International Journal of Heat and Mass Transfer 86 (2015) 667680

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/ijhmt

The study of a thermoelectric generator with various thermal conditions

of exhaust gas from a diesel engine
Byung deok In a, Hyung ik Kim a, Jung wook Son a, Ki hyung Lee b,
a
b

Graduate School of Mechanical Engineering, Hanyang University, 1271 Sa3-dong, Sangrik-gu, Gyeonggi-do 426-791, Republic of Korea
Mechanical Engineering, Hanyang University, 1271 Sa3-dong, Sangrik-gu, Gyeonggi-do 426-791, Republic of Korea

a r t i c l e

i n f o

Article history:
Received 23 April 2014
Received in revised form 11 March 2015
Accepted 15 March 2015
Available online 1 April 2015
Keywords:
Thermoelectric generator
Seebeck effect
Heat sink shape
Heat transfer coefcient

a b s t r a c t
It is well known that internal combustion engines have approximately 30% thermal efciency, and the
remaining energy is lost as heat through both exhaust gas and coolant. To improve upon this efciency,
technology for recycling waste energy is desirable. Increasing the thermal efciency of engines would
also reduce CO2 emissions. Recently, waste energy recovery systems have drawn a great deal of attention
for their potential ability to address stringent emission regulations as well as fossil fuel depletion.
Thermoelectric generators (TEGs) can be applied to produce electricity from waste heat; however, there
are still remaining many problems to be solved. It would be signicant not only to improve the efciency
of the TEG, but also to optimize the operating conditions. Therefore, this study provides useful information on optimization of the TEG system, including optimization of the system layout, and conguration of
TEG. In addition, the heat transfer characteristics of thermoelectric modules were predicted by theoretical analysis and compared to experimentally measured heat transfer characteristics. The performance of
thermoelectric generators heated by exhaust gas from an actual engine was investigated. From these
results, we claried factors affecting the optimization of thermoelectric module according to operating
conditions.
2015 Elsevier Ltd. All rights reserved.

1. Introduction
Recently, the automotive industry in the world emphasizes
more and more about the importance of CO2 gas reduction as the
global warming is sharply progressed. Besides, the fuel efciency
due to the oil price became the most important issue at present.
It is well known that the thermal efciency of internal combustion
engines currently being used in the automobiles cannot exceed
30%, only one-third of the fuel or combustion energy is converted
into power energy, and the rest is wasted. The greatest losses
escape into the exhaust gas and coolant, and the most reliable
way to improve fuel efciency is to minimize these losses [1]. To
improve the fuel efciency and the combustion performance of
engines, the earlier efforts were on the optimization of fuel mixing
process or combustion system. Recently, a technology using waste
heat from exhaust gas and coolant has drawn considerable attention. A waste heat recovery system is the technology needed to
improve automotive fuel efciency by recovering heat wasted in
the form of engine coolant and exhaust gas and converting it back
into useable power. The ability to improve fuel efciency by
Corresponding author.
E-mail address: hylee@hanyang.ac.kr (K.hyung Lee).
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2015.03.052
0017-9310/ 2015 Elsevier Ltd. All rights reserved.

recovering waste heat from automotive engines represents a major

opportunity to reduce CO2 emissions [2]. Therefore, it is necessary
to develop a competitive and reliable key technology for the recovery of waste heat exhausted from engines [3]. As demand for cars
with high gas mileage continues to increase as a result of fossil fuel
depletion, the waste heat recovery system is the good way to
improve fuel efciency in the cars by recovering heat which otherwise disposed as an engine cooling water and exhaust gas [4].
However, earlier researches for improving fuel efciency by recovering waste energy were mainly focused to the thermoelectric
power generation system which uses a waste energy from iron
works and incinerators. On the other hand, the researches about
engine waste heat recovery system are not being so progressed.
Up to now, many efforts were usually concentrated on the
optimization of combustion phenomenon to improve the fuel efciency of engines instead of this technology [5]. Among them, a
steam generation is currently one of the main technologies used
for waste heat recovery [6]. In this study, we focus on the
thermoelectric generation technology, which can convert heat
energy into electric energy, thereby simplifying the system [7].
In a thermoelectric module, when the temperatures of its opposite sides are different, power is produced by the Seebeck effect [8].
Conversely, if the module receives a power input, it actively moves

