INTERNATIONAL JOURNAL OF MECHANICAL ENGINEERING AND TECHNOLOGY (IJMET)

INTERNATIONAL JOURNAL OF MECHANICAL ENGINEERING AND TECHNOLOGY (IJMET) Proceedings of the 2 nd International Conference on Current Trends in Engineering and Management ICCTEM -2014 ISSN 0976 6340 (Print) ISSN 0976 6359 (Online) Volume 5, Issue 9, September (2014), pp. 167-173 IAEME: www.iaeme.com/ijmet.asp Journal Impact Factor (2014): 7.5377 (Calculated by GISI) www.jifactor.com IJMET I A E M E THERMAL ANALYSIS OF AIR FLOW IN A CPU CABINET WITH MOTHERBOARD AND HARD DISK AS HEAT SOURCES SUBHAS. L. HUNASIKATTI 1, SUNEEL. M. P 2, P. S. KULKARNI 3, G. S. SHIVA SHANKAR 4 1, 4 Department of Mechanical Engineering, S.I.T, Tumkur, Karnataka, India 572103 2, 3 Department of Aerospace Engineering, IISc, Bangalore, Karnataka, India 560012 ABSTRACT The present work investigates the numerical simulation of thermal analysis of mixed convection air flow in a CPU Cabinet. The simulation is focused on the non-uniformly heated mother board temperature distribution. In the present work three cases have been studied, 1) Placing the CPU in vertical position, 2) Placing the CPU in horizontal position and 3) Providing exhaust fan on top. The work also includes studies of effectiveness of different inlets provided. The temperature distribution of the components and streamlines were investigated in order to get a clear picture of which case is more effective for cooling of the mother board. The simulation was carried out using a standard commercial CFD code-ansys-fluent. It is found that horizontal position results in reduction of motherboard average temperature of 0.1 0 C as compared to vertical position. It is also observed that bottom opening has very less effect on motherboard temperature. Keywords: CPU Cooling, Electronic Cooling, Forced Cooling of Electronic Devices and Mixed Convection. 1. INTRODUCTION Every system has a critical temperature and it is very crucial to avoid reaching this critical temperature. When passing the manufacturer's critical temperature, the system starts to fail or even make the system to work off. As a result, the heat transfer in the electronic systems must be examined before designing the application [1]. There are many ways in which possible air-cooling design can fail; inadequate coolant flow, poor flow distribution, poor mixing and low heat transfer coefficient or power dissipation higher than the expected. To analyze these types of causes CFD approach is good [2]. The cooling of electronic systems is essential in controlling the temperature and avoiding any hot spots. Cooling of electronic equipment is one of the most important factors to be considered in designing electronic devices and in making them energy efficient. Desktop computers are one among the most commonly used devices in the world. Hence making the desktop computers, energy efficient can effectively reduce the electricity consumption. The power consumption due to exhaust fans being used in the cabinets becomes significant when considered in mass. The maximum amount of heat dissipation in the CPU cabinet is because of processor, chipset and hard disk drive. Many researchers have studied heat transfer effects around the electronic devices. T. Y. Tom Lee, used the CFD technique to evaluate the temperature and velocity fields of air flow in a computer system enclosure [3]. Yu and Webb have mentioned total cabinet power dissipation being 313 W [4]. They have also given the details of power dissipation from individual components such as Memory, Chipset, AGP and PCI cards. Another computational study has been carried out by Konstantinos-Stefanos P.NIKA S and Andreas D. PANAGIOTOU [1]. In this study the investigation has been made regarding the air flow circulation, and the temperature distribution profiles occurred over non-uniformly heated printed circuit board. Based on the position, geometrical characteristics and the number of fans have been altered to study the thermal effects. In their study the inlet fans and their 167

