- Email: [email protected]
The development and performance of the high-power LED radiator
International Journal of Thermal Sciences 113 (2017) 65e72
Contents lists available at ScienceDirect
International Journal of Thermal Sciences journal homepage: www.elsevier.com/locate/ijts
The development and performance of the high-power LED radiator Mimi Wang, Hanzhong Tao*, Zishuai Sun, Changmi Zhang School of Energy Science and Engineering, Nanjing Tech University, Nanjing, Jiangsu, 211800, China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 April 2016 Received in revised form 7 October 2016 Accepted 15 November 2016
The performance of a kind of pure aluminum heat sink with radicalized straight fins and a kind of aluminum heat sink with heat pipe are investigated through numerical simulation method. Based on the correctness of numerical simulation model verified by the experiment, the optimal design is preceded. The correlations for Nu without and with heat pipe are obtained. Results show that the junction temperature of LED components is efficiently reduced by raising the radiator height (100e200 mm), increasing the number of fins (16e32), increasing the fin height respectively and the limitations are strengthened at the same time. The influence of fin thickness considered in this paper on Nusselt number of the radiator is less than 1%. By the introduction of the heat pipe, the equivalent coefficient of rolled aluminum heat pipe composite radiator fin upper part increased from 70.8% to 73.1%, temperature difference between the top and bottom fins is reduced, heat dissipation is strengthened. At the same time, the increase of the length of heat pipe radiator is favorable to the heat dissipation of radiators and the reduction in radiators weight. Those all provide the theoretical basis for the design and prioritization of high-power LED module radiator. © 2016 Elsevier Masson SAS. All rights reserved.
Keywords: High-power LED Radiator Numerical simulation Heat pipe
1. Introduction LED (Light Emitting Diode), as the new type of luminescent solid material [1], has been widely applied in the signal, alphanumeric display [2,3], automobile [4], back light [5,6], and other fields. As lighting equipment, it was proposed due to the development of white-light LED in recent years [7,8]. And it has received extensive attention and concern since proposed [9]. As the lighting device, especially the mining lamp, street light and other powerful light sources, the power is often up to tens of watts or even hundreds of watts. With the increase of temperature, the light conversion efficiency of LED lamp declines, wavelength becomes longer and the forward voltage drops [10]. In this circumstance, the heat emitted from LED can't be ignored. If the heat cannot be dissipated efficiently, the performance of the device will decline even the device is burned. To make the matter worse, the life and property will be threatened. To obtain higher cost performance, dissipate the heat from LED effectively, control the junction temperature and improve the life and brightness of the use are the keys to making high-power LED lighting equipment further popularization and application [11].
* Corresponding author. E-mail address: [email protected] (H. Tao). http://dx.doi.org/10.1016/j.ijthermalsci.2016.11.012 1290-0729/© 2016 Elsevier Masson SAS. All rights reserved.
Heat pipe, a kind of high-efficient heat transfer component, has been widely used in process industry [12], new energy, aerospace, electronic heat dissipation and so on. Lan et al. [13] have already tried to apply heat pipe into high-power LED. The results showed that heat pipe radiator can reduce the junction temperature of 24.3 C, the thermal resistance of 0.91 C/W compared with common radiator under the speed of the wind at 7 m/s. Li et al. [14] proposed a copper-water loop heat pipe (LHP) heat sink for highpower integrated LED. And under natural convection, the total thermal resistance range of copper-water loop heat pipe is 1.0 to 0.4 C/W when heating loads range is from 30 W to 300 W. Lin et al. [15] indicated that the temperature of the LED significantly decreased when a plate oscillating heat pipe (OHP) was used. Zeng et al. [16] developed a novel phase change heat sink to improve the thermal performance of high-power LED package. Lu et al. [17] proposed a novel flat heat pipe (FHP) to improve the thermal characteristics of high power LED. The junction temperature, total thermal resistance of LED, different filling rates and inclination angles of the heat pipe were investigated experimentally. Cheng et al. [18] used the finite element method (FEM) to invest multi-fin heat sink. Park et al. [19] developed and numerically simulated a cooling system which consists of a chimney and a radial heat sink to improve the cooling efficiency. Jang et al. [20] numerically studied the orientation and geometric parametrics effect for a cylindrical heat sink used to cool a LED light bulb. And a correlation which is
66
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
v w
Nomenclature A cp D d Gr H L N Nu p Pr Q R T DT u ! V
area (m2) specific heat at constant pressure diameter of heat sink (mm) diameter of mandril of heat sink (mm) Grashof number Height of heat sink (mm) characteristic length number of fins on heat sink Nusselt number pressure Prandtl number heat power (W) thermal resistance temperature ( C) temperature difference ( C) x-velocity component
y-velocity component z-velocity component
Greek symbol d thickness of the fin (mm) h equivalent factor l conductivity coefficient (W/m$K) m dynamic viscosity n kinematic viscosity r Density (kg/m3) Subscripts a ambient f Fin hp Heat pipe j junction max maximum 1 bottom of fins 2 top of fins
velocity vector
used to predict the Nusselt number around an inclined cylindrical heat sink was proposed. Li et al. [21] numerically simulated the copper/water miniature heat pipe (mHP) to cooling high power multi-chip LED packaging. In the present study, a natural convection heat sink used to cooling high-power LED module is developed by combining heat pipe technology with rolling aluminum heat sink. The full passive high-power LED array is realized. The thermal performance was investigated by experiment and finite element numerical simulation. The effects of the radiator height, the number of fins, fin height, fin thickness, heat pipe and heat pipe length on thermal performance were studied in this paper. The correlations to predict the Nu without and with heat pipe are proposed. And the mathematical and physical model for further research is provided. 2. Model description 2.1. Geometric model The radiator model, with straight fins along the diameter orient, consists of mandrel and radial type fins (presented in Fig. 1). The whole contour is cylindrical (represented in Fig. 1(a)). D and H are the outer diameter and the height of the heat sink; d, H, and N are the thickness, length and number of the fins. For the radiator model with heat pipe: Hhp is the length of heat pipe, Dhp is diameter, other
sizes are the same as previously mentioned. The properties of geometric configuration are listed in Table 1. 2.2. Physical and numerical model The base plate is placed between LED array and the radiator, both of which are contact with each other tightly. The heat emitted from LED during working process is conducted to the radiator through the base plate and then is spread into the environment by the extended surface. The parameters of radiator's material and air properties are presented in Table 2. The assumptions of the model are listed as follow: (1) The air is continuous Newtonian incompressible fluid. The flow is the fully developed steady state and laminar flow; (2) Radiation heat is neglected; (3) Physical properties parameters of radiator's material are constant and the air property obeys the Boussinesq assumption; (4) The heat pipe, in the present study, is simplified as a solid with both the density and the specific heat of 1 and equivalent heat conductivity of 20000W=ðm,KÞ. The governing equations are as follows, Air side,
Fig. 1. Radiator model.
