Physics Procedia 22 (2011) 150 – 156

2011 International Conference on Physics Science and Technology (ICPST 2011)

Analysis of Thermal Field on Integrated LED Light Source Based on

COMSOL Multi-physics Finite Element Simulation
Jingsong Lia, Qingxin Yanga,b, Pingjuan Niua, Liang Jina, Bo Menga, Yang Lia, Zhaoxia
Xiaoa, Xian Zhanga
a
Key Laboratory of Advanced Electrical Engineering and Energy Technology, Tianjin Polytechnic University, Tianjin 300387˗
b
Institute of Electrical and appliancesˈBeijing Jiaotong UniversityˈBeijing 100044ˈChina.

Abstract

This paper obtained the average integrated heat transfer coefficient for the thermal resistance of a classic of integrated LED light
source and its cooling fin-root on the basis of thermal circuit method. Simulation analysis on its steady-state temperature field
distribution using COMSOL Multi-physics finite element method was carried out. This method has high precision and intuitive
simulation results. The iteration method of the Numerical Analysis is introduced into method for the first time. The results have
significant promotion on the LED cast light structure optimization and the affection of reduced heat coupling on the light temperature
distribution. The comparison between thermocouple experimental data and calculation results proved the correctness and validity of the
proposed method. This experimental study plays a guiding role to thermal analysis and design of other integrated lights.
© 2011 Published by Elsevier B.V. Open access under CC BY-NC-ND license.
© 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of [Jingsong Li]
Selection and/or peer-review under responsibility of Garry Lee.
Keywords: LED integrated light source; thermal circuit method; COMSOL Multi-physics FEM; temperature distribution; thermal couple method;

1. Forewords

Light Eitting Dode is a class of semiconductor devices which can convert light energy directly into electrical energy [1-2].
It has aroused great public attention and is considered as the fourth generation of green lighting, meanwhile, the most
likely alternative to incandescent lighting [3]. However, cost and heat dissipation are two technology bottlenecks which
limit the LED industrial development. In particular, the development of LED regard heat problem as a fatal blemish. The
heat of LED generated in the process at work will inevitably lead to the change of its light, electricity, color performance,
etc, which has Serious impact on LED performance, if these build up without rapid dissipation. To this end, scholars from
various countries did a lot of research [4-9].
From the object of study, the main studies focus on the single chip, and for more complex structure, multi-chip package
integrated light source heat research is still relatively small. Only Lane Kim and Moo Wan Shin [10] and others studied
thermal properties of high power LED light source of multi-chip package. The results show that thermal resistance
decreases with the increase of the number of chip package, and derived the function of the number of chip and package
thermal resistance on this basis. Qi Zhang and Xu Chen[11] etc, who used finite element method to conduct thermal field
simulation of LED lights, reached a stable temperature distribution in the lights and its cooling structure has been
optimized through orthogonal design on this basis.
This is an analysis for a multi-chip integrated LED light source of thermal field (temperature field). Researchers are
mainly done in the following areas:(1) The first time introduced the iteration method in "Numerical Analysis"[12] to
compute the thermal resistance and average integrated cooling fin heat transfer coefficient of a class of LED lamps work

* Corresponding author. Tel.:+86 13702070369; fax: +022-83955735.

E-mail address: ljs13702070369@hotmail.com.

1875-3892 © 2011 Published by Elsevier B.V. Open access under CC BY-NC-ND license.
Selection and/or peer-review under responsibility of Garry Lee.
doi:10.1016/j.phpro.2011.11.024
 Jingsong Li et al. / Physics Procedia 22 (2011) 150 – 156 151

in the thermal stability on the basis of thermal circuit method in Heat Transfer. This provides a basis for subsequent
finite element simulation. (2) Using the COMSOL Multi-physics finite element method to establish the finite element
model for simulation of the LED integrated light. By analyzing the simulation results, obtained spotlights temperature
distribution. (3)By Thermocouple method, temperature measurements of such LED lights work in the thermal stability was
taken. The temperature measurements and finite element method simulation results were compared to verify the accuracy
and feasibility of the finite element method.

2. Basic Principles

2.1 Thermal circuit method

Aaccording to the principles of heat transfer [2] and integrated optical source thermal circuit method (1),
the derived formula for calculating the conduction thermal resistance of the sheet structure is as follow.
G (1)
R th
OA
Total system thermal resistance is the sum of the three series thermal resistance which can be expressed as a formula:
 Rtotal Rchip  RPCB  Rheat sin k (2)
Where Rtotal is the total system thermal resistance, Rchip is the LED chip thermal resistance, RHeat sin k is the radiator
thermal resistance, and RPCB is the PCB board thermal resistance. See Fig.1.

