Experimental study of the effects of structured surface geometry on water spray cooling performance in non-boiling regime

Abstract

Experiments were conducted to study the effects of enhanced surfaces on heat transfer performance during water spray cooling in non-boiling regime. The surface enhancement is straight fin. The structures were machined on the top surface of heated copper blocks with a cross-sectional area of 10 mm×10 mm. The spray was performed using Unijet full cone nozzles with a volumetric flux of 0.044–0.053 m³/(m²·s) and a nozzle height of 17 mm. It is found that the heat transfer is obviously enhanced for straight fin surfaces relative to the flat surface. However, the increment decreases as the fin height increases. For flat surface and enhanced surfaces with a fin height of 0.1 mm and 0.2 mm, as the coolant flux increases, the heat flux increases as well. However, for finned surface with a height of 0.4 mm, the heat flux is not sensitive to the coolant volumetric flux. Changed film thickness and the form of water/surface interaction due to an enhanced surface structure (different fin height) are the main reasons for changing of the local heat transfer coefficient.

Keywords

spray cooling / finned surface / heat transfer

Introduction

Many emerging technologies with higher transistor integration densities, such as microwave systems, defense laser, modern electronics and power devices, have increased the chip level heat fluxes to more than 100 W/cm² [1]. As a traditional cooling technology, air cooling cannot adequately satisfy such high-flux heat removal and will have to be replaced or supplemented by other effective cooling solutions. Alternative cooling technologies, involving jet impingement, micro-channel flow, and spray cooling, are receiving greater attention from researchers. Of the above cooling technologies, spray cooling may be the best competing option for thermal management of high heat flux systems, which could provide high heat flux in excess of 1000 W/cm² with water as coolant while allowing uniformity of heat removal at small fluid inventory [2]. Despite those advantages, spray cooling cannot be widespread in industrial applications because of poor understanding of the underlying mechanism and the key parameters that influence cooling performance.

Most of spray-cooling studies have been performed on the influence of physical as well as geometrical parameters on cooling heat flux [2,3]. Both physical and geometrical parameters affect spray-cooling performance [4-8]. The physical parameters include mean droplet size, droplet flux, droplet velocity, volumetric flux, etc., while the geometrical parameters are composed of spray cone angle, orifice-to-surface distance, surface type and inclination angle. Besides, the subcooling of coolant and the nozzle type also influence cooling performance.

Of the above factors, geometrical parameters, especially the surface type, are the least of the concerns by researchers and, therefore, has the least literature. Existing studies and literature mainly focus on surface roughness and enhanced structure effects on CHF. Slik et al. [9] have studied the impact of structured surface geometry on spray cooling in boiling regime. The surface enhancements consist of cubic pin fins, pyramids and straight fins. It is found that each of these surfaces increases in evaporation efficiency at CHF compared to the flat surface. Sodtke and Stephan [10] have studied the spray cooling on microstructured surfaces (micropyramids with different heights) and have found that the heat flux is enhanced due to an increase in the three-phase contact line. Coursey et al. [11] have studied the spray cooling of high aspect ratio open microchannels with a fin length of 0.25, 0.5, 1, 3 and 5 mm and have found that, in the single-phase regime, longer fins have a better performance. The relative performance enhancement decreases with the increase of fin length, indicating that the optimal fin length would be only slightly longer than 5 mm for 0.35 mm in width and 0.5 mm in fin width channels. Kim et al. [12] have studied the spray cooling of plain and microporous coated surfaces. However, the above studies focus only on evaporative spray cooling, and the conclusions are only appropriate for high temperature. Non-boiling regime cooling performances, as a part of considerable importance in the whole spray cooling, intrigued very few researchers.

The present study is to examine the effects of structured surface geometries on the spray-cooling performance in non-boiling regime. Experimental results illustrate that enhanced surfaces can provide significantly larger heat transfer relative to smooth surfaces. Three enhanced surface geometries (straight fins with height 0.1, 0.2 and 0.4 mm respectively) have been tested to determine heat flux as a function of surface geometry, surface temperature and coolant volumetric flux with distilled water as the coolant.

Experimental apparatus and spray parameters

The experimental apparatus used in the present study consists of a spray system, a data acquisition system and a heater source, as shown in Fig. 1. The experimental setup provides the opportunity to vary the surface heat flux, the water mass flow rate, the droplet diameter and velocity, and the distance between the orifice and the surface as well.

