Applied Thermal Engineering 25 (2005) 3149–3162

www.elsevier.com/locate/apthermeng

Computational model for refrigerators based on

Peltier effect application
D. Astrain *, J.G. Vián, J. Albizua
Departamento de Ingenierı́a Mecánica, Energética y de Materiales, Universidad Pública de Navarra,
UPNa. Pamplona, Spain

Received 25 May 2004; accepted 8 April 2005

Available online 4 June 2005

Abstract

A computational model, which simulates thermal and electric performance of thermoelectric refrigera-
tors, has been developed. This model solves the non-linear system that is made up of the thermoelectric
equations and the heat conduction equations providing values for temperature, electric consumption, heat
flow and coefficient of performance of the refrigerator. Finite differences method is used in order to solve
the system and also semi empirical expressions for convection coefficients.
Subsequently a thermoelectric refrigerator with an inner volume of 55 · 10
3 m3 has been designed and
tested, whose cold system is composed of a Peltier pellet (50 W of maximum power) and a fan of 2 W. An
experimental analysis of its performance in different conditions has been carried out with this prototype,
which, in his turn, has been useful for assessing the accuracy of the developed model. The built thermoelec-
tric refrigerator prototype, offers advantages with respect to vapour compression classical technology such
as: a more ecological system, more silent and robust and more precise in the control of temperatures which
make it suitable for camping vehicles, buses, special transports for electro medicine, etc.

2005 Elsevier Ltd. All rights reserved.

Keywords: Thermoelectricity; Simulation; Refrigeration; Peltier

*
Corresponding author. Tel.: +34 948 169597; fax: +34 948 169099.
E-mail address: david.astrain@unavarra.es (D. Astrain).

1359-4311/$ - see front matter
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2005.04.003
3150 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162

Nomenclature

Ataco heat extender transversal area (m2)

Aais inter insulation area that separates both cold and hot sides (m2)
C calorific capacity (W/s K)
COP coefficient of performance
cp specific heat at constant pressure (J/kg K)
E voltage difference (V)
e insulation thickness (m)
EAB Seebeck electromotive force (V)
hint convective heat transfer coefficient for the inside (W/m2 K)
hext convective heat transfer coefficient for the outside (W/m2 K)
I electric current (A)
J current density (A)
k thermal conductivity (W/mK)
kAl aluminium thermal conductivity (W/mK)
kais insulator thermal conductivity (W/mK)
L length (m)
L length of heat extender (m)
Nu Nusselt number
Pr Prandtl number
_Q heat flow per unit time, heating power (W)
Q_ P Peltier heating power (W)
Q_ H Peltier pellet hot side heat flow (W)
Q_ C Peltier pellet cold side heat flow (W)
q_ r Thomson heat flow per unit volume (W/m3)
q_ J Joule heat generation per unit volume per unit time (W/m3)
q* heat generation per unit volume per unit time (W/m3)
R thermal resistance (K/W)
Rais thermal resistance of the refrigerator insulation (K/W)
Rtaco thermal resistance of the heat extender (K/W)
Rp thermal resistance of the Peltier pellet (K/W)
Rais inter thermal resistance of the insulator that separates both hot and cold dissipaters
(K/W)
Re electric resistance (X)
Re Reynolds number
S surface (m2)
T absolute temperature (K)
t temperature (C)
Tamb ambient temperature (K)
Tdh hot side dissipater temperature (K)
Thp Peltier hot side temperature (K)
 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162 3151

Tm Peltier pellet intermediate node temperature (K)

Tcp Peltier cold side temperature (K)
Th ext heat extender temperature (K)
Tdc cold side dissipater temperature (K)
Tint inner temperature (K)
U Heat transfer global coefficient (W/m2 K)
VPeltier Peltier supplied voltage (V)
V volume (m3)
W_ e electrical power consumed by Peltier (W)
$ gradient
a Seebeck coefficient (V/K)
q electric resistivity (Ohm m)
r Thomson coefficient (V/K)
s time (s)
d density (kg/m3)
pAB Peltier coefficient at the junction of materials A and B (V)
c specific weight (N/m3)