geometrical characteristics have been altered in order to identify which is the best case that produced the maximum heat transfer results [1]. In past studies some people have tried with structured grid for their simulation [3, 5] only a few of them have used unstructured grid [1]. The structured mesh has been given better results as compared to unstructured mesh. Different turbulence models have been tested and compared [1] [6], for these kind of problems k-ɛ Turbulence model produces the best approach as far as the comparison between numerical and experimental data [1]. From the literature survey it is found that there are no studies on the heat dissipation and fluid flow variations inside the CPU cabinet in vertical and horizontal positions and there are no studies on which inlet opening is more effective for cooling of the motherboard. This is the motivation of the present work. In the present work, we have considered the 3-D, steady state, mixed convection air cooling of a non-uniformly heated mother board inside the cabinet. Heat dissipation from the mother board and hard disk drive have been studied. The simulations are carried out using industrial standard commercial CFD code ANSYS-Fluent. 2. PROBLEM DESCRIPTION The objective of the present work is to analyze the airflow pattern in a CPU Cabinet and temperature distribution of motherboard for different orientation and different location of openings. The problem essentially needs geometry simplification to reduce the complexities which may arise in grid generation. It is also required to use multiblock approach for grid generation since the problem involves many heat sources and boundary conditions. The orientation of the CPU Cabinet will be altered in order to identify which orientation is the best case that produced the maximum heat dissipation. The inlet opening of the CPU-Cabinet will be altered in order to identify which inlet opening is more effective for cooling of the mother board. The detailed geometry is shown in Fig.1. 3. NUMERICAL METHODOLOGY Figure 1: Basic Geometry of CPU Cabinet ANSYS-FLUENT 14.5 commercial package software has been used for simulation. The current study uses the finite volume method and the equations solved are the continuity, momentum and energy equation. From the past studies, we have considered k-ε is the best turbulence model for this kind of problem as compared to the other turbulence models and the same has been used in the present work. The present study is performed under steady state conditions and air is assumed as incompressible. Boussinesq approximation is used to account buoyancy. SIMPLE algorithm is used for pressure-velocity coupling. 4. GOVERNING EQUATIONS RANS (Reynolds-averaged Navier-Stroke) equations are solved for steady state flow. The equations for continuity, momentum and energy are shown below. Where equations 1 is the continuity equation and equations 2, 3 and 4 are the momentum equation in x, y and z direction respectively and the equation 5 is the energy equation: = 0 (1) 168 (2)

(3) (4) (5) Where in equation 3 is a source term which takes gravity effects into account. 5. BASIC ASSUMPTIONS In order to simplify the present work the following basic assumptions made are. The motherboard is considered as non-uniformly heated. Radiation heat transfer is neglected because of the domination of forced convection and relatively low temperature differences inside the cabinet. The surrounding air pressure and temperature are considered as 1 atmosphere and 35 0 C. Mother board thickness is considered as 5mm. Conduction through the cabinet wall is neglected. CPU total heat dissipation is taken as 80 W/m 2. 6. BOUNDARY CONDITIONS The exhaust fan and processor fan are 80 mm in diameter. As per the manufacturer specifications the velocity of the 80 mm diameter fan is 2.85 m/s. The cabinet is simulated as zero wall thickness, but only the wall-faces to form the boundary with no slip condition. The maximum amount of heat dissipation from the CPU is 80 W and the chipset is 10 W. The hard disk drive dissipates 20 W of heat. 7. GRID DEPENDENCE STUDY A grid dependence study has been carried out successfully based on the following parameters. Motherboard average temperature, Exhaust fan average temperature, velocity; SMPS exhaust average temperature, velocity and inlet average velocity. Figure 2: Mesh Generated by ICEM CFD In the present case, the multiblock structured grid has been used. Total 9 different grid was tested, out of them 4- cases with first element height from the wall is 0.5mm and 5-cases with first element height from the wall is 0.1mm with growth rate at 1.2. 169