Fig.1. Thermal circuit diagram of a typical LED lighting system

Similar to the current source, the total heat is equal to the sum of each heat when the heat source consist of the multi-
chip integrated light source comes in parallel. The reciprocal of the total parallel resistance equals to
the reciprocal of thermal resistance. Namely:
) total n ˜ ) (3)

1 
Rchiptotal Rchip
n
2.2 Comprehensive heat transfer coefficient calculation of LED project lamp

For the integrated heat transfer coefficient, the paper used empirical formula [5] and combined iteration methods for
calculation; we can calculate the average heat transfer coefficient D .
Iterative methods [12] of the specific steps are as follows:
(1) Assuming temperature difference 't .
In a given environment temperature cases of 20 qC , first assume the average temperature difference between the
environment temperature and radiator fin-root. By the temperature definition formula 't t Heat sin k  tD , assume the
average temperature of the fin-root radiator when LED reaches heat stable state.
(2)Calculating the average integrated heat transfer coefficient D . (See reference [5])
(3)Calculate the new temperature difference 't .
Substitute the integrated heat transfer coefficient D from above into the Fin-root thermal resistance calculation
formula (1) to obtain the radiator fin-root thermal resistance. According to formula (2), we can get the average temperature
difference between the radiator fin-root and the environment under the dissipation power known conditions.
152 Jingsong Li et al. / Physics Procedia 22 (2011) 150 – 156

(4) Iterative calculation of temperature difference 't .

Repeat steps (2), (3), compare the temperature 't difference between before and after. The error meets the
requirements when the temperature difference 't is convergence and tends to a stable value. The temperature
difference 't at this point, is he average temperature difference between ambient temperature and Radiator fin-root when
the system reaches heat stable state.
(5) Solving the average temperature of project-light lamp for each part.
According to the average temperature difference 't from iteration, we obtained the fin-root convection heat thermal
resistance from formula (1). In the case of knowing each part of the thermal resistance of cast light and ambient
temperature tD , calculate temperature difference between each part step by step starting from the fin and obtained the
average temperature for each part. The average temperature of all parts of the spot lights was calculated through the use of
iterative method. See Table.1.
Table.1. Datasheet of the LED projection lamp thermal circuit

The parts Coefficient of Comprehensive heat Thermal resistance Average

thermal conductivity transfer coefficient temperature
(K/W)
(W/(m*k) ) (W/(m*k) ) ( qC )
Heat source 0.571 60.84
of LED chip
üü üü
387 üü
PCB board 0.5 üü 0.013 52.34
144 üü
The
aluminum 144 üü 0.002 50.71
floor of
radiator
Radiator of 144 5.881 1.170 49.53
the fin-root

From the values above, we can obtain the average integrate heat transfer coefficient 5.88W / (m2 ˜ K ) .

3. The finite element analysis and modeling

3.1 LED temperature field of steady state [13]

According to the Fourier law [14] and the steady temperature field basic principle we can get steady state temperature
field under three dimensional Poisson equation of the general form as follows:
’ ˜ O’T  U (1  DT ) J 2 G (4)
0
The coefficient of thermal conductivity O , resistance coefficient Uo and resistance temperature coefficient D of material
in the formula are all function of coordinate x, y, and z. See references [13], [14].

3.2 The finite element model [14-16]

In this paper, COMSOL Multi-physics finite element method was used for the analysis of the temperature field of
certain types of LED lamps. See Fig.2.
 Jingsong Li et al. / Physics Procedia 22 (2011) 150 – 156 153

Fig.2. LED project lamp model (1/4)

4. Software-based simulation

4.1 Simulation Solution

Click on the solver, COMSOL Multi-physics can automatically get results. Project-light lamp quarter model of
temperature distribution is as shown in figure 3 (a) ~ 3(c) blow:

Fig.3. (a) isometric view of the LED project lamp temperature distribution (1/4 model); (b) front view of the LED project lamp temperature distribution
(1/4 model)

Fig.3. (c) back view of the LED project lamp temperature distribution (1/4 model)