1. Introduction

Peltier technology has experienced a major advance in recent years, basically due to the devel-
opment of semiconductors and the incorporation of the thermoelectric devices into the civil mar-
ket. As a result of environmental problems produced by CFC gas, the development of equipment
based on this technology has been fostered. Nowadays thermoelectric refrigeration has a place in
medical applications, scientific equipment and in other devices where a high precision temperature
control is essential, in this sense it is worth to consider [1] and [2]. Nevertheless there are papers
about applications with a huge future potential, like the ones presented in [3] and [4] which
already concern companies. These include for instance: domestic or automobile industry air-con-
ditioned systems, portable refrigerators, domestic refrigerators, transport of perishable products,
vending machines, food display cases etc., where this technology has to compete against tradi-
tional vapour compression cooling systems.
However, whenever thermoelectric-based applications are going to be developed, the problem
representing the equipment calculation and design appears. A theoretical expression exists for the
COP of thermoelectric refrigerators [5], but this expression needs the Peltier pellet sidesÕ temper-
ature as boundary conditions, which are an unknown quantity. In addition, even assuming that
those temperatures are known, the offered results do not agree with the reality due to the hypoth-
eses considered to achieve an analytic solution.
In this work a computational model has been developed, which allows the simulation of those
complete thermoelectric devices, analysing the most important parameters for the optimisation,
the same as calculating and designing thermoelectric refrigerators. In the same way, an applica-
tion of this technology to a refrigerator of 55 · 10
3 m3 has been developed and experimentally
analysed.
3152 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162

2. Aims

Our principal aims are:

• Development of a computational calculation model that simulates the thermoelectric refriger-

ators thermal and electric performance.
• Experimental validation of the computational model by building a thermoelectric refrigerator
in order to this.

Secondary aims are:

• Analysis of the refrigerator prototype constructed and its application advantages and
disadvantages.

3. Computational model

A calculation model that simulates the thermal and electric performance of whole refrigerators
based on thermoelectric technology has been implemented. The model inputs are: refrigerator
material and geometry, Peltier pellet type (dimensions and materials), electrical voltage supplied
to the Peltier pellet and the hot and cold side dissipater thermal resistances. After the simulation
the model returned outputs are: temperatures, heat flows, Peltier electric consumption and COP
of the refrigerator.
The model solves the thermoelectric phenomena corresponding equations, which are Seebeck
effect, Peltier effect, Joule effect and Thompson effect [6] and the transitory regime heat conduction
equation, whose expressions are:

dEAB
aAB ¼ ¼ aA
aB ð1Þ
dT
Q_ P ¼ pAB I ¼ IT ðaB
aA Þ ð2Þ
q_ r ¼ rrT  J ð3Þ

J2
q_ J ¼ ð4Þ
q
ot
d  cp  ¼ rðkrtÞ þ q ð5Þ
os

The assumed hypotheses that have been taken in account in the model are:

• The materials are considered isotropic.

• The Thompson effect is negligible in comparison with both Peltier and Joule effects, as it is
shown in [6].
 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162 3153

A finite differences implicit method has been used for the numerical solution of Eq. (5), which
allows one to obtain node i temperature ti 0 at the instant s = n + 1 for the one-dimensional case, as
a function of the temperature of node i at the instant s = n and temperatures of the closest points
at the instant s = n + 1.
The discretization of the thermoelectric refrigerator is shown in Fig. 1, using the symbols of an
electrical analogy. The resistance Rtor has been introduced so that the model takes in account the
losses produced between dissipaters, the cold side and the hot side one, by the union bolts. In the
same way the resistance Rais inter considers the losses in the part of the insulator that separates both
dissipaters (the cold and the hot one) being a critical zone.
The temperatures corresponding to the nodes of the Fig. 1 are:

T1 = T9 = Tamb: Ambient temperature.