668

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

Nomenclature
A
AB
D
Dg
f
Hf
h
k
L
LHS
P
q_

area [m2]
area of heat sink base [m2]
diameter of the exhaust pipe [m]
n to n spacing [m2]
friction factor
n height [m]
heat transfer coefcient [W/m2/K]
thermal conductivity [W/m/K]
length of thermoelectric generator [m]
length of heat sink [m]
overall generated electricity [W]
internally generated heating per unit volume [W/m3]

heat energy by means of the Peltier effect [9]. Keeping in mind the
efciency improvement of the TEG, we examine the thermoelectric
module with regard to power generation [10]. Since power is generated by the temperature difference between the sides of the
thermoelectric module without the need for any moving parts,
the power generation device has a simple, structure and can have
long life. However, thermoelectric modules presently have less
generation efciency than steam generation modules, presenting
a hurdle for commercialization. Therefore, thermoelectric generation has mostly been used in specialized electric equipment in
aerospace and military applications. Research has advanced the
ability to use thermoelectric modules for power generation in
automotive applications [3]. However, so far thermoelectric
generators have rarely been applied in automobiles because the
ow rate and the exhaust gas temperature change according to
the engines operating conditions. To implement thermoelectric
modules in automobiles, rst of all, analysis of the characteristics
of thermoelectric module is required [4]. Also, performance
improvements are needed that can enhance engine performance.
However, data about optimum performance to implement
thermoelectric module in the automobile is lacking. Besides, information about necessary elements required for performance
improvement is quite poor. Thus, it is necessary to clarify the characteristics of thermoelectric modules and investigate the possibility of applying them to the engine system. In a previous study,
basic data were obtained about the heat transfer characteristic of
a thermoelectric module by means of theoretical analysis as a prerequisite stage to adopt it in automobile engines [11]. The effect of
the thermoelectric modules shape upon heat transfer characteristics has been claried [12]. Further, the exhaust system of an actual
engine was equipped with thermoelectric modules, and their characteristics of were investigated under different drive conditions to
evaluate the possibility of implementing thermoelectric modules
in the engine was evaluated through these experiments [13].
Finally, by providing factors affecting the optimization of the
thermoelectric module for various drive conditions, know-how
was established that was needed to develop devices to converts
waste energy from exhaust gas into electric energy. In addition,
we also provide valuable information in terms of the optimization
of the TEG system, including optimizing the system layout, exhaust
manifold design, and conguration.

2. Experimental apparatus and procedure

2.1. Theoretical analysis of the thermoelectric module
In this study we investigated the principle and operation characteristics of the thermoelectric module, which is a most important

Q
WHS

thermal power [W]

width of heat sink [m]
seebeck coefcient [V/K]
seebeck parameter of TEG [1/K]
heat-transfer ratio of heat absorber to heat sink
thickness of thermoelectric generator [m]
Biot number of TEG
viscosity coefcient
density
inner spread, i.e. of the spreader of the heat absorber
outer spread, i.e. of the spreader of the external heat
sink

a
a
c
d
k

l
q
i,s
o,s

part of the TEG. The Reynolds number can be determined based on

the ow rate of exhaust gas and the diameter of the exhaust pipe,
as expressed in the following Eq. (1).

RE

qVD
l

Then, the friction coefcient can be calculated using Eq. (2) if

The Reynolds number is less than 100,000.

0:3164
RE1=4

The pressure loss, which strongly affects the heat transfer, can
be determined by using the friction coefcient as given in Eq. (3).

Dp f

L qV 2
D 2

Since ow rate increases with decreased friction loss and engine

load increases according to drive conditions, pressure loss shows
an increasing trend.
In the case of incompressible uid, the pressure drop that
occurs in the circular cylinder can be applied in the energy equation as below.

V 21 p1
V2 p
z1 2 2 z2 h L
2g q
2g q

here, hL is the pressure drop. In this case, because the sectional area
of circular cylinder is constant,

V1 V2

Therefore, Eq. (4) simplies to

hL

Dp ch

assuming that the front and rear ends of the exhaust pipe are of the
same height (Z1 = Z2). The energy loss can be expressed in proportion to Dp. In other words, the lost energy is the difference between
inow energy and outow energy. It is considered that this energy
is transferred to thermoelectric module through heat sink because
this energy has an effect on the heat transfer. Therefore, the characteristics of the thermoelectric module were analyzed based on pressure difference in this study.
Also, characteristics of heat transfer were calculated using the
mechanism of the Seebeck effect as shown in Fig. 1.
The heat conduction equation can be derived based on the
internal energy generation of the thermoelectric module [14].

@2T
q_

0
@x2 kTEG

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

669

Fig. 1. Heat transfer mechanism and Seebeck effect for the thermoelectric module [14].

Fig. 2. Photograph of thermo-electric module showing the cold side (left) and hot side (right).

