Novel Lightweight Metal Foam Heat Exchangers

Novel Lightweight Metal Foam Heat Exchangers David P. Haack 1, Kenneth R. Butcher 1, T. Kim 2 and T. J. Lu 2 1 Porvair Fuel Cell Technology, Inc., 700 Shepherd St., Hendersonville, NC 28792, USA 2 Department of Engineering, University of Cambridge, Trumpington Street, Cambridge, UK ABSTRACT An overview of open cell metal foam materials with application to advanced heat exchange devices is presented. The metal foam materials considered consist of interconnected cells in a random orientation. The manufacture of metal foam materials into complex heat exchange components is described. Experiments with flat foam panels brazed to copper sheets show increasing heat removal effectiveness with decreasing foam pore size at equivalent coolant flow rates. The high-pressure drop experienced with flow through small pore-size metal foam materials makes the use of larger pore size material more attractive. The paper will demonstrate that in certain configurations, particularly difficult geometries, confined spaces, high temperatures and demanding environments, metal foam is an excellent heat exchange medium. INTRODUCTION New materials are needed in the development of advanced, compact, and lightweight thermal systems to satisfy new demands from emerging technologies. Three such technologies are multi-functional fuel processors that demand the capability for simultaneous heat exchange and chemical reaction; highly efficient radiators for high power heat rejection at low temperature differentials; and high-density microelectronic circuits requiring high rates of heat removal. Porvair Fuel Cell Technology is investigating metal foam as a medium to use in the manufacture of highly effective, high temperature-capable, geometry-flexible, and multi-functional heat exchange devices. Two mechanisms have been found to be important to the heat transfer enhancement associated with the use of metal foam materials: specifically, interactions between the solid foam material and a through-flowing fluid [1-4]; and the importance of achieving a quality metal-tofoam bond [3, 5]. Applications in electronics cooling and compact heat exchangers have been investigated, revealing promising advances in the rate of heat removal or transfer under experimental conditions. Some applied research has been performed, applying metal foam in unique designs to radiators and advanced reactors [6]. While most of the test results are proprietary, applied research is at Porvair Fuel Cell Technology. Metal foam heat transfer: A literature survey Metal foam materials have been investigated for use in heat exchange applications in the open literature. Studies have attempted to describe thermal transport in ceramic and metallic foams on a basic and applied basis. Younis and Viscanta [7] have measured the volumetric heat transfer of ceramic foam materials, and developed a Nusselt number correlation fit to the experimental data. The volumetric heat transfer rates measured were higher than those for packed beds or sintered metals. Calmidi and Mahajan [3] studied the solid-to-fluid thermal transport from a heated metal plate brazed to aluminum metal foam. Results indicated a significant contribution of the thermal transport resulted from solid plate to foam contact, and subsequent fluid-to-foam thermal transport. High quality joints are indicated as being important to effective thermal transport. In a similar study, Kim et al. [8] examined metal foam thermal transport between two isothermal plates. Aluminum foam filled the space between the plates, which were heated with flowing water. The foam was put into mechanical contact by pressing to minimize thermal contact resistance. Pressure drop friction factors and modified Colburn J-factors were measured, and comparisons were made to conventional louvered fins for overall performance. Results indicated that the foam material offered better heat transfer performance compared to a louvered array, but at a greater pressure drop. Methods of reducing pressure drop or improving the foam to solid contact were not investigated. Lu et. al. [2] have developed a model describing metal foam heat transfer, where the foam is modeled by inter-connected cylinders. The analysis was extended to electronics cooling, and to multi-layer heat exchangers, with good performance predicted. Bastarows [5] studied single side heating of a foam-filled channel for an electronics cooling application. The experimental method utilized both conductive epoxy bonding and brazing of the metal foam to heated plates. Results indicated that brazed foam materials are much more effective at heat removal than epoxy-bonded samples. Measured heat exchange performance indicated 3 times more heat removal capability compared to a conventional fin-pin array. Outline of paper This paper will discuss open-cell metal foam fabrication techniques, the use of metal foam in complex heat exchanger designs, and the potential to mass manufacture metal foam 1 of 7