T2 = Tdh = Temperature of the hot side dissipater.
T3 = Thp = Temperature of the Peltier pellet hot side.
T4 = Tm = Temperature of the Peltier pellet intermediate node.
T5 = Tcp = Temperature of the Peltier pellet cold side.
T6 = Ttaco = Temperature of the aluminium heat extender.
T7 = Tdc = Temperature of the cold side dissipater.
T8 = Tint = Inner refrigerator temperature.

In the discretization of Fig. 1, the Peltier pellet is represented by three nodes: one for the hot
side (node 3), where the Peltier effect is produced and heat is liberated; another one for the cold
side (node 5), where the Peltier effect is produced and heat is absorbed and the third one an inter-
mediate node (node 4), where the heat generated by the Joule effect is introduced. In this way the
((2) and (4)) expressions change for each node into the following:

Fig. 1. Scheme of the computational model discretization.

3154 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162

Q_ 3 ¼ a  I  T 3 ð6Þ
Q_ 5 ¼
a  I  T 5 ð7Þ
Q_ 4 ¼ I 2  Re ð8Þ
The discretization allows that the heat generated by the Joule effect is distributed over each side
of the Peltier pellet, according to its temperatures.
On the other hand, the 3, 4 and 5 nodes are separated by two thermal resistances which
represent the Peltier pellet thermal conductivity. This value is also very important, since there will
be an important conduction effect in this Peltier pellet because of the different temperature of both
sides.
In the selected Peltier pellet there are 127 couples of semiconductors that are brought
together by a junction by means of copper flat plates, which in turn are brought together
with a ceramic material. As in the model the Peltier pellet has been discretized in two elements,
the half of the Peltier resistance is calculated, using this value for both nodes and allowing the
calculation of the total thermal resistance. Consequently, when the Peltier pellet thermal resis-
tance is calculated the half of the semiconductors length was taken and the equivalent resis-
tance for two parallel elements is calculated, REQBi2 Te3 . The copper plate resistance is added to
the equivalent one, RCu. With this entire sum the resistance of one of the 127 couples can be
calculated:

RPar ¼ RCu þ REQBi2 Te3 ð9Þ

The equivalent resistance for the 127 couples in parallel is afterwards calculated and then the cera-
mic material resistance is added, RAlumina, that is:

RPar
RP ¼ RAlumina þ ð10Þ
N

The thermal resistances of the Peltier pellet are illustrated in Fig. 2, where the thermal resistance
of each semiconductor (half-length) is indicated by Rsc.
The thermal resistance of the thermoelectric refrigerator insulator is calculated in the model
with the following expression:
1
Rais ¼ ð11Þ
U S
where S is the walls surface, U is the global coefficient of heat transfer between the inside and out-
side of the refrigerator, provided by the expression:
1
U¼ ð12Þ
1 e 1
þ þ
hint k hext
The expression used to calculate the inner and outer convection coefficients is the one experimen-
tally proved in [7] for a flat plate, without considering the viscous dissipation and in laminar
regime since the air velocities are small enough:
 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162 3155

Fig. 2. Scheme of the thermal resistances of the Peltier pellet.

1=2
NuL ¼ 0.664  Pr1=3  ReL
" #
0.6 6 Pr 6 50
ð13Þ
Re < Rex;c  5  105

The contact resistance has been obtained from the work of Ritzer and Lau (1994) [8], whose value
is 0.03 W/K. The rest of the model thermal resistances, are expressed by:
Aluminium heat extender resistance:
Ltaco
Rtaco ¼ ð14Þ
k Al  Ataco
Intermediate insulator resistance:
Ltaco
Rais inter ¼ ð15Þ
k ais  Aais inter
The heat capacities, which are necessary for the calculation in the transitory regime, are
described by the expression:
C ¼ V  d  cp ð16Þ

