10 PPI metal foam as alternative for louvered fins in a HVAC heat exchanger

S De Schampheleire1, P De Jaeger1,2, H Huisseune1, B Ameel1, C TJoen1, M De Paepe1

1

Department of Heat, Mass and Combustion Mechanics, Ghent University, SintPietersnieuwstraat 41, 9000 Ghent, Belgium
2

NV Bekaert SA, Bekaertstraat 2, 8550 Zwevegem, Belgium

E-mail: Sven.DeSchampheleire@ugent.be
Abstract. A commercial heat exchanger with louvered fins (fin pitch: 1.4mm) and a prototype with open-cell metal foam (10 PPI) are compared. Heat transfer and pressure drop measurements were performed in a wind tunnel through a modified Wilson plot technique with air velocities between 1.1 and 3.1 m/s. The comparison is made with a well-defined performance evaluation criterion for foam. The performance of both heat exchangers is the same for velocities lower than 1.1 m/s. The louvered fin heat exchanger performs however better at higher velocities. Finally, the influence of the thermal contact resistance is investigated based on literature. Comparing the airside convective resistance only, the metal foam heat exchanger performs better at high velocity range (2.75 3.1 m/s). It is demonstrated that more research is needed in improved contact technologies for metal foam.

1. Introduction As HVAC (Heating, Ventilation and Air Conditioning) applications constitute 30% of the total energy use in society [1], it provides a huge potential for energy reduction. HVAC systems typically use liquid-to-air heat exchangers. For such systems, the airside convective resistance is dominant, representing up to 75-90% of the overall thermal resistance [2]. Decrease of this resistance is typically accomplished with fins, where louvered fins are the current state-of-the-art. Another material that recently got lot of attention is open-cell metal foam. In this study, aluminium foam (see Fig. 1) is considered because of its high thermal conductivity (Al1050, k = 200 W/mK). The foams used in this study are in-house manufactured, basically by replicating an organic preform via an investment casting process. The organic preform defines the PPI value (Pores Per linear Inch) of the foam. This PPI value serves as a classifying parameter of the foam and is directly linked to the specific surface area [3]. In this paper, 10 PPI foam is used. The replica of the thickened preform foam is the result of surface energy minimization, inducing interesting functional and structural properties: e.g. high porosity (93.7% used for this study), high strength and toughness, excellent fluid mixing, high specific surface area, high gas permeability, low weight, 3D structure (shape-ability). As the good mixing properties heavily disturb the boundary layers, pressure drop is frequently quoted as primary disadvantage. Therefore the aim of this paper is to experimentally study the thermo-hydraulic characteristics of a commercial available heat exchanger with adapted inclined louvered fins and a prototype heat exchanger with foam. For such a complex material and full scale application, typically the modified Wilson plot technique is considered. This black box method relates kinetic variables (heat transfer,

pressure drop) to kinematic variables (temperatures, velocity), by means of proportionality constants (thermal resistance, friction factor). The depending variables which are related via the proportionality constants are averaged, e.g. averaged inlet and outlet temperatures, mean mass flow rate... Such a comparison (between foam and louvered fins) has been already made by Sertkaya et al. [5] in 2012. However, it is not clear what the uncertainty on the reported data is: mass flow rate is based on a local velocity measurement, the thermocouples are not calibrated, the convection coefficient is incorrectly reported and the pressure drop is not according to our measurements (it differs more than a factor 10). Furthermore, this paper provides new insights as thermal contact resistance is taken in account.