In order to maintain the atmospheric pressure, an open spray system is devised, and the liquid coolant is supplied from a low-temperature sink which is used prior to heat and cool the coolant to the desired nozzle inlet temperature. A variable-speed, magnetically-coupled centrifugal pump which can supply the maximal pressure of 1080 kPa and the maximal flow rate of 4 L/min is used to circulated the fluid. The pumped liquid first passes through a filter, to remove any entrained impurities, followed by rotameters, a pressure transducer and a temperature transducer, where rate flow, tube pressure and nozzle inlet temperature are measured, respectively. The fluid then enters into the nozzle and is sprayed downwards. After the heat exchange, the fluid is drained to the lower portion of a large reservoir by gravity. Finally, it passes through the heater exchange to bring the surplus heat by the tap water and return to the low-temperature sink.

The heat is supplied to the 10 mm×10 mm test surface by five 300-W cartridge heaters embedded in an oxygen-free copper, as show in Fig. 2. To minimize the heat loss, all surfaces of the copper block are insulated except the test surface. Before fabricating, the heater is numerically studied using a three-dimensional heat diffusion model in FLUENT. It is found that the heater exhibits a good one-dimensional heat transfer characteristic, as shown in Fig. 3. Six type-K (Chromel–Alumel) thermocouples with a wire diameter of 0.127 mm and a bead diameter of 0.25 mm are embedded 10 mm below the test surface to measure the surface temperature. Knowing the distance between the thermocouples and that between the thermocouple and the heater surface, it is possible to predict the heat flux and the surface temperature using the one-dimensional Fourier law of heat conduction.

The data acquisition system includes an Agilent 34970A digital acquisition/control system and a personal computer which records and processes the signal from the thermocouples, the temperature and pressure from the sensors in the loop.

The experiments are performed using Unijet full cone nozzles supplied by Spraying System Company. The key spray parameters measured by PDPA are listed in Table 1. The droplet flux is calculated by

(1)N={\displaystyle \frac{\mathrm{6}Q}{\mathrm{\pi }{d}_{\mathrm{32}}^{\mathrm{3}}}}.

In the course of the experiment, a regular operating procedure must be observed to ensure the consistency between tests corresponding to different operating conditions. Before the experiments, the distilled water is pumped into the pipe without the nozzle to remove impurities adhering to the pipe, and the test smooth surface is polished with 240-grit sandpaper and then cleaned with acetone to remove oxides and residues.

To achieve the optimum orifice-to-surface distance where the system has maximum heat flux, all the experiments are conducted when the spray is configured such that the impact area just inscribes a square test surface. Experiment curves are produced by raising the electrical power input to the heater in small increments and by recording the heat flux from the test surface. The surface temperature is assumed to be a steady state when the change of the surface temperature is less than 0.1°C within 20 min. The steady state for each run is obtained in about 1 to 4 hours.

Enhanced surfaces

The dimensions of the enhanced surface geometries studied are the straight-finned surface, as shown in Fig. 4(b), while the photograph of the enhanced surface is shown in Fig. 4(a). a, c, and b are the structure height, width and channel width, respectively. To study the enhanced surface geometry effects on spray cooling performance, experimental data of three kinds of structure heights (a=0.1, 0.2 and 0.4 mm) are presented to compare with that of flat surface (no fins present).

Results and discussion

In the experiment, the main variable observed is the temperature of heated elements. Both the surface temperature and heat flux are calculated using those temperature readings assuming one-dimensional heat transfer and using Fourier law.

(2)q^{″}=k{\displaystyle \frac{\mathrm{\Delta }T}{\mathrm{\Delta }x}},

where k is the thermal conductivity of the heat plate, ΔT is the temperature difference between the thermocouple levels, and Δx is the distance between the thermocouples. The average surface temperature on the hot surface T_wall can be obtained by

(3)T_{\mathrm{w}\mathrm{a}\mathrm{l}\mathrm{l}}=T_{\mathrm{1}m}-{\displaystyle \frac{q^{″}\mathrm{\Delta }{x}_{\mathrm{1}-\mathrm{w}\mathrm{a}\mathrm{l}\mathrm{l}}}{k}},

where T_1m is the arithmetic means of the temperatures indicated by the two thermocouples at the upper plane, and Δx_1-wall is the distance between the first thermocouple and the test surface. The average heat transfer coefficient is a very important parameter which can be calculated by

(4)h={\displaystyle \frac{q^{″}}{T_{\mathrm{w}\mathrm{a}\mathrm{l}\mathrm{l}}-T_{\mathrm{s}\mathrm{p}\mathrm{r}\mathrm{a}\mathrm{y}}}}.

