* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download C044
Survey
Document related concepts
Radiator (engine cooling) wikipedia , lookup
Insulated glazing wikipedia , lookup
Space Shuttle thermal protection system wikipedia , lookup
Building insulation materials wikipedia , lookup
Thermoregulation wikipedia , lookup
Dynamic insulation wikipedia , lookup
Solar air conditioning wikipedia , lookup
Heat equation wikipedia , lookup
Intercooler wikipedia , lookup
Heat exchanger wikipedia , lookup
Solar water heating wikipedia , lookup
Cogeneration wikipedia , lookup
Underfloor heating wikipedia , lookup
Copper in heat exchangers wikipedia , lookup
R-value (insulation) wikipedia , lookup
Transcript
Proceedings of NHTC'00: National Heat Transfer Conference Pittsburgh, Pennsylvania, August 20-22, 2000 3 4 th NHTC 2000 - 12104 THE EFFECTS OF FLOW RATE AND SODIUM LAUREL SULFATE SURFACTANT CONCENTRATION ON SINGLE- AND TWO-PHASE HEAT TRANSFER CHARACTERISTICS IN A MICROCHANNEL Michael S June Graduate Student Kate Gleason College of Engineering, Rochester Institute of Technology, Rochester, NY 14623 ABSTRACT The processing speed of modem microchips and the density of components at the card level has been increasing at a steady rate. With these increases, come an increase in heat generation and a need to dissipate very high heat fluxes. Heat sinks that employ very narrow channels, on the order of 58 600 m wide have been shown to dissipate heat fluxes as high as 10,000 kW/m 2 (Tuckerman, 1984) while maintaining chip temperatures below 130 C. The current study investigates the heat transfer characteristics of single and two-phase flows in a 200 m wide channel. Water was used as a working fluid at three flow rates and at three surfactant concentrations. The results showed a strong dependence on flow rate in the single phase region, and up to 9120 kW/m 2 was dissipated while maintaining a surface temperature of 115 C. The data also suggest a reduction in heat transfer efficiency with the addition of the surfactant sodium laurel sulfate (0.01%) in the single phase. There was a delay in the onset of nucleate boiling, however as expected for the same surfactant concentration. In the two-phase region, there was no significant change in the heat transfer characteristics with the addition of surfactants. INTRODUCTION Driven by the needs of the electronics industry, investigations have been underway to find more efficient ways to transfer heat away from electronic components. This has been of particular interest in the electronics packaging industry. Silicon chips must be packaged prior to being installed on a circuit board. The package provides protection, electrical connection to the circuit board, and heat dissipation. Both the reliability of the chip and the reliability of the package depend on temperature. The processing speed of the chip can also be Satish Kandlikar* Professor and Head Mechanical Engineering Department, Kate Gleason College of Engineering, Rochester Institute of Technology Rochester, NY 14623 *-corresponding author affected by high temperature. Chips operate best below certain prescribed temperatures, usually around 130 degrees Celsius. As the processing speed of a chip increases, so does the generation of heat. Due to the high heat fluxes generated by modem chips and the ever increasing density of electronic components on circuit boards it was estimated in the early 1980s that there would be a need to dissipate as much as 100 W/cm 2 by the end of the decade Prior to 1981, the highest cooling rates were achieved by specialized cooling units made by IBM and Honeywell. These devices could dissipate only 20 W/cm 2, which was considered as the practical upper limit to heat dissipation in electronics packaging. In their pioneering works, Tuckerman and Pease (1982, 1984) showed that enhanced heat transfer performance could be achieved using heat exchangers incorporating very small channels for the working fluid to flow. They performed optimizations that showed 58 m (0.002 in) to be the ideal channel width for heat transfer with water as a working fluid. Channel width was optimized by minimizing hydraulic diameter ( D h ) in order to minimize convective thermal resistance. The lower bound of 58 m was determined based on the viscosity of the working fluid, water, and a channel length of lcm. With these dimensions, a system pressure of 50 psi would need to be maintained for laminar flow. These parameters were chosen to achieve the goal of compactness, and to keep system pressures comparable to contemporary cooling systems. Other researchers use the term microchannel to refer to channels as large as 600 m in width. They have been fabricated from various materials such as silicon, aluminum, and indium phosphide, all yielding similar heat transfer characteristics. The heat transfer performance of microchannels is significantly higher than conventional sized channels, but 1 Copyright © 2000 by ASME present