materials. In addition, the heat exchange properties of a high temperature-capability metal foam alloy (Porvair Fuel Cell Technology s FeCrAlY material) will be presented. OPEN-CELL POROUS METAL FOAMS: FABRICATION AND ASSEMBLY Cellular morphology of metal foams The metal foam structure, shown in Fig. 1, consists of ligaments forming a network of inter-connected dodecahedral-like cells. The cells are randomly oriented and mostly homogeneous in size and shape (Fig. 1a, a result of the manufacturing method used to create the metal foam precursor material). The triangular-shaped edges of each cell are hollow (Fig. 1b), a result of the manufacturing technique. Pore size may be varied from approximately 0.4 mm to 3 mm, and the net density from 3% to 15% of a solid of the same material. Metal foam from Porvair Fuel Cell Technology is available in alloys and single-element materials. Common materials include copper, stainless steel, and high temperature iron-based alloys (e.g., FeCrAlY). Metal foam thermal properties Heat transfer enhancement using porous metal foams depends on both the cellular structure of the foam material, and the thermal properties of the metal foam. Metal foam thermal conductivity is dependent upon the overall density of the piece and the metal from which the foam is made. Conductive pathways through the porous material are limited to the ligaments of the material. Experimental measurements have determined a functional relationship between the foam thermal conductivity and density as λ s ρ r 1.8 < λ f < λ s ρ r 1.65, where ρ r is the foam relative density, λ f is the foam conductivity, and λ s is the solid conductivity [6]. Higher material conductivity is associated with higher density materials, and significant increase in thermal conductivity results from an increase in material density. On the other hand, heat transfer by metal foams due to thermal dispersion effects is proportional to cell size [4]. Metal foam fabrication and capabilities Metal foams have been manufactured for many years using a variety of novel techniques. Metallic sintering, metal deposition through evaporation, electrodeposition or chemical vapor decomposition (CVD), and investment casting (among numerous other methods) have created open cell foams. In foam creation through metal sintering, metallic particles are suspended in slurry and coated over a polymeric foam substrate. The foam skeleton vaporizes during heat treatment and the metallic particles sinter together to create the product. This method is thought to be the most costeffective and the most amenable to mass production. The CVD method utilizes chemical decomposition of a reactive gas species in a vacuum chamber to deposit material onto a heated substrate (polymer or carbon/graphite, depending upon the temperature of the deposition process). Production rates are limited in this method by the rate at which material is deposited on the substrate. Highly refractory metals and ceramics may be created with this method with high quality. Molten metal infiltration is utilized to make aluminum and copper foam materials [9]. With this method, a foam precursor is coated with a ceramic casing and packed into casting sand. The casting assembly is heated to decompose the precursor and harden the casting matrix. Molten metal is then pressure infiltrated into the casting, filling the voids of the original matrix. After solidification, the material is broken free from the mold. The method has the advantage of being capable of producing a pieces in widely used metals and alloys with solid struts. However, the process requires several processing steps and specialized equipment, and does not lend itself to rapid production processes. High volume manufacturing Of the methods suitable to produce metal foam materials, the metal sintering method offers the most promise for mass production. Necessary production equipment is easily automated and yields high-quality, low-cost metal foam materials for use in a variety of applications. Capability of manufacturing complex assemblies To effectively use metal foam materials in heat exchange devices it is necessary to combine the material with tubes and sheets for flow control and heat transfer. Development efforts have taken place at Porvair Fuel Cell Technology to successfully combine a variety of metal foam materials with solid structures. Several proprietary