Average Temperature in 0 C 33 32 31 30 0.5mm first element height 0.1mm first element height 0 5 10 15 20 25 Number of Elements in Lakhs Figure 3: Variation of Average Motherboard Temperature for Different Grid Size The cells were denser in the region near to the boundary surfaces and coarse in the middle region of the domain. Finally the 5-Lakh element denser grid has been used for further study, since it was found to take less computational space and time. Fig.3. shows the variation of average motherboard temperature for different grid size. It can be seen that there will not be much change of the results after increasing the mesh elements above 5-Lakhs. 8. RESULTS AND DISCUSSION In this work three different cases have been studied by placing the CPU in vertical and horizontal orientations, the details of the cases studied are listed in TABLE 1. Temperature distribution and corresponding streamlines are used for the thermal analysis of CPU cabinet. Table 1: Analysis of Simulated Cases Cases simulated Description of each case A Simulation of CPU Cabinet-Vertically oriented. B Simulation of CPU Cabinet-Horizontally oriented. C Simulation of CPU Cabinet-Modified based on analysis of case A and case B. 8.1 Case A: Simulation of CPU Cabinet with vertical orientation The first case refers to simulation of CPU Cabinet with vertical orientation. In general, a Cabinet has two fans, exhaust fan and processor fan. One is located at the back side of the cabinet and the other one is placed over the processor heat sink as fluid outlet. The temperature distribution over the surface of the motherboard is illustrated in fig. 4. The simulation showed that the maximum temperature is up to 65.9 0 C and it is concentrated over the center of the chipset, around 55.3 0 C temperature takes place at the center of the processor, this is due to the maximum amount of heat dissipation is from the processor and chipset. The overall average temperature of the motherboard is 37.7 0 C. The stream lines shown in Fig.5. is the clear picture about how the air is moving inside the cabinet and also it shows the variation of motherboard temperature due to variation of local air velocity and temperature. Figure 4: Temperature distribution contours for vertical orientation Figure 5: Air streamlines for vertical orientation 170

8.2 Case B: Simulation of CPU Cabinet- horizontal orientation The second case refers to the simulation of CPU Cabinet with placing it in horizontal orientation, keeping all other parameters same as that of case A except acceleration due to gravity in the negative x-direction. The temperature distribution over the surface of the motherboard is shown in Fig.6. The simulation showed that the maximum temperature of the chipset is around 63.3 0 C and that of motherboard is around 55.3 0 C. Again, this is due to higher amount of heat dissipation from the chipset and motherboard as compared to other parts (Hard disk drive, DVD, etc.). The overall average temperature of the motherboard is 37.6 0 C.This shows that 0.1 0 C less than that of case A, in this case the maximum amount of fresh air will contact over the surface of the motherboard (refer Fig.7) and also case B shows better cooling of the motherboard than case A. Figure 6: Temperature distribution contours for horizontal orientation Figure 7: Air streamlines for horizontal orientation 8.3 Case C: Simulation of CPU Cabinet-vertical orientation and changing exhaust fan location In case A and B, the maximum quantity of hot air after hitting the surface of the motherboard and hard disk drive will move towards the top portion of the cabinet. So we think that instead of placing an exhaust fan at the back side of the cabinet, placing it at the top portion of the cabinet then the fresh air comes in giving better cooling effect. The temperature distribution and airflow inside the cabinet in this case is illustrated in Fig.8 and Fig.9 respectively. The total average temperature of the motherboard that can be obtained is 37.7 0 C. Figure 8: Temperature distribution contours for exhaust fan on top Figure 9: Air streamlines for exhaust fan on top 8.1.1 Effectiveness of left side inlet in CPU Cabinet In the above simulated cases the left side inlet is mainly helpful for entering fresh air into the cabinet. When the fresh air enters into the cabinet it will directly hit over the surface of the motherboard and extract the heat from the mother board after that it will spread in different location of the cabinet and finally it will move out through the exhaust 171