Figure 1. In-house made 10 PPI (on top) and 20 PPI aluminium foam

Figure 2. Experimental test set-up

2. Test rig and procedure 2.1. Experimental test set-up and measurements The test rig consists of an open air wind tunnel and a closed hot water cycle (see Figure 2). A centrifugal fan sucks air through a calibrated nozzle (1), where the pressure drop is measured over the nozzle with a 0.5% accurate differential transducer and the airside mass flow rate determined according to the ISO5167 standard. The frequency controlled fan covers an airside mass flow rate between 0.15 and 0.41 kg/s, corresponding with an air speed in the test section of 1.1 to 3.1 m/s (in 6 steps). To obtain a uniform velocity profile at the test section a diffuser section is used, followed by a settling chamber, a flow straightener and a double sinusoidal contraction section. The uniformity of the inlet airside velocity was confirmed with hotwire measurements. The heater (2) has a maximal power of 9 kW and is PID controlled. The water pump (3) has a relay control. A motorized three way valve (4) controls the mass flow rate to the heat exchanger, keeping it below 50% of the maximal flow rate of the pump. By using the water heating system in this way, a very stable temperature can be maintained, as the water circuit acts as a hot buffer with a large recirculating flow. The water mass flow rate is measured by a coriolis mass flow meter (PROMASS 80-Endhauser, (5)) and hold constant on 0.04 kg/s (representing with a velocity in the tubes of 0.5 m/s). Based on the calibration sheet, the relative error on the mass flow rate is less than 0.3%. All temperatures are measured with K-type thermocouples, which were calibrated using a temperature calibrator furnace (DBC 150, Druck). The reference temperature is measured with a FLUKE (type 1523) temperature reader (with accuracy of 0.015C). The uncertainty of the thermocouples is found to be 0.07C (and conservatively taken on 0.1C). The water temperature is measured with four thermocouples, two in the inlet collector and two in the outlet collector. In this paper, the water inlet temperature is held constant on 70C and the measurements were performed under steady state-condition (once the variation was lower than 0.07C over 300s). All measurements are done at least 2 times in random sequence. Every 3 seconds a measurement is recorded and a least

150 measurements were taken per air mass flow rate set point. Other tested water inlet temperatures (50C-60C-80C) were tested as well and found within the accuracy of the measurements. Two different heat exchangers (see Fig. 4) were tested using this set-up. Both have a cross sectional area of 256x447 mm, a flow depth of 24 mm and two staggered tube rows. The heat exchangers are so called low capacity units, which means that not all the tubes are connected to the header. In this case water flows through only 18 of the 24 tubes, as shown in Fig. 4. Although this unit normally is installed at an angle, it is placed vertically in the wind tunnel, so that both heat exchangers are tested in the same conditions. Measures are taken to prevent bypass flow: the whole surroundings of the heat exchanger are filled with silicones. The tubes of the louvered fin heat exchanger originally have an outside diameter of 7 mm, and are inserted into the 7.2 mm diameter holes in the fins before being expanded. The louvered fins are installed with a fin pitch of 1.4mm (fin thickness: 0.115mm), a transversal tube pitch of 21mm and a longitudinal tube pitch of 12mm. For the metal foam heat exchanger the holes are first drilled to 7mm, using EDM (Electrical Discharge Machining) to finish the surface, than the tubes are expanded in these holes (using an air-pressurised bullet). As the foam will deform, this results in a good thermal contact and an external diameter of 7.2mm (as well). Measurement of the airside outlet temperature is special. Based on the geometry of the heat exchanger (Fig. 4), the temperature gradient in a horizontal plane is expected to be constant. This was confirmed using an infra red (IR) camera from Dias Infrared Systems (see Fig. 3). Therefore an equidistant single row of 9 thermocouples was mounted vertically across the test section to compute area-mean outlet temperatures. The camera is used to verify the thermocouple measurements and the temperature distributions. Good calibration for the IR camera and careful set up is required. In preparation of this experiment we considered the recommendations of the 3 ASTM standards. E.g. one must avoid reflections from walls, lights, nearby setups... Next, a correct surface emissivity value is required to determine the temperature from the acquired radiation. Therefore, we decided to place a thin slice of metal foam at the test section exit: this material also has a rather high, homogeneous and constant emissivity (0.82), allowing proper IR images. The emissivity was tuned with a calibrated thermocouple attached to the foam slice. The averaged air temperature based on the 9 thermocouples shows good agreement with the IR measurements (see Table 1). Although the averaged output of the 9 thermocouples is used in further calculations. Table 1. Measurement of the IR camera at 70C water inlet temperature (accuracy 1C). thermocouples IR capture (in C) (in C) (in kg/s)

Figure 3. IR image of test section (in C) at = 0.256 ; The capture is filtered in Matlab.

2.2. Wilson plot technique The used data reduction method, according to the modified Wilson plot technique, is explained in the literature paper of TJoen et al. [6]. In this paper, the heat balance (difference between and ) closes within 3% and both the LMTD (Log Mean Temperature Difference) and method are used to verify a correct determination of the overall thermal resistance. In the LMTD method, the correction factor F for the selected cross flow configuration ranges between 0.925 and 0.98. Internal Reynolds numbers for this case where higher than 8500. is determined in a way to minimize the uncertainty, using the literature paper of Park et al. [7]. The Wilson plot technique is based on the separation of the overall thermal resistance into: (1) convective heat transfer on waterside and on airside, (2) conductive resistance through the copper tubes, (3) contact resistance. The fouling resistances are neglected as filters are used. This is represented in Eq. (1).