components have been constructed combining tubes and other solid materials to construct advanced, multifunctional heat exchange devices for a variety of customers. An important consideration in the formation of the advanced heat exchangers is the quality of the bond joint between foam and solid material through which heat is transferred [5]. Figure 2 is a photograph of a developmental component consisting of tubes imbedded in a metal foam matrix, generating an advanced high-temperature radiator. Metallurgical bonding between the tube wall and the foam matrix was achieved by sintering during material heat treatment. Figure 3 shows an SEM micrograph of the joint region, which shows good bonding between the foam and tube. Assemblies have also been manufactured through a proprietary co-sintering technique. Figure 4 shows an example of a foam-filled tube manufactured with this method. Complex assemblies combining metal foam with metal packaging are in the design stage to create an advanced two-phase heat exchange component for use in fuel cell fuel processing systems at Porvair Fuel Cell Technology. HEAT TRANSFER MEASUREMENTS An experimental program was performed at Cambridge University for the purpose of measuring the heat transfer effectiveness of metal foams under varying flow conditions. Experimental procedure The experimental apparatus mainly consists of four sections: coolant supplier, test section, test model, and data acquisition system. A photograph of the test rig with a sample inserted in its test section is shown in Figure 5. Air is used as a coolant and is forced through the channel 2 of 7

inlet by a suction type air blower. A total of four static pressure taps are placed along the flow direction on the upper copper skin. An asymmetrical isoflux (constant wall heat flux) boundary condition is imposed on the lower copper skin by a heating element (silicone-rubber etched foil from Watlow TM Inc.). Five thin foil (each 0.05 mm thick) T-type copper-constantan thermocouples (from Rhopoint Inc.) are inserted on the lower skin along the flow direction. There are two additional T-type thermocouples, positioned separately at the inlet and outlet of the test section to measure the coolant temperature at each location. A temperature scanner with reading resolution of ±0.1K is used to record and analyze temperature readings from all thermocouples simultaneously. The experiments were run for several minutes until the flow inside the channel became hydraulically and thermally stabilized. All measurements were performed under steady state conditions. A Pitot tube was positioned before the test section to measure stagnation and static pressures at the inlet. Because the blockage ratio of the pitot tube (tube diameter (0.51 mm) to channel height (12 mm)) is small, wall interference from the Pitot tube is expected to be negligible. Measurement uncertainties During experimentation, measurements were repeated until significant data repetition was ensured (i.e., 5% uncertainty interval). An uncertainty analysis was performed following the method suggested by Kline and McClintock [10]. The maximum heat loss through the insulation materials was estimated to be less than 2 percent of input heat flux. The heat loss through the perspex side-walls was estimated to be negligible through a conduction heat loss analysis. The thermal conductivity k f of air varies slightly in the operating temperature range of 300.0 K to 350.0 K. An arithmetic mean value is used for k f, with uncertainty estimated to be within 6.6 %. From these, the uncertainty in the measured heat transfer coefficient and Nusselt number was estimated to be less than 7.0% and 9.6%, respectively, whilst the uncertainty in the pressure drop and friction factor measurements was estimated to be less than 5.0% and 7.8%, respectively, using a root-sum-square method. Test samples The samples fabricated by Porvair Fuel Cell Technology consisted of FeCrAlY metal foam bonded on top and bottom to a thin copper sheet (1 mm thick). Bonding was achieved by brazing. Metal foam pore size included 10, 30 and 60 PPI (pores per inch). Foam relative density was set to 5%, 7.5% and 10 %. Table 1 shows the specifications for each heat exchange test sample. The thermal conductivity of solid FeCrAlY alloy is taken to be 16 W / mk. The sandwiched foam specimens were trimmed to fit into the test section of a heat sink channel of size 0.127 m (W) 0.127 m (L) 0.012 m (H). RESULTS AND DISCUSSION Permeability and inertial coefficient The measured pressure drop across the FeCrAlY foam samples is presented in Figure 6 as a function of mean flow velocity U m (at the test section inlet). The modified Darcy equation: dp