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
67
Table 1 Properties of geometric configuration.
Height of heat sink/heat pipe/fin Diameter of heat sink Fin height Thickness of fin Length of heat pipe Diameter of heat pipe Number of fins
Symbol
Value
H D Hf
100 160 55 2 90 20 16
d Hhp Dhp N
120 170 60 2.2 110
140 180 65 2.4 130
160 190 70 2.6 150
180 200 75 2.8 170
20
24
28
32
Table 2 Properties of physical model.
Aluminum Air Heat pipe
l, W=ðm,KÞ
cP , J=ðkg,KÞ
r, kg=m3
m, kg=ðm,sÞ
121 0.0261 20000
900 1005.0 1
2700 e 1
e 1.84 105 e
200 80 3.0 190
mm mm mm mm mm mm pitches
bottom temperature of fins T1, top temperature of fins T2 and environmental temperature Ta were measured by thermocouple (see Fig. 3). The stable performance of radiator was tested. The temperature was measured by T thermocouple of 0.2 mm and the data was collected every 5 min using AT4310.
4. The grid effectiveness analysis Continuity equation,
! V$ r V ¼ 0
(1)
Momentum equations,
! vp V$ð ruV Þ ¼ þ mV2 u vx
(2)
! vp ¼ þ mV2 n þ ð r ra Þg V$ rn V vy
(3)
! vp V$ð rwV Þ ¼ þ mV2 w vz
(4)
5. Results and discussion 5.1. Parameter definition
Energy equation,
! k 2 V$ r V T ¼ V T cp
(5)
The maximum temperature of numerical simulation is equivalent to the core temperature measured by thermocouple of LED array.
Tj ¼ Tmax
(7)
The maximum temperature rise is the most important control parameter of the cooling system. And it is defined as follows:
Solid side, Fourier heat conduction law,
V2 T ¼ 0
The computational domain was meshed by hexahedral unstructured grid in this paper. When the number of grid increases from 892173 to 1138240, Nu decreases from 364.87 to 364.78 (see Fig. 4), the deviation of which is only 0.0026%. So we suppose that the number of grid do not have the significant effect on the calculation results when the grid is over one million.
(6)
The SIMPLE algorithm was used to solve the problem. 2.3. Boundary conditions (1) Inlet: pressure inlet; (2) The non-slip wall condition is applied to the surface between air and radiator; (3) LED array is surface source; (4) Outlet: pressure outlet.
DTmax ¼ Tj Ta
(8)
Thermal resistance is defined as follows:
DTmax
Rth ¼
Q
¼
Tj Ta Q
(9)
Heat transfer coefficient is defined as follows:
h¼
Q Q ¼ Atot DTmax Atot Tj Ta
(10)
where Atot is defined as follows: 3. Experiment description The experiment to verify the reliability and validity of the numerical method is carried out. A 100 1 W LED array is used as lighted element. Also, LED uses GaN as based material which has been widely used as tested elements. As presented in Fig. 2(a), 10 10 arrays were installed on the Al base. LED arrays are 25 25 mm2 while Al board's outline is 40 40 mm2. The rated heat is Q ¼ 60 W. Al board is put in the Al heat radiator. The structure of cooling fin and central stick is presented in Fig. 2(b). The experimental condition is shown in Fig. 2, the concrete structure parameters of radiator is presented in Table 3. The center temperature Tj between Al based plate and fins,
Atot ¼
pd2 4
þ Nðd þ HÞðD dÞ þ pdH
(11)
The Nusselt number for the present heat sink is defined as follows:
Nu ¼
hL
l
(12)
where l is the coefficient of thermal conductivity of air, the characteristic length L is the height of the heat sink. Temperature difference between top and bottom of fins is defined as follows:
68
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
Fig. 2. Heat sink for test.
average temperature difference between the surfer of the heat sink 2 and the ambience, DT ¼ T1 þT 2 Ta ; the characteristic length L is the height of radiator; n is kinetic viscosity of air.
Table 3 Properties of test heat sink. Material
Case1
D N H d Hhp Dhp
Case2
Al
Al-Heat pipe
160 mm 28 180 mm 50 mm e e
160 mm 28 180 mm 50 mm 170 mm 20 mm
5.2. Experimental results validation The experiment is performed firstly to confirm the correctness of physical model and numerical model. Experimental testing environment is in a place without interference of natural convection. When the LED begins to glow, the data were collected until the temperature is stable in 30 min. The junction temperature data collected for Case1 and Case2 were 66.2258 C and 48.816 C (Fig. 5), respectively. The highest junction temperatures of numerical simulation were 68.1728 C and 53.256 C with the error of 2.94% and 9.1% respectively. It establishes foundation for further numerical simulation study. 5.3. Optimization results The performance of radiator by numerical simulation is investigated. The number of fins, fin thickness, long heat pipe length and so on were considered in this paper, and the correlations for predicting the Nusselt number with and without the heat pipe are obtained. 5.3.1. The correlation for Nu without and with heat pipe Fig. 6 shows the relationship between Nu number and Gr$Pr when heat pipe is not considered. As Nu number increase, Gr$Pr increase. The relationship between Gr$Pr and Nu can be fitted to
Fig. 3. Thermocouple arrangement.
