim ([plow , p-high ]) xlabel('Time {\itt} (s)','FontSize ',12,'Fontweight ', 'b') case 5 % Temporal data, heat flux and inlet/exit pressure xI = 29; to = 0; x-min = 20; x-max = 35; y-minl = 120; y-maxl = 150; y-min2 = 1.6; y-max2 = 2.8; figure (5); subplot (1 ,2,1) plot(t{xl}-tO ,Q{xl},'-sb' ,'LineWidth ',1.5 , ... 'Color ' ,[0.8 0 0.2] ,... 'MarkerEdgeColor ' ,[0.8 0 0. 2] ,... 'MarkerFaceColor ' ,[0.8 0 0. 2] ,... 'MarkerSize ' ,2); title (['Structured 10-40, q"=' sprintf('%0.lf' ,q(xl)) ' W/cm^2']) xlabel ('Time {\itt} (s) ','FontSize' ,12 , 'Fontweight ', 'b') y labeI ('Heat Flux {\it q"} (W/cm^ 2) ','FontSize ',12 , 'Fontweight ' 'b') xlim ([x-min ,x-max]) subplot (1 ,2 ,2) plot (t{x1}-tO ,Pi{xl},'-sb' , 'LineWidth ' ,1.5 ... 'Color ' ,[0 0.65 0.32] ,... 'MarkerEdgeColor ' ,[0 0.65 0. 35], ... 'MarkerFaceColor ',[0 0.65 0. 35] , ... 'MarkerSize ' ,2); title (['Structured 10-40, q"=' sprintf('%0.lf' ,q(xl)) ' W/cm^2') xlim (x.min ,x-max]) hold on, plot (t{x1}-tO ,Po{x1},'-sb' , 'LineWidth ' ,1.5 'Color ',[0.8 0 0.2],... 'MarkerEdgeColor ' ,[0.8 0 0. 2] 'MarkerFaceColor ' ,[0.8 0 0. 2] 'MarkerSize ',2); legend ('P in ' , 'P out ') ylabel('{\itP} (kPa)','FontSize',12,'Fontweight','b') ,12, 'Fontweight ','b') xlabel('Time {\itt} (s) ','FontSize' hold off figure (4); h = subplot(1,2,1); 45 plot (t {x1}-tO, T3{x},'-sb' , 'LineWidth ' ,1.5 'Color' ,[0.8 0 0.2] .... 'MarkerEdgeColor ' ,[0.8 0 0. 2] 'MarkerFaceColor ' ,[0.8 0 0. 2 'MarkerSize ' 2); grid minor ' W/cm^2']) ([ 'Structured 10-40, q"=' sprintf('%O.1f ',q(xl)) title xlim ([x-min,x-max]) ylabel('{\itT_3} (^oC) ','FontSize ' ,12,'Fontweight ','b') 'FontSize ',12, 'Fontweight ','b') xlabel('Time {\itt} (s)', set (h, 'LineWidth' ,1); h=subplot (1 ,2 ,2) 'LineWidth' ,1.5 ... plot (t{xl}-tO, dP{xl},'-sb ... 0.2] 0 'Color ' ,[0.8 'MarkerEdgeColor ' ,[0.8 0 0. 21 ... 'MarkerFaceColor '40.8 0 0.2] 'MarkerSize ' ,2); (['Structured 10-40, q"=' sprintf('%0.lf' ,q(xl)) title xlim ([x-min ,x-max]) ylabel ( '{\ it\DeltaP} (kPa) ','FontSize ' ,12 , 'Fontweight ' xlabel('Time {\itt} (s)' ,'FontSize',12,'Fontweight' ,'b') set (h, 'LineWidth ' ,2); W/cm^2'1) , 'b') end 5.2.2 LabVIEW Figure 5.1: Screenshot of the front panel of the LabVIEW file used to monitor the entire system loop. 46