(1)

Figure 4. Schematic sketch of the studied heat exchanger (on the left), with metal foam (in the middle) and louvered fins (on the right) as heat enhancing materials. The external convection coefficient ( ) from the third term on the rhs in Eq. (1) is frequently reported as a Nusselt number to compare with literature. However, the authors must find a way to calculate the exterior surface area of the metal foam ( ). Several researchers have determined this through a destructive method, with high errors [8]. This results in high uncertainties for the reported Nusselt number and only a limited ability for comparison. Therefore, as proposed by Moffat et al. [9], the parameters and are simply hold as one entity. The fin efficiency ( ) for metal foam can be calculated as proposed by Ghosh et al. [10], the authors call it a foam efficiency as metal foam cannot be seen as a fin. However, they used a simplified cubic model to reconstruct the foam and as the used heat exchangers in this study are low capacity units, the correlation of Ghosh cannot be used. The external convective resistance is therefore taken as a whole. Furthermore, for the moment, the external convective resistance and the contact resistance are also taken as one entity, noted as ( ). In a latter paragraph, and will be decoupled. 2.3. Definition of a performance evaluation criterion for foam A good Performance Evaluation Criterion (PEC) is necessary to make a proper comparison between two heat exchangers, especially for foam this is not evident. An area goodness factor, define as j/f, is commonly applied in metal foam research [11]. However, the area goodness criterion is associated with the assumption of fixed pressure drop [12]. To solve this, authors like TJoen et al. [6] refer to plain fins, so that changes in pressure drop are negligible. In this study however it is not possible to have a plain fin reference, because of the selected heat exchanger. Additionally, this criterion is less good for metal foams: when a comparison is made between a thick and a thin strut (e.g. according to a changing porosity), the thinner strut will have a bigger j/f, because h=Nu*k/df (strut diameter). This

makes the heat exchanger more efficient, but because no account of the mantle surface A is made, the absolute heat transfer will decrease. A volume goodness factor is a more general approach: the heat transfer surface is plotted against the friction power consumption per unit volume. It can however only be used in geometries with a same hydraulic diameter. Due to this latter, it cannot be used to compare metal foams with louvered fins. The PEC used in this paper is a representation of the thermal conductance on the Kays and London friction factor [12], , taken into account pressure drop and heat transfer. 2.4. Error analysis To ensure the quality of the measurements an extensive error propagation is made. The uncertainties on the thermodynamic properties of water are calculated according to the The International Association for the Properties of Water and Steam (IAPWS IF-97). The uncertainties on the density of the air are calculated using the ideal gas law. Standard error propagation rules as described by Moffat [9] were used to determine the total uncertainty. Although the result from the and LMTD method are the same, use is made of the LMTD method to minimize the uncertainty as in [6]. For the louvered fin heat exchanger, the relative uncertainty on ranges from 4.4% to 6.0%; on around 1%; on from 5.4% to 5.7%; on from 6.8% to 8.3%; on from 1.5% to 4.5%; on the friction factor from 9.6% to 12.1% and on from 11.9% to 15.0%. For the foam heat exchanger, the relative uncertainty on ranges from 4.5% to 5.8%; on from 0.6% to 2.2%; on from 3.4% to 5.2%; on from 5.16% to 5.6%; on from 6.5% to 7.6%; on from 1.5% to 20.6% (high uncertainties in low velocity range); on the friction factor from 10.4% to 15.5% and on from 11.9% to 16.5% (average of 13.3%). 3. Results 3.1. Comparison of both heat exchangers For the determination of the friction factor, the minimal flow area in case of the metal foam heat exchanger is calculated by subtracting the heat transferring surface (for the first tube row in the flow direction) from the total flow area and multiply this with the foams porosity. The surface -to-volume ratio of 440m/m ( ) for this 10 PPI foam is reported in [3]. The total surface area is thus determined by multiplying this specific surface area with the foam volume. Furthermore is calculated with following definition of Vmax, where AHX is the front surface of the heat exchanger (Eq. (2)). (2) For the louvered fins, the friction factor is determined by following Wang et al. [13]. Both friction factors were plotted in function of the mass flow rate in Fig. 5. The friction factor of the louvered fin heat exchanger is surprisingly higher than of the metal foam heat exchanger, indeed if we consider Fig. 4 we can see that the 10PPI foam structure is more open. The friction factors correspond to a pressure drop of respectively 23.9 Pa to 94.48 Pa for the louvered fins and 9.7 Pa to 62.06 Pa for the foam. Remarkably, the friction factor of the metal foam heat exchanger is relatively flat. This is caused by the highly unsteady flow regime occurring within the foam. In laminar flow regime a von Krmn street (unsteady-oscillating) will be induced, whereas for turbulent regime more unsteady vortices will be induced. For this study, the Reynolds number based on the strut diameter and the interfacial velocity ( Darcy velocity, ) ranges from 69.09 to 182.05. The interfacial velocity is calculated as the , on the porosity. The strut diameter is calculated from the measured strut cross