All heat flux data are based on the projected area of 1cm², instead of the wetted area s listed in Table 2.

Heat flux as a function of surface temperature and water volumetric flux are illustrated in Fig. 5. It is found that, for flat surface (Fig. 5(a)), as the volumetric flux increases, the heat transfer increases. The heat transfer mechanism in the non-boiling regime mainly includes forced convection and film evaporation. It is known from Table 1 that the augmentation of the operating pressure resulted in the aggrandizement of the volumetric flux, the velocity of droplets and droplet flux. At higher droplet velocity, the effect of washout becomes violent, and the heat transfer increases. However, at higher volumetric flux, a thicker liquid film is formed on the target surface; therefore, the heat transfer due to the film evaporation reduces. Because of the simultaneity of the two effects mentioned above and the inability to control drop velocity and volumetric flux, it is difficult to study the influence of a certain parameter on heat transfer. Mudawar and Valentine [6] argue that the volumetric flux, not the droplet velocity, is the dominant factor for heat transfer performance. Larger droplet flux can cause two enhancement effects: ① more droplets arriving at the heater surface bring about enhancement of the mixing effect; ② the discrete and random nature of larger droplet disputes the steady boundary layer of the liquid film and augments the disturbance. On the other hand, higher droplet velocity leads to the turbulence in the film. Thus, the combined effect of mixing, disrupting the boundary layer and the turbulence in the liquid film causes enhancement in the heat transfer rate.

For enhanced surface with a height of 0.1 mm and 0.2 mm (Figs. 5(b) and (c)), as the volumetric flux increases, the heat transfer increases as well, which shows that the dependence of heat transfer enhancement on fin height is a function of volumetric flux. However, for finned surface with a height of 0.4 mm, it seems that the heat flux is not sensitive to water volumetric flux. Spray cooling has been observed to be a convective boiling process that is largely dominated by single-phase convection, especially in no-boiling regime. The enhanced structures used in this study can be considered as finned heat sinks. Addition of finned heat sinks to convectively cooled surfaces is known to decrease the single-phase convective thermal resistance to heat transfer by increasing the total wetted area. If the heat flux were to scale with the total wetted surface area, then it would be expected that the channel with a height of 0.4 mm would have the highest increment of heat flux, followed by channels with a height of 0.2 mm and 0.1 mm, respectively. However, the data from Fig. 5(d) indicate that the heat transfer does not scale directly with the total wetted surface. This is due to the fact that the addition of enhanced surface geometries greatly modifies the nature of the fluid/solid contact and, thus, fluid motion. If the channel is too high, the fin ribs separate the water film, and the water film will be much thicker inside the channel. Even if the fin doesn’t separate the flow, for a certain flow rate, as fin height increases, the flow velocity in microchannels decreases; therefore, the lower heat transfer could be expected.

Heat flux as a function of surface temperature and channel height is rearranged and presented in Fig. 6. Compared with smooth surface, it is found that the heat transfer is distinctly enhanced for structured surfaces. For the case of the surface temperature of 50°C, water volumetric flux of 0.044 m³/(m²·s), with a fin height of 0.1, 0.2 and 0.3 mm, the heat flux increases by about 39%, 53%, and 31.6%, respectively. When the volumetric flux increases to 0.053 m³/(m²·s), for a fin height of 0.2 mm (channel 2), the maximum increment will be 67% compared with smooth surface data, as shown in Table 3. The heat transfer increment decreases for a fin height of 0.4 mm (channel 3). It is likely that the surface with a fin height of 0.2 mm gets the best heat transfer performance.

As mentioned above, the higher the fin height, the bigger the total wetted surface. For channel 1 (0.1-mm fin height), compared with smooth surface, the wetted surface increment is about 38%, while the heat transfer increment is about 39%-41%. The heat transfer is roughly proportional to the total wetted surface. When the fin height enlarges, the heat transfer increment is not as big as the wetted surface. For channel 2 (0.2 mm fin height), the wetted surface increment is about 76%, while the heat transfer increment is about 53%-72%. As the fin height becomes 0.4 mm, the wetted surface increment is about 152%, while the heat transfer increases 18%-31.6% only. This is due to the fact that for 0.1 and 0.2 fin surface, the enhanced structure offers the water film, and part of the film is thinned due to gravity and surface tension. The thinner or bigger the area of the film, the higher the heat transfer. The enhanced mechanism may be governed by disrupted film or enlarged film area. When the fin height becomes 0.4 mm, however, as the volumetric flux increases, the heat transfer increment decreases, which is quite different from the cases of 0.1 and 0.2 mm. The local heat transfer coefficient may decrease for this kind of enhanced structure. On the other hand, as the fin height increases, the local temperature difference between the wall surface and the fluid is also changed, which maybe the other reason for the decrease of the heat transfer increment.