drawbacks as well. They are difficult to fabricate and clog easily. For these reasons, the trend has been toward larger channels, which are still small enough to maintain the high performance of microchannels. These are sometimes referred to as millichannels. Turbulent Flow in Microchannels In general, for internal flows through conventional sized pipes and ducts, flows are considered to be laminar when the Reynolds number is less than 2300. In microchannels it has been observed that fully turbulent flow occurs at Reynolds numbers between 1000 to 1500 (Peng and Wang, 1993, Wang and Peng, 1994, Sivagnanam, Balakrishnan, and Varma, 1994). Effect of Surfactant The addition of surfactants to the working fluid has been found to slightly increase the heat transfer coefficient and delay the onset of nucleate boiling. Kandlikar (1999) presents a table summarizing results of studies on the effects of various surfactants on heat transfer. While most studies have found a slight increase in heat transfer with surfactants such as ethylene glycol, there is one study by Wang and Harnett (1994) that found a slight decrease in heat transfer with the addition of sodium laurel sulfate. It was also stated by Dhir (1999), that as surface wetability increases, the boiling curve tends to shift to the right. Increasing wetability is an effect of surfactants. Objective The objective of the current work was to construct an apparatus which could be built from easily obtainable components, and would allow the study of heat transfer characteristics of water in a microchannel. In addition, the apparatus needed to allow for visualization of the flow. The effect of flow rate, and surfactant concentration were studied while using a microscope to view the flow and observe the presence and nature of bubble formation. NOMENCLATURE A Cross sectional area of a resistor Ac Channel cross sectional area B Bias CHF Critical heat flux Cp Specific heat D Diameter Hydraulic diameter Dh h Heat transfer coefficient in the microchannel k ONB P Average heat transfer coefficient in the microchannel Thermal Conductivity Onset of nucleate boiling Perimeter P Pr q" R R" R2 Re 1% SLS To Tlmtd To Tsat Tw U O~ /.t/ /.iv P Precision Prandtls number Heat flux Resistance Total thermal resistance Curve fit Reynolds number Initial resistance Sodium laurel sulfate Initial temperature Log mean temperature difference Outlet temperature Saturation temperature Heating element (wall) temperature Total uncertainty Temperature coefficient of resistivity Viscosity of liquid water Viscosity of water vapor Density PROCEDURE The apparatus was intentionally kept very simple. Available literature contains descriptions of many microchannel fabrication techniques such as chemical etching and microsawing, all requiring larger budgets and possibly outside vendors. The objective was to build a simple inexpensive apparatus that could be built from easily obtainable components and that could be built in an average shop with no exotic tooling. In addition, the apparatus had to be easily disassembled for reasonably efficient repair in case of heating element burn out. The apparatus designed for this experiment meets these needs. Apparatus The base and test section were built primarily of glass and acrylic. An acrylic base was bonded to a piece of ordinary 6.35 mm (1/4 inch) thick glass, which had holes to allow the water to pass through at the inlet and the outlet of the channel (Figure 1). The glass also had holes for the bolts which applied pressure to hold the various layers together. Larger holes were bored into the acrylic, concentric to the water inlet and outlet holes in the glass. Brass inserts were machined and bonded into these holes to serve as attachment points for the plumbing fittings and as electrical conductors to the heating element. The heating element (Omega cat. no. SPPL-008, 0.008 in diameter pure Platinum wire) was threaded through the brass inserts so that it ran along the surface of the glass between the holes, and was secured inside the brass inserts with set screws. Microscope slides (VWR Scientific cat. no. 48300-025, 25 X 75 X t mm) were placed tightly against both sides of the heating element to form the channel. Another ~Ain thick piece 2 Copyright © 2000 by ASME of glass matching the perimeter of the microscope slides was placed on top forming the top of the channel and providing a window into the channel for observation. The channel dimensions were such that any attempts to place adhesives between the glass layers resulted in excess adhesive moving into the channel by capillary action. This immediately clogged the channel rendering it useless. Pressure, therefore was necessary to hold the layers together and to insure water would flow in the channel and not between the layers. This was accomplished by milling a pocket and a window in another piece of acrylic to hold the topmost piece of glass, and bolting this structure to the base. Plumber's putty was