ISSN: ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 7, January 2014

Transcription

1 ISO 91:28 Certified Volume 3, Issue 7, January 214 Forced Convection Cooling of a 3D Protruding Heaters Array wit Laminar Flow Of Dielectric Fluid Felipe Baptista Nisida & Tiago Antonini Alves Federal University of Tecnology Paraná/Campus Ponta Grossa, Brazil Abstract In te present work a numerical analysis was performed to investigate te forced convection eat transfer wit laminar flow of te dielectric fluid Novec TM 75 from a 3D protruding eaters array mounted in cross-stream direction on an adiabatic substrate of a orizontal rectangular cannel using te ANSYS/Fluent TM 15. software. A uniform eat generation rate was assumed in te 3D protruding eaters and te cooling was performed by means of a forced fluid flow wit constant properties in te laminar regime under steady state conditions. At te cannel inlet, te flow velocity and temperature profiles were assumed uniform. Te governing equations and teir boundary conditions were numerically solved in a single domain troug a coupled procedure using te Control Volumes Metod. Te SIMPLE algoritm was used to solve te pressure-velocity couple. Te discretization of te convective-diffusive terms was performed using te Second-order upwind sceme. Due to te non-linearity of te momentum equation, te correction of te velocity components and te pressure were under-relaxed to prevent instability and divergence. After a computational mes independence analysis, te numerical simulations were obtained and displayed as a 3D non-uniform mes wit 212,67 control volumes. Tis computational mes was more concentrated near te solid-fluid interface regions due to te larger primitive variable gradients in tese regions. To obtain te numerical results, typical properties values and geometry dimensions found in forced convection cooling wit dielectric fluid of electronics components mounted in a printed circuit board were used. An investigation was done on te effects of te ynolds numbers ranging from 1 to 3. Te termal parameters of interest, suc as, temperature distribution, local and average adiabatic Nusselt numbers, local and average eat transfer coefficients, convective and overall termal conductance, were found and compared, wen possible, wit te available results in te literature for te air as te cooling fluid. Index Terms Array of 3D Protruding Heaters, Dielectric Fluid, Forced Convection, Laminar Flow, exponentially wit te work temperature tat sould not exceed a value between 85 C and 1 C [2]. Te possible causes of te failures are te diffusion of te semiconductor material, te cemical reactions, te movement of te glued materials and te termal tensions [3]. In special applications, e.g., supercomputers were te eat generation is excessive and te space used for eat transfer is limited, te use of non-conventional and ig cost cooling tecniques is required. Dielectric fluids are utilized for te proper termal control of te electronic packaging in question. In a dielectric fluid cooling system, one problematic factor tat causes concern is te maintenance, because of te importance of te fluid s discard and te risk of intoxication as a result of andling it. Terefore, te selection of a eat transfer fluid for semiconductor processing equipment and electronics cannot be treated wit minor importance anymore, because environmental problems became a critical factor in te decisions of manufacture operations and project of computers. Tere is te need for ig performance and long term solutions, aiming for a low maintenance necessity and tis way causing a smaller environment impact [4]. In te present work, problems motivated by te Level 2 of electronic packaging, associated wit te termal control of an array of 3D protruding eaters mounted on a printed circuit board (PCB) were considered, as sown in Fig. 1 [5]. A dielectric fluid was considered as te cooling fluid. Te available space for te eaters can be limited and te cooling process must be done troug forced convection wit moderate velocities due to operational limitations and noise reduction. Under suc conditions, tere may not be enoug space to work wit eat sinks in tese concentrate eat dissipation components. Tese components can be simulated by 3D protruding blocks mounted on a substrate [6]. I. INTRODUCTION In te last decade, academic researces and scientifictecnological efforts were developed in order to enance te cooling tecnologies of electronic equipment because, wit te innovation of te modern electronic tecnology, it became faster, smaller and incorporated more functions, resulting in a significant unavoidable increase in te volumetric eat generation rate. Tat is te case of smart pones, notebooks, tablets and computers [1]. Te failure factor of te electronic devices in general increase almost Fig. 1. An array of 3D protruding eaters mounted on a PCB. Te dielectric fluid selected to perform tis work was te Novec TM 75 tat is utilized for termal tests and immersion 48