dx 1 = 1 µu m K ρ I C U m K µ is subsequently used to determine the permeability, K, and inertial coefficient, I C, for each sample, with µ and ρ denoting separately the dynamic viscosity and density of air. The results are listed in Table 1, and are found to be similar to those reported for aluminum foams [3-4]. Pressure drop is found to be highly dependent upon material pore size, and less dependent upon material density. Thermal resistance and heat removal performance The surface temperature T w (x) of the copper plates was measured from thermocouples, where x denotes the longitudinal axis. Linear variation of T w with x was observed under the isoflux boundary condition. The thermal performance of FeCrAlY foams as a heat sink medium was characterized by the local heat transfer coefficient h and local Nusselt number Nu defined as: (1) q h(x) = T w (x) T in (2) Nu(x) = h(x) k f / D h (3) where q, T in, k f and D h are heat flux, coolant temperature at inlet, coolant thermal conductivity, and hydraulic diameter of the heat sink channel. These are averaged over the sample length to obtain the mean heat transfer coefficient and mean Nusselt number as: h = 1 L 0 L h(x)dx (4) Nu = 1 L 0 L Nu (x)dx (5) Reynolds number is based on the measured permeability, Re K = ρu m K /µ. Fig. 7a plots the averaged Nusselt number as a fucntion of Re K for the FeCrAlY samples with a fixed pore density (30 PPI) but different relative densities, whilst Fig. 7b plots Nu as a function of Re K for samples at 10% and 15% relative densities and different pore densities. The results for all FeCrAlY samples are summarized in Figure 8, and compared with those taken from [3] for aluminium foams. The results of Fig. 7a indicate that for a foam with a fixed pore density, a higher relative density is favored for improved heat transfer rate, although the corresponding flow resistance is higher. On the other hand, at a fixed relative density and a fixed value of Re K, the 60 PPI foam sample removes more heat than the 10 PPI sample (Fig. 7b). The importance of the material density can also be seen in this figure through a 3 of 7

comparison of 30 ppi material performance at a slightly lower relative density (10% compared to 15%). The higher density large pore material was found to outperform the slightly less dense smaller pore material. Figure 7b also shows that, at a given pumping power (implied from permeability), the maxium value of Re K, (Re K ) max, is limited by the flow resistance of each sample, with (Re K ) max increasing as the flow resistance is decreased (or, equivalently, as the peamibility is increased). Consequently, the highest Nusselt numbers obtained from the 10 PPI foam with 15% relative density nearly double that of the 60 PPI foam having the same relative density, indicating that the former has the best heat transfer efficiency at a specified pumping power. Comparison was made with selected test data for aluminum foams having solid struts (these foams were processed via the expensive investment casting technique). In general, at a fixed value of Reynolds number Re K under forced air convection, the FeCrAlY samples remove 30~50 % of heat that is removed by aluminum foams of a similar pore size and density, although the thermal conductivity of FeCrAlY (~16 W / mk ) is an order of magnitude smaller than that of pure aluminum (~200 W / mk ). However, if air is replaced by water as the coolant, it is expected that FeCrAlY foams and aluminum foams will have similar heat transfer characteristics, because the thermal conductivity of high porosity foams, whether ceramic or metal, is roughly the same as the thermal conductivity of the coolant [4]. CONCLUSION New applications for highly effective, multi-functional heat exchange devices are driving the development of metal foam components. Metal foam materials have the potential to increase heat transfer rates from solid surfaces by conducting heat to the material struts and inducing a high interaction between the struts and a through-flowing fluid. New manufacturing techniques developed at Porvair Fuel Cell Technology allow effective, low-cost, high-volume manufacturing, and new assembly techniques are being developed to manufacture complex assemblies of foam and solid metals to form heat exchange devices. Several prototype devices have been constructed for industry to increase performance and reduce size, cost and weight. Heat transfer and pressure drop measurements reveal that high rates of heat removal are possible with FeCrAlY foam, a high-temperature metal. While small pore-size material is advantageous for achieving high rates of heat removal, pressure drop will be higher. Larger pore size materials can