Here, q_ is heat energy generated per unit volume, and I and d are
the current and the thickness of thermoelectric module, respectively. If the high temperature side and low side are assumed to
be same, the boundary conditions are as follows.

T0 T i;s

at

x0

Td T o;s

at

xd

Eq. (7)(9) can be combined to yield the following equation.

Tx T i;s 

q_
T o;s  T i;s qd2 =2k
x2
x
2kTEG
d

10

Thermal power output from the high-temperature side and the

low-temperature side can be expressed as in the following equations by considering the Seebeck effect of the respective sides.

QH

kTEG ATEG
1
T i;s  T o;s aT i;s I  Rint I2
2
d

11

670

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

Table 1
Properties of HZ-20.
Physical properties
Width & length
Thickness
Weight
Compressive yield stress
Number of active couples
Electrical properties
Power
Load voltage
Current
Open circuit voltage
Efciency
Thermal properties
Max. intermittent temperature
Min. intermittent temperature

Value
7.5 cm
0.5 cm
115 grams
70 MPa
71 couples
Value
19 Watts
2.38 Volts
8 Amps
5 Volts
4.5%
Value
400 C
None

Table 2
Engine specications.
Displacement

2199 cc

Bore
Stroke
Compression ratio
Max. power (PS/rpm)
Max. Torque (kg m/rpm)

85.4 mm
96 mm
16
200/3800
44.3/1800

QC

T o;s 

kTEG ATEG
1
T i;s  T o;s aT o;s I Rint I2
d
2

12

Here, ATEG can be expressed as the area of the heat sink in the
case of the high-temperature side, and as the area of the
thermoelectric module at the low temperature side. Eq. (7) can
be expressed differently as follows.

@T
h1 A1 T i  T i;s kTEG ATEG
jx 0 a tx 0
@x
h2 A2 T o;s  T j kTEG ATEG

@T
jx d at
@x

xd

T i;s

dATEG
h2 A2




2
h1 A1 d
2kdTEG q_ hh12 AA12 kTEG
Ti Tj
ATEG
h1 A1
h2 A2

h1 A1 d
kTEG ATEG

h 1 A1
dATEG
h 1 A1
q_
T i;s
Ti Tj
h 2 A2
h2 A2
h 2 A2

16

This equation can be expressed in terms of calories at the hightemperature and low-temperature sides as follows.

Q H h1 A1 T i;s  T o;s aT i;s I

@T
jx 6 aT o;s I
@x
h1 A1 T i;s  T o;s dATEG q_ aT o;s I

17

Q C kTEG ATEG
13

14

When Eqs. (7), (13) and (14) are substituted into Eq. (10), the
following equation results.

Fig. 4. Shape of heat sink.

15

18

Therefore, the total generated power can be expressed as

follows.

P Q H  Q C aIT i;s  T o;s  dATEG q_

19

Also, if the above equation is dened as

h1 A1
h1 A1 d
a2 d
;k
; a
kTEG ATEG
4kTEG ATEG Rint
h2 A2

Fig. 3. Schematic diagram of engine testing device using thermoelectric modules.

20

671

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680
Table 3
Experimental conditions and temperature conditions.
CASE

RPM

Torque (kg m)

Temperature (C)

Flow rate (kg/s)

1
2
3
4
5
6
7

1500
1500
1500
1500
1500
1500
1500

6
8
10
12
14
16
18

250
293
342
397
437
466
487

0.01527
0.01729
0.01816
0.02455
0.02939
0.03420
0.03765

(a) Triangular pillar Heat sink as forward direction

Table 4
Properties of HZ-14.
Physical properties
Width & length
Thickness
Weight
Compressive yield stress
Number of active couples
Electrical properties
Power
Load voltage
Current
Open circuit voltage
Efciency
Thermal properties
Max. intermittent temperature
Min. intermittent temperature

(b) Triangular pillar Heat sink as reverse direction

Value
6.3 cm
0.5 cm
82 grams
70 MPa
49 couples
Value
14 Watts
1.65 Volts
8 Amps
3.5 Volts
4.5%
Value
400 C
None

Therefore, the maximum electric power generated is calculated

using Eq. (24).
(c) Rectangular pillar Heat sink

Fig. 5. Heat sink installation condition.

Pmax
and if Eq. (20) is substituted into Eqs. (15) and (16), Eq. (21)
results.
2

a T i;s  T o;s  c k 1T i;s  T o;s kT i  T j 0

T i;s  T o;s

q
c k 1  c k 12  4a2 kT i  T j
2a

21

When Eq. (22) is rearranged, it is ultimately dened as Eq. (23).