DTf ¼ T1 T2
(13)
Equivalent coefficient of fins is an important parameter to characterize the efficient areas and the ratio actual heat dissipation areas. The definition of the top and bottom fins' equivalent heat conductivity are defined as follows:
Tj þ T1 2 Ta 100% Tj Ta
(14)
Tj þ T2 2 Ta h2 ¼ 100% Tj Ta
(15)
h1 ¼
Grashof number:
Gr ¼
gaV DTH3
n2
(16)
where, g is gravity; aV is the volume expansion ratio of air; DT is the
Fig. 4. Mesh independence test.
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
Fig. 5. The junction temperature curve.
Fig. 7. The junction temperature vs. height of radiator.
Fig. 6. Nusselt number vs. Grashof number and Prandtl number (N ¼ 28, d ¼ 2.8 mm, H ¼ 180 mm, Hf ¼ 55, without heat pipe).
Fig. 8. Nusselt number vs. height of radiator.
69
Equation (17), the regression coefficient of which is 0.996. And the correlation for considering heat pipe is presented in Equation (18).
Nu ¼ 0:666ðGr,PrÞ1=3 33:84
(17)
Nu ¼ 0:178ðGr,PrÞ1=3 þ 112:162
(18)
5.3.2. The effect of the radiator height on the thermal performance The effect of the radiator height on the junction temperature is shown in Fig. 7. As the heat sink height increases, the heat dissipation improves; the junction temperature decreases even though the tendency becomes gentle. In these three different kinds of powers, radiator height increased from 100 mm to 140 mm, the junction temperature dropped by an average rate of 0.14 C/mm, and when the radiator height increases from 140 mm to 200 mm, the junction temperature only drops by an average rate of 0.07 C/ mm. It can be seen from Fig. 8 that as H increases, Nusselt number increases; however, the tendency becomes gentle. At the same time, as the power increases, Nusselt number increases too. Fig. 9 shows the temperature of radiator has the trend to decrease; however, the tendency slowly becomes gentle. The radiator with H ¼ 100 mm has the distance with 66 mm to the bottom. Also, the
Fig. 9. Center temperature of radiator vs. height.
other one with H ¼ 200 mm has the distance with 90 mm to the bottom. The axial center temperature changes were all less than 0.046 C/mm, which indicates that the effect of enhanced heat dissipation by increasing the radiator height to increase cooling area is weakened.
70
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
Fig. 10. Junction temperature vs. radiator fin number.
Fig. 12. Junction temperature vs. fin height.
5.3.3. The effect of the number of fins on the thermal performance Fig. 10 displays the decrease in junction temperature with the increase in the number of fins. The number of fins increased from 16 to 24 resulting in an average junction temperature dropped 9.26 C; and when the number of fins increased from 24 to 32, the junction temperature average reduces only 2.93 C. With the number of fins increases, Nusselt number increases, as presented in Fig. 11, which indicates that the effect of enhanced heat dissipation by increasing the number of fins to increase cooling area is weakened. 5.3.4. The effect of the fin height on the thermal performance It is an effective way to extend the cooling area by increasing the fin height. As the height of fins increases, the junction temperature increases, as presented in Fig. 12, and the Nusselt number decreases (Fig. 13). It is because, with the increase of fin height, the increasing rate of total cooling area is higher than the decrease rate in maximum temperature according to Eqs. (9) and (11).
Fig. 13. Nusselt number vs. fin height.
5.3.5. The effect of the fin thickness on the thermal performance Fin thickness from 2 mm to 3 mm is considered in this paper. As shown in Figs. 14 and 15, fin thickness gradually increase from 2 mm to 3 mm, the variation range of the junction temperature and the Nusselt numbers are less than 1%. It indicates that the fin
thickness does not have a significant impact on the cooling effect of the radiator, it is because the changes in the thickness of the fin do not have a significant effect on the total cooling area when keeping other conditions are the same.
Fig. 11. Nusselt number vs. radiator fin number.
Fig. 14. Junction temperature vs. thickness of radiator fin.
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
Fig. 15. Nusselt number vs. thickness of radiator fin.
5.3.6. The effect of the heat pipe and heat pipe length on the thermal performance Fig. 16 shows that LED junction temperature average decrease about 8 C by introducing the heat pipe radiator. From Fig. 17 we can see that at the edge of pure rolling aluminum radiator fins has large temperature in the height direction. The maximum temperature difference between the upper and lower edges of the fins is about 2.6 C. However, the rolled aluminum upper and lower composite heat pipe radiator fin edges maximum temperature difference is about 0.17 C, which is only 0.6% of pure rolling aluminum radiators at the same conditions. The decrease of the temperature difference between the upper and lower part of the radiator can be attributed to the introduction of the heat pipe. The upper and lower equivalent coefficients described in Fig. 16 is shown in Table 4. On the one hand, its equivalent coefficients of top and bottom are all improved; on the other hand, the top fins almost have same equivalent coefficients with bottom fins. The heat of the bottom radiator spreads rapidly to the top radiators by heat pipe, the axial isothermal of radiators was improved which is more favorable to dissipate heat. In this paper, the influence of the length of rolled aluminum heat pipe composite radiator on heat dissipation was investigated. Fig. 18 shows that under the same condition, the increase of length heat pipe is favorable to the reduction of LED junction temperature.
71
Fig. 17. Fin top temperature vs. height difference.
Table 4 Equivalent thermal conductivity of Al radiator and Al-heat pipe radiator (Q ¼ 60 W).
The lower part of equivalency factors (%) The upper part of equivalency factors (%)
Al
Al-Heat pipe
72.9 70.8
73.2 73.1
Fig. 18. Junction temperature vs. length of heat pipe. (H ¼ 180 mm, D ¼ 160 mm, d ¼ 50 mm).