2 ISO 91:28 Certified Volume 3, Issue 7, January 214 cooling of electronics, sold by te brand 3M TM Novec Engineered Fluids. Tis fluid was cosen due to te adequacy wit te temperature range used and te environmentally friendly properties, assuming, nowadays, te position of one of te dielectric fluids tat causes te least environmental impact. Te fluids Novec TM are a group of materials wit low Global Warming Potential (GWP) and ave excellent properties for eat transfer applications, suc as, dielectric properties, wide range of boiling points and good materials compatibility, in addition to demand little maintenance and to offer safe performance. Tey ave ig resistivity and will not damage electronic equipment or integrated circuits in te event of a leak or oter failure. Furter information about te Novec TM 75 is presented in te manufacturer s catalog [7]. II. LITERATURE REVIEW Nakamura et al. [8] performed experimental studies to investigate te airflow and te eat transfer around a 3D protruding cubic eater mounted on a substrate. Te different temperatures at te eater surface and at te adiabatic substrate were measured under te constant eat flux condition. Te beavior of te flow around te eater surface and te substrate was presented qualitatively. Te pressure distributions and te local Nusselt number in te region close to te cubic eater and to te adiabatic substrate were presented for different ynolds numbers. Te caracteristics of te eat transfer at te eater and at te substrate were correlated wit te caracteristics of te flow around tem. Witin te investigation range, a correlation for te average Nusselt number at te 3D protruding eater was expressed by Nu m = Nakajima et al. [9] presented numerical results of te laminar flow and te eat transfer of rectangular 3D protruding eaters mounted on te surface of a cannel. Tey studied te case of tree rows of 3D protruding eaters. A comparison between te numerical and te experimental results of te laminar flow streamlines was sown. Te numerical investigation was executed for ynolds numbers tat ranged from 1 to 5, considering te Prandtl number equal to.7. Te main caracteristics of te flow around te 3D protruding eaters were te formation of orsesoe vortices and te recirculation regions. Te local friction coefficient distributions at te bottom plate upstream te eaters and at te eater walls were presented. Te temperature distribution at te eater surfaces was also presented. Te eat transfer coefficient varied noticeably at te different surfaces of te eaters and wit te cange of. Te average Nusselt increased wit te increase in te ynolds number. Yagoubi & Velayati [1] numerically studied te eat transfer and te developing flow around a row of cubes in te transversal direction to te airflow, representing 3D protruding eaters mounted on a plate (configuration similar to te Fig. 1). Te main caracteristics of te flow around te eaters were presented. Te velocity profiles and pressure distribution were sown upstream and downstream te cubic eater. Te friction coefficient beavior was presented in function of te ynolds number for different eigts of te parallel plate cannel. Te average friction coefficient decreased wit. Te numerical results found for te average Nusselt number in a 3D protruding eater were compared wit te experimental results of [8] and [11]. Nisida & Alves [12] performed a numerical analysis of te forced convection eat transfer of a row of 3D protruding eaters mounted on te bottom wall of orizontal rectangular cannel utilizing te air as te work fluid. An investigation was done on te effects of te ynolds numbers ranging from 1 to 3. Te beavior of te laminar airflow around te protruding eaters was sowed troug te streamlines. Te temperature distribution, local and average adiabatic Nusselt numbers, local and average eat transfer coefficients, convective and overall termal conductance, were sown. Te average adiabatic Nusselt number was correlated wit deviations not greater tan.5% troug Nu ad = Oter works available in te consulted literature tat contributed wit te air forced convection eat transfer from protruding eater(s) were [13]-[41]. III. MODEL DESCRIPTION Te basic configuration representing te treated problem