A 3-D Numerical Simulation of Temperature Distribution across the Contact Surface of a Small-scale Hot Press Machine for PEFC Applications Naren Chaithanee 1, Patcharawat Charoen-amornkitt 2, Kotchakarn Nantasaksiri 2 and Nuttapol Limjeerajarus 2, * 1 Automotive Engineering Program, Faculty of Engineering, Thai Nichi Institute of Technology 2 Research Center for Advanced Energy Technology, Faculty of Engineering, Thai Nichi Institute of Technology 1771/1 Pattanakarn, Suan Luang, Bangkok 10250, THAILAND *Corresponding Author: Email: nuttapol@tni.ac.th, Tel.: +66-2763-2600 Ext. 2922, Fax.: +66-2763-2600 Ext. 2900 Abstract The polymer electrolyte fuel cell (PEFC) is one of the most promising renewable energy converters for future power generation. To make a high performance PEFC, membrane electrode assembly (MEA), which is the heart of a PEFC, must be assembled at appropriate conditions (pressure and temperature). The most common method of PEFC MEA assembly is the hot press method by which the temperature of contact surfaces of a hot press machine must be controlled to be uniformly distributed at the glass transition temperature of a Nafion polymer. However, the contact surfaces are required to be as smooth as possible, and thus no thermocouple can be placed on the contact surfaces. Therefore, in this study, a 3-D numerical simulation has been developed on ANSYS software to study the temperature distribution across a contact surface of a small-scale hot press machine specifically designed and developed in our laboratory. The accuracy of the model was validated with experimental data obtained from thermocouple-placeable points around the pressing block. Finally, for PEFC MEA applications, a suitable waiting time required to ensure the uniform temperature distribution across the 49 cm 2 in the middle of the contact surface was suggested to be 1250 s. Keywords: PEFC, Membrane electrode assembly, Hot press machine, Temperature distribution, Numerical simulation 1. Introduction Global warming is widely known as a major problem of our society. One important cause of which is the continuing increase in carbon dioxide level in the atmosphere, which is mainly produced by the fossil fuel-based energy converter. Thus, finding a new cleaner power source is becoming a main interest in a research study. Polymer electrolyte fuel cell (PEFC) has received a high attention as a strong contender of an alternative power source for automotive and stationary applications because of their high energy conversion efficiency, zero greenhouse gas emission, low operating temperature and pressure, high power densities, low noise and fast start ups. One of the key components of a PEFC is the membrane electrode assembly (MEA) at which the electrochemical reaction occurs and the
electricity is generated. To obtain a high performance PEFC, the MEA is normally assembled by the hot press method at an appropriate pressure and temperature [1]. Especially for the temperature, it needs to be distributed uniformly across the contact surfaces at the glass transition temperature of Nafion polymers, which vary from 110 130 C depending on their types [2, 3], so that the catalyst layers and the membrane can be well connected. The distribution of the temperature, however, cannot be controlled easily since the contact surfaces are required to be as smooth as possible. Therefore, a thermocouple cannot be placed on the contact surfaces and thus, the temperature of the surfaces cannot be measured. Since the temperature cannot be known, this study attempts to create a 3 dimensional model by using finite element method (FEM) in ANSYS software to investigate the distribution of the temperature across the contact surface. The simulation results would suggest the required waiting time, which ensures that the temperature is going to distribute uniformly. 2. Theory The temperature distribution in the heating block due to the conductive heat flow is governed by the Fourier-Biot equation [4] as follows: ρc ( T ) + ((K )T) = t e gen (1) where ρ is the density, c is the specific heat, T is the temperature, t is the time, K is the conductivity vector, and e gen is the heat generation rate per unit volume. The heat loss due to the convection heat transfer mechanism on the exposed surfaces is governed by: {q} T {η} = h f (T s T B ) (2) where {q} is the heat flux vector, {η} is the unit normal vector, h f is the convection heat transfer coefficient (or film coefficient), T s is the surface temperature, and T B is the bulk temperature of the adjacent fluid, i.e., air. The heat transfer coefficients of the surfaces were obtained from Eq. (3), as follows: Nu = h fl c k f (3) where Nu is the Nusselt number, L c is the characteristic length, and k f is the thermal conductivity of adjacent fluid. Since there is no external supply to force the adjacent fluid of the heating block to move and the surface temperature is not high, natural convection is considered as a type of this convection mechanism. Thus, for the side surfaces, Nusselt number can be calculated by using the Churchill and Chu s relation in Eq. (4) [5]. where, Nu = {0.825 + 1/6 0.387Ra L [1+(0.492/Pr) 9/16 ] 8/27} (4) Ra L = Gr L Pr = gβ(t s T B )L c 3 ν 2 Pr (5) where Ra L is the Rayleigh number, Pr is the Prandtl number, Gr L is the Grashof number, g is the gravitational acceleration, β is the coefficient of volume expansion, and ν is the kinematic viscosity of adjacent fluid. The following Eqs. (6) and (7) describe natural convection of the upper surface and the lower surface, respectively [6]. and, Nu = 0.54Ra L 1/4 (6) Nu = 0.27Ra L 1/4 (7)