T i;s  T o;s

c k 1 

q
c k 12  4a2 kT i  T j
2a

23

q2
c k 12  4a kT i  T j
4a
24

Also, generated total heat transfer amount was calculated using

the following equations.

Qs
22

kTEG ATEG

d

c k 1 

D2g Dp
C p T i  T o
1 t f =Dg 12l LHS

qW HS Hf

25

Total heat transfer can be calculated using Eq. (25). This equation can be arranged as Eq. (26).

Q l 1:208kF W HS Hf

Ti  To
1 t f =Dg

!1=3
PrLHS  p

qm2 D2g

Fig. 6. Differential pressure gage installed at front end and rear end of thermoelectric module.

26

672

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

Fig. 7. Schematic diagram of engine testing device using thermoelectric generator.

Fig. 8. Schematic diagram of each module installation on thermoelectric generator.

In addition, the heat transfer coefcient is dened as in Eq. (27).

Q max
AB T i  T o

27

2.2. Experimental procedure according to heat sink shape

The power generation characteristics of modules operated by
engine exhaust heat were examined by means of experiments
using HZ-20 (Hi-Z Inc.) to review the possibility of applying
thermoelectric modules in actual engines (Fig. 2).
In case of the cold side, a constant-temperature water bath (CPT
Inc. CDRC-8) was used to maintain coolant temperature and constant ow (12 L/min) in this experiment.
The thermoelectric module was a Bi2Te3-type module with
dimensions 75  75  5 mm. Table 1 lists properties of the
thermoelectric module. Electric load (KIKUSUI PLZ334 W) was
used to measure the characteristics of thermoelectric generator.
The generated voltage, temperature, and air ow were recorded
using a data collector (cDAQ-9178) while input current was

increased sequentially. Temperatures at the hot side (Th) and cold

side (Tc) were measured by thermocouples; grooves were
machined into both ends of the thermoelectric module and the
thermocouples (K-type) were attached inside the grooves. During
measurement, a thermal compound was applied at the contact
point of the aluminum block and module to minimize heat resistance and to avoid a gap at the contact.
The experiment was performed using actual exhaust gas to
evaluate the performance of the thermoelectric modules. A commercial 2.2-L diesel engine was used in the experiment; Table 2
lists its specications.
Thermoelectric modules were installed at the rear end of the
catalyst on the exhaust line as shown in Fig. 3. Thermoelectric
modules were xed on the same line with the exhaust pipe and
positioned at the rear end of the catalyst, considering the characteristics and structure of commercial exhaust lines of cars during
the experiment. In addition, temperatures from the front and rear
ends of the thermoelectric module were measured.
The uidity and the exhaust gas temperature are the most
important factors in the automotive waste heat recovery system.

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

(a) Temperature from triangular prism Heat sink

installed as forward direction of exhaust

673

measured by attaching thermocouples to both ends of the

thermoelectric module. Temperature and ow of the waste gas of
each part were measured using Labview, and data were obtained
in real time. Engine speed (rpm) and torque were adjusted using
a power controller.
In this series of experiments, the heat sink at the thermoelectric
module was varied. Triangular prism and square pillar heat sinks
were replaced for the experiment, as shown in Fig. 4. The triangular
prism heat sinks were installed to face the forward and reverse
directions of the exhaust gas ow. Fig. 5 shows each heat sink
installation condition.
The differential pressure at the front and rear ends of the
thermoelectric module was measured to investigate the ow rate
of the exhaust gas for each installation condition. The differential
pressures from the front and rear ends of the thermoelectric module were measured using the method shown in Fig. 6. Increasing
differential pressures between the front and rear ends of the
thermoelectric module indicate decreasing ows of exhaust gas;
this implies that the exhaust gas remained longer inside the
thermoelectric module, which increased the heat transfer capability of the exhaust gas inside the heat sink.
The temperature at the front end of the thermoelectric module
must be higher than 250 C for the module to perform optimally.
Input resistance into the thermoelectric module was also adjusted
within 3 O to investigate its effect upon thermoelectric module
performance. Table 3 lists experimental conditions.
2.3. Experimental method with a thermoelectric generator

(b) Temperature from triangular prism installed as

backward direction of exhaust

A thermoelectric generator was fabricated in this study by combining a thermoelectric module, and the generation characteristics
of the thermoelectric modules were investigated. A total of 20
thermoelectric modules were used to make the thermoelectric
generator, including ten HZ-20 modules and ten HZ-14 modules.
It is known that there is the temperature difference between front
end and rear end of generator. Since the temperature of rear side is
low, the performance of HZ-14 can be achieved in this condition
because the temperature of front side is too high for operating
range of HZ-14. Thus, the HZ-14 which has lower performance
than HZ-20 was installed at the rear end of generator. The HZ-14
module was used in order to maintain temperature conditions,
because the temperature of the exhaust gas decreases the farther
it is from the exhaust port. Table 4 lists the properties of the HZ14 module. Table 4 lists the properties of the HZ-14 module.