As increasing 7.4% of the heat pipe length, the average junction temperature decreases 4%. And with the increase in heat pipe length, the weight of rolled aluminum heat pipe composite radiator reduced 4.77 kg/m in this study.
6. Conclusions Al-heat pipe radiator with high thermal conductivity is developed. The thermal performance of the Al-heat pipe radiator is investigated numerically based on the correctness of the numerical model verified by the experiment. The correlations for Nu without and with heat pipe are obtained.
Fig. 16. Junction temperature vs. length of heat pipe without heat pipe and with heat pipe separately.
(1) As the increase of the radiator height, the heat dissipation effect is improved and the Nusselt number increases with a downward trend.
72
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
(2) Under the conditions of different power, the junction temperature of LED assembly increases with a downward trend as the increase in the number of fins. The enhance heat dissipation effect by increasing the number of fins to increase cooling area is weakened. (3) LED component junction temperature decreases with the increase in fin height. Nusselt number has the same tendency. (4) Fin thickness investigated in this paper does not have a significant impact on the radiator cooling effect, the variation range of LED component temperature and the Nusselt numbers are less than 1%. (5) LED junction temperature reduces an average of about 8 C by applying Al-heat pipe radiator, the equivalent coefficient of the top fins increases which are in favor of dissipating heat. (6) The heat pipe length of Al-heat pipe radiator was investigated. The results show that under the same condition, the increase of length heat pipe is favorable to reduce the LED junction temperature and weight of radiator. A kind of pure aluminum heat sink with radicalized straight fins and a kind of aluminum heat sink with heat pipe are developed in this paper. On one hand, the high-power LED radiator forms and extended cooling area methods are enriched. On the other hand, considering that there are limitations to increasing the area only by changing geometric parameters. The heat dissipation effect is enhanced by using heat pipe with high thermal conductivity and it provides theoretical foundation and guidance for the design of high-power LED component radiators. References [1] Burroughes JH, Bradley DDC, Brown AR, Marks RN, Mackay K, Friend RH, et al. Erratum: light-emitting diodes based on conjugated polymers. Nature 1990;347:539e41. [2] Dasgupta PK, Eom I-Y, Morris KJ, Li J. Light emitting diode-based detectors. Anal Chim Acta 2003;500:337e64.
_ [3] Ryskevi c N, Jursenas S, Vitta P, Bakiene_ E, Gaska R, Zukauskas A. Concept design of a UV light-emitting diode based fluorescence sensor for real-time bioparticle detection. Sensors Actuators B Chem 2010;148:371e8. [4] Huang MS, Hung CC, Fang YC, Lai WC, Chen YL. Optical design and optimization of light emitting diode automotive head light with digital micromirror device light emitting diode. Optik Int J Light Electron Opt 2010;121:944e52. [5] Chen CF, Wu CC, Wu JH. Modified wide radiating lenses of the power-chip light emitting diodes for a direct-lit backlight. Optik Int J Light Electron Opt 2010;121:847e52. [6] Beeson S, Mayer JW. Patterns of light. New York: Springer; 2008. [7] Y. Shimizu, K. Sakano, Y. Noguchi, T. Moriguchi. Indium, gallium, aluminum nitride; garnet containing rare earth oxide. In Google Patents, 1999. [8] S. Tasch, P. Pachler, G. Roth, W. Tews, W. Kempfert, D. Starick. Light source comprising a light-emitting element. In Google Patents, 2004. [9] Xu J, Xiong Y, Chen S, Guan Y. A lamp light-emitting diode-induced fluorescence detector for capillary electrophoresis. Talanta 2008;76:369e72. [10] Senawiratne J, Chatterjee A, Detchprohm T, Zhao W, Li Y, Zhu M, et al. Junction temperature, spectral shift, and efficiency in GaInN-based blue and green light emitting diodes. Thin Solid Films 2010;518:1732e6. [11] Vamvounisa G, Aziz H, Hu NX, Popovicb ZD. Temperature dependence of operational stability of organic light emitting diodes based on mixed emitter layers. Synth Met 2004;146:69e73. [12] Zhang H, Zhuang J. Research, development and industrial application of heat pipe technology in China. Appl Therm Eng 2003;23:1067e83. [13] Lan K, Choi JH, Sun HJ, Shin MW. Thermal analysis of LED array system with heat pipe. Thermochim Acta 2007;455:21e5. [14] Li J, Lin F, Wang D, Tian W. A loop-heat-pipe heat sink with parallel condensers for high-power integrated LED chips. Appl Therm Eng 2013;56: 18e26. [15] Lin Z, Wang S, Huo J, Hu Y, Chen J, Zhang W, et al. Heat transfer characteristics and LED heat sink application of aluminum plate oscillating heat pipes. Appl Therm Eng 2011;31:2221e9. [16] Zeng D, Liu Y, Jiang L, Li L, Gang X. A novel manufacturing approach of phasechange heat sink for high-power LED. Energy Procedia 2012;17:1974e8. [17] Lu XY, Hua TC, Wang YP. Thermal analysis of high power LED package with heat pipe heat sink. Microelectron J 2011;42:1257e62. [18] Cheng HH, Huang DS, Lin MT. Heat dissipation design and analysis of high power LED array using the finite element method. Microelectron Reliab 2012;52:905e11. [19] Park S, Jang D, Yook SJ, Lee KS. Optimization of a chimney design for cooling efficiency of a radial heat sink in a LED downlight. Energy Convers Manag 2016;114:180e7. [20] Jang D, Park SJ, Yook SJ, Lee KS. The orientation effect for cylindrical heat sinks with application to LED light bulbs. Int J Heat Mass Transf 2014;71:496e502. [21] Li D, Zhang GQ, Pan K, Ma X. Numerical simulation on heat pipe for high power LED multi-chip module packaging. In: International conference on electronic packaging technology and high density packaging; 2009. p. 393e7.