placed under the holding fixture and outside the perimeter of the microscope slides to plug the ends of the channels. To ensure a good seal, silicon bathtub caulk was applied along the perimeter of the holding fixture between the holding fixture and the base. The water flowed up through the brass fitting in the base where it made a 90-degree turn into the channel. Here it ran over the heating element in the channel making another 90degree tum at the end and exiting through another brass fitting in the base. Electrical current was supplied to the heating element also through the brass fittings (Figure 1). Drawings of the apparatus are shown in Figures 1 and 2. A schematic diagram of the flow path is shown in Figure 2A. 6.35 nn (1/4 in) Glass Plunbee's 75 x 25 X I mM Microscope ~ \ II J • ~--Microchan.el o,eo3e nn (.OOB ~n) Di~ Neoting Etement I 635 ~ ( 1 / 4 i~> 6LQSS Plunber's Putty Holding Fixture Sibcon ou~w II 1' II 25. . . . . ~ . . . . . . . . . . ~oss i in Figure 2 End view of the apparatus with detail of channel cross section A t m o s pto Outlet heric Pressure glides Hoidimg F i x t u r e ..... Mr II ~ 3 r 3~ d L / Z5.4 nm (l in) AcryLic Figure 2A Schematic diagram of flow path Water Weter Figure 1 Side view of the experimental apparatus Test Section The cross sectional dimensions of the test section were determined by the diameter of the heating element and the thickness of the microscope slides. Pure platinum thermocouple wire was chosen as the heating element because of its ability to withstand high temperatures and because it's readily available in a variety of sizes. In preliminary experiments, 0.002 inch wire was used as the heating element, but it proved too difficult to work with, and the channels frequently clogged. The wire used in the final apparatus was .008 inches in diameter, which resulted in a channel width of 203.2 p.m ( 0.008 inches) which is roughly 3.5 times wider than Tuckerman's optimized channels (Tuckerman, 1984). Since the cross section of the heating element is round and makes up the bottom of the channel, the channel is not rectangular. This is not expected to affect the results significantly, as only one side of the channel is affected. It was also assumed that the enhancement in heat transfer characteristics due to the size of the channel would dominate any due to this departure from rectangularity. The cross section 3 Copyright © 2000 by ASME of the channel is shown in detail in Figure 3. The cross sectional area of the channel was calculated by considering a rectangle of the same width as the channel, from the top of the channel down to the point of contact tangent to the heating element. One half of the cross sectional area of the heating element was then subtracted from the area of the aforementioned rectangle. C r o s s S e c t i o n a l A r e a : O.1489mm^2 P e r i m e t e r : 2147mm, Dh: 0277mm / / drilled near the bottom for an outlet and another was drilled at the top for an air inlet. Brass fittings were threaded into the holes on the outside for the water outlet and the air inlet. Another fitting was threaded to the same hole as the air inlet, but on the inside of the vessel. This provided an attachment for an ordinary water balloon. The air was delivered from a regulator to the balloon, which forced the water out of the vessel without introducing air into the water. This allowed the amount of dissolved air in the water to be minimized. The pump is shown in Figure 4. /- ~- 02032nm / Air Inlet / Figure 3 Cut away cross section of channel with dimensions Water It was assumed that the edges of the microscope slides were in close contact with the heating element. Therefore, the width of the channel was the same as the diameter of the heating element. It was assumed that there was water undemeath the heating element, but it was also assumed that the water underneath was not flowing with a significant velocity. Since the water underneath the heating element was not flowing, it reached an equilibrium condition and did not contribute significantly to heat transfer from the heating element. It was further assumed that the lower half of the heating element was well insulated by the glass base and channel sides. It was therefore assumed that heat flux to the working fluid occurred only through the upper half of the surface area of the heating element. Heat loss through the apparatus was minimized by the insulating properties of glass and acrylic. Extra fiberglass insulation was placed around the apparatus to further minimize heat loss, especially around the brass fittings. This was verified by flowing heated water through the test section and determining the energy loss from the temperature drop across the apparatus. Energy loss was negligible at less than onepercent throughout the power range. Pump A pressure vessel was fabricated from a piece of ¾ inch thick walled tubing, 6 inches in diameter and 8 inches long. Steel end caps were fabricated and bolted to each end