sectional area in the middle of the strut. According to experiments by Seguin et al. [14-15] on porous media, laminar regime end at a pore Reynolds number of 180, whereas a value of 900 corresponds to a stabilization of the vortex velocity gradient fluctuating rate. As a pore Reynolds number of 1256 corresponds in the case of 10 PPI foam with a strut Reynolds number of 69.66, the regime in this study

can be considered highly unsteady. This results in a flat friction profile. This limits further improvements to decrease the friction factor at higher velocities. This phenomenon is also experienced with the louvered fin heat exchanger: at higher mass flow rates (> 0.4 kg/s) the friction factor is also flatting through the unsteady vortices creation. However, the airside resistance of the foam heat exchanger is significantly higher. This is related to the lower available heat transfer area of the foam structure. The used 10 PPI foam has a heat transferring surface of 1.016 m, calculated with the measured surface-to-volume ratio in [3]. This is only 66% of the heat transferring surface of the current state-of-the-art louvered fin heat exchanger (1.544 m). The trend of both series is very similar. For both units the airside heat transfer resistance is the dominating component, accounting up to 92% of the total resistance. The PEC is shown in Fig. 6, and as can be seen, the performance of the louvered fin heat exchanger is better at higher velocity. There is less difference at lower velocities (24% when comparing the average values). For an air velocity of 3.1 m/s the difference in performance is 59%. The reason for this evolution is related to the stepper decline of the airside resistance and the friction factor for the louvered fin heat exchanger.
0.25 Friction factor [-] 0.2 0.15 0.1 0.05 0 0.1 0.2 0.3 0.4 Airside mass flow rate [kg/s] 0.5 PEC [W/K] 1500 1000 500 0 0.1 0.2 0.3 0.4 0.5 Airside mass flow rate [kg/s] louvered fins metal foam 2500 2000 louvered fins metal foam

Figure 5. Kays and London friction factor in function of the airside mass flow rate

Figure 6. The performance factor PEC in function of the airside mass flow rate

To increase the performance of the metal foam heat exchanger, one can consider using other foam types. For example, a 20 PPI foam heat exchanger ( = 720m/m [3]) will have a higher heat transfer surface (1.66 m) which is very comparable to that of the louvered fin unit. But this denser foam will result in a similar increase in the pressure drop. Out of preliminary pressure drop measurements in the same test rig as Fig. 2, the average pressure drop increase in the obtained Reynolds range is 46.1%, whereas the specific surface area increases 64% (720/440). This makes 20 PPI foam interesting in future work. So changing the foam type will improve the situation, but considering the big differences at high velocities, additional measures will be required. 3.2. Influence of the contact resistance As we did not measure the thermal contact resistance in the tested heat exchangers, it is hard to exactly assess the impact of the resistance. However, based on literature we can make a few recommendations. In this paper the louvered fin is made with a collar. Elsherbini and Jacobi [16] reported a collared contact resistance of 9.44 kW m-2 K-1 (press-fit) for a louvered fin heat exchanger. As the fin collars completely overlap the tubes, the contact resistance is 6.108 10-4 K/W. With increasing airside mass flow rate, this represents a relative contribution of the contact resistance over the overall thermal resistance ranging from 7.7% to 11.1%. This is in good agreement with the work of Kim and Jeong et al. [17, 18]. Because of the many small spot-contacts, the pressed-fitted metal foam heat exchanger is expected to have higher contributions. Based on the values of De Jaeger et al. [19], the contact resistance is evaluated to be 0.011 K/W with an uncertainty on of 11.4%, resulting in a

contribution of the contact resistance to the overall thermal resistance from 47.9% to 68.4%. This is comparable with the finding of TJoen et al. [6], using a single epoxy contact. By plotting for both heat exchangers as a function of the airside mass flow rate, the convective resistance of the metal foam actually is lower than that of the louvered fin heat exchanger at higher air velocities (see Fig. 7). The convective resistances cross at an airside velocity of 2.75 m/s. This means that from that point on, the convection coefficient is higher in the metal foam heat exchanger. But, due to the high contact resistance, the metal foam heat exchanger has a lower overall heat transferring performance ( ). This highlights that the contact resistance is in fact the major problem in introducing the metal foam into heat exchangers. However if we also consider the uncertainty on no conclusions can be taken, because the uncertainties for the metal foam heat exchanger are high. The maximum relative uncertainty on is 44.9%, caused by the high uncertainty on the contact resistance.