4.2 Results and post-processing

According to the symmetry of spotlights temperature distribution, the model is extended along the symmetry plane.
We can get the complete temperature distribution model as shown in figure 4(a) ~ 4(b). LED spotlights
under thermal steady state as shown in figure 5 show the highest temperature.
154 Jingsong Li et al. / Physics Procedia 22 (2011) 150 – 156

Fig.4. (a) front view of the LED projection lamp temperature distribution (whole model); (b) back view of the LED projection lamp temperature
distribution (whole model)

Fig.5. Nodes with MAX temperature of the LED project lamp

4.3 Data analysis

By observing the comparative figure 3(a) ~ 5 which shows the temperature of the LED lamps for light distribution
data we can analysis the simulation results from the following two aspects.
(1)From the temperature distribution uniformity, we can see that overall temperature of spotlights distribution is not
uniform. The highest temperature is about 56 qC , which is located in the center project-light lamp of two LED chips place,
as the mark shown in figure 5 MAX. Minimum temperature is less than 54 qC (easy to know by the temperature scale
color), temperature differs more than 2 qC . Set ambient temperature as 20 qC , the maximum temperature rise is 36 qC , set
the ratio of the maximum temperature within lighting fixtures and the maximum of the surrounding environment
temperature as W , there are:
2
W u 100% | 5.6%
 36 
The total power of this type of LED project-light lamp is about more than 30 watts, W is about 5.6%. This type
of LED light fixtures has a good temperature distribution. What need to be improved is that the speed of temperature rise
is slow.
(2)Viewing the problems of the heat source coupling: Cast light by the overall temperature distribution shows: in the
adjacent smaller area, the closer to the geometric center of the lamp, the higher the temperature are. As shown in figure 3(c)
and figure 4(b) the arrow shown; the closer the chips to the heat source, the higher the temperature are. See the direction of
arrow in figure 3(a).
The temperature distribution is the result of the coupled heat source. This is because, in the central of this type of LED
lamps area, more heat distributed and relatively concentrated. Thermal coupling phenomenon is more obvious. From
the temperature form, regular pattern was showed the closer to the source of heat, the closer to the lamp center, the higher
the temperature rise is.

5. Experimental verification

5.1 Analysis of temperature data

Figure 6 shows the comparison total diagram of the measured temperature with a thermocouple and the finite
element simulation. The figure which marks the number of temperature measurement points measures average
temperature, the original one shows finite element simulation of the temperature.
 Jingsong Li et al. / Physics Procedia 22 (2011) 150 – 156 155

Fig.6. Temperature measured in the testing points and FEM of the temperature (total figure)

For each data differs not significantly and has no obvious fluctuations, we only take three groups of measurements. The
environment temperature is (20 r 1) qC when measuring. Observing data in the table, we know that three groups of data
values measured are not significant differed. Take each number of average temperatures and compared with finite element
simulation temperature, we can know that the minimum temperature is 0.55 qC and the maximum temperature difference
is 2.64 qC . The maximum temperature difference point lies on the 10 sampling point.

5.2 Comparison of three methods

Comparing the temperature curves in figure7, it is known that: temperature trends of the three methods are basically
the same. The highest temperature distributions are consistent and all are located in project-light lamp canter of the chip in
the heat source. The highest temperature derives from the thermal circuit method, while the rest are relatively in low
temperature. It is because the thermal resistance data is provided by the LED chip manufacturers. See Fig.7.

Fig.7. Temperature curves of the testing points by the three methods

Combining with the three methods above, through the contrast analysis, we can draw a conclusion. There are:
(1)The finite element method in the process of modelling ignored the chip internal structure and chip PCB, as well as
the air gaps between the PCB board and the aluminium plate, the impact on the accuracy of the simulation results is not
big.
(2)The finite element method simulation of the temperature and measured temperature data fits better. The temperature
distribution rule is consistent.
(3)Because of the finite element method simplified when modelling, we ignored the chip internal structure and all the
parts of the contact between air gaps. This caused the chip temperature slightly lower than the experimental data. The
biggest simulation error on the maximum temperature rise of this method is less than 7.3%, which accounted for only
4.7% of the actual temperature; this can meet the engineering accuracy requirements.
In conclusion, the experimental method of temperature measurement proved that LED thermal field analysis results
based on COMSOL Multi-physics finite element simulation are accurate, feasible and effective.