Since the type of convection heat transfer is natural convection, therefore the thermal radiation needs to be taken into account. The thermal radiation energy exchange between the surfaces and its surrounding (Q S ) was calculated by: Q s = A s εσ(t 4 4 s T surr ) (8) where A s is the surface area through which thermal radiation takes place, ε is the emissivity of the surface, σ = 5.67 x 10-8 W/m 2 K 4 is the Stefan-Boltzmann constant, and T surr is the temperature of the surroundings, which is assumed to be constant and is equal to the ambient temperature. Therefore, the heat transfer mechanisms of the heating block, which included in the model, are displayed in Fig. 1. It has been widely known that a straight rod heater, which is used for heating the heating block, cannot uniformly produce the temperature distribution. Therefore, the heat generated from the heater may not be constant through all contact areas and thus, the temperature distribution needs to be carefully checked to calculate for the heat flow through each area by using this following relation. Q i,j = A iδt i,j t j (9) where A is the heating area, ΔT is the difference between starting temperature and ending temperature of the heater, t is the time of heating from starting temperature to ending temperature, i indicates the section of heating area, and j indicates the time. The assumptions that were made on the model are listed below: Constant convection heat transfer coefficient irrespective of the change in the film temperature. The convection heat transfer coefficient was calculated by assuming the surface temperature to be constant through the surface area and equal to mean temperature. The view factor was negligible. Thermal grease is completely filled in the air gap between the heater and the heating block. The emissivity of the heating block is assumed to be constant, irrespective of the direction and wavelength (i.e., diffuse and gray surface). Isotropic and homogenous properties of all parts were assumed. Radiation and convection in other parts were negligible. Fig. 1 Schematic of the heat transfers of the heating block 3. Research Methodology 3.1 Experimental setup The heating block, which is the focus of this research, was made of ST-37 steel and coated by hot-dip galvanization, as displayed in Fig. 2. Since the heat flux along the straight rod heater
(mm) TSF010 is not uniform, the thermocouples were placed, as shown in Fig. 3, to collect the data every 10 seconds for 50 seconds. After the results were obtained, the heat flow through each area of the heater were calculated by using Eq. (9) and cross-multiplication. (a) (b) Fig. 2 The contact surface and the heating block Fig. 3 Positions on a heater at which thermocouples were attached to determine heat flux in each section of the heater The heating system was consisted of three 300W heaters. Since the thermocouples (k-type) could not be placed directly onto the contact surface, they were instead placed to measure the temperature at 1-mm beneath the contact surface (T th ), front side surface (T 1 ), right side surface (T 2 ), rear side surface (T 3 ), and left side surface (T 4 ) of heating block (as shown in Fig. 4). The temperatures were collected every 10 seconds for 500 seconds and have been used for validating the model. Finally, time counting were done for achieving the on/off signal of heating. The temperature and the signal were collected for 30 minutes and used for validating the model. Thermocouple Fig. 4 a) Schematic of thermocouple position on the heating block b) Section A-A of the heating block 3.2 Simulation The purpose of the simulation of this study is to investigate the temperature distribution across the contact surface of the heating block for suggesting an appropriate waiting time. The 3-D geometric model was mainly built in CATIA software after that the model was imported and modified in ANSYS software. For the meshing, a total of 130,835 elements was used in the model. The element types were a combination of tetrahedral, hexahedral, wedge, and pyramid elements in which the tetrahedral was the main element type. Since, in practice, thermal grease was filled in the air gap between the heater and the heating block in order to enhance the conduction heat transfer, the thermal grease was also included in the model. The convection heat transfer coefficients of the upper surface, the 4 side surfaces, and the lower surface were
calculated and set at 8.081, 8.076, and 4.041 W/m 2 K, respectively. Since the spectrometer was not available in our laboratory, the emissivity was assumed to be constant (c.a. 0.4) based upon the value of a galvanized surface suggested in the previous study [7]. Once 5 points temperature results were validated, the heat flows were input varying with time as collected from the experiment. Finally, from observing the temperature distribution across the contact surface, the waiting time for which the distribution does not differ for more than ± 5 C within 49 cm 2 in the middle of the surface can be suggested. The computational time was about 30 minutes and 1 hour for validation and heat flow variation, respectively, in Intel Xeon E5-2620 2 GHz processor of 32 GB RAM and 4 GB graphic card memory. A0 A1 A2 A3 Fig. 5 Temperature distribution of the heater in different times 4. Results and discussion 4.1 Experimental results The temperature distribution of the heater, as shown in Fig. 5, were divided into 4 sections (A0 A3). From calculation, the heat flow through each section in every 10 seconds, which are displayed in Fig. 6, did not significantly change with time. The average values of heat flow were 27 W, 93 W, 591 W, and 189 W in the area A0, A1, A2, and A3, respectively. 4.2 Simulation results For validation, a good agreement has been clearly seen between the simulation and the experimental results of the temperatures at the 5 different points (T 1 T 4, and T th ), as depicted in Fig. 7 Fig. 6 Calculated heat flow at different times There was a difference between the simulation and the experimental results because of the model assumption that treats convection heat transfer coefficient as a constant. However, the convection heat transfer coefficient actually varies with temperature, and the temperature varies with position. As the film temperature is higher, the convection heat transfer coefficient increases.