(c) Temperature from rectangular pillar Heat sink

Fig. 9. Exhaust gas temperature by different experiment conditions (Tc = 80 C, HZ20).

Therefore, a constant-temperature water bath was used to maintain Tc. In addition, the electric load was again used to measure
the performance of thermoelectric power generator. Th and Tc were

Fig. 10. Comparison of differential pressures under different experimental conditions (Tc = 80 C, HZ-20).

674

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

(a) Forward direction of the triangular pillar Heat sink

(a) Forward direction of the triangular pillar Heat sink

(b) Reverse direction of the riangular pillar Heat sink

(b) Reverse direction of the triangular pillar Heat sink

(c) Rectangular pillar Heat sink

Fig. 11. Voltage comparison by different experimental conditions (Tc = 80 C, HZ20).

The current and voltage generated by each module were measured after constructing an electric circuit. A power value was calculated based on the measured current and voltage by using Eq.
(24). Fig. 7 shows the prototype thermoelectric generator installed
on the exhaust pipe of the engine. A rectangular pillar-shaped heat
sink was used in this experiment; it showed the highest heat conductivity with good heat transfer performance. The temperature at
the cold side was adjusted and maintained by the temperature
controller. The generated power was calculated using the current
and voltage data collected from the thermoelectric generator. In
addition, the temperature difference between the hot side and cold

(c) Rectangular pillar Heat sink

Fig. 12. Current comparison result under each experiment condition (Tc = 80 C,
HZ-20).

side was adjusted by changing Tc. The power generation of the

thermoelectric generator was measured according to the differential temperature range.
The engine speed was xed at 1500 rpm as noted in Table 3, and
the load was increased from 618 kg m in increments of 2 kg m.

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

675

Fig. 14. Reynolds number of exhaust gas (Tc = 80 C, HZ-20).

(a) Forward direction of the triangular pillar Heat sink

(b) Reverse direction of the triangular pillar Heat sink

Fig. 15. Friction factor of exhaust gas (Tc = 80 C, HZ-20).

(c) Rectangular pillar Heat sink

Fig. 13. Comparison of power generation under each experiment condition
(Tc = 80 C, HZ-20).

Fig. 16. Comparison of theoretical power and experimental power (Tc = 80 C, HZ20).

This experimental condition was used to maintain an appropriate

exhaust gas temperature, because of the internal heat inside the
thermoelectric module; if the exhaust gas temperature is too high,

the thermoelectric module will break. With the PEG setting, the
power generation characteristics of the thermoelectric generator
were examined for various Tc: 80, 100, and 120 C.

676

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

Fig. 17. Total heat transfer of theoretical result and experimental result (Tc = 80 C,
HZ-20).

Fig. 18. Heat ux of theoretical result and experimental result (Tc = 80 C, HZ-20).

As in the previous experiment, module numbers from 1 to 10

were assigned to HZ-20, and module numbers from 11 to 20 were
assigned to HZ-14. Fig. 8 shows each module installation condition.
3. Experimental results and discussion

Fig. 19. Power generation performance under different experimental condition

(Tc = 80 C).

3.1. Experimental results according to heat sink type

Temperatures were measured for each heat sink installation
condition. The minimum temperature of the front end of the
thermoelectric module was around 250 C, and the maximum temperature was around 430 C (Fig. 8). The temperature difference
between the front and rear ends of the thermoelectric module
averaged around 30 C. The differential pressures under different
experimental conditions were compared; with increased engine
load, the differential pressure of the thermoelectric module
increased (Fig. 9). The increase in the differential pressure indicates
unstable exhaust gas ow, which implies that heat transfer to the
heat sink would further enhance the heat transfer. The differential
pressure was the lowest when the backward-facing triangular
prism heat sink was used, followed by the forward-facing triangular prism heat sink. The greatest pressure differential was observed
for the rectangular pillar heat sink. This is because exhaust gas ow

can be interrupted by the heat sink. In the case of the triangular

prism heat sink, the resistance of ow was low when the inclined
face of the heat sink faced the exhaust gas ow. On the other hand,
in the case of the rectangular pillar heat sink, the ow resistance
was high because the heat sink directly contacted the exhaust
gas ow, increasing interruption of the exhaust gas ow. This
implied that there is a pressure differential between various heat
sink shapes. This phenomenon can be used to increase the rate of
heat transfer by slowing the exhaust gas ow. Heat transfer was
also improved by increasing the sectional area that was able to
transfer the heat from the heat source to the thermoelectric module, increasing the heat release rate by improving the heat transfer
in the perpendicular direction and thereby increasing the thermal
diffusivity inside the heat sink. Therefore, the rectangular heat sink
had the best heat transfer capability and improved the performance of the thermoelectric module.