Contents lists available at ScienceDirect
International Journal of Thermal Sciences journal homepage: www.elsevier.com/locate/ijts
The development and performance of the high-power LED radiator Mimi Wang, Hanzhong Tao*, Zishuai Sun, Changmi Zhang School of Energy Science and Engineering, Nanjing Tech University, Nanjing, Jiangsu, 211800, China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 April 2016 Received in revised form 7 October 2016 Accepted 15 November 2016
The performance of a kind of pure aluminum heat sink with radicalized straight fins and a kind of aluminum heat sink with heat pipe are investigated through numerical simulation method. Based on the correctness of numerical simulation model verified by the experiment, the optimal design is preceded. The correlations for Nu without and with heat pipe are obtained. Results show that the junction temperature of LED components is efficiently reduced by raising the radiator height (100e200 mm), increasing the number of fins (16e32), increasing the fin height respectively and the limitations are strengthened at the same time. The influence of fin thickness considered in this paper on Nusselt number of the radiator is less than 1%. By the introduction of the heat pipe, the equivalent coefficient of rolled aluminum heat pipe composite radiator fin upper part increased from 70.8% to 73.1%, temperature difference between the top and bottom fins is reduced, heat dissipation is strengthened. At the same time, the increase of the length of heat pipe radiator is favorable to the heat dissipation of radiators and the reduction in radiators weight. Those all provide the theoretical basis for the design and prioritization of high-power LED module radiator. © 2016 Elsevier Masson SAS. All rights reserved.
Keywords: High-power LED Radiator Numerical simulation Heat pipe
1. Introduction LED (Light Emitting Diode), as the new type of luminescent solid material [1], has been widely applied in the signal, alphanumeric display [2,3], automobile [4], back light [5,6], and other fields. As lighting equipment, it was proposed due to the development of white-light LED in recent years [7,8]. And it has received extensive attention and concern since proposed [9]. As the lighting device, especially the mining lamp, street light and other powerful light sources, the power is often up to tens of watts or even hundreds of watts. With the increase of temperature, the light conversion efficiency of LED lamp declines, wavelength becomes longer and the forward voltage drops [10]. In this circumstance, the heat emitted from LED can't be ignored. If the heat cannot be dissipated efficiently, the performance of the device will decline even the device is burned. To make the matter worse, the life and property will be threatened. To obtain higher cost performance, dissipate the heat from LED effectively, control the junction temperature and improve the life and brightness of the use are the keys to making high-power LED lighting equipment further popularization and application [11].
* Corresponding author. E-mail address: [email protected] (H. Tao). http://dx.doi.org/10.1016/j.ijthermalsci.2016.11.012 1290-0729/© 2016 Elsevier Masson SAS. All rights reserved.
Heat pipe, a kind of high-efficient heat transfer component, has been widely used in process industry [12], new energy, aerospace, electronic heat dissipation and so on. Lan et al. [13] have already tried to apply heat pipe into high-power LED. The results showed that heat pipe radiator can reduce the junction temperature of 24.3 C, the thermal resistance of 0.91 C/W compared with common radiator under the speed of the wind at 7 m/s. Li et al. [14] proposed a copper-water loop heat pipe (LHP) heat sink for highpower integrated LED. And under natural convection, the total thermal resistance range of copper-water loop heat pipe is 1.0 to 0.4 C/W when heating loads range is from 30 W to 300 W. Lin et al. [15] indicated that the temperature of the LED significantly decreased when a plate oscillating heat pipe (OHP) was used. Zeng et al. [16] developed a novel phase change heat sink to improve the thermal performance of high-power LED package. Lu et al. [17] proposed a novel flat heat pipe (FHP) to improve the thermal characteristics of high power LED. The junction temperature, total thermal resistance of LED, different filling rates and inclination angles of the heat pipe were investigated experimentally. Cheng et al. [18] used the finite element method (FEM) to invest multi-fin heat sink. Park et al. [19] developed and numerically simulated a cooling system which consists of a chimney and a radial heat sink to improve the cooling efficiency. Jang et al. [20] numerically studied the orientation and geometric parametrics effect for a cylindrical heat sink used to cool a LED light bulb. And a correlation which is
66
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
v w
Nomenclature A cp D d Gr H L N Nu p Pr Q R T DT u ! V
area (m2) specific heat at constant pressure diameter of heat sink (mm) diameter of mandril of heat sink (mm) Grashof number Height of heat sink (mm) characteristic length number of fins on heat sink Nusselt number pressure Prandtl number heat power (W) thermal resistance temperature ( C) temperature difference ( C) x-velocity component
y-velocity component z-velocity component
Greek symbol d thickness of the fin (mm) h equivalent factor l conductivity coefficient (W/m$K) m dynamic viscosity n kinematic viscosity r Density (kg/m3) Subscripts a ambient f Fin hp Heat pipe j junction max maximum 1 bottom of fins 2 top of fins
velocity vector
used to predict the Nusselt number around an inclined cylindrical heat sink was proposed. Li et al. [21] numerically simulated the copper/water miniature heat pipe (mHP) to cooling high power multi-chip LED packaging. In the present study, a natural convection heat sink used to cooling high-power LED module is developed by combining heat pipe technology with rolling aluminum heat sink. The full passive high-power LED array is realized. The thermal performance was investigated by experiment and finite element numerical simulation. The effects of the radiator height, the number of fins, fin height, fin thickness, heat pipe and heat pipe length on thermal performance were studied in this paper. The correlations to predict the Nu without and with heat pipe are proposed. And the mathematical and physical model for further research is provided. 2. Model description 2.1. Geometric model The radiator model, with straight fins along the diameter orient, consists of mandrel and radial type fins (presented in Fig. 1). The whole contour is cylindrical (represented in Fig. 1(a)). D and H are the outer diameter and the height of the heat sink; d, H, and N are the thickness, length and number of the fins. For the radiator model with heat pipe: Hhp is the length of heat pipe, Dhp is diameter, other
sizes are the same as previously mentioned. The properties of geometric configuration are listed in Table 1. 2.2. Physical and numerical model The base plate is placed between LED array and the radiator, both of which are contact with each other tightly. The heat emitted from LED during working process is conducted to the radiator through the base plate and then is spread into the environment by the extended surface. The parameters of radiator's material and air properties are presented in Table 2. The assumptions of the model are listed as follow: (1) The air is continuous Newtonian incompressible fluid. The flow is the fully developed steady state and laminar flow; (2) Radiation heat is neglected; (3) Physical properties parameters of radiator's material are constant and the air property obeys the Boussinesq assumption; (4) The heat pipe, in the present study, is simplified as a solid with both the density and the specific heat of 1 and equivalent heat conductivity of 20000W=ðm,KÞ. The governing equations are as follows, Air side,
Fig. 1. Radiator model.