for one of te 3D protruding eaters is indicated in Fig. 2. In tis case, te cannel as a eigt, H, lengt, L, and widt, W. Te substrate as te same lengt and widt as te cannel wit a tickness, t, and termal conductivity k s. Te eater as a lengt, L, eigt, H, widt, W and it is located at a distance, L u, from te cannel entrance. Te space between te eaters is 2W s. A uniform eat generation rate was assumed for eac of te protruding eaters and te cooling process occurred troug a forced laminar flow wit constant properties under steady state conditions. In te cannel inlet, te velocity profile (u ) and te temperature profile (T ) of te flow were considered uniform. Bot top and bottom cannel surfaces were adiabatic. A. Problem Formulation Te matematical model of te present problem was performed for a single domain: te solid regions (protruding eater and substrate) and te fluid flow in te cannel. Due to te problem symmetries, te conservation equations were formulated for te domain wit lengt, L, widt, W/2 and eigt, (H + t). Te governing equations of te considered domain cover te principles of mass, momentum and energy conservation, Esq. (1), (2) and (3), respectively, under steady state conditions, constant properties and negligible viscous dissipation. Te occasional effects of te natural convection, radiation, oscillation in te flow are not being considered in tis modeling, a typical procedure adopted in similar problems, e.g., [37], [42]-[44]. Mass Conservation (Continuity Equation) u. (1) Momentum Conservation (Navier-Stokes Equation) 49

3 ISO 91:28 Certified Volume 3, Issue 7, January uu p u. (2) L Nu, (8) k Energy Conservation (Energy Equation) p 2 c u T k T S. (3) In te energy equation, δ = 1 in te 3D protruding eater region and δ = in te substrate and te fluid regions. Te boundary conditions of te flow were uniform velocity (u ) at te cannel inlet, and null velocity at te solid-fluid interfaces (no-slip condition). At te cannel outlet, te flow ad its diffusion neglected in te x-direction for tree velocity components. Te termal boundary conditions considered were uniform temperature (T ) at te cannel inlet and negligible termal diffusion in te x- direction at te cannel outlet. Te top and bottom surfaces were adiabatic. Perfect termal contact condition was considered at te interface 3D protruding eater-substrate. Symmetry boundary condition (periodic condition) was applied for te velocity and temperature fields at te lateral boundaries of te solution domain (same geometry and eat dissipation for all te 3D protruding eaters). B. Termal Parameters of Interest Te solution of te governing equations output te velocity and pressure distributions in te considered domain. Te numerical solutions of te primary variables distribution (u, v, w, p) were utilized to define te derived quantities. Te ynolds number in te cannel was based on te 3D protruding eater eigt (H ) expressed by u H. (4) u H Te local eat transfer coefficient, (ξ), was defined based in te difference between te local temperature of te eater surface, T (ξ), and te inlet temperature of te fluid in te cannel T, T q f T, (5) were, q f "(ξ) represents te local eat flux at te eater surface for te fluid flow. Wit te definition of te local eat transfer coefficient, Eq. (5), te eater lengt L was selected as te caracteristic lengt for te local Nusselt number at te eater. Nu L k. (6) Te average eat transfer coefficient and te average Nusselt number of te eaters were respectively defined as were, A cv is te eater surface area in contact wit te fluid flow. Te convective termal conductance (UA) cv is defined as UA cv q f T T were in tis case, U cv coincident wit te eat transfer coefficient (Eq. (7)). Te overall termal conductance (UA) is expressed by UA T q T were, A is te total area of te 3D protruding eater. C. Numerical Solution (9) (1) Te governing equations and teir boundary conditions were numerically solved utilizing te Control Volume Metod [45] troug te ANSYS/Fluent TM 15. software. Te SIMPLE (Semi-Implicit Metod for Pressure Linked Equations) algoritm was used to treat te pressure-velocity couple. Te discretization of te diffusive-convective terms was done troug a Second-order upwind sceme. Te boundary conditions for te laminar flow and te eat transfer were applied at te boundaries of te analyzed domain. Te numerical procedures assumed were verified troug a comparison wit te numerical results of te termo-fluid-dynamic parameters presented by [46]. After a mes independency study, te numerical results were obtained wit a 3D non-uniform mes containing 212,67 control volumes. Tis mes was more concentrated in te regions near te solid-fluid interfaces due to te larger gradients in te primitive variables of tese regions, as sown in Fig. 3. Due to te non-linearity in te momentum equation, te velocity components and te pressure correction were under-relaxed to prevent instability and divergence. Te numerical computations were made wit te use of a microcomputer equipped wit an Intel TM Core i7 3.6 GHz processor and 16 GB of RAM. Te processing time of a typical solution was approximately 15 (fifteen) minutes. A T T cv q f, (7) Fig. 3. 3D non-uniform mes (3D perspective view). 5