With the discretization shown in Fig. 1, the following equations system is set out:
ds _
½M  ½T 0i  ¼ ½T i  þ  ½Qi  ð17Þ
Ci
which can be written in matrix form as follows:
3156 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162
2 3
1 þ C1dsR12 ds
C 1 R12
0 0 0 0 0 0 0
6
 7
6 7
6 ds
C 2 .R12
ds 1
þ R123 þ 1 ds
0 0 0 ds ds
0 7
6 C2 R12 C 2 R23 C 2 R27 C 2 R28 7
6
 7
6 7
6 0 ds ds 1
þ R134 þ 1 ds
0 0 0 0 0 7
6 C 3 R23 C3 R23 C 3 R34 7
6
 7
6 7
6 0 0 ds ds 1
þ R145 þ 1 ds
0 0 0 0 7
6 C 4 R34 C4 R34 C 4 R45 7
6
 7
6 7
6 0 0 0 ds ds 1
þ R156 þ 1 ds
0 0 0 7
6 C 5 R45 C5 R45 C 5 R56 7
6
 7
6 7
6 0 0 0 0 ds ds 1
þ R167 þ1 ds
0 0 7
6 C 6 .R56 C6 R56 C 6 R67 7
6
 7
6 7
6 0 ds
0 0 0 ds ds 1
þ R178 þ 1 ds
0 7
6 C 7 .R72 C 7 R67 C7 R67 C 7 R78 7
6
 7
6 7
6 0 ds
0 0 0 0 ds ds 1
þ R189 þ 1 ds 7
6 C 8 R82 C 8 R78 C8 R78 C 8 R89 7
4 5
ds
0 0 0 0 0 0 0 C 9 R89
1 þ C9dsR89
2 3 2 3 2 3
T 01 T1 0
6 07 6 7 6 7
6T 2 7 6T 2 7 6 0 7 6
7
6 7 6 7 6
6 07 6 7 6 0 7 7
6T 3 7 6T 3 7 6
6 7 6 7 6 dsQ_ 3 7 7
6 07 6 7 6
6 T 4 7 6 T 4 7 6 C3 7
6 7 6 7 6 _ 7 7
6 7 6 7
 6 T 05 7 ¼ 6 T 5 7 þ 6 dsQ4 7
6 7 6 7 6 C4 7
6T 0 7 6T 7 6 7
6 67 6 67 6 _57
dsQ
6 7 6 7 6 6 C5 7
7
6T 0 7 6T 7 6
6 77 6 77 6 0 7 7
6 7 6 7 6 7
6T 0 7 6T 7 6
4 85 4 85 4 0 7 5
T 09 T9 0

Provided that this system is not linear, since the heat flows from the thermoelectric effects (expres-
sions (3)–(5)) depend on the temperature, it has been solved by means of an iterative method,
programmed in Matlab, calculating the heat flows at the instant s = n from the obtained tem-
peratures at the instant s = n
1. In this way, the matrix [M] and the vector ½Q_ i  are again
calculated for each iteration. Since both capacities C1 and C9 in the matrix [M] are the ambient ones,
they are considered an infinite value; so that fractions in the matrix depending on both values are
zero.

4. Experimental prototype of a thermoelectric refrigerator

A thermoelectric refrigerator with an inner volume of 55 · 10
3 m3 has been designed and built
in this work. It is composed of a Peltier pellet with 50 W of power and the dimensions:
40 · 40 · 4 mm, two fins dissipaters (one for the hot side and another one for the cold side) a
fan of 2 W and an aluminium heat extender which separates both sides, the hot one from the cold
one.
A thermal scheme of the thermoelectric refrigerator and a photograph of the prototype are
respectively shown in Figs. 3 and 4.
The Peltier pellet is supplied with a continuous current (Vmax = 12 V) so that heat is liberated by
one side and absorbed by the other one. The first law of thermodynamics for steady state
(expressed by powers) applied to a refrigerating machine, is the following:
Q_ H þ W_ e ¼ Q_ C ð18Þ
 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162 3157

Fig. 3. Thermal sketch of the thermoelectric refrigerator.

Fig. 4. Photograph of the thermoelectric refrigerator.

The prototype has been built in order to validate the computational calculation model which
has been developed. Nevertheless, this application of thermoelectricity is of interest by itself for
small hotel refrigerators and for camping vehicles, buses and other special vehicles, since it offers
several advantages with respect to vapour compression technology. For instance: lower level of
noise and vibrations, a greater useful life, it does not use refrigerants and provides a greater con-
trol of temperatures. Several tests has been carried out with this prototype in the climatic chamber
whose analysis is presented in Section 5.