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
67
Table 1 Properties of geometric configuration.
Height of heat sink/heat pipe/fin Diameter of heat sink Fin height Thickness of fin Length of heat pipe Diameter of heat pipe Number of fins
Symbol
Value
H D Hf
100 160 55 2 90 20 16
d Hhp Dhp N
120 170 60 2.2 110
140 180 65 2.4 130
160 190 70 2.6 150
180 200 75 2.8 170
20
24
28
32
Table 2 Properties of physical model.
Aluminum Air Heat pipe
l, W=ðm,KÞ
cP , J=ðkg,KÞ
r, kg=m3
m, kg=ðm,sÞ
121 0.0261 20000
900 1005.0 1
2700 e 1
e 1.84 105 e
200 80 3.0 190
mm mm mm mm mm mm pitches
bottom temperature of fins T1, top temperature of fins T2 and environmental temperature Ta were measured by thermocouple (see Fig. 3). The stable performance of radiator was tested. The temperature was measured by T thermocouple of 0.2 mm and the data was collected every 5 min using AT4310.
4. The grid effectiveness analysis Continuity equation,
! V$ r V ¼ 0
(1)
Momentum equations,
! vp V$ð ruV Þ ¼ þ mV2 u vx
(2)
! vp ¼ þ mV2 n þ ð r ra Þg V$ rn V vy
(3)
! vp V$ð rwV Þ ¼ þ mV2 w vz
(4)
5. Results and discussion 5.1. Parameter definition
Energy equation,
! k 2 V$ r V T ¼ V T cp
(5)
The maximum temperature of numerical simulation is equivalent to the core temperature measured by thermocouple of LED array.
Tj ¼ Tmax
(7)
The maximum temperature rise is the most important control parameter of the cooling system. And it is defined as follows:
Solid side, Fourier heat conduction law,
V2 T ¼ 0
The computational domain was meshed by hexahedral unstructured grid in this paper. When the number of grid increases from 892173 to 1138240, Nu decreases from 364.87 to 364.78 (see Fig. 4), the deviation of which is only 0.0026%. So we suppose that the number of grid do not have the significant effect on the calculation results when the grid is over one million.
(6)
The SIMPLE algorithm was used to solve the problem. 2.3. Boundary conditions (1) Inlet: pressure inlet; (2) The non-slip wall condition is applied to the surface between air and radiator; (3) LED array is surface source; (4) Outlet: pressure outlet.
DTmax ¼ Tj Ta
(8)
Thermal resistance is defined as follows:
DTmax
Rth ¼
Q
¼
Tj Ta Q
(9)
Heat transfer coefficient is defined as follows:
h¼
Q Q ¼ Atot DTmax Atot Tj Ta
(10)
where Atot is defined as follows: 3. Experiment description The experiment to verify the reliability and validity of the numerical method is carried out. A 100 1 W LED array is used as lighted element. Also, LED uses GaN as based material which has been widely used as tested elements. As presented in Fig. 2(a), 10 10 arrays were installed on the Al base. LED arrays are 25 25 mm2 while Al board's outline is 40 40 mm2. The rated heat is Q ¼ 60 W. Al board is put in the Al heat radiator. The structure of cooling fin and central stick is presented in Fig. 2(b). The experimental condition is shown in Fig. 2, the concrete structure parameters of radiator is presented in Table 3. The center temperature Tj between Al based plate and fins,
Atot ¼
pd2 4
þ Nðd þ HÞðD dÞ þ pdH
(11)
The Nusselt number for the present heat sink is defined as follows:
Nu ¼
hL
l
(12)
where l is the coefficient of thermal conductivity of air, the characteristic length L is the height of the heat sink. Temperature difference between top and bottom of fins is defined as follows:
68
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
Fig. 2. Heat sink for test.
average temperature difference between the surfer of the heat sink 2 and the ambience, DT ¼ T1 þT 2 Ta ; the characteristic length L is the height of radiator; n is kinetic viscosity of air.
Table 3 Properties of test heat sink. Material
Case1
D N H d Hhp Dhp
Case2
Al
Al-Heat pipe
160 mm 28 180 mm 50 mm e e
160 mm 28 180 mm 50 mm 170 mm 20 mm
5.2. Experimental results validation The experiment is performed firstly to confirm the correctness of physical model and numerical model. Experimental testing environment is in a place without interference of natural convection. When the LED begins to glow, the data were collected until the temperature is stable in 30 min. The junction temperature data collected for Case1 and Case2 were 66.2258 C and 48.816 C (Fig. 5), respectively. The highest junction temperatures of numerical simulation were 68.1728 C and 53.256 C with the error of 2.94% and 9.1% respectively. It establishes foundation for further numerical simulation study. 5.3. Optimization results The performance of radiator by numerical simulation is investigated. The number of fins, fin thickness, long heat pipe length and so on were considered in this paper, and the correlations for predicting the Nusselt number with and without the heat pipe are obtained. 5.3.1. The correlation for Nu without and with heat pipe Fig. 6 shows the relationship between Nu number and Gr$Pr when heat pipe is not considered. As Nu number increase, Gr$Pr increase. The relationship between Gr$Pr and Nu can be fitted to
Fig. 3. Thermocouple arrangement.