4 ISO 91:28 Certified Volume 3, Issue 7, January 214 IV. RESULTS AND DISCUSSION In order to obtain te numerical results, typical design and properties values found in cooling applications of electronic components mounted on a circuit printed board [47]. Te geometric configurations sowed in Fig. 2 were assumed considering a space of H =.254 m between te parallel plates. Te cooling fluid considered in te current study was te dielectric fluid Novec TM 75. Te 3D protruding eaters were considered to be made of pure aluminum and te substrate, adiabatic. Te properties of te fluids and solids were considered constants at 3 K [11]. Te termopysical properties of te dielectric fluid according to te manufacture s catalog [7] were equal to c p = 1,128 J/kg.K, k =.65 W/m.K, μ =.124 Pa.s, ρ = 1,614 kg/m 3 and Pr = Te dissipation rate in eac eater was 2W wic corresponds to a volumetric eat generation rate of 94,55.5 W/m 3. Te effects of te ynolds numbers =1,15, 2, 25, and 3 were investigated. According to Morris & Garimella [48], te flow is laminar in te cannel for tis range of. In Fig. 4, te streamlines around a 3D protruding eater, in a perspective view, are presented for ynolds numbers of 1 and 3, and in Fig. 5, tese streamlines are presented in more detail for te region upstream te protruding eater. Te main caracteristics of te laminar flow are te orsesoe vortices wic start upstream te eater and develop around te eater lateral surfaces; a small recirculation upstream te 3D protruding eater; te detacment of te fluid boundary layer at te top of te eater causing a recirculation; and a large recirculation region downstream te eater due to te flow reattacment. It is interesting to state tat te fluid flow development around te 3D protruding eater lateral surfaces does not freely appen due to te small space between te eaters. (a) = 1 (b) = 3 Fig. 5. Streamlines around a 3D protruding eater (in a perspective 3D view detail) [49]. A numerical study of te laminar flow of te dielectric fluid Novec TM 75 around te 3D protruding eaters was presented in detail in [49]. In tis work, te autors analyzed te streamlines, te velocity profile, te mean friction coefficient, te pressure distribution and total pressure drop in te cannel, te required pumping power, and te Darcy-Weisbac friction factor. Considering te adiabatic substrate, te eat transfer at te surfaces of te 3D protruding eater to te dielectric fluid flow appens only troug forced convection caracterizing a convective cooling process. Te isotermal maps for = 1 and 3 are sown in Figs. 6, 7, and 8 for te xy, xz and yz-planes, respectively. Te protruding eater can be considered wit a uniform temperature due to its ig termal conductivity. Furtermore, te influence of te laminar flow around te 3D eater in te temperature distribution can be clearly noticed. (a) = 1 (b) = 3 Fig. 8. Isotermal map at te yz plane wit x = 2.375H. (a) = 1 (b) = 3 Fig. 4. Streamlines around a 3D protruding eater (in a perspective 3D view) [49]. Figs. 9(a), 9(b), and 9(c) sow te local adiabatic Nusselt number distributions along te lines ABCD, EFGH and IJKL at te 3D protruding eater surfaces, respectively, in function of te ynolds number. Nu ad (ξ) increases wit ynolds number. Te results of te average eater temperature, average adiabatic Nusselt number, convective termal conductance, eat transfer coefficient, overall termal conductance and overall eat transfer coefficient are sown in Table 1 in function of te ynolds number considering te forced convection cooling process wit laminar flow of te dielectric fluid Novec TM 75. In order to associate te 51