To further know this, let’s check the effect of enhanced structure on heat transfer coefficient. Assuming that the enhanced surface is the combination of the smooth surface and the cuboid rib, the average heat transfer coefficient of the fin rib can be written as [13]

(5)h={\displaystyle \frac{Q}{A_{\mathrm{b}}{\theta }_{\mathrm{b}}+n{A}_{\mathrm{f}}{\theta }_{\mathrm{m}}}}={\displaystyle \frac{q{A}_{\mathrm{w}\mathrm{a}\mathrm{l}\mathrm{l}}}{A_{\mathrm{b}}{\theta }_{\mathrm{b}}+n{A}_{\mathrm{f}}{\theta }_{\mathrm{m}}}},

where θ is the temperature difference between the fin surface (t) and the coolant (t_L):

(6)\theta =t-t_{\mathrm{L}},

θ_b is the temperature at the fin rib bottom (near smooth surface) , and θ_m is the average value of θ

(7){\theta }_{\mathrm{m}}={\displaystyle \frac{\mathrm{1}}{a^{\prime }}}{\displaystyle {\int }_{\mathrm{0}}^{a^{\prime }}{\theta }_x}dx,

where n is the number of the ribs (in this study, it is 18), A_b is the total area between the rib, A_f is the area of each rib for heat convection, x is the distance along the height from the rib bottom, anda^{\prime }=a+c/\mathrm{2}.

The relationship between θ_b and θ_x can be written as [13]:

(8){\displaystyle \frac{{\theta }_{\mathit{x}}}{{\theta }_{\mathrm{b}}}}={\displaystyle \frac{\cosh [m(a^{\prime }-x)]}{\cosh (m{a}^{\prime })}},

where m=(hp/λA_c), P is the perimeter of the rib cross section, λ is the conductivity of the rib, and A_c is the top area of the rib; thus

(9){\theta }_{\mathrm{m}}={\displaystyle \frac{\tanh (m{a}^{\prime })}{a^{\prime }m}}{\theta }_{\mathrm{b}}\mathrm{.}

The relationship between θ_m and θ_b can be calculated and shown in Table 4. It is found that as the fin height increases, θ_m/θ_b decreases slightly.

The average heat transfer coefficient as a function of surface temperature for different channel height and volumetric flux 0.044 and 0.049 m³/(m²·s) is shown in Fig. 7. It is found that enhanced surface enlarges not only the wetted area but also the average heat transfer coefficient when the fin height is 0.1 and 0.2 mm. It is because the enhance structure enlarges the film area due to liquid surface tension. The larger the film area, the thinner the film. Hence, smaller heat transfer resistance could be achieved [9, 10]. Among the smooth surface and three enhanced surfaces, the 0.1-mm fin structure gets the best heat transfer coefficient, followed by the 0.2-mm fin structure and the smooth surface, while the 0.4-mm enhanced surface gets the worst heat transfer performance. The relative performance enhancement decreases with the increase of the fin length, indicating that the optimal fin height is about 0.2 mm. As the fin height is 0.4 mm, the fin rib separates the flow. Due to surface tension and fluid viscosity, the flow in the separated channel is not as quick as that on the smooth surface. The local heat transfer coefficient will decrease because of its lower fluid velocity and thicker water film. It could be concluded that if the enhanced surface enlarges and thins the film, the heat transfer coefficient will be enhanced. If the enhanced surface separates the film (reduced the film area) or blocks the flow (lower velocity), the heat transfer will decrease. Therefore, it is easy to understand why the surface with a fin height of 0.1 mm gets the biggest heat transfer coefficient, while the surface with a fin height of 0.4 mm gets the worst heat transfer performance, and its average heat transfer coefficient is even smaller than that of the smooth surface.

Figure 7 also shows that the heat transfer coefficient increases slightly with the wall temperature. In other words, the heat transfer coefficient changes with the surface temperature. It is quite different from other ways of heat transfer where the heat transfer coefficient is irrelevant to the wall temperature and mainly depends on the properties and the motion state of the coolant. Such abnormal phenomena can be attributed to the evaporation of the water film when the wall temperature increases. However, the conclusion is inconsistent with the study by Oliphant [14] who ignored the evaporation of the liquid film.