achieve higher Nussalt numbers at high rates of flow with relatively low fan power required. Increasing material density was found to increase Nussalt numbers at a given rate of coolant flow. Bare metal conductivity of aluminum is approximately ten times that of FeCrAlY, however, the heat transfer performance of aluminum foam is only 2-3 times greater than FeCrAlY. This demonstrates that the foam structure, and therefore the turbulence induced in the process fluid substantially improves heat transfer performance. Future work will examine the performance of copper materials in a similar arrangement. A theoretical model will be developed to enable flexible design of advanced heat exchange concepts using metal foam materials. ACKNOWLEDGEMENTS This work was supported by Porvair Fuel Cell Technology, Inc., the US Office of Naval Research (ONRIFO/ONR Contract No. N00014-01-1-0271), and by UK Engineering and Physical Scientific Research Council (EPSRC grant number EJA/U83). The authors would like to thank Mr. Alberic du Chene of Cambridge University for providing Figure 1. REFERENCES 1. A.-F. Bastawros, A.G. Evans, and H.A. Stone, Evaluation of Cellular Metal Heat Dissipation Media, Technical Report MECH-325, DEAS, Harvard University, March 1998. 2. T.J. Lu, H.A. Stone and M.F. Ashby, Heat transfer in open-cell metal foams, Acta Mater 46 (1998) 3619-3635. 3. V.C. Calmidi and R.L. Mahajan, Forced Convection in High Porosity Metal Foams, Trans. of ASME, J of Heat Transfer 122 (2000) 557-565. 4. M.L. Hunt and C.L. Tien, Effects of Thermal Dispersion on Forced Convection in Fibrous Media, Int. J. Heat Mass Transfer 31 (1988) 301-309. 5. A.-F. Bastawros, Effectiveness of Open-cell Metallic Foams for High Power Electronic Cooling, IMECE Paper, Thermal Management of Electronics, ASME Proc. HTD-361-3/PID-3, 211-217. 6. M.F. Ashby, A. Evans, N.A. Fleck, L.J. Gibson, J.W. Hutchinson, H.N.G. Wadley, Metal Foams: A Design Guide, Butterworth-Heinemann, Boston, ISBN 0-7506- 7219-6. 7. L.B. Younis and R. Viskanta, Experimental determination of the volumetric heat transfer coefficient between stream of air and ceramic foam, Int. J. of Heat Mass Transfer, 36 (1993) 1425-1434. 8. S.Y. Kim, J.W. Paek, B.H. Kang, Flow and Heat Transfer Correlations for Porous Fin in a Plate-Fin Heat Exchanger, Trans. Of the ASME, J. of Heat Transfer, 122 (2000) 572-578. 9. G.J. Davies, S. Zhen, Review: Metallic foams: their production, properties and applications, J. Material Sci., 18 (1983) 1899-1911. 10. S.J. Kline and F.A. McClintock, Describing Uncertainties in Single-Sample Experiments, Mechanical Engineering (1953) 3-8. 4 of 7

Table 1. Specifications of FeCrAlY foams Sample No. S-1 S-2 S-3 S-4 S-5 S-6 S-7 Pore size (PPI) 10 10 30 30 60 60 30 Relative density (%) 5 15 5 10 5 15 7.5 Permeability, K ( 10 7 m 2 ) 1.67 0.5 1.0 0.5 0.33 0.11 1.0 Inertial Coefficient, I C 0.093 0.13 0.15 0.164 0.24 0.49 0.2 (a) (b) Figure 1. SEM images of reticulated metal foam structure (FeCrAlY). Interconnected tortuous pathways create turbulence in through-flowing fluids. Figure 2. Metal foam compact heat exchanger for high temperature service. Foam material is PFCT s FeCrAlY. 5 of 7

Figure 3. SEM micrograph of a foam strut sintered to a solid tube. Bonding region shows metallurgical sintering between foam and solid. Figure 4. Example assemblies manufactured in a proprietary co-sintering technique (patent applied for). Figure 5. Test apparatus showing parallel plates for flow, and a foam sample with insulation. 120 80 S-1 S-2 S-3 S-4 S-5 S-6 S-7 60 ppi 60 ppi 5% relative density 15% 10% relative density density 30 ppi 10% relative density 30 ppi 7.5% relative density 10 10 ppi ppi 15% 10% relative relative density density dp/l [kpa/m] 60 30 ppi 5% relative density 40 20 10 ppi 5% relative density 0 0 2 4 6 8 10 12 14 16 U m Figure 6. Static pressure drop per unit length results from the Porvair foam samples. 6 of 7

400 400 350 300 S-3 S-4 S-7 350 300 10ppi 15% relative density Nu 250 200 150 10% Relative density Nu 250 200 150 60ppi 15% relative density 50 7.5% Relative density 5% Relative density 50 150 200 250 300 Re K (a) (b) Figure 7. Average Nusselt number of Porvair foams (FeCrAlY) as a function of Reynolds number: 50 10% relative density (a) fixed pore size (30 PPI ) (b) 10% and 15% relative densities. ReK S-2 S-4 S-6 50 150 200 Nu 0 900 800 700 600 500 400 S -1, (10 ppi, 5% ) S -2, (10 ppi, 15% ) S -3, (30 ppi, 5% ) S -4, (30 ppi, 10% ) S -5, (60 ppi, 5% ) S -6, (60 ppi, 15% ) S -7, (30 ppi, 7.5% ) S am ple 1 of ER G S am ple 3 of ER G S am ple 4 of ER G 300 200 50 150 200 250 300 Re K Figure 8. Comparison of FeCrAlY foams (Porvair) with aluminum foams (ERG, data from [3]). 7 of 7