5 Nu ad Nu ad Nu ad T K ISSN: ISO 91:28 Certified Volume 3, Issue 7, January 214 numerical values, te results considering air as te work fluid are also presented in Table 1 [12]. Te properties of te air were considered constant, obtained at 3K [11] B independent of te fluid considered, indicating te drop in te average eater temperature. Furtermore, te magnitudes related to te Novec TM 75 (dielectric liquid) are greater tan te ones related to te air Air (Nisida & Alves [12]) Novec TM A E (a) Line ABCD F C D Nu ad Fig. 1. Average eater temperature. 6 5 Air (Nisida & Alves [12]) Novec TM H J (b) line EFGH G K Fig. 11. Average adiabatic Nusselt number I (c) line IJKL Fig. 9. Local adiabatic Nusselt number distribution Te magnitudes of te average eater temperatures, considering te convective cooling process of te dielectric fluid Novec TM 75, are smaller tan te ones obtained wen considering te air forced cooling process due to te fact tat te termal conductance (eat transfer coefficients), tat are associated wit te Novec TM 75 (dielectric liquid), are greater tan te ones related to te air. In Fig. 1 te average eater temperature distribution is presented in function of te ynolds number parameterized in te work fluids. As expected, te average temperature decreases wit te increasing ynolds independent of te cooling fluid. Fig. 11 illustrates te beavior of te average Nusselt number wit te ynolds number parameterized in te work fluids. Tis important termal parameter increases wit ynolds L Te results of te average adiabatic Nusselt number can be correlated for Novec TM 75 wit deviations not greater tan 1.5% by. 42 Nu (11) ad, dielectric fluid A general correlation for te average Nusselt number considering te 3D protruding eater mounted on a adiabatic substrate, wit deviations not greater tan 2.%, can be expressed by Nu ad Pr (12) Te beavior of te overall termal conductance, (UA), in function of te ynolds number parameterized in te work fluids is sown in Fig. 12. (UA) increases wit indicating a greater eat excange. In tis case, since te substrate is adiabatic, te convective termal conductance and te overall termal conductance are identical, tat way, (UA) = (UA) cv. Additionally, te magnitudes related to te Novec TM 75 are greater tan te ones related to te air. However, te eat transfer coefficient (Fig. 13) and te overall eat transfer coefficient (Fig. 14) are different. Te eat transfer coefficients increases wit te ynolds number indicating a greater eat excange between te protruding eater and te 52

6 U W/m 2 K U cv W/m 2 K UA W/K ISSN: ISO 91:28 Certified Volume 3, Issue 7, January 214 fluid flow wit te greater mass flow in te cannel. Moreover, te U values associated wit te Novec TM 75 are greater tan te ones associated wit te air Air (Nisida & Alves [12]) Novec TM Fig. 12. Overall termal conductance. Air (Nisida & Alves [12]) Novec TM Fig. 13. Heat transfer coefficient Air (Nisida & Alves [12]) Novec TM Fig. 14. Overall eat transfer coefficient. V. CONCLUSION In te present work a numerical analysis was performed to investigate te forced convection eat transfer wit laminar flow of te dielectric fluid Novec TM 75 from a 3D protruding eaters array mounted in cross-stream direction on an adiabatic substrate of a orizontal rectangular cannel (Fig. 1) utilizing te ANSYS/Fluent TM 15. software. A uniform eat generation rate was assumed at te 3D protruding eaters and te cooling process occurred troug a forced laminar flow wit constant properties under steady state conditions. In te cannel entry, te velocity and temperature profiles were uniform. Te conservation equations and teir boundary conditions were numerically solved witin a single domain tat includes te solid and fluid regions troug a coupled procedure utilizing te Control Volume Metod. Te occasional effects of te natural convection, radiation and fluid flow oscillation were not considered in te problem modeling. Due to te problem symmetries, te basic configuration of te problem was reduced to te one in Fig. 2. Te SIMPLE algoritm was used to treat te pressure-velocity couple. Te discretization of te diffusive-convective terms was done troug a Second-order upwind sceme. Due to te non-linearity of te momentum equation, te correction of te velocity components and te pressure were under-relaxed to prevent instability and divergence. Te verification of te adopted numerical procedures was performed troug a comparison of te termo-fluid-dynamic parameters of te numerical results wit te ones presented in [46]. After a study of te computational mes independence, te numerical results were obtained, displayed as a 3D non-uniform mes wit 212,67 control volumes (Fig. 3). Tis computational mes was more concentrated near te solid-fluid interface regions due to te larger primitive variable gradients in tese regions. In order to obtain te numerical results, typical design and properties values found in cooling applications of electronic components mounted on a circuit printed board. Te geometric configurations