Conclusion

Experiments were conducted to study the effects of straight fin surfaces and volumetric flux on heat transfer during water spray cooling in non-boiling regime. The results show that the spray cooling on enhanced surfaces can significantly improve the cooling performances compared to smooth surfaces at the same spray parameters and wall condition. For the case of surface temperature of 50°C, water volumetric flux of 0.044 m³/(m²·s) with a fin height of 0.1, 0.2 and 0.3 mm, the heat flux increases by about 39%, 53%, and 31.6%, respectively. When the volumetric flux increases to 0.053 m³/(m²·s), for a fin height of 0.2 mm, the maximum increment will be 67% compared with that of the smooth surface. It is found that the heat transfer does not scale directly with the total wetted surface. The relative performance enhancement decreases with the increase of fin length, indicating that the optimal fin height is about 0.2 mm. Different fin heights change the film/surface interaction. For enhanced surfaces with a fin height of 0.1 and 0.2 mm, the spray produces a thinner film on the surface through which conduction occurs. The thinner or the bigger the film, the higher the heat transfer coefficient. However, for the case of enhanced surfaces with a fin height of 0.4 mm, the water film cannot be as thick as the fin height; thus, compared with the smooth surface and the surface with a fin height of 0.1 and 0.2 mm, the water/surface interaction changes. This may be the main reason for the fact that the local heat transfer coefficient decreases.

References

[1] Amon C H, Murthy J, Yao S C, Narumanchi S, Wu C-F, Hsieh C-C. MEMS-enabled thermal management of high-heat-flux devices EDIFICE: Embedded droplet impingement for integrated cooling of electronics. Experimental Thermal and Fluid Science, 2001, 25(5): 231-242

[2] Rini D P, Chen R-H, Chow L C. Bubble behavior and nucleate boiling heat transfer in saturated FC-72 spray cooling. Journal of Heat Transfer, 2002, 124(1): 63-72

[3] Chen R H, Chow L C, Navedo J E. Effects of spray characteristics on critical heat flux in subcooled water spray cooling. International Journal of Heat and Mass Transfer, 2002, 45(19): 4033-4043

[4] Chen R H, Chow L C, Navedo J E. Optimal spray characteristic in water spray cooling. International Journal of Heat and Mass Transfer, 2004, 47(23): 5095-5099

[5] Estes K A, Mudawar I. Correlation of Sauter mean diameter and critical heat flux for spray cooling of small surfaces. International Journal of Heat and Mass Transfer, 1995, 38(16): 2985-2996

[6] Mudawar I, Estes K A. Optimizing and predicting CHF in spray cooling of a square surface. Journal of Heat Transfer, 1996, 118(3): 672-679

[7] Cabrera E, Gonzalez J E. Heat flux correlation for spray cooling in the nucleate boiling regime. Experimental Heat Transfer, 2003, 16(1):19-44

[8] Rybicki J R, Mudawar I. Single-phase and two-phase cooling characteristics of upward-facing and downward-facing sprays. International Journal of Heat and Mass Transfer, 2006, 49(1): 5-16

[9] Silk E A, Kim J, Kiger K. Spray cooling of enhanced surfaces: Impact of structured surface geometry and spray axis inclination. International Journal of Heat and Mass Transfer, 2006, 49(25,26): 4910-4920

[10] Sodtke C, Stephan P. Spray cooling on micro structured surfaces. International Journal of Heat and Mass Transfer, 2007, 50(19,20): 4089-4097

[11] Coursey J S, Kim J, Kiger K T. Spray cooling of high aspect ratio open microchannels. Transactions of the ASME, 2007, 129(8): 1052-1059

[12] Kim J H, You S M, Choi S U S. Evaporative spray cooling of plain and microporous coated surfaces. International Journal of Heat and Mass Transfer, 2004, 47(14-16): 3307-3315

[13] Incropera F P, Dewitt D P, Bergman T L, Lavine A S. Fundamentals of Heat and Mass Transfer. 6th ed. John Wiley & Sons, Inc., 2007, 90-103

[14] Oliphant K, Webb B W, McQuay M Q. An experimental comparison of liquid jet array and spray impingement cooling in the nonboiling regime. Experimental Thermal and Fluid Science, 1998, 18(1): 1-10

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg