A Flat Plate Heat Pipe with Screen Meshes for the Cooling of Electronic Components 0th IHPS, Taipei, Taiwan, Nov. 6-9, 20 Frédéric Lefèvre a, Jean-Baptiste Conrardy a, Martin Raynaud b and Jocelyn Bonjour a a Université de Lyon, CNRS INSA-Lyon, CETHIL, UMR5008, F-6962, Villeurbanne, France Université Lyon, F-69622, France b Thales Communications, Thales, 9270 Colombes, France Tel: (33) 4 72 43 82 5 ; Fax: (33) 4 72 43 88 e-mail: frederic.lefevre@insa-lyon.fr ABSTRACT Experimental investigations of a flat plate heat pipe (FPHP) with two screen mesh layers as capillary structure are presented. The heat pipe is made of copper and filled with methanol. Water heat exchangers are used as heat sinks but also aluminum radiators cooled by natural convection of air in order to show the performance of the heat pipe in experimental conditions representing a practical case. The proposed application is the cooling of several electronic components mounted on a printed circuit board. The technological solution chosen to manufacture the FPHP is very simple and cheap. Experimental results are compared to an analytical thermal and hydrodynamic model in order to estimate the physical properties of the capillary structure, i.e. its equivalent thermal conductivity and its permeability. Both experimental parameters are introduced in the model to estimate the performance of the heat pipe in a real application with several electronic components. Keywords: flat plate heat pipe; screen mesh layers; grooves; experimental study; electronic cooling. INTRODUCTION The interest in the use of heat pipes for thermal management of electronic components is recognized for both aerospace and terrestrial applications, especially with increasing heat flux requirements. In many applications, heat pipes are circular and are used to transport heat from one single heat source to one single heat sink. Flat plate heat pipes (FPHP) have the same function, but offer a wide cross-section, which allows reducing their thickness without reducing their thermal performance. They are particularly interesting when several heat sources have to be cooled, for examples several electronic components on an electronic card. The thermal performance of a FPHP depends mainly on the capillary structure, which is usually made of grooves, meshes or sintered powder. FPHP can also be realized using several micro heat pipes placed parallel one to another at a small distance. Lots of theoretical works have been published on flat plate heat pipes with rectangular grooves or also on micro heat pipes arrays. Indeed, with these simple capillary structures, it is possible to calculate the shape of the liquid-vapor interface at each point of the capillary structure. Thus, an accurate description of the physical mechanisms involved in the heat pipe can be obtained. Furthermore, some recent experiments were performed on transparent FPHP, in which the liquid-vapor interface was measured along the grooves. As a result, both theoretical and experimental approaches can be compared [,2]. For more complex capillary structures such as meshes or sintered copper wick, a fine description of the liquid-vapour interface inside the porous media is difficult to model and there is no measurement techniques reported to measure it in the literature. Theoretical model are generally based on Darcy s law, with the permeability and the equivalent thermal conductivity of the capillary structure as main parameters [3]. Experimentally, temperature measurements are used to estimate the maximum power and the overall thermal resistance. In addition, some papers present transparent heat pipes to visualize phenomena occurring in working conditions [4-6]. Nevertheless, a fine description of the liquid-vapour interface is not yet available. This article reports experiments performed with one flat plate heat pipe having rather large dimensions. Its capillary structure is made of 2 layers of screen meshes. The meshes are not sintered, but plated against the wall by means of a coarse screen mesh, as recommended in [7]. This technological solution makes it very easy to manufacture. The heat pipe has been tested with various natures of the heat sinks (water heat - 29 -