sowed in Fig. 2 were assumed considering a space of H =.254 m between te parallel plates. Te effects of te ynolds number, based on te protruding eaters eigt, were inspected for = 1, 15, 2, 25, and 3. Te flow in te cannel was always laminar for te range of investigated. Te beavior of te laminar flow around te 3D protruding eaters was sowed troug te streamlines. Te streamlines around a protruding eater were presented for ynolds numbers of 1 and 3 (Figs. 4 and 5). Te main caracteristics of te laminar flow were te orsesoe vortices wic start upstream te eater and develop around te eater lateral surfaces; a small recirculation upstream te protruding eater; te fluid boundary layer detacment at te top of te eater causing a recirculation; and a large recirculation region downstream te eater due to te flow reattacment. More information about te laminar flow of te dielectric fluid Novec TM 75 around te 3D protruding eaters can be found in [49]. For te forced convection cooling process (adiabatic substrate), te isotermal maps for = 1 and 3 considering te xy, xz and yz- planes were presented (Figs. 6, 7, and 8). Te local Nusselt number distributions along te lines ABCD, EFGH and IJKL at te surfaces of te 3D protruding eaters in function of te ynolds number were presented in Figs. 9(a), 9(b), and 9(c), respectively. Te main termal parameters of interest, average eater temperature (Fig. 1), average adiabatic Nusselt number (Fig. 11), overall termal conductance (Fig. 12), eat transfer coefficient (Fig. 13) and overall eat transfer coefficient (Fig. 14) were presented in function of te ynolds number in Table 1 considering air and dielectric fluid Novec TM 75. Te average adiabatic 53

7 ISO 91:28 Certified Volume 3, Issue 7, January 214 Nusselt number was correlated wit deviations not greater tan 2.% troug Eq. (12). Te magnitudes of te average eater temperatures, considering te convective cooling process of te dielectric fluid Novec TM 75, are smaller tan te ones obtained wen considering te air convective cooling process due to te fact tat te termal conductance (eat transfer coefficients), tat are associated wit te Novec TM 75 (dielectric liquid), are greater tan te ones related to te air. Finally, it is interesting to state tat te fluid flow development around te 3D protruding eaters lateral surfaces does not freely appen due to te small space between te protruding eaters. Te fluid dynamic symmetry conditions of te blocks were dominant and te corresponding flow and te termal wake were different tan a single 3D protruding eater wit free domain in te transversal direction to te flow. VI. ACKNOWLEDGMENT Te autors gratefully acknowledge te Federal University of Tecnology Paraná/Campus Ponta Grossa. REFERENCES [1] F. B. Nisida, Análise numérica do escoamento laminar e da transferência de calor de aquecedores 3D protuberante utilizando diferentes fluidos de resfriamento, Trabalo de Conclusão de Curso, Universidade Tecnológica Federal do Paraná, Ponta Grossa, Brasil, 18 p., Dec [2] G. P. Peterson, and A. Ortega, Termal control of electronic equipment and devices, In J. P. Hartnett, and T. F. Irvine, (Eds.) Advances in eat transfer, Oxford, ENG: Academic Press, pp , 199. [3] Y. A. Çengel, and A. J. Gajar, Heat and mass transfer: fundamentals and applications, New York, USA: McGraw- Hill, 96p., 214. [4] F. B. Nisida, and T. A. Alves, Conjugate forced convectionconduction eat transfer using different cooling fluids in cannel flow, 15t International Heat Transfer Conference, Kyoto, JAP, IHTC , to be publised in 214. [5] T.A. Alves, sfriamento conjugado de aquecedores discretos em canais, Tese de Doutorado em Engenaria Mecânica, Faculdade de Engenaria Mecânica, Universidade Estadual de Campinas, Campinas, Brasil, 129 p., July 21. [6] T. A. Alves, and C. A. C. Altemani, Conjugate cooling of a protruding eater in a cannel wit distinct flow constraints, Global Journal of searces in Engineering A, vol. XIII, pp. 9-21, Dec [7] 3M TM, Termal Management Fluids, Sep. 29. [8] H. Nakamura, T. Igarasi, and T. Tsutsui, Local eat transfer around a wall-mounted cube in te turbulent boundary layer, International Journal of Heat and Mass Transfer, vol. 44, pp , Sep. 21. [9] M. Nakajima, H. Yanaoka, H. Yosikawa, and T. Ota, Numerical simulation of tree-dimensional separated flow and eat transfer around staggered surface-mounted rectangular blocks in a cannel. Numerical Heat Transfer, Part A, vol. 47, pp , July 25. [1] M. Yagoubi, and E. Velayati, Undeveloped convective eat transfer from an array of cubes in cross-stream direction, International Journal of Termal Sciences, vol. 