with rubber gaskets between the mating surfaces. A hole was Outlet Figure 4 Air pressure-driven pump The heating element temperature could not be measured with a thermocouple as it would become an efficient path for heat loss. Instead, the resistance of the element was measured and from this, the average temperature over the length of the element was determined Water was pumped using a peristaltic pump from a constant temperature bath through the apparatus. To achieve better accuracy during the calibration, current was supplied using a Keithley model 224 Current Supply. The Hewlett Packard power supplies used in the experiment were unable to supply a steady current in the milliamp range needed for calibration. Calibration was performed using a 75 milliamp current with the corresponding voltage drop across the heating element measured with the a Keithley model 199 System DMM/Scanner with 6 '/2 digit readout. From this data, resistance was determined at the temperature of the water pumped through the apparatus. Temperatures were chosen to correspond to the expected temperature range of the heating element during the experiment. The calibration plot is shown in Figure 5. 4 Copyright © 2000 by ASME The equation the best fit line through the calibration data was determined to be R = 0.00025(T - To) + 0.06867 (RZ=0.995) (1) where To = 27.2 o C. From Equation 1, a can be determined to be 0.003640 which is a 7.3% difference from the value given in the CRC Handbook of Chemistry and Physics (1986) of 0.003927 (over the range of temperatures from 0-100 C.). 0.08 ..... y: ...... 0.078 " 0.076 E ~e 0.074 • ~ 0.072 • 0.07 0.068 0 5 10 15 20 25 30 35 40 45 T-To (C) Figure 5 Calibration curve The working fluid in these experiments was either deionized water, or de-ionized water mixed with the surfactant, sodium laurel sulfate (SLS, CAS number 151-21-3). The working fluids were degassed using a vacuum pump and were left under negative pressure until out-gassing ceased. For solutions with 0.05% SLS and 0.10% SLS, degassing was not effective as emulsions were formed which made it impossible to tell when out-gassing had ended. Consequently, an experiment was performed using working fluids that were not degassed, with de-ionized water and de-ionized water with 0.01% SLS. This ensured that the level of dissolved gasses in the two working fluids were comparable and therefore did not affect the comparison. The solutions of 0.05% and 0.10% SLS were made by dissolving 0.500g and 1.000g respectively of solid sodium laurel sulfate in room temperature de-ionized water. The final volume of each mixture was 1.0L therefore the solution concentrations are given as percent by weight. The 0.01% solution was prepared by dissolving 10.0g of SLS into 100ml of de-ionized water, and diluting the subsequent mixture to a final concentration of 0.01% by weight. All working solutions were filtered through a 5/Jz., Gelman Scientific in-line filter prior to reaching the test section. Prior to recording data, the system was allowed to reach a steady. Steady state was considered to have been reached when there was not more than a 0.2 C change in temperature over a 1-minute interval. In most cases steady state was reached within seconds and lasted until an air pressure variation in the pump was encountered. Since the results were sensitive to flow rate, any fluctuation in flow rate caused departure from steady state. During certain phases of boiling, temperature variations were not due to changes in pump air pressure fluctuations, and could not be kept within 0.2 C. Steady state in these cases was considered to be the point at which the output temperature stopped showing an upward trend and fluctuated about a steady mean. Once steady state was established, readings of voltage and temperatures were taken approximately every 15 seconds, for a total of 10 readings for every power setting. The quantities read were, • Voltage across the heating element. • Voltage across the shunt resistor (for determining current through the heating element). • Temperature of the working fluid at the inlet of the channel. • Mixed mean temperature of the working fluid at the exit of the channel. • Volume of water leaving the apparatus over a 1-minute time interval. In addition, the following dimensions were measured using micrometers and scales where appropriate. Ten measurements of each dimension were averaged to obtain the final results. • Length of the heating element. • Length of the channel. • Diameter of the heating element. (width of the channel) • Thickness of the microscope slides (height of the channel) RESULTS Effect of Flow Rate on Single Phase Data For single phase heat transfer, heat fluxes as high as 5000kW/m 2 were achieved while keeping the heating element temperature under 100 C. Figure 6 shows the heating element temperature as a function of heat flux for three flow rates. The three flows are all turbulent and in the single-phase region. Clearly, as flow