DTf ¼ T1 T2
(13)
Equivalent coefficient of fins is an important parameter to characterize the efficient areas and the ratio actual heat dissipation areas. The definition of the top and bottom fins' equivalent heat conductivity are defined as follows:
Tj þ T1 2 Ta 100% Tj Ta
(14)
Tj þ T2 2 Ta h2 ¼ 100% Tj Ta
(15)
h1 ¼
Grashof number:
Gr ¼
gaV DTH3
n2
(16)
where, g is gravity; aV is the volume expansion ratio of air; DT is the
Fig. 4. Mesh independence test.
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
Fig. 5. The junction temperature curve.
Fig. 7. The junction temperature vs. height of radiator.
Fig. 6. Nusselt number vs. Grashof number and Prandtl number (N ¼ 28, d ¼ 2.8 mm, H ¼ 180 mm, Hf ¼ 55, without heat pipe).
Fig. 8. Nusselt number vs. height of radiator.
69
Equation (17), the regression coefficient of which is 0.996. And the correlation for considering heat pipe is presented in Equation (18).
Nu ¼ 0:666ðGr,PrÞ1=3 33:84
(17)
Nu ¼ 0:178ðGr,PrÞ1=3 þ 112:162
(18)
5.3.2. The effect of the radiator height on the thermal performance The effect of the radiator height on the junction temperature is shown in Fig. 7. As the heat sink height increases, the heat dissipation improves; the junction temperature decreases even though the tendency becomes gentle. In these three different kinds of powers, radiator height increased from 100 mm to 140 mm, the junction temperature dropped by an average rate of 0.14 C/mm, and when the radiator height increases from 140 mm to 200 mm, the junction temperature only drops by an average rate of 0.07 C/ mm. It can be seen from Fig. 8 that as H increases, Nusselt number increases; however, the tendency becomes gentle. At the same time, as the power increases, Nusselt number increases too. Fig. 9 shows the temperature of radiator has the trend to decrease; however, the tendency slowly becomes gentle. The radiator with H ¼ 100 mm has the distance with 66 mm to the bottom. Also, the
Fig. 9. Center temperature of radiator vs. height.
other one with H ¼ 200 mm has the distance with 90 mm to the bottom. The axial center temperature changes were all less than 0.046 C/mm, which indicates that the effect of enhanced heat dissipation by increasing the radiator height to increase cooling area is weakened.
70
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
Fig. 10. Junction temperature vs. radiator fin number.
Fig. 12. Junction temperature vs. fin height.
5.3.3. The effect of the number of fins on the thermal performance Fig. 10 displays the decrease in junction temperature with the increase in the number of fins. The number of fins increased from 16 to 24 resulting in an average junction temperature dropped 9.26 C; and when the number of fins increased from 24 to 32, the junction temperature average reduces only 2.93 C. With the number of fins increases, Nusselt number increases, as presented in Fig. 11, which indicates that the effect of enhanced heat dissipation by increasing the number of fins to increase cooling area is weakened. 5.3.4. The effect of the fin height on the thermal performance It is an effective way to extend the cooling area by increasing the fin height. As the height of fins increases, the junction temperature increases, as presented in Fig. 12, and the Nusselt number decreases (Fig. 13). It is because, with the increase of fin height, the increasing rate of total cooling area is higher than the decrease rate in maximum temperature according to Eqs. (9) and (11).
Fig. 13. Nusselt number vs. fin height.
5.3.5. The effect of the fin thickness on the thermal performance Fin thickness from 2 mm to 3 mm is considered in this paper. As shown in Figs. 14 and 15, fin thickness gradually increase from 2 mm to 3 mm, the variation range of the junction temperature and the Nusselt numbers are less than 1%. It indicates that the fin
thickness does not have a significant impact on the cooling effect of the radiator, it is because the changes in the thickness of the fin do not have a significant effect on the total cooling area when keeping other conditions are the same.
Fig. 11. Nusselt number vs. radiator fin number.
Fig. 14. Junction temperature vs. thickness of radiator fin.
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
Fig. 15. Nusselt number vs. thickness of radiator fin.
5.3.6. The effect of the heat pipe and heat pipe length on the thermal performance Fig. 16 shows that LED junction temperature average decrease about 8 C by introducing the heat pipe radiator. From Fig. 17 we can see that at the edge of pure rolling aluminum radiator fins has large temperature in the height direction. The maximum temperature difference between the upper and lower edges of the fins is about 2.6 C. However, the rolled aluminum upper and lower composite heat pipe radiator fin edges maximum temperature difference is about 0.17 C, which is only 0.6% of pure rolling aluminum radiators at the same conditions. The decrease of the temperature difference between the upper and lower part of the radiator can be attributed to the introduction of the heat pipe. The upper and lower equivalent coefficients described in Fig. 16 is shown in Table 4. On the one hand, its equivalent coefficients of top and bottom are all improved; on the other hand, the top fins almost have same equivalent coefficients with bottom fins. The heat of the bottom radiator spreads rapidly to the top radiators by heat pipe, the axial isothermal of radiators was improved which is more favorable to dissipate heat. In this paper, the influence of the length of rolled aluminum heat pipe composite radiator on heat dissipation was investigated. Fig. 18 shows that under the same condition, the increase of length heat pipe is favorable to the reduction of LED junction temperature.
71
Fig. 17. Fin top temperature vs. height difference.
Table 4 Equivalent thermal conductivity of Al radiator and Al-heat pipe radiator (Q ¼ 60 W).
The lower part of equivalency factors (%) The upper part of equivalency factors (%)
Al
Al-Heat pipe
72.9 70.8
73.2 73.1
Fig. 18. Junction temperature vs. length of heat pipe. (H ¼ 180 mm, D ¼ 160 mm, d ¼ 50 mm).