fan. The Fig.10 shows the distribution of air inside the cabinet. These stream lines show that the left side inlet is mainly helpful for cooling of the motherboard, not much affecting for cooling of the hard disk drive. Figure 10: Air streamlines for left inlet opening. 8.1.2 Effectiveness of bottom inlet in CPU Cabinet As like left side opening the bottom side opening is also helpful for entering fresh air into the cabinet. When the fresh air enters into the cabinet some portion of the air will hits the bottom surface of the hard disk drive and extract the heat from that surface, After hitting the surface of the hard disk drive the surface gets cooled, air extract the heat from that surface, air becomes hot. There is no contact between air and motherboard. Finally the hot air moves out through the exhaust fan and processor fan. The fig. 11 shows the distribution of air inside the CPU Cabinet. These stream lines show that bottom side inlet is mainly helpful for cooling of the hard disk drive not much affecting for cooling of the motherboard. The comparison of temperature at different locations for all simulated cases is shown in TABLE 2. Fig. 11: Air Streamlines for Bottom Inlet Opening Table 2: Different Components Average Temperature at Different Cases Variables Case A * Case B * Case C * Processor average temperature 48.4 0 C 48.5 0 C 48 0 C HDD average temperature 35.2 0 C 35.2 0 C 35.2 0 C Chipset average temperature 62.6 0 C 62.4 0 C 64.4 0 C Motherboard average temperature 37.7 0 C 37.6 0 C 37.7 0 C Processor fan exhaust average temperature 35.3 0 C 35.3 0 C 35.3 0 C Exhaust fan average temperature 35.2 0 C 35.2 0 C 35.3 0 C (*A, B, C - refer table 1.) 172

CONCLUSION This case study demonstrates the capability of computational fluid dynamics software in predicting flow field and heat transfer in an air-cooled computer system. The conclusions from the present case study are: Horizontal case shows average motherboard temperature 0.1 0 C, chipset average temperature 0.2 0 C less than that in the vertical case. The study suggests less fan power requirements in case of horizontal position as compared to vertical position. Left side inlet is mainly helpful for cooling of the motherboard and not much affecting for cooling of the hard disk drive. Bottom side inlet is mainly helpful for cooling of the hard disk drive and not affecting for cooling of the motherboard. Top exhaust fan case shows average motherboard temperature is same as in the vertical case and 0.1 0 C more than that in the horizontal case; processor average temperature is 0.4 0 C less than that in the vertical case and 0.5 0 C less than that in the horizontal case; chipset average temperature 0.2 0 C less than that in the vertical case and is same as that in horizontal case. The study suggests less fan power requirements in case of a top exhaust fan as compared to the existed a vertical case and more fan power requirements in case of a top exhaust fan as compared to the existed horizontal case. REFERENCES [1] Konstantinos-Stefanos P. Nikas and Andreas D. Panagiotou, Numerical Investigation of Conjugate Heat Transfer in a Computer Chassis, Columbia International publishing journal of Advanced Mechanical Engineering, (2013)1, 40-57. [2] Robert J. Moffat, Getting the Most out of your CFD Program, The Eighth Inter Society Conference on Thermal and thermomechanical Phenomena in Electronics Systems, 2002, 9-14. [3] T. Y. Tom Lee, and Mali Mahalingam, Application of a CFD Tool for System-Level Thermal Simulation, IEEE Transactions on components, packaging, and Manufacturing Technology-Part A, (17)4, 1994. [4] C. W. Yu, and R. L. Webb, Thermal Design of a Desktop Computer System Using CFD Analysis, Semiconductor Thermal Measurement and Management, Seventeenth IEEE Symposium, 2001. [5] Christopher W. Argento, Yogendra K. Joshi, and Michael D. Osterman, Forced Convection Air-Cooling of a Commercial Electronic Chassis: An Experimental and Computational Case Study, IEEE Transactions on Components, Packaging, and Manufacturing Technology-Part A, 1996, 248 257. [6] Masud Behnia, Wataru Nakayama, and Jeffrey Wang, CFD Simulations of Heat Transfer from a Heated Module in an Air Stream: Comparison with Experiments and a Parametric Study, 1998 InterSociety Conference on Thermal Phenomena. [7] Rebecca Biswas, Evaluation of Airflow Prediction Methods in Compact Electronic Enclosures, Master s Thesis, San Jose State University, 1998. [8] Seri Lee, Optimum Design and Selection of Heat Sinks, Eleventh IEEE SEMI-THERM Symposium, 1995. [9] http://www.qats.com/cms/2011/06/13/selecting-a-fan-for-your-thermal-anagement-system-part-1-of-2/, (20/08/2014). 173