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

677

3.2. Comparison of theoretical and experimental results

For various operating conditions, as the engine load was
increased, the Reynolds number increased (Fig. 13). This is because
the ow rate and velocity of exhaust gas increase when the engine
load is increased; also, increasing the engine load increases the turbulence of the exhaust gas ow. It was also conrmed that the friction coefcient decreased as the turbulence was strengthened (see
Fig. 14).

Fig. 20. Comparison of total power under different experimental condition

(Tc = 80 C).

Fig. 10 plots the generated voltages measured for the various

input resistances of the thermoelectric module depending on the
heat sink shape; the greatest voltage (about 2.7 V) was generated
when the rectangular pillar heat sink was used, followed by the
forward-facing triangular prism heat sink (about 2.5 V) and the
reverse-facing triangular prism heat sink (about 2.4 V). As discussed above, this performance difference might be due to differences in heat transfer arising from differences in the ow of
exhaust gas inside the thermoelectric module. However, the
increase in voltage diminished as the input resistance increased.
In addition, the highest voltage level for each of the heat sink
shapes was observed when the input resistance was around 3.3 O.
The results of the current measurements from the thermoelectric module when input resistance was varied for the three heat
sink shapes and installation conditions are discussed as follows.
The average current measured from the heat sink was greatest
for the rectangular pillar, followed by the forward-facing triangular
prism and the reverse-facing triangular prism. The same trend was
observed in voltage measurements. The measured current was
greatest (around 5 A) for the rectangular pillar heat sink, followed
by 4.7 A for both forward- and reverse-facing triangular prism heat
sinks. As the input resistance was increased, the current decreased.
However, the increment was gradually reduced afterwards. Also,
the greatest current for each experimental condition was recorded
when the input resistance was around 0.3 O (see Figs. 11 and 12).
The results of the current measurements from different
thermoelectric modules with different heat sink shapes and input
resistances are discussed as follows. As with the voltage and current measurements, the power was greatest for the rectangular pillar-shaped heat sink (around 6.2 W); the forward-facing triangular
prism heat sink generated the second most power (around 5.5 W),
followed by the reverse-facing triangular prism heat sink (around
5.5 W). Power decreased as input resistance increased, but the
increments of this reduction gradually decreased. The greatest
power for each experimental condition was measured when the
input resistance was around 0.3 O. The voltage and current measurements differed due to the correlation between voltage and current expressed by P = VI.
Practically, the allowable safe temperature of the thermoelectric module is around 250 C, and the maximum temperature it
can withstand is around 450 C. The exhaust gas temperature
was higher than 250 C under most of the load conditions tested
during the engine experiments, which might have reduced the performance of the thermoelectric modules.

(a) Tc at 80

(b) Tc at 100

(c) Tc at 120

Fig. 21. Current characteristics under different Tc.

678

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

(a) Tc at 80

(b) Tc at 100

(a) Tc at 80

(b) Tc at 100

(c) Tc at 120
Fig. 22. Voltage characteristics by variation of Tc.

To validate the theoretical analysis, theoretical value and most

result of actual value were compared using these calculation equations. The theoretical and experimental values for the rectangular
pillar-type heat sinks varied considerably from each other
(Fig. 15). This might have been because the total generated heat
could not be transferred to the thermoelectric module; most of
the heat was lost and only some was converted into power. It was
also conrmed that as the temperature difference increased owing
to increases in engine load, the generated power increased, with
efciency increasing from around 7.9% to around 17.5%.
Therefore, it was revealed that the temperature difference between
the high and low sides was used to generate power in the

Fig. 23. Power characteristics by different Tc.

thermoelectric module, and that the magnitude of the temperature

difference had the greatest effect on the efciency of the
thermoelectric module. Fig. 16 indicates the total amounts of heat
transfer obtained by calculation and experimentation, thus comparing the theoretical and experimental values. In the theoretical
values, as the engine load was increased, the heat transfer increased
because the temperature difference was increased. On the other
hand, the experimental results showed that as the engine load
was increased, the increment of total heat transfer decreased.
This might be due to reduced performance as the temperature at
higher side approached the upper limit of the allowable