exchangers or air radiators). Experimental results are compared to an analytical model in order to estimate the thermal characteristics of the meshes. 2. EXPERIMENTAL SET-UP The flat plate heat pipe under investigation is shown in figure. It is made of one copper plate of thickness.5 mm and surface 267 5 mm². The FPHP is hermetically sealed on its upper face with a transparent plate, which allows observations inside the system. A Viton flat ring of thickness 2 mm is placed between the copper and the transparent plates to ensure tightness. The capillary structure, of dimensions 230 90 mm 2, is made of 2 CuSn 325 square screen mesh layers. The capillary screen meshes are plated against the copper plate by means of a coarse screen mesh, which creates the vapor space (figure 2). Eight micro grooves are etched at the back of the copper plate in order to embed 00 µm diameter thermocouples (uncertainty lower than 0.2 K), using silver lacquer in order to reduce the contact resistance (figure ). Their voltage is recorded by a Keithley 2700 multimeter. A filling copper pipe closed by a valve is sealed on the transparent plate at the extremity of the FPHP. The heat source is a thick resistor film of dimensions 55. 96.5 mm 2 and resistance 90.6 Ω, which is supplied by a stabilized DC power supply. Electric power is obtained by measuring the voltage drop across the resistor and the intensity with a calibrated resistance. Thus, the uncertainty for the heating power is negligible. The heat source is located in the middle of the heat pipe. The bottom face of the copper plate is insulated. Two different heat sinks are tested (figure 3): Two aluminum radiators (40 70 mm 2 ) cooled by natural convection of air at ambient temperature. Two board card retainers are screwed on each radiator to maintain the heat pipe by means of two small fins located on each side of the FPHP (figure ). The contact area between the board card and the heat pipe is equal to twice 05 5 mm². Two water heat exchangers of heat transfer area 30 90 mm 2. The water flow rate is constant and the inlet temperature is controlled by means of a thermostatic bath in order to have a constant working temperature when the heat transfer rate increases. FPHP cooled by water heat exchangers 267 Screen mesh location 0 Heat source Water heat exchangers 5 95 FPHP cooled by air radiators Heat source Board card retainers 6 Viton flat ring Transparent closing plate Grooves for thermocouples Copper plate Figure. Schematic of the FPHP d w CuSn 325 screen mesh d = 36 µm w = 42 µm Coarse screen mesh d = 600 µm w = 2800 µm Figure 2. Coarse and fine screen meshes Aluminum radiator Figure 3. Cooling configurations Aluminum radiators are used to simulate the use of a flat plate heat pipe in a practical application, namely the cooling of electronic cards, which are most of the time located inside a metallic rack. The rack is used as a radiator to dissipate heat to the ambient air. The idea is to insert the FPHP into the metallic box exactly like an electronic card, using card retainers. In the purpose of working in controlled conditions, water heat exchangers are also used to estimate the heat transfer characteristics of the FPHP. - 30 -
3. EXPERIMENTAL RESULTS The FPHP is thermally insulated during thermal tests. First, it is degassed and filled. In order to promote surface wetting, the copper plate is first cleaned. The FPHP and the working fluid are degassed carefully to eliminate the non-condensable gases. The method of evacuating the non-condensable gases from the working fluid is based on the fluid solidification under vacuum. The fluid contained in a heated vessel vaporizes, releases non-condensable gases and solidifies in a second vessel dipped into liquid nitrogen. The non-condensable gases are evacuated by vacuum pumps. The FPHP is degassed by heating during vacuum pumping at 0-5 mbar. heat pipe behavior. The following results are obtained with water heat exchangers, which were preferred for practical reasons for that experimental campaign as they have a much shorter time response than air radiators. Figure 5 presents the maximal temperature difference between the heat source and the heat sinks in horizontal location vs. heat transfer rate. Linear characteristics are obtained until a threshold heat transfer rate of 82 W, for which the thermal resistance is equal to 0.4 K/W a value that must be compared to the thermal resistance of the copper plate alone (0.62 K/W). 50 Figure 4 presents temperature profiles in horizontal orientations with two types of heat sinks: two water heat exchangers cooled by a thermostatic bath at a temperature close to 20 C or two aluminum radiators cooled by natural convection of air at ambient temperature (~25 C). The heat transfer rate is equal to 5 W. Tmax (K) 40 30 20 60 0 Temperature ( 蚓 ) 50 40 30 20 0 air radiators water heat exchangers 0 0 50 00 50 200 Q (W) Figure 5: Maximal temperature difference vs. heat transfer rate; horizontal location 4. COMPARISON WITH AN ANALYTICAL MODEL Heat source 0-20 20 60 00 40 80 220 260 Position (mm) Figure 4: Influence of the heat sink nature on the heat pipe characteristics (Q=5 W) The overall shapes of temperature fields are very similar for these two cases. The only difference is the mean temperature level, which is close to the heat sink temperature for water heat exchangers and 30 K higher for air radiators. This difference is due to the nature of the heat sinks and not to the heat pipe: the thermal resistance of air radiators is much higher than that of water heat exchangers. As a consequence, water heat exchangers impose their temperature at the condenser. However, this comparison underlines that the nature of the heat sink does not change In this part, an analytical model of the FPHP is presented. This model couples a 3D thermal model of the FPHP wall with a 2D hydrodynamic model of the liquid and vapour flows inside the FPHP. The model is compared to the experimental results, in order to evaluate the thermal properties of the meshes. 