44, pp , Aug. 25. [11] T. L. Bergman, A. S. Lavine, F. P. Incropera, and D. P. Dewitt, Fundamentals of eat and mass transfer, New Jersey, USA: Jon Wiley & Sons. 18 p., April 212. [12] F. B. Nisida, and T.A. Alves, Forced convection cooling of 3D protruding eaters wit laminar flow in a rectangular cannel, International Journal of Emerging Tecnology and Advanced Engineering, vol. 4, pp , Jan [13] E. M. Sparrow, J. W. Ramsey, and C. A. C. Altemani, Experiments on in-line pin fin arrays and performance comparisons wit staggered arrays, Journal of Heat Transfer, vol. 12, pp. 44-5, Feb [14] E. M. Sparrow, J. E. Neitammer, and A. Caboki, Heat transfer and pressure drop experiments in air-cooled electronic-component arrays, Journal of Termo pysics and Heat Transfer, vol. 25, pp , Oct [15] D. E. Arvizu, and R. J. Moffat, Experimental eat transfer from an array of eated cubical elements on an adiabatic cannel wall, Termo sciences Division searc port HMT 33, Stanford University, [16] R. J. Moffat, D. E. Arvizu, and A. Ortega, Cooling electronic components: forced convection experiments wit an air-cooled array, Heat Transfer in Electronic Equipment ASME HTD, vol. 48, pp , [17] G. L. Lemann, and R. A. Wirtz, Te effect of variations in stream-wise spacing and lengt on convection from surface mounted rectangular components, Journal of Electronic Packaging, vol. 111, pp , Mar [18] Y. Asako, and M. Fagri, Tree-dimensional eat transfer analysis of arrays of eated square blocks, International Journal of Heat and Mass Transfer, vol. 32, pp , Feb [19] S.V. Garimella, and P.A. Eibeck, Heat transfer caracteristics of an array of protruding elements in single pase forced convection, International Journal of Heat and Mass Transfer, vol. 33, pp , Feb [2] S. V. Garimella, and P. A. Eibeck, Fluid dynamic caracteristics of te flow over an array of large rougness elements, Journal of Electronic Packaging, vol. 113, pp , Dec [21] R. J. Moffat, and A. M. Anderson, Applying eat transfer coefficient data to electronics cooling, Journal of Heat Transfer, vol. 112, pp , Nov [22] R. J. Moffat, and A. M. Anderson, Convective eat transfer from arrays of modules wit non-uniform eating: experiments and models, Termo sciences Division searc port HMT 43, Stanford University, 199. [23] A. M. Anderson, and, R. J. Moffat, Te adiabatic eat transfer coefficient and te superposition kernel function: part 2 - modeling flat pack data as a function of cannel turbulence, Journal of Electronic Packaging, vol. 114, pp , Mar [24] R. A. Wirtz, and W. Cen, Laminar-transitional convection from repeated ribs in a cannel, Journal of Electronic Packaging, vol. 114, pp , Mar

8 ISO 91:28 Certified Volume 3, Issue 7, January 214 [25] G. L. Lemann, and J. Pembroke, Forced convection air cooling of simulated low profile electronic components: part 1 - base case, Journal of Electronic Packaging, vol. 113, pp , Mar [26] M. Fagri, and Y. Asako, Prediction of turbulent tree-dimensional eat transfer of eated blocks using low ynolds number two-equation model, Numerical Heat Transfer, Part A, vol. 2, pp , Jan [27] H. J. Hussein, and R. J. Martinuzzi, Energy balance for te turbulent flow around a surface mounted cube placed in a cannel, Pysics of Fluids, vol. 8, pp , Mar [28] E. R. Meinders, T. H. Van Der Meer, and K. Hanjalic, Local convection eat transfer from an array of wall-mounted cubes, International Journal of Heat and Mass Transfer, vol. 41, pp , Jan [29] E. R. Meinders, and K. Hanjalic, Vortex structure and eat transfer in turbulent flow over a wall-mounted matrix of cubes, International Journal of Heat and Fluid Flow, vol. 2, pp , June [3] E. R. Meinders, G. M. P. Van Kempen, L. J. Van Vliet, and T. H. Van Der Meer, Measurement and application of an infrared image restoration filter to improve te accuracy of surface temperature measurements of cubes, Experiments in Fluids, vol. 26, pp , Jan [31] M. Nakajima, and T. Ota, Numerical analysis of treedimensional unsteady flow and eat transfer around a surface-mounted exaedron in a cannel, Transactions of te Japan Society of Mecanical Engineers B, vol. 65, pp , Sep [32] M. Molki, and M. Fagri, M. Temperature of in-line array of electronic components simulated by rectangular blocks, Electronics Cooling, vol. 6, pp , May 2. [33] B. Niceno, A. D. T. Dronkers, and Hanjalic, K. Turbulent eat transfer from a multi-layered wall-mounted cube matrix: a large eddy simulation, International Journal of Heat and Fluid Flow, vol. 23, pp , April 22. [34] W. Nakayama, and S. H. Park, Cconjugate eat transfer from a single surface-mounted block to forced convective