Fig. 7 Comparison of experimental and numerical results on 5 point temperatures of the heating block Therefore, the difference between the value of actual convection heat transfer coefficient and the constant one is larger at the higher temperatures, and so does the discrepancy between the simulation and experimental results. The effect of the unrealistic convection heat transfer coefficient, which results in the discrepancy, was much greater on the surfaces (T 1 T 4 in Fig. 6a d, respectively) than that inside the heating block (T th in Fig. 6e) where the heat conduction mechanism is dominant.
Prior to the prediction of temperature distribution across the contact surface, the model must be validated for its accuracy on which the simulation is done under the real on/off heating operation of the heating block. In this regard, the temperature T th was selected for the validation. The on/off signal made the temperature profile zig-zag near the desired temperature (130 C), as seen in Fig. 7. Overall, Fig. 8 illustrates that our model can satisfactorily simulate the real on/off heating operation of the heating block. Fig. 8 Comparison of experimental and numerical results on T th when controlling the temperature at 130 C Since the MEA size used in our lab is only 5 cm 2, an area of 49 cm 2 in the middle of the contact surface is sufficient for the MEA to be placed. The temperature fluctuation within ±5 C across that 49 cm 2 area is acceptable for our application. Therefore, from simulation, the highest and the lowest temperatures of the middle area at different times were plotted and presented in Fig. 9. It was found that a suitable waiting time for heating was at least 1250 seconds, approximately. Fig. 9 Maximum and minimum temperatures on the 49 cm 2 area in the middle of the contact surface over time This result was in concurrence with the temperature distribution across the middle area of the contact surface at different times after 1250 seconds, as displayed in Fig. 10. The results in Fig. 10 showed that the temperature in the middle area was uniformly distributed within the acceptable range of ±5 C. The maximum temperature difference was only about 4 C and 6 C when the heater was off and on, respectively (Fig. 10a and b). Therefore, from the simulation, the suitable waiting time for achieving a uniform temperature distribution at about 130 C across the 49 cm 2 in the middle of the contact surface was suggested to be 1250 s. Nevertheless, the accuracy of the model can be improved by taking realistic convection heat transfer coefficient and the view factor of heat radiation into account. This modification has been studied and is ongoing in our group. In addition, a comparison with other techniques, such as infrared thermometer, could also be done further to confirm the accuracy of the model.
(a) C (b) C Fig. 10 Temperature distribution across the middle area of the contact surface at a) 1250 s (heater-off) and b) 1300 s (heater-on) 5. Conclusion In summary, a 3-D numerical model of a heating block of a hot press machine for PEFC application has been developed and validated. The model was used to investigate the temperature distribution across the contact surface. The results revealed that the suitable waiting time of 1250 seconds is needed to ensure that the temperature at about 130 C will be uniformly distributed within an acceptable range of ±5 C. 6. Acknowledgement The authors would like to express our special thanks for the financial support funded by Thai- Nichi Institute of Technology. 7. References [1] Therdthianwong, A., Manomayidthikarn, P., and Therdthianwong, S. (2007). Investigation of membrane electrode assembly (MEA) hotpressingparameters for proton exchange membrane fuel cell, Energy, vol.32, December 2007 (12), pp. 2401 2411. [2] Jalani, N. H., Dunn, K., and Datta, R. (2005). Synthesis and characterization of Nafion - MO 2 (M = Zr, Si, Ti) nanocomposite membranes for higher temperature PEM fuel cells, Electrochimica Acta, vol.51 (3), October 2005, pp. 553 560. [3] Jung, H.- Y., and Kim, J. W. (2012). Role of the glass transition temperature of Nafion 117 membrane in the preparation of the membrane electrode assembly in a direct methanol fuel cell (DMFC), International Journal of Hydrogen Energy, vol.37 (17), September 2012, pp. 12580 12585. [4] ANSYS, Inc. (2009). Theory Reference for the Mechanical APDL and Mechanical Applications, ANSYS, Inc., Southpointe. [5] Churchill, S. W., Chu H. H. S., (1975). Correlating Equations for Laminar and Turbulent Free Convection from a Vertical Plate, International Journal of Heat Mass transfer, vol.18 (11), November 1975, pp. 1323-1329. [6] Cengel, Y. A., and Ghajar, A. J. (2011). Heat and Mass Transfer: Fundamentals and Applications, 4 th Edition, ISBN: 978-007-131112-0, McGraw - Hill, Singapore. [7] Incropera, F. P., Dewitt, D. P., Bergman T. L., and Lavine, A. S. (2007). Introduction to heat transfer, 5 th Edition, ISBN: 978-047-179472-1, John Wiley & Sons, Asia.