4. Model equations The model has already been described in previous articles [3,8] to which the reader should refer for further details. The temperature field in the wall can be expressed by Fourier s series: - 3 -
T T sat + B 0n n= + B = B n= m= mn m0 m= (z)cos(nπy) (z)cos (z)cos(mπx) ( mπx) cos( nπy) () where T sat is the saturation temperature. B m0, B 0n and B mn depend on the location and on the heat transfer rate of the heat sources and heat sinks. They also depend on the thickness and the equivalent thermal conductivity of the capillary structure. Once the temperature field is calculated, it is possible to calculate the evaporation and condensation mass flow rates at each point of the system. A 2D hydrodynamic model, based on Darcy s law and on mass balance equations, is used to calculate the liquid and vapor velocities in the capillary structure and in the vapor channel, and the liquid and vapor pressure fields in the system. For the liquid, the pressure field is obtained by Fourier s series: Cm0cos( mπx) m= P µ = l l + C0ncos( nπy) Kρ ( ) ( ) lhlvh (2) g n= + C π π n= m= mn cos m x cosn y where K is the permeability of the capillary structure, µ l the dynamic viscosity, h lv the latent heat of vaporization of the liquid and h g the thickness of the wick. C m0, C 0n and C mn are calculated from the thermal model. The boundary conditions are a zero velocity at the extremities of the FPHP in x and y directions. A similar expression can be obtained for the vapour flow. 4.2 Comparison of the experimental results with the model The comparison between experimental data and the model allows calculating the equivalent thermal conductivity and the permeability of the capillary structure. In the model, the thickness of the capillary structure is equal to 40 µm for 2 screen meshes. liquid and vapor pressure drops: P = P + P. The capillary pressure of the cap l v 2σ σ meshes is equal to Pcap = =, where r eff reff w is the effective radius and w the distance between two fibers. For the screen meshes and the fluid used in this work, P cap is equal to about 2000 Pa. In the model, the permeability is adjusted until the pressure drops in the liquid and the vapor, calculated for a heat transfer rate just lower than the heat transfer rate at the dry out, is equal to the capillary pressure. Figure 6 presents an example of the comparison between the model and the measured temperatures for a heat flux of 82 W. The equivalent thermal conductivity that fits at best the experimental data is equal to 0.23 W/mK. It is rather constant for heat transfer rates between 45 W and 82 W. It has to be noted that this equivalent thermal conductivity is close to that of methanol. Tmax (K) 6 4 2 0 8 6 4 2 0 0 50 00 50 200 250 Position (mm) Figure 6: Comparison between analytical model and experimental data (Q= 82 W) The permeability of the meshes is estimated to 0.55 0-0 m² by comparison with the results at the heat transfer rate threshold (i.e. at Q = 82 W, cf. section 3). 4.3 Comparison of the estimated properties of the meshes with theoretical laws For a mesh with a fiber diameter equal to d and a distance between two fibers equal to w (figure 2), the permeability K can be expressed as [9]: The permeability of the meshes is estimated by comparing the capillary pressure to the sum of the - 32 -
K 2 3 d ϕ = 22 ϕ ( ) 2.05πNd ϕ = 4 N = d + w (3) where ϕ is the porosity. The equivalent thermal conductivity depends on both the liquid and the mesh thermal conductivities λ l and λ s [9]: l[ ( λl + λs ) ( ϕ) ( λl λs )] [( λ + λ ) + ( ϕ) ( λ λ )] λ λ eq = (4) l s l s The permeability calculated with equation (3) is equal to 0.8 0-0 m² and the equivalent thermal conductivity calculated with equation (4) is equal to 0.44 W/mK. The permeability estimated experimentally is 5 times higher than that calculated theoretically while the experimental equivalent thermal conductivity is 2 times lower. Both discrepancies can be explained by an underestimation of the real thickness of the screen meshes. Indeed, increasing the thickness of the capillary structure in the analytical model leads to decrease the permeability and increase the equivalent thermal conductivity respectively. 