air flow in a cannel, Journal of Heat Transfer, vol. 118, pp , May [35] T. A. Alves, and C. A. C. Altemani, Termal design of a protruding eater in laminar cannel flow, 14t International Heat Transfer Conference, Wasington, USA, IHTC , Aug. 21. [36] M. A. Barbur, F. B. Nisida, and T. A. Alves, Análise numérica do resfriamento por convecção forçada de um aquecedor 3D protuberante em um canal orizontal de placas paralelas com escoamento laminar, XXXIV Iberian Latin- American Congress on Computational Metods in Engineering, Pirenópolis, Brasil, CIL , Nov [37] Y. Zeng, and K. Vafai, An investigation of convective cooling of an array of cannel-mounted obstacles, Numerical Heat Transfer, Part A, vol. 55, pp , June 29. [38] T. A. Alves, and C. A. C. Altemani, sfriamento convectivo de um aquecedor protuberante num canal de placas paralelas com escoamento laminar, VII Congresso Nacional de Engenaria Mecânica, São Luis, Brasil, CONEM , July 212. [39] F. B. Nisida, M. A. Barbur, and T.A. Alves, sfriamento por convecção forçada de aquecedores 3D protuberantes em um canal de placas paralelas com escoamento laminar, XI Congresso Ibero-americano de Engenaria Mecânica, La Plata, Argentina, Nov [4] T. A. Alves, and C. A. C. Altemani, Convective cooling of tree discrete eat sources in cannel flow, Journal of te Brazilian Society of Mecanical Sciences and Engineering, vol. XXX, pp , July 28. [41] T. A. Alves, and C. A. C. Altemani, Conjugate cooling of a discrete eater in laminar cannel flow, Journal of te Brazilian Society of Mecanical Sciences and Engineering, vol. XXXIII, pp , July 211. [42] T. A. Alves, and C. A. C. Altemani, An invariant descriptor for eaters temperature prediction in conjugate cooling, International Journal of Termal Sciences, vol. 58, pp , Aug [43] J. Davalat, and Y. Bayazitoglu, Forced convection cooling across rectangular blocks, Journal of Heat Transfer, vol. 19, pp , May1987. [44] S. Ramadyani, D. F. Moffat, and F. P. Incropera, Conjugate eat transfer from small isotermal eat sources embedded in a large substrate, International Journal of Heat and Mass Transfer, vol. 28, pp , Oct [45] S. V. Patankar, Numerical eat transfer and fluid flow, New York, USA: Hemispere Publising Corporation, 197 p., 198. [46] ANSYS/Fluent TM, Tutorial, 211: solving a conjugate eat transfer problem using ANSYS/Fluent TM, pp. 1-3, 211. [47] A. Bar-Coen, A. A. Watwe, and R. S. Praser, Heat transfer in electronic equipment, In Bejan, A., Kraus, A. D., eds, Heat transfer andbook, pp New Jersey, USA: Jon Wiley & Sons., cap.13, 23. [48] G. K. Morris, and S. V. Garimella, Termal wake downstream of a tree-dimensional obstacle, Experimental Termal Fluid Science, vol.12, pp , Dec [49] F. B. Nisida, and T. A. Alves, Laminar flow of a dielectric fluid around an array of 3D protruding eaters, International Journal of Engineering and Innovative Tecnology, vol. 6, pp , Dec AUTHOR BIOGRAPHY Felipe Baptista Nisida is a Mecanical Engineer graduated by Federal University of Tecnology Paraná/ Campus Ponta Grossa (UTFPR/Ponta Grossa). He spent one year in an intercange program at te University of Kansas to complement is Mecanical Engineering degree. He is studying to get is Master degree in Mecanical Engineering by UTFPR/Ponta Grossa (felipenisida@otmail.com). Tiago Antonini Alves is a Mecanical Engineer graduated by São Paulo State University/Campus Ila Solteira Unesp/Ila Solteira (24), as a Master degree in Mecanical Engineering by Unesp/Ila Solteira (26), and is Doctor of Science in Mecanical Engineering by State University of Campinas - Unicamp (21). Professor and Coordinator of te Mecanical Engineering Graduation at Federal University of Tecnology Paraná/Campus Ponta Grossa (UTFPR/ Ponta Grossa). Tiago as experience in Termal Sciences, mainly in eat transfer, termodynamic and fluid mecanics. His researces consist mainly of convection, conduction, termal control of electronic equipments, numerical and experimental analysis (tiagoaalves@utfpr.edu.br). 55

9 ISO 91:28 Certified APPENDIX Volume 3, Issue 7, January 214 L = 6.5H u,t y k k H H =.3H x L u = 2H L d = 3.75H t =.1H x W =.6H k s W = H z L =.75H W s =.2H Fig. 2. Basic configuration representing te problem for one of te 3D protruding eaters (a) = 1 (b) = 3 Fig. 6. Isotermal map at te xy-plane wit z =. (a) = 1 (b) = 3 Fig. 7. Isotermal map at te xz plane wit y =.16H.. 56

10 ISO 91:28 Certified Volume 3, Issue 7, January 214 Table 2. Termal parameters of interest Novec TM 75 Air [12] T [K] (UA) Nu cv U cv (UA) U ad [W/K] [W/m 2 K] [W/K] [W/m 2 T [K] (UA) K] Nu cv U cv (UA) U ad [W/K] [W/m 2 K] [W/K] [W/m 2 K]