6. Conclusion
156 Jingsong Li et al. / Physics Procedia 22 (2011) 150 – 156

This paper launched the research to LED integrated light source thermal issues on the principle of heat transfer.
Through the using of the thermal circuit method to establish the integrated LED light source and thermal resistance model
and the using of iterative method to get every part of the thermal resistance, average temperature and radiator fin-root
average of integrated heat transfer coefficient, we obtained the basis for finite element temperature field simulation. We
established the LED lamp board finite element model, and ultimately obtained the simulation results of LED lamp on the
temperature field through the setting of temperature field partial differential equations and boundary conditions. Then use
thermocouple method under steady-state for temperature measurement. Through the finite element simulation results
which are being compared and analyzed, verified the accuracy and effectiveness of the finite element computation method
and results. The experimental study played a guiding role on the thermal field analysis and design of the other integrated
light sources.

References

[1]Deqing Yang. LED and its Application [M].Beijing: People's Posts and Telecommunications Publishing Press, 2007:13-48.
[2]Xingwu Mao, YanWen Zhang, Jianjun Zhou,etc. A New Generation of Green Light LED Source and its Application Technology [M].Beijing:
People's Posts and Telecommunications Publishing Press, 2008:101-139.
[3]Hong Liu [D]. National Development Forecasting Model Research of LED Lighting [D]. Tianjin: Tianjin University, 2005.
[4]Yuandeng Chen, Yu Chen. LED Manufacturing Technology and Application (Second Edition) [M]. Beijing: Publishing House of Electronics
Industry.
[5]Peter Baureis. Compact Modeling of Electrical, Thermal and Optical LED Behavior. Proceedings of ESSDERC 2005, 2005, 2005:145-148.
[6]Jim Browne. Evaluation of an Advanced Thermal Transfer Adhesive for Thermal Management Applications[C].Sixteenth IEEESEMI-THERMN
Symposium. Rhodes, Greece:IEEE, 2000:182-187.
[7]Farn-Shiun Hwu,Jyh-Chen Chen. A Numerical Study of Thermal and Electrical Effects in a Vertical LED Chip [J]. Journalof the
Electrochemical Society, 2010, 157(1):H31-H37.
[8]Zetao Ma, Xiaojun Wang. The Thermal Analysis of a High Power LED [J]. Photoelectric Device, 2006, 27(1): 16-19 (in Chinese).
[9]Chun-Jen Weng. Advanced Thermal Enhancement and Management of LED Package [J].International Communications in Heat and Mass
transfer, 2009, 36(3):245-248.
[10]Lan Kim, Moo Whan Shin. Thermal Resistance Measurement of LED Package with Multichip [J].IEEE Transactions on Components and
Packaging Technologies, 2007, 30(4):632-636.
[11]Qi ZhangˈXu Chen. The Modular Design and Study of LED Street Lamps [J]. China Lighting Electric, 2010, 5:1-4 (in Chinese).
[12]Qingyang Li, etc. Numerical Analysis [M]. Wuhan: Haiphong University of Science and Technology press, 2006, 187-249.
[13]Hui Ma, Gang Wang. Basic Operational Guidelines and FAQ of COMSOL Multi-physics [M].Beijing: People's traffic press.2009:111-178.
[14]WeiLi Yan, Qingxin Yang, Youhua Wang, etc. Numerical Analysis of Electrical Engineering Magnetic Field [M]. Beijing: China Machine
Press, 2005:216-268.
[15]Bo Pang. Thermal Field Analysis of Integrated LED Light Source [M]. Tianjin: Hebei University of Technology, 2010.
[16]William B. J. Zimmerman, COMSOL Multi-physics. Multi-physical Finite Element Modelling and Analysis [M], Beijing people's traffic press,
2007:18-143.

Biographical sketch

This research is the key project of Tianjin application foundation and frontier technology research plan (Tianjin
natural science fund). (10JCZDJC15400)
Jinsong Li (1987- ), male, the bachelor of engineering. Graduated from the institution of electrical engineering and its
automation of Chengdu University of Technology in 2010. Postgraduate student majoring engineering electromagnetic
field at Tianjin Polytechnic University since 2010, mainly engaged in
LED optoelectronic integrated thermal and magnetic properties research of the interaction mechanism and wireless energy
transfer and conversion of Electromagnetic - Mechanical synchronous resonance research works.Tel:+86
13702070369;E-mail:ljs13702070369@hotmail.com.
Qingxin Yang (1961- ), male, professor, doctorial tutor and headmaster of Tianjin Polytechnic University, mainly
engaged in engineering magnetic field numerical calculation and the magnetic technology research.