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

679

temperature range for the thermoelectric module. Also, the heat

ux was calculated as shown in Fig. 17; the theoretical heat ux
increased as the temperature difference was increased. On the
other hand, increment of experimental result showed a decreased
trend. The heating characteristics of the thermoelectric module
were analyzed through theory and experiment, considering various
factors that affect the heat exchanger by adopting heat transfer theory, and not only by analyzing the effects of the pressure difference.
3.3. Experimental results of thermoelectric generation
A thermoelectric generator was fabricated and its generation
characteristics were investigated. The rectangular pillar-shaped
heat sink was adopted for these experiments based on the results
of the previous experiments. Fig. 18 shows the power generation
performance under different experimental conditions. Since
thermoelectric modules from 1 to 10 had high performance, they
showed higher generation characteristics than modules from 11
to 20. Among all the thermoelectric modules, generation characteristics were improved as the position approached the exhaust port of
the engine. The generated current ranged from a maximum of
around 6.1 A to a minimum of around 2.2 A. The voltage ranged
from a maximum of around 2.4 V to a minimum of around 1.43 V.
Generated power ranged from a maximum of around 14.5 W to a
minimum of around 3.2 W. These results showed that the generation performance of the thermoelectric module increased as the
distance from the exhaust port of the engine decreased. In addition,
greater differences between Th and Tc corresponded to improved
generation performances of the thermoelectric modules.
However, the increase in the generation performance diminished
as Tc increased. This might have been caused by excessive exhaust
gas temperature at the hot side, which would have caused the performance of the thermoelectric module to deteriorate.
Fig. 19 shows the total power generated under different experimental conditions. The total power generation steadily increased
from Case 1 to Case 5, and then became almost steady-state from
Case 5 to Case 7. This might be because the excessive exhaust
gas temperature (which reached the durable temperature of the
thermoelectric module) negatively affected the modules performance. Total generation from Case 1 was around 80 W, and that
from Case 7 was around 260 W. These results imply that greater
temperature differences between the hot side and the cold side
improve the performance of the thermoelectric generator.
However, excessive temperature at the hot side reduces the efciency of the thermoelectric module, causing the power generation
performance to be rather poor. Therefore, the optimum high temperature at the hot side is an important factor that affects the
maximum power generation performance of the TEG.
Fig. 20 shows the current characteristic of the thermoelectric
generator versus Tc. As the exhaust gas temperature at the hot side
was increased, the generated current increased. Reversely, current
generation increased as Tc was decreased. That is to say, the performance of the thermoelectric module increased as the difference
between Th and Tc was increased. Because the optimized Tc of the
thermoelectric module is around 80 C, the generation performance was higher when Tc was 80 C, as shown in the gure.
However, if Th becomes too high at the hot side, the thermoelectric
module efciency decreases, which results in decreased power
generation. Therefore, it is important to maintain an adequate temperature at the hot side of the thermoelectric module.
Fig. 21 shows the voltage characteristic of the thermoelectric
generator versus Tc. Like the current characteristics, the generated
voltage increased as the exhaust gas temperature at the hot side
was increased and as Tc was decreased. However, the increment
of voltage decreased as Th was increased. This might be due to a
decrease in efciency as thermoelectric modules deteriorated.

Fig. 24. Comparison of total power under the different experiment condition.

Therefore, we conclude that appropriate temperature at the hot

side is an important factor that affects the performance of the
thermoelectric module.
Fig. 22 shows the power characteristics of the thermoelectric
generator versus Tc. As shown in the gure, the generated power
increased when Tc was decreased. This might be due to the temperature difference from the cold side with increases of Th. In addition,
the generated power increased for low-temperature conditions at
the cold side, which was attributed to a greater temperature difference between the two sides of the thermoelectric module.
Fig. 23 shows the total power generated under different experimental conditions. Total power values increased from Case 1 to
Case 2, but the increment from Case 5 to Case 7 was small. This
might be due to excessive exhaust gas temperature (which is close
to the durable temperature of the thermoelectric module), which
negatively affects the thermoelectric module performance. Also,
the total generated power increases as the temperature of the cold
side decreases. The increase of power generation capability was
attributed to increased difference between Th and Tc. Though the
generated power increased as the difference between Th and Tc
was increased, excessive Th decreased the generation performance
due to impaired durability of the thermoelectric module.
Therefore, appropriate Th is an important factor to achieve maximum power generation. Maintaining Tc is equally important
(Fig. 24).
4. Conclusions
This study measured the performance of thermoelectric modules in an actual engine system, using different heat sink shapes.
In addition, the performance of a prototype thermoelectric generator was measured under various experimental conditions. Our
results are as follows.
(1) Measurements of thermoelectric modules showed that current, voltage, and power characteristics were adequate only
for input resistances less than 3 O, and measurement values
below that were insufcient. Therefore, the adequate input
resistance of the tested thermoelectric modules was determined to be less than 3 O.
(2) The power generation was greatest when a rectangular pillar
heat sink was used, followed by a triangular prism heat sink
installed to face the direction of the exhaust ow; the least
power was generated when the heat sink used was a
triangular prism heat sink installed to face the reverse direction of the exhaust gas ow. The power generated from the