rate is increased, higher heat fluxes can be dissipated. Figure 7 shows the average heat transfer coefficient (~) for the same three turbulent flows in Figure 7. All flows show a nearly constant value for ~, as expected, and ~-increases with increasing flow rate. 5 Copyright © 2000 by ASME 200 9000 0~180 8000 ~ 160 • 0.46 mils (Re = 1200) • 0.83 mils (Re = 2000) 7000 ~14o • 1.0 mils (Re = 2600) • 0.46 mils (Re = 1200) 110.83 mils (Re = 2000) • 1.0 mils (Re = 2600) ~---6000- 12o ! = ~ 50 ~ 90 ~ 40 < eee•ee • • • 2000 3000 4000 Heat Flux (kWlm z) 5000 00 6000 40 60 80 100 120 Figure 8Boiling curve including single phase data for three flow rates N~'140 E 16o. ~120 ¢ • 0.46 ml/s (Re = 1200) • 0.46 mils (Re = 1200) • 0.83 ml/s (Re = 2000) • 1.O mils (Re = 2600) J 100 • 0.83 mils (Re = 2000) .y120 8 90 • 1.0 m l / s (Re = 2600) o 50 m o <J 80 40 & I 20 o: Average Heating ElementTemperature(°C) Figure 6 Average heating element temperature at various heat fluxes with flow rate as a parameter ~ 20 .:.~.'.." I000- i" #* 2000- 1000 "~,m. g Io 3000- 20 z • &&~• 5000x= E 4000- ~ t00 • •! tI000! • e l • elle • •emcee• 1 • • 20 < • 0 00 1000 2000 3000 4000 5000 6000 7000 8000 9000 Heat Flux (kWIm 2) 01000 2000 3000 4000 5000 6000 m=Fh~0et~r~ Figure 9 Average heat transfer coefficient at various heat fluxes with flow rate as a parameter for single phase data Figure 7 Average heat transfer coefficient at various heat fluxes with flow rate as a parameter for single phase data Effect of Flow Rate on Two-Phase Data For two-phase heat transfer, heat fluxes as high as 8000kW / m 2 were achieved while keeping the heating element temperature under 106 C. Figure 8 shows the heat flux as a function of heating element temperature for three flow rates in the single and two-phase regimes. Figure 9 shows for the three flow rates in single and two-phase heat transfer. Single-phase heat transfer shows an increase with increasing flow rate. Values of h become coincident in the fully developed boiling region. Figure 10 shows ~ as a function of cooling water inlet temperature subtracted from heating element temperature. Once again, h is directly dependent on flow rate. Hysterisis can be seen in the 0.83 ml/s (8.3E -4 k g / s ) flow rate data in the twophase region. 120 A 100 == 60 ~ 60 I- 40 • 0.46 ml/s (Re = 1200) • 033 ml/s (Re = 2000) • 1.0 mils (Re = 2600) • g E 0 eO A& • • ~,A= 0 • • ¢~ • • • • 20 0 20 40 60 80 T.,. - Ti.l.* (°C) Figure 10 Average heat transfer coefficient for single-and two-phase data with flow rate as a parameter 6 Copyright © 2000 by ASME Effect of Surfactant on Single Phase Data Figure 11 shows heat flux as a function of heating element temperature. There is no apparent enhancement of heat transfer characteristics shown from the addition of surfactants in this data. 6000 0.46 mils (Re = 1200) Water 5000 • 0,46 mils (Re = 1200) 0.05% SLS ~P'4000 41 • • 1,) ".:, • 0.46 mils (Re = 1200) 0,10% SLS o• • • ~3000 ,"C lw* Z 2000 Higher Heat Flux Data Figure 13 and 14 show the results of the repeat experiment comparing de-ionized water to a lower SLS concentration. Figure 13 shows heat fluxes on the order of 6 0 0 0 k W / m 2 were dissipated while keeping the heating element temperature below 100 C. Figure 14 shows g as a function of of heat flux, and like the previous surfactant data, show decreasing ~ with surfactant. Figure 13 shows heat flux as a function of heating element temperature. Here the delay in the onset of nucleate boiling can be observed, but hysterisis is not observed. Bubbles were observed in the test section, but appeared to be forming only in areas where the seal was not tight. The bubbles appeared to be going off to one side and may have been escaping between the layers of glass. 1000 10000 9000 20 40 60 80 A v e r a g e H e at in g E l e m e n t T e m p e r a t u r e (*C) 100 120 • 0.46 mils (Re = 1200) 0.01% SLS 8000 • 0.46 ml/s (Re = 1200) Water 7000 Figure 11 Heat flux as a function of average heating element t e m p e r a t u r e with surfactant concentration as a parameter. I ~ 6000 5000 4000 |* Z 3000 2000 1000 20 Effect of Surfactant on T w o - P h a s e Data Figure 12 shows h as a function of heat flux. The effect of the surfactant is unclear from this data, but significant hysterisis can be seen in the two-phase region. The data shows a possible downward trend in g as surfactant concentration is increased. 40 60 80 100 A v e r a g e Heating E l e m e n t Te mpe r a tur e (~C) 120 140 Figure 13 H e a t flux as a function of average heating e l e m e n t t e m p e r a t u r e with surfactant concentration as a parameter 140 • 0.46 mils (Re = 1200) 0.01% Surfactant ~E ~ 120 • • 0,46 mils (Re = 1200) NO Surfactant 90 • ,F • 0.46 ml/s (Re = 1200) Water ~= 8 0 . Q '0 • 0.46 ml/s (Re = t 200) 0,05% Suffactant 100 80" ¢ • 0.46 ml/s (Re = 1200) 0.10% Suffactant 70 60- ° 60. ~ 40 " • • • • g ~ 20- 50 # g 40 0 < 3c 0 1000 2000 3000 4000 5000 6000 2000 4000 6000 H e a t F l u x ( k W l m 2) 8000 10000 Figure 14 H e a t transfer coefficient as a function of heat flux with surfactant concentration as a p a r a m e t e r H e a t Flux ( k W l m 2) Figure 12 Average heat transfer coefficient as a function of heat flux with surfactant concentration as a parameter. 