As increasing 7.4% of the heat pipe length, the average junction temperature decreases 4%. And with the increase in heat pipe length, the weight of rolled aluminum heat pipe composite radiator reduced 4.77 kg/m in this study.
6. Conclusions Al-heat pipe radiator with high thermal conductivity is developed. The thermal performance of the Al-heat pipe radiator is investigated numerically based on the correctness of the numerical model verified by the experiment. The correlations for Nu without and with heat pipe are obtained.
Fig. 16. Junction temperature vs. length of heat pipe without heat pipe and with heat pipe separately.
(1) As the increase of the radiator height, the heat dissipation effect is improved and the Nusselt number increases with a downward trend.
72
M. Wang et al. / International Journal of Thermal Sciences 113 (2017) 65e72
(2) Under the conditions of different power, the junction temperature of LED assembly increases with a downward trend as the increase in the number of fins. The enhance heat dissipation effect by increasing the number of fins to increase cooling area is weakened. (3) LED component junction temperature decreases with the increase in fin height. Nusselt number has the same tendency. (4) Fin thickness investigated in this paper does not have a significant impact on the radiator cooling effect, the variation range of LED component temperature and the Nusselt numbers are less than 1%. (5) LED junction temperature reduces an average of about 8 C by applying Al-heat pipe radiator, the equivalent coefficient of the top fins increases which are in favor of dissipating heat. (6) The heat pipe length of Al-heat pipe radiator was investigated. The results show that under the same condition, the increase of length heat pipe is favorable to reduce the LED junction temperature and weight of radiator. A kind of pure aluminum heat sink with radicalized straight fins and a kind of aluminum heat sink with heat pipe are developed in this paper. On one hand, the high-power LED radiator forms and extended cooling area methods are enriched. On the other hand, considering that there are limitations to increasing the area only by changing geometric parameters. The heat dissipation effect is enhanced by using heat pipe with high thermal conductivity and it provides theoretical foundation and guidance for the design of high-power LED component radiators. References [1] Burroughes JH, Bradley DDC, Brown AR, Marks RN, Mackay K, Friend RH, et al. Erratum: light-emitting diodes based on conjugated polymers. Nature 1990;347:539e41. [2] Dasgupta PK, Eom I-Y, Morris KJ, Li J. Light emitting diode-based detectors. Anal Chim Acta 2003;500:337e64.
_ [3] Ryskevi c N, Jursenas S, Vitta P, Bakiene_ E, Gaska R, Zukauskas A. Concept design of a UV light-emitting diode based fluorescence sensor for real-time bioparticle detection. Sensors Actuators B Chem 2010;148:371e8. [4] Huang MS, Hung CC, Fang YC, Lai WC, Chen YL. Optical design and optimization of light emitting diode automotive head light with digital micromirror device light emitting diode. Optik Int J Light Electron Opt 2010;121:944e52. [5] Chen CF, Wu CC, Wu JH. Modified wide radiating lenses of the power-chip light emitting diodes for a direct-lit backlight. Optik Int J Light Electron Opt 2010;121:847e52. [6] Beeson S, Mayer JW. Patterns of light. New York: Springer; 2008. [7] Y. Shimizu, K. Sakano, Y. Noguchi, T. Moriguchi. Indium, gallium, aluminum nitride; garnet containing rare earth oxide. In Google Patents, 1999. [8] S. Tasch, P. Pachler, G. Roth, W. Tews, W. Kempfert, D. Starick. Light source comprising a light-emitting element. In Google Patents, 2004. [9] Xu J, Xiong Y, Chen S, Guan Y. A lamp light-emitting diode-induced fluorescence detector for capillary electrophoresis. Talanta 2008;76:369e72. [10] Senawiratne J, Chatterjee A, Detchprohm T, Zhao W, Li Y, Zhu M, et al. Junction temperature, spectral shift, and efficiency in GaInN-based blue and green light emitting diodes. Thin Solid Films 2010;518:1732e6. [11] Vamvounisa G, Aziz H, Hu NX, Popovicb ZD. Temperature dependence of operational stability of organic light emitting diodes based on mixed emitter layers. Synth Met 2004;146:69e73. [12] Zhang H, Zhuang J. Research, development and industrial application of heat pipe technology in China. Appl Therm Eng 2003;23:1067e83. [13] Lan K, Choi JH, Sun HJ, Shin MW. Thermal analysis of LED array system with heat pipe. Thermochim Acta 2007;455:21e5. [14] Li J, Lin F, Wang D, Tian W. A loop-heat-pipe heat sink with parallel condensers for high-power integrated LED chips. Appl Therm Eng 2013;56: 18e26. [15] Lin Z, Wang S, Huo J, Hu Y, Chen J, Zhang W, et al. Heat transfer characteristics and LED heat sink application of aluminum plate oscillating heat pipes. Appl Therm Eng 2011;31:2221e9. [16] Zeng D, Liu Y, Jiang L, Li L, Gang X. A novel manufacturing approach of phasechange heat sink for high-power LED. Energy Procedia 2012;17:1974e8. [17] Lu XY, Hua TC, Wang YP. Thermal analysis of high power LED package with heat pipe heat sink. Microelectron J 2011;42:1257e62. [18] Cheng HH, Huang DS, Lin MT. Heat dissipation design and analysis of high power LED array using the finite element method. Microelectron Reliab 2012;52:905e11. [19] Park S, Jang D, Yook SJ, Lee KS. Optimization of a chimney design for cooling efficiency of a radial heat sink in a LED downlight. Energy Convers Manag 2016;114:180e7. [20] Jang D, Park SJ, Yook SJ, Lee KS. The orientation effect for cylindrical heat sinks with application to LED light bulbs. Int J Heat Mass Transf 2014;71:496e502. [21] Li D, Zhang GQ, Pan K, Ma X. Numerical simulation on heat pipe for high power LED multi-chip module packaging. In: International conference on electronic packaging technology and high density packaging; 2009. p. 393e7.