0.01 0.009 0.008 0.007 0.006 0.005 0.004 0.003 0.002 0.001 0 0.1 0.2 0.3

louvered fins metal foam

Rext [K/W]

0.4

0.5

Airside mass flow rate [kg/s]

Figure 7. The external convective resistance in function of the airside mass flow rate (pressed-fit) 4. Conclusions Two HVAC heat exchangers were compared in a wind tunnel experiment, using a purely thermodynamic PEC. The following conclusions were drawn: It was found that the required pumping power for the metal foam heat exchanger is lower than for its louvered fin counterpart. Furthermore, there is observed that the friction factor flattens-out through the induced highly unsteady flow in the foam, even at low velocities (1.1m/s). This limited further improvements for the decreasing friction factor at higher velocities. According to the predefined PEC, taken as the ratio of the thermal conductance to pressure drop, the current state-of-the-art louvered fin heat exchanger outperformed the metal foam one. However, at low velocities (<1.5 m/s) this difference is acceptable. Furthermore, metal foam has other added values that have to be taken into account: e.g. weight, shape-ability, etc. The contribution of the thermal contact resistance to the overall thermal resistance is experienced to be very high for the metal foam pressed-fit heat exchanger. Contributions up to 58% of the overall thermal resistance were calculated. Whereas the louvered fin heat exchanger, with collared fins, only experience a contribution up to 11.1%. This preliminary work shows the necessity to optimize the metal foam heat exchanger by exploring improved contact technology (e.g. brazing or epoxy bonding) and using a denser metal foam (e.g. 20 PPI). This will be studied in future work. Acknowledgments The authors want to express gratitude to Bekaert for the close cooperation and financial support. They also want to thank Robert Gilles and Patrick De Pue for the technical support.

References [1] Bressand F, Farrell D, Hass P, Morin F, Nyquist S, Remes J, Roemer S, Rogers M, Rosenfeld J and Woetzel J 2007 Curbing global energy-demand growth: The energy productivity opportunity, San Francisco, McKinsey&company, pp. 48. [2] He J, Liu L and Jacobi A M 2010 J. Heat. Transf. 132 071801 [3] De Jaeger P, TJoen C, Huisseune H, Ameel B and De Paepe M 2011 J. Appl. Phys. 109 103519 [4] Bonnet J P, Topin F and Tadrist L 2008 Transport Porous Med. 73 233 [5] Sertkaya A A, Altinisik K and Dincer K 2012 Exp. Therm. Fluid. Sci. 36 86 [6] T'Joen C, De Jaeger P, Huisseune H, Van Herzeele S, Vorst N and De Paepe M 2010 Int. J. Heat. Mass. Tran. 53 3262 [7] Park Y G 2010 Exp. Therm. Fluid. Sci. 34 720 [8] Liu P S 2010 Phil. Mag. Lett. 90 447 [9] Moffat R J, Eaton J K and Onstad A 2009 J. Heat. Transf. 131 011603 [10] Ghosh I 2009 Int. J. Heat. Mass. Tran. 52 1488 [11] Kim S Y, Paek J W and Kang B.H. 2000 J. Heat. Transf. 122 572 [12] Shah R K and Sekulic D P 2003 Fundamentals of heat exchanger design (John Wiley & sons) [13] Wang C C, Lee C J, Chang C T and Lin S P 1999 Int. J. Heat. Mass. Tran. 42 1945 [14] Seguin D, Montillet A and Comiti J 1998 Chem. Eng. Sci. 53 3751 [15] Seguin D, Montillet A, Comiti J and Huet F 1998 Chem. Eng. Sci. 53 3897 [16] ElSherbini A I, Jacobi A M and Hrnjak P S 2003 Int. J. Refrig. 26 527 [17] Kim C N, Jeong J and Youn B 2003 Int. J. Refrig. 26 900 [18] Leong J, Kim C N and Youn B 2006 Int. J. Heat. Mass. Tran. 49 1547 [19] De Jaeger P, TJoen C, Huisseune H, Ameel B and De Paepe M 2012 Int. J. Heat. Mass. Tran., accepted. DOI: 10.1016/j.ijheat masstransfer.2012.06.043