680

B.deok In et al. / International Journal of Heat and Mass Transfer 86 (2015) 667680

rectangular pillar heat sink was around 6.2 W under the

highest load condition, followed by around 5.5 W for the forward- and reverse-facing triangular prism heat sinks.
(3) Under each experimental condition, increasing the engine
load increased the differential pressures at both ends of
the thermoelectric module. The lowest differential pressure
was observed from the triangular prism heat sink installed
in the reverse direction of exhaust gas ow, followed by
the forward-facing triangular prism heat sink. Among the
experimental conditions, the greatest differential pressure
was recorded for the rectangular pillar heat sink.
(4) The performance of the thermoelectric modules increased as
the temperature of the hot side was increased, until the durable temperature of the thermoelectric module was
exceeded, at which point further increases impaired the
power generation performance. When the cold side temperature was increased, its temperature difference from
the hot side decreased, which ultimately reduced the power
generation performance of the thermoelectric modules.
The experimental results illustrated that the temperature difference between the two ends of the thermoelectric module and
the differential pressure of the exhaust gas are the key factors in
the modules power generation performance.
Conict of interest
Non declared.
Acknowledgements
This work was supported by Mid-career Researcher Program
through National Research Foundation of Korea (NRF) grant funded

by
the
Korea
government
(MSIP).
2014R1A2A2A01005055, NRF 2010-0011537)

(Nos.

NRF

References
[1] K. Kim, K. Choi, K. Lee, K. Lee, Active coolant control strategies in automotive
engines, IJAT 11 (6) (2010) 767772.
[2] J. Ringler, M. Seifert, V. Guyotot, W. Hubner, Rankine cycle for waste heat
recovery of IC engines, (2009), SAE Paper No. 200901-0174.
[3] D. Crane, G. Jackson, D. Holloway, Towards optimization of automotive waste
heat recovery using thermoelectrics, (2001), SAE Technical Paper 2001011021.
[4] R. Stobert, D. Milner, The potential for thermo-electric regeneration of energy
in vehicles, (2009), SAE Technical Paper 200901-1333.
[5] K. Kim, K. Choi, Y. Kim, K. Lee, K. Lee, Feasibility study on a novel cooling
technique using a phase change material in an automotive engine, Energy 35
(2010) 478484.
[6] H. Teng, G. Regner. C. Cowland, Waste heat recovery of heavy-duty diesel
engines by organic rankine cycle Part I: Working uids for WHR-ORC, (2007),
SAE Paper No. 200701-0543.
[7] R. Stobert, D. Milner, The Potential for Thermo-Electric Regeneration of Energy
in Vehicles, (2009), SAE Technical Paper 200901-1333, 2009.
[8] C. Wu, Analysis of waste-heat thermoelectric power generators, Appl. Thermal
Eng. 16 (1) (1996) 6369.
[9] D.M. Rowe, Thermoelectric Handbook Macro to Nano, Taylor and Francis
Group Press, Wales, 2007.
[10] J.C. Bass, N.B. Elsner, F.A. Leavitt, Performance of the 1 kW thermoelectric
generator for diesel engines, in: Proceedings of the 13th International
Conference on Thermoelectrics, (1994), pp. 295298.
[11] K.M. Saqr, M.K. Mansour, M.N. Musa, Thermal design of automobile exhaust
based thermoelectric generators: objectives and challenges, Int. J. Automot.
Technol. 9 (2) (2008) 155160.
[12] Chien-Chang Wang, Chen-I Hung, Wei-Hsin Chen, Design of heat sink for
improving the performance of thermoelectric generator using two-stage
optimization, Energy 39 (2012) 236245.
[13] F. Brito, J. Martins, L. Goncalves, R. Sousa, Temperature Controlled Exhaust
Heat Thermoelectric Generation, (2012), SAE 201201-1214.
[14] Sheng-Chung Tzeng, Tzer-Ming Jeng, Yi-Liang Lin, Parametric study of heattransfer design on the thermoelectric generator system, Int. Commu. Heat
Mass Transfer 52 (2014) 97105.