Critical Heat Flux Measurement and Model for Refrigerant-123 under Stabilized Flow Conditions in Microchannels

The present work is aimed toward understanding the effect of flow boiling stability on critical heat flux (CHF) with Refrigerant-123 (R-123) in microchannel passages. Experimental data and theoretical model to predict the CHF are the focus of this work. The experimental test section has six parallel microchannels with each having a cross sectional area of 1054 × 157 μm 2 . The effect of flow instabilities in microchannels is investigated using flow restrictors at the inlet of each microchannel to stabilize the flow boiling process and avoid the backflow phenomena. This technique resulted in successfully stabilizing the flow boiling process as seen through a high-speed camera. The present CHF result is found to correlate to mean absolute error (MAE) of 24.1% with a macroscale empirical equation by Katto [13]. A theoretical analysis of flow boiling phenomena revealed that the ratio of evaporation momentum to surface tension forces is an important parameter. For the first time, a theoretical CHF model is proposed using these underlying forces to represent CHF mechanism in microchannels, and its correlation agrees with the experimental data with MAE of 2.5%. INTRODUCTION Advancements in microprocessors and other high power electronics have resulted in increased heat dissipation from those devices. In addition, to reduce cost, the functionality of microprocessor per unit area has been increasing. The increase in functionality accompanied by reduction in chip size has caused its thermal management to be challenging. In order to dissipate the increase in heat generation, the size of conventional fin-type heat sinks has to be increased. As a result, the performance of these high heat flux generating electronics is often limited by the available cooling technology and space to accommodate the larger conventional air-cooled heat sinks. One way to enhance heat transfer from electronics without sacrificing its performance is the use of heat sink with many microchannels and liquid water passing through it. Because of the small size of microchannel heat sink, the performance of a computer system can also be increased by incorporating more microprocessors at a given space without the issue of overheated or burned-out chips. Flow boiling in microchannels is being studied worldwide because of its potential in high heat flux cooling. Comparing to single-phase flow, flow boiling is advantageous because it utilizes the heat of vaporization of a working fluid. Because of that, given a mass flow rate, the heat flux in flow boiling is much higher than that of single phase. In addition, flow boiling in microchannel heat sink can provide approximately uniform fluid and solid temperatures, and it can also be directly coupled with a refrigerant system to provide a lower coolant temperature. In designing a two-phase microchannel heat sink, it is necessary to know its critical heat flux (CHF). This is because CHF determines the upper thermal limit on the microchannel operation, and the rapid rise in operating temperature after CHF is detrimental to electronics. That is why CHF data and a good understanding of CHF in microchannels are needed before the application of two-phase microchannel heat sink can be implemented. Furthermore, very few experimental CHF data have been reported in microchannels. Hence, the objective of the present work is to experimentally investigate the CHF of saturated flow boiling in microchannels using R-123 as the working fluid. 1 Copyright © 2006 by ASME The present experiment involves the collecting of CHF data over the ranges of mass flux and heat flux supplied to the microchannels. For the first time, a theoretical CHF model is proposed using these underlying forces to represent CHF mechanism in microchannels. The predicted results from the model are then compared to the present experimental CHF data. Similarly, the CHF model is also used to predict Qu and Mudawar’s [1] water CHF data, and the predicted results are compared to their experimental CHF data. In their experiment, Qu and Mudawar obtained the CHF data using 21 parallel channels with each channel having a cross-sectional area of 215 × 821 μm 2 . The operating conditions from Qu and Mudawar’s and present experiments can be found in Table 1. Table 1 Operating conditions CHF Data by Fluid Operating conditions G (kg/m 2 s); q" (kW/m 2 ); x; Tin (°C); Pin (kPa) Qu and Mudawar [1] Water G = 86-368; q" = 264.2-542.0; x = 0-0.56; Tin = 30; Pin = 121.3-139.8 