4.4 Results for the cooling of several electronic components Figure 7 presents the temperature field for the FPHP in horizontal orientation with three heat sources and two heat sinks. The properties of the capillary structure are those estimated in the previous sections, i.e. an equivalent thermal conductivity equal to 0.23 Wm - K - and a permeability of.25 0-0 m². The heat sink size and location are the same as the water heat exchangers on each side of the FPHP. The total heat transfer rate is equal to 75 W and is supposed to be equally distributed into 2 dissipative components of area 0 0 mm² each and one dissipative component of area 20 20 mm². The maximum temperature difference reaches 20 K. Just as an illustration, one can note that an aluminium plate ( =65 Wm - K - ) of thickness 3 mm (i.e. more than 4 times thicker than a FPHP of 3 mm) would present the same thermal performance (figure 8). The maximum temperature is reached under the small heat sources, where the heat flux is the highest. Figure 7: Temperature field on the FPHP with 3 electronic components Figure 8: Temperature field on an aluminum plate of thickness 3 mm Figure 9 presents the liquid pressure field inside the screen meshes. The maximum pressure difference inside the capillary structure is lower than 700 Pa, which is below the capillary limit (2000 Pa). Figure 9: Liquid pressure in the screen meshes 5. CONCLUSION Experimental data obtained with a flat plate heat pipes with 2 screen mesh layers were presented. The realization of the capillary structure of the FPHP is very simple and cheap. The results show that the FPHP thermal performance does not change with the nature of the heat sinks: experimental results are similar using air radiators as cold sources rather than water heat exchangers. The comparison with an analytical model shows that the equivalent thermal conductivity of the meshes is close to the thermal conductivity of the methanol. This result is not as high as one could expect in comparison with theoretical results. On the contrary, the permeability estimated experimentally is higher than that calculated - 33 -
theoretically. Both results can be partially explained by an underestimation of the real thickness of the meshes. Hydrodynamic and thermal simulations show the ability of the flat plate heat pipe to cool several electronic components located on a printed circuit board. The maximum heat transfer rate in horizontal location is around 80 W for a 0.2 m long card. NOMENCLATURE B 0m, B 0n, B mn : Fourier coefficients for T C 0m, C 0n, C mn : Fourier coefficients for P d: fibre diameter, m h g : thickness of the wick, m h lv : latent heat of vaporization, J.kg - K: permeability, m² P: pressure, Pa Q: heat transfer rate, W T: temperature, K w: distance between 2 fibers, m x, y, z: coordinates, m Greek symbols λ: thermal conductivity Wm - K - µ: dynamic viscosity, Pa.s ρ: density, kg.m -3 σ: surface tension, N.m - ϕ: porosity Subscripts cap: capillary eff: effective eq : equivalent l: liquid max: maximum sat: saturation s: solid electronic components, International Journal of Heat and Mass Transfer, Vol. 49, pp. 375-383, 2006. [4] Kempers, R., Robinson, A.J., Ewing, D., Ching, C.Y., Characterization of evaporator and condenser thermal resistances of a screen mesh wicked heat pipe, International Journal of Heat and Mass Transfer, Vol. 5, pp. 6039-6046, 2008. [5] Wong, S.C., Liou, J.H., Chang, C.W., Evaporation resistance measurement with visualization for sintered copper powder evaporator in operating flat plate heat pipes, International Journal of Heat and Mass Transfer Vol. 53, pp. 3792 3798, 200 [6] Li, C., Peterson, G.P., The effective thermal conductivity of wire screen, International Journal of Heat and Mass Transfer, Vol.49, pp. 4095-405, 2006. [7] Rosenfeld, J. H., Gernert, N. J., Sarraf, D. B., Wollen, P., Surina, F., Fale, J., Flexible heat pipe, United States Patent 6,446,706, 2002. [8] R. Révellin, R. Rullière, F. Lefèvre, J. Bonjour, Experimental validation of an analytical model for predicting the thermal and hydrodynamic capabilities of flat micro heat pipes, Applied Thermal Engineering, Vol 29, pp. 4-22, 2009. [9] Faghri A., Heat pipe science and technology. Washington, Taylor & Francis, (994) 874 p. REFERENCES [] Lefèvre, F., Rullière, R., Pandraud, G., Lallemand, M., Prediction of the temperature field in flat plate heat pipes with micro-grooves - Experimental validation, International Journal of Heat and Mass Transfer, Vol. 5, pp. 4083-4094, 2008. [2] Lips, S., Lefèvre, F., Bonjour, J., Physical mechanisms involved in grooved flat heat pipes: experimental and numerical analyses, International Journal of Thermal Sciences, accepted in February 20. [3] F. Lefèvre, M. Lallemand, Coupled thermal and hydrodynamic models of flat micro heat pipes for the cooling of multiple - 34 -