7 Copyright © 2000 by ASME DISCUSSION Comparison to available literature must be made with some qualification. Available literature deals with the cooling of a heat source (microchip) via a heat sink, whereas this study deals with the direct cooling of a heating element. In the available literature, therefore, the heat source is separated from the working fluid. The heat sink conducts the heat to the working fluid on three sides of the channel. In the current study, the heat source is directly immersed in the working fluid, which is heated by only one wall of the channel. 0.12 - 0.097 0,1 0,085 ~'~'0.08 0.072 ~.~ 0.04 0.02 - - Phillips Dissipated Heat Flux Despite the aforementioned differences, the data are similar to published results. Figure 15 shows heat fluxes dissipated at given surface temperatures for the current study as well as previous studies by Tuckerman (1984) and Phillips (1987). As seen in Figure 15, the heat fluxes dissipated in the current study are within 14% of the heat fluxes dissipated in current studies at similar surface temperatures. Figure 16 shows a comparison of total modified thermal resistance between the current study and previous studies by Tuckerman (1984) and Phillips (1987). The results of the current study differ from the results of Phillips (1987) by as much as 35%, however differences in flow rate may account for the difference in results. Results reported by Phillips (1987) were for Re = 1500 while the results of the current study are for Re = 1200. Tuckerman June Figure 16 Comparison of total thermal resistance to previous studies Comparison to Correlations Figure 17 compares the results of the current study, to correlations given by Dittus-Bolter (1930), and Peng and Wang (1993). Both correlations are for single phase, turbulent flow, however Peng and Wang (1993) formulated their correlation specifically for microchannel flows, citing that Dittus-Bolter under predicted the results of microchannel phenomena. The Dittus-Bolter correlation is given by, Nu = 0.023 Re 4/5 Pr °'4 (2) 11000 (110,10590) Phillips • 10500 (130,10000) T u c k e r m a n ~ 10000 A ~lE 9500 ! (115, 9120) June and the Peng and Wang correlation is given by, • TM e (3) 9ooo 8500 8000 7500 7000 20 40 60 80 100 120 140 Heating Element Temperature (C) As seen in Figure 17, data match Peng and Wang for Reynolds numbers of 1100 and 1921, but results of the current study differ from both correlations for Reynolds number of 2556. This is due to the scatter in the data at that Reynolds number. The data average to a higher value than was expected. Figure 15 Comparison of heat flux results to previous studies. 8 Copyright © 2000 by ASME O Peng and Wang A .= • June x Dietus BoRer ~) lo =, ,< Average RE Figure 17 Comparison of average Nusselt numbers to correlations Power data were compared to values predicted by a simple energy balance for given heating element temperatures. As expected, the experimental data showed slightly higher than predicted power was needed to obtain a given heating element temperature, due to losses. Figure 18 shows the energy balance results and experimental results for power at given heating element temperatures. Effect of Surfactant Figure 12 shows inconclusive results for the effects of the surfactant, sodium laurel sulfate on single phase heat transfer. This is most likely due to the differences in dissolved gasses in each working fluid. A study by Adams, Ghiaasiaan, and AbdelKhalik (1999) showed that dissolved air in water could enhance the heat transfer as much as 17%. Due to the foaming of the surfactant when placed under negative pressure, the working fluids could not be held under negative pressure for equal amounts of time and were therefore degassed to different extents. Also it was thought that the surfactant concentrations were too high at 0.05% and 0.10%. For this reason the experiment was repeated for a non-degassed working fluid at a concentration of 0.01% sodium laurel sulfate. This was compared to the same de-ionized water that was used to prepare the surfactant solution to ensure that the levels of dissolved gasses were as similar as possible between the deionized water and the surfactant solution. Results of this experiment are given in Figures 13 and 14 where it can be seen that the heat transfer coefficient was lower for the working fluid with the surfactant in single-phase heat transfer. For two-phase heat transfer, Figure 13 shows a delay in the onset of nucleate boiling as predicted by Dhir (1999). Hysterisis is also observed in Figure 13 for the de-ionized water, but not for the surfactant data. This is most likely due to having too few data for higher heat fluxes in the surfactant data. Hysterisis in the de-ionized water data is from the sudden release of dissolved air in the working fluid at the onset of nucleate boiling. 0.05000 0.04500 0.040Oo o 0.03500 g 0.03OO0 i 0.02500 UNCERTAINTY Estimates of uncertainty in this experiment were calculated using a 95% confidence interval. Uncertainty was determined from estimates of bias (B) and precision (P) by the following, U = [B2 + p2 ]1/2 (4) o O ExperimentalData • Energy Balance o" o 0.02000 _o 0.01500 o 0.01000 i 8 0.00500 0,00000 20 40 60 80 100 120 Heating ElementTemperature (C) Figure 18 Experimental results compared to results predicted with energy balance In general, bias was considered to be the least count of the measuring instrument, calibration data as supplied by the manufacturer or, in the case of the thermocouples, calibration done in-house. Precision was determined by multiplying the standard deviation of 10 measurements by 2.26 (from a standard Students T-Table). In general, 30 would be the optimal number of measurements, however difficulties in maintaining steady state for a long period of time prevented that number of measurements. Propagation of error was determined by solving the propagation equation of Kline and McClintock (1955). 9 Copyright © 2000 by ASME A summary of bias, precision and total uncertainty values determined for this experiment is given in Table 1. Table 1 Summary of bias, precision and total uncertainty for a representative subset of the data. q" Data Set 0.01% SLS 0.46 ml/s Re = 1200 Water 0.46 ml/s Re= 1200 (kResult W/m2) test section. The number of minor losses between the pressure reading and the test section needs to be minimized, and more accurate measurements need to be taken on either side of the test section. The challenge is eliminating or accurately accounting for the losses, since the working fluid must flow through at least one large reduction and make two 90-degree turns into and out of the test section. T~ B P U Result ~c) I*/,) (P)'/.(Uo) Result B [ P U B (°/o} ('/.) I*/.) (kW/m2k) (%) (%) ~%) 212 255 297 343 395 499 622 0.019 0.019 0.019 0.019 0.018 0.018 0.018 3.19 3.14 3.14 3.14 3.14 3.14 3.13 3.19 3.14 3.14 3.14 3.14 3,14 3.13 95.3 90.5 95.3 97.7 100.3 103.0 105.8 3.1 3.3 3.1 3.1 3.0 2.9 2.8 2.18 0.62 0.97 0.82 0.50 0.55 0.27 3.83 3.37 3.29 3.18 3.03 2.96 2.85 33 42 47 52 59 72 87 4.93 5.33 5.10 5.01 4.95 4.76 4.73 4.81 3.40 3,66 3,49 3.61 3.40 3.20 6.88 6.32 6.28 6.10 6A3 5.85 5,71 136 169 200 244 290 333 445 561 624 760 912 0.020 0.020 0.019 0.019 0.019 0.019 0.018 0.018 0.018 0.018 0.018 3.19 3.14 3.13 3.14 3,14 3.16 3.14 3.14 3.13 3.13 3.14 3.19 3.14 3.13 3.14 3.14 3.16 3.14 3.14 3.13 3.13 3.14 60.7 65.1 70.8 82.0 90.3 93.7 99.3 102.9 104.4 108.5 115.0 4.9 4.6 4.2 3.7 3.3 3.2 3.0 2.9 2.9 2.8 2.6 2.93 0,60 0.37 0.57 0.62 1.45 0.67 0.55 0.33 0.26 0,56 5.74 4.65 4.25 3.70 3.38 3.51 3.10 2.97 2.89 2.78 2.67 42 46 48 47 48 53 66 82 90 107 127 9,94 8,80 7.64 6.16 5.35 5.20 4.86 4.95 4.95 4.92 5.21 6.73 3.36 3.22 3.28 3.33 3.95 3.34 3.30 3.21 3.31 4.59 12.01 9.42 8,29 6.98 6.31 6,53 5.90 5.95 5,90 5.93 6.95 CONCLUSION The goal of constructing a low cost apparatus for investigation of flow boiling phenomena in microchannels, was achieved, however further improvements are possible. The small cross section heating element, being only 1.8 cm long, had a very low resistance that was difficult to measure accurately. In addition, the low thermal coefficient of resistance of Platinum, combined with the low resistance, made detecting the changes in resistance even more difficult. It is suggested that a different material be found, with a change to a much larger area. This would approximate the chip/heat sink model, followed by the majority of the researchers in the current literature. This would necessitate the construction of a microchannel heat sink rather than just a single channel. Further investigation is needed into the effect of the surfactant concentrations on the heat transfer and boiling phenomena. In this study, and one other study (Wang and Harnett, 1994) heat transfer was actually reduced by the addition of SLS, while in other studies, the heat transfer was increased. It is possible that above some limiting concentration, the heat transfer is adversely affected by the surfactants, while below that concentration, heat transfer is enhanced. All investigations to date have been unsuccessful in visualizing bubbles in the channel during boiling. The nature of the boiling curve, confirms two-phase heat transfer, but bubbles have been elusive. This apparatus would be ideal used in conjunction with a high-speed camera, under magnification, or with strobe light photographic techniques. Finally, saturation temperature could not be determined due to inaccuracies in calculating the pressure drop across the REFERENCES Adams, T.M. Ghiaasiaan, S.M. Abdelkhalik, S.I., 1999, "Enhancement of Liquid Forced Convection Heat Transfer in Microchannels due to the Release of Dissolved Noncondensibles", Int. J. Heat Mass Transfer, Vol. 42 No. 19, Oct. 1999, pp. 3563-3573. Ammerman, C.N., You, S.M., 1996, "determination of the Boiling Enhancement Mechanism Caused by Surfactant Addition to Water", J. of Heat Transfer, Vol. 118, pp. 429-435. Dittus, F.W. and Boelter, L.M.K, 1930, University of California, Berkely, Publications on Engineering, Vol. 2, p. 443. Dunskus, T., and Westwater, J.W., 1961, "The Effect of Trace Additives on the Heat Transfer to Boiling Isopropanol", Chem. Engng. Prog. Symp. Ser. No. 32, Vol. 57. pp. 173-181. Jontz, P.D., and Meyers, J.E., 1960, "The Effect of Dynamic Surface Tension on Nucleate Boiling Coefficients", AIChE J., Vol. 6, No. 1, pp. 34-38. Kandlikar, S., Alves, L., 1998, "Effects of Surface Tension and Binary Diffusion on Pool Boiling of Dilute Solutions - An Experimental Assessment", Paper presented at the ASME National Heat Transfer Conference, Albuquerque, NM, June. Kandlikar, S., Dhir, V.K., Shoji, M., 1999, "Handbook of Phase Change: Boiling and Condensation", Taylor and Francis, Philadelphia PA. Kochaphakdee, P., and Williams, M.C., 1970, "Enhancement of Nucleate Pool Boiling with Polymeric Additives," Int. J. Heat and Mass Transfer, Vol. 13, pp. 835-848. Koh, J.C.Y. Colony, R., 1974, "Analysis of Cooling Effectiveness for Porous Material in a Coolant Passage", Transaction of ASME Journal of Heat Transfer, pp. 324-330, August. Koh, J.C.Y. Colony, R., 1986, "Heat Transfer of Microstructures for Integrated Circuits", Int. Comm. Heat Mass Transfer, Vol. 13, pp. 89-98. Kline, S.J. and McClintock, F.A., 1953, "Describing Uncertainties in Single Sample Environments", Mechanical Engineering, Vol. 75, pp. 3-8. Lowery, A.J., and Westwater, J.W., 1957, "Heat Transfer to Boiling Methanol - Effect of Added Agents", Ind. And Eng. Chem., Vol. 49, pp. 1445-1448. Malyshenko, S.P., 1994, "Effect of a Curved Pase Boundary on Surface Tension and Kinetics of Nucleation in Liquids", High Temperature, Vol. 32, No. 5, pp. 671-678. 10 Copyright © 2000 by ASME Peng, X.F. Hu, H.Y. Wang, B.X., 1998, "Flow Boiling Through V-Shape Mierochannels", Experimental Heat Transfer, Vol. 11, No. 1, Jan-Mar 1998, pp. 87-90. Peng, X.F. and Wang,B.X., 1993, "Forced Convection and Flow Boiling Heat Transfer for Liquid Flowing Through Microchannels", Int. J. Heat Mass Transfer, Vol. 36, pp. 34213427. Phillips, R. J., 1987, "Forced-Convection Liquid Cooled Microchannel Heat Sinks", Masters Thesis, Massachusetts Institute of Technology. Reihl, R.R. Seleghim, P. Jr. Ochterbeck, J.M., 1998, "Comparison of Heat Transfer Correlations for Single- and Two-Phase Microchannel Flows for Microelectronics Cooling", Thermomechanieal Phenomena in Electronic Systems-Proceedings of the Intersociety Conference 1998. IEEE, Piscataway, N J, USA, pp. 409-416. Roll, J.B., and Meyers, J.M., 1964, "The Effect of Surface Tension on Factors in Boiling Heat Transfer", AIChE J., Vol. 10, No. 4, pp. 530-534. Shaw, B.T., and Darby, R., 1973, "The Effect of Surfactant on Evaporative Heat Transfer in Vertical Film Flow", Int. J. Heat and Mass Transfer, Voh 16, pp. 1889-1903. Sivagnanam, P; Balakrishnan, A. R. and Varma, Y. B. G., 1994, "On the Mechanism of Subcooled Flow Boiling of Binary Mixtures", Int. J. Heat Mass Transfer, Vol. 37, No. 4, pp. 681689. Tuckerman, D.B. and Pease, R.F., 1982, "Optimized Convective Cooling Using Micromachined Structure", Journal of Electrochemical Society, Vol. 129, No.3, p. C98. Tuckerman, D. B., 1984,"Heat Transfer Microstructures for Integrated Circuits", Ph.D. Dissertation, Stanford University. Tuekerman, D.B. and Pease, R.F., 1991, "High Performance Heat Sinking for VLSI", IEEE Electronic Device Letters, EDL2, pp. 126-129. Tzan, Y.L., and Yang, Y.M., 1990, "Experimental Study of Surfactant Effects on Pool Boiling Heat Transfer," Journal of Heat Transfer, Vol. 112, pp. 207-212. Wang, C.H. and Harnett, 1994, "Pool Boiling of Heat Transfer from a Horizontal Wire to Aqueous Surfactant Solutions", Heat Transfer 1994, Proceedings of the 10th International Heat Transfer Conference, Brighton, Institute of Chemical Engineers/Institute of Mechanical Engineers, pp. 177-182. Wang, B.X. and Peng, X.F., 1994, "Experimental investigation on Liquid Forced-Convection Heat Transfer Through Microchannels", Int. J. Heat Mass Transfer, Voh 37, pp. 73-82. Wozniak, G., Wozniak, K., and Berglet, H., 1996, "On the Influence of Buoyancy on the Surface Tension Driven Flow Around a Bubble on a Heater Wall", Experiments in Fluids, Voh 21, pp. 181-186. Yang, Y.M., and Maa, J.R., 1983, "Pool Boiling of Dilute Surfactant Solutions", Journal of Heat Transfer, Trans. ASME, Vol. 105, pp. 190-192. Yin, X. Bau, H.H., 1997, "Uniform Channel Micro Heat Exchangers", Journal of Electronic Packaging, Vol. 119, pp. 89-94. Zhuang, Y. Ma, C.F. Qin, M., 1997, "Experimental Study on Local Heat Transfer with Liquid Impingement Flow in TwoDimensional Micro-channels", Int. J. Heat Mass Transfer, Vol. 40, No. 17, pp. 4055-4059. 11 Copyright © 2000 by ASME