present work R-123 G = 410.5-533.8; q" = 136.3-201.3; x = 0.79-0.93; Tin = 17.2; Pin = 162.8-248.3 NOMENCLATURE b – height of microchannel, m C – constant in present CHF model CHF – critical heat flux D – characteristic dimension, m d – diameter of circular channel, m FI – force due to inertia, N FM – force due to momentum change, N FS – surface tension force, N F' – force per unit length, N/m G – mass flux, kg/m 2 s I – electrical current, A MAE – mean absolute method P – pressure, kPa ∆P – pressure drop, kPa PDE – pressure drop element q" – heat flux, kW/m 2 CHF q ′ ′ – critical heat flux, kW/m qin – power input to the test section, W qloss – heat loss from the test section, W T – temperature, °C ∆TAmb– differential temperature for use in performing test section heat loss calibration, °C Ts – surface temperature, °C V – average velocity based on average density, ρ , m/s We – Weber number, (G 2 D)/(ρσ) x – thermodynamic quality Greek Symbols ρ – density, kg/m 3 ρ – average density, kg/m3 θ – dynamic receding contact angle, degrees σ – surface tension, N/m Subscripts amb – ambient CHF – critical heat flux, kW/m 2 exp – experimental G, g – gas or vapor I – inertia in – inlet L, f – liquid loss – unrecoverable loss M – due to momentum change pred – predicted s – microchannel surface S – surface tension LITERATURE REVIEW Because of the limited number of investigations on CHF in microchannels, experimental studies related to both minichannels and microchannels will be reviewed. Minichannels cover the range from 200 μm to 3 mm channel diameter. Bowers et al. [2] experimentally studied CHF in circular channels with diameters of 2.54 mm and 0.510 mm using R-113 as the working liquid. The heated length of the channels is 10 mm. In their experiment, CHF is found to be independent of the inlet subcooling at low flow rates due to fluid reaching the saturation temperature in a short distance into the heated channels. Roach et al. [3] used uniformly heated channels to experimentally investigate CHF. The four different channels, all 160 mm in length, are: two circular with 1.17 mm and 1.45 mm diameter, and two other flow channels in microrod bundle with a triangular array and 1.131 mm hydraulic diameter. One of the microrod bundles is uniformly heated over its entire surface and the other is heated only over the surfaces of the surrounding rods. The authors found that the CHF occurs at high flow quality of 0.36 and higher, indicating dry-out as the CHF mechanism. In addition, the CHF increases with increasing mass flux and pressure, and depend on channel diameter. 2 Copyright © 2006 by ASME Table 2 Summary of studies on CHF in small channels Author/ year Fluid Operating conditions G (kg/m 2 s); q" (kW/m 2 ); x; Tin (°C); Pin (kPa) Channel geometry (mm) Remarks Bowers and Mudawar 1994 [2] R-113 G = 10-490; q" = 200-2000; x = 0-1 Circular, multi-channels, d = 2.54 and 0.51, horizontal Higher CHF can be obtained in microchannels, and presented correlation for CHF. Roach et al. 1998 [3] Water G = 250-1000; q" = 860-3698; x = 0.36-1.2; Tin = 49 to 72.5; Pin = 407-1204 Circular, d = 1.17 and 1.45; microrod bundle with a triangular array, Hydraulic diameter = 1.131, vertical CHF occurs at high flow quality between 0.36 and higher, indicates dry-out; CHF increase with increasing mass flux or pressure, and depends on channel diameter. Jiang et al. 1999 [4] Water P = 50-320 V-grooved, Hydraulic diameter = 0.04 and 0.08, horizontal CHF condition depends on the flow rate and the channel size. Yu et al. 2002 [5] Water G = 50-200; q" = 20-320; x = 0.5-1.0; Tin = ambient to 80; Pin = 200 Circular, d = 2.98, horizontal CHF occurs at high quality between 0.51.0; CHF quality decrease with decreased mass flux. Qu and Mudawar 2003 [1] Water G = 86-368; q" = 264.2-542.0; x = 0-0.562; Tin = 30 and 60; Pin = 121.3-139.8 Rectangular, 0.215 × 0.821, 21 parallel channels, horizontal Flow instability greatly amplified nearing CHF; CHF increases with mass flux; A new CHF correlation is proposed. Koşar et al. 2005 [6] Water G = 41, 83, 166 and 302; q" = 280-4450; x = 0-0.9; ∆P = 0.4, 0.8, 1.7 and 3.0. Rectangular, Hydraulic diameter = 0.223, 5 parallel channels, horizontal CHF increases with mass flux and decreases with exit quality; CHF data correlated well with conventional correlation; A new CHF correlation is proposed. Jiang et al. [4] investigated the CHF condition in diamondshaped channels with hydraulic diameter ranging from 0.04 mm to 0.08 mm using water as the working fluid. The authors suggest that the evolution of the phase change from liquid to vapor in microchannels is different from conventional channels. They found that the CHF condition depends on the flow rate and the channel size. The authors speculated that in such small channels, bubble formation may be suppressed and recommended flow visualization studies to determine the governing heat transfer mechanism. Yu et al. [5] found that CHF occurs at high flow quality between 0.5 and 1.0 for water, and such qualities are higher than those found in larger diameter tubes at higher pressures and mass fluxes. The CHF quality was found to decrease with decreasing mass flux, and this trend is opposite to the one found in larger tubes. Their experiments were performed using a horizontal tube with 2.98 mm inside diameter and 910 mm heated length. Qu and Mudawar [1] measured CHF for a water-cooled heat sink containing 21 parallel 0.215 mm x 0.821 mm channels. The authors found that flow reversal caused by flow instabilities have resulted in a CHF independent of inlet temperature but which increases with increasing mass velocity. Koşar et al. [6] found that CHF increases with mass flux and decreases with vapor mass fraction at the exit. The summary of the above literature review can be found in Table 2. EXPERIMENTAL FACILITY The experimental setup developed by Kuan and Kandlikar [7] is used in the present work. The experimental setup is designed to provide R-123 at a constant flow rate and temperature to the test section. The inputs to the test section are the working fluid (R-123) and the converted heat energy from the supplied electric current. The outputs from the test section are the heated working fluid and the heat loss from the test section. 3 Copyright © 2006 by ASME Test Section Design Microchannels are fabricated on a copper block. The copper is an Electrolytic Tough Pitch alloy number C11000 which is 99.9 percent copper and 0.04 percent oxygen (by weight). It has a thermal conductivity of 388 W/mK at 20°C. An optically clear glass (fused silica) is then placed on top of the copper block to serve as a transparent cover through which flow patterns can be observed. The glass cover has a thermal conductivity of 1.3 W/m·K. Figure 1 shows a schematic of the copper block with the glass cover. The resistive cartridge heater provides a uniform heat flux to the copper block. The length and width of the copper block are 88.9 mm × 29.6 mm. The thickness of the copper block and the glass cover are 19.1 mm and 12.3 mm respectively. The cross section of each of the microchannels measures 1.054 mm × 0.157 mm, and their edge to edge spacing is 0.589 mm. The length of each microchannel is 63.5 mm. There are a total of six microchannels on the copper block. The glass cover is being held onto the copper block by ten mounting screws, and the force provided by the screws and a thin layer of vacuum grease are enough to seal the microchannels from the ambient environment. The assembled test section is shown in Fig. 2. The mounting screws are secured onto the phenolic plate that is placed on the bottom of the copper block. The phenolic plate also acts as an insulating layer on the bottom surface of the copper block. It has the same length and width as the copper block, but its thickness is 12.7 mm. It is a laminate of paper and has a thermal conductivity of 0.2 W/m·K. The copper block is cleaned in an ultrasonic bath using water before it is assembled with the glass cover and the phenolic plate. Two layers of six thermocouples each are placed into the sides of the copper block along the length of the microchannels. The thermocouple layers are 3.18 mm apart from each other, and the top layer is placed at 3.18 mm below the top surface of the microchannels. The thermocouples are inserted into the copper block until it reaches half the width of the copper block. The thermocouple layers are inserted into the copper block from the opposite directions. The thermocouples from both layers are placed at the same locations along the length of the microchannels. The locations, as measured from the inlet of the microchannels and along the its length are 6.35 mm, 19.05 mm, 25.40 mm, 38.10 mm, 44.45 mm, and 57.15 mm [8]. Figure 1 Schematic of copper block with glass cover Figure 2 Test section assembly To reduce heat transfer in the manifold, the inlet manifold is machined into the glass cover and the working fluid is delivered at the very beginning of the microchannels as shown in Fig. 3. Only the pressure drop across the inlet and outlet manifolds is measured because the actual pressure drop in the microchannels could not be easily measured. Glass Cover