5. Results and discussion

With the aim of verifying the precision of the calculation of the insulator resistance (expression
(11)–(13)), a series of tests have been carried out with the refrigerator prototype, to obtain this
3158 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162

value experimentally and to compare it with the model-calculated results. The obtained results
and the linear adjustment of the experimental points are shown in Fig. 5. The obtained experimen-
tal resistance is 2.09 K/W, which agrees with the calculated one: 2.08 K/W.
With the object of assessing the accuracy of the developed calculation model, an experimental
validation has been carried out with the built prototype, which has been introduced in a climatic
chamber to keep the ambient temperature and the humidity controlled. In the study the following
parameters have been changed, in both the prototype and the model:

• Electric power of the Peltier device, by means of the variation of the supplied voltage VPeltier.
• Thermal resistance of the cold side dissipater, changing the type of heat dissipater.
• Thermal resistance of the hot side dissipater, changing the air flow which gets across the fins
dissipater, by means of the variation of the fan voltage.
• Ambient temperature.

The comparison between the temperatures calculated by the computational model and the
experimental values obtained with the prototype are shown in Fig. 6 as a function of time for
a representative case of the 13 cases studied. Specifically, the studied configuration is: inner fins
without a fan and outer fins with a fan of 12 V, for an ambient temperature of 296 K and a rel-
ative humidity of 60%.
The results show agreement of the model with the experimental values, both in steady state and
in transition.
A summary of all the tests carried out for the model validation is shown in Tables 1 and 2,
where the temperatures obtained with the model and prototype are compared at steady state.
The ambient temperature was changed from 296 until 305.4 K. The Peltier supply voltage, the
outer fan voltage and the inner dissipater configuration have been altered in the same way, man-
aging to capture the influence of the dissipater thermal resistances and the Peltier pellet power.
The results that have been obtained changing the Peltier supply voltage from 6 V to 12 V are
presented in Table 1 for the following configuration: an outer fins dissipater with a 12 V fan
(3 W), and a inner fins dissipater without a fan. The tests is repeated for each Peltier voltage and
with three different ambient temperatures. The results show good prediction of the computational

50
45
40
Thermal drop (K)

35
30
25
20
15
10
5
0
0 5 10 15 20 25
Heat flow (W)

Fig. 5. Experimental results of the thermal drop as a function of the heat flow.
 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162 3159

40

35

30

25
Temperature (ºC)

20

15

10

-5

-10

-15
0 100 200 300 400
Time (min)

Model hot peltier temperature Model cold peltier temperature

Model inner temperature Prototype hot peltier temperature
Protoype cold peltier temperature Prototype inner temperature

Fig. 6. Comparison of the computational model temperatures with the experimental ones.

Table 1
Comparison between the model-prototype results for a configuration of outer fins and a fan with 12 V and inner fins
without a fan
VPeltier Temperature (K) Tamb = 296 K Tamb = 300.4 K Tamb = 305.4 K
Model Prototype Model Prototype Model Prototype
6V Tint 283.1 282.8 287.3 287.2 291.9 292.1
Thp 302.5 300.2 307.1 304.8 312.2 309.8
Tcp 272.8 273.1 276.7 277.4 281.1 282.2
(Tamb
Tint) 12.9 13.2 13.1 13.2 13.5 13.3
9V Tint 279.5 278.7 283.5 283.3 287.7 287.8
Thp 307 303.9 311.4 308.6 316.3 313.4
Tcp 267.2 266.6 270.6 271.1 274.4 275.4
(Tamb
Tint) 16.5 17.3 16.9 17.1 17.7 17.6
12 V Tint 278.2 277 282.1 281.2 286 285.7
Thp 312.4 308.6 317.1 313.3 321.7 317.8
Tcp 265.2 264 268.4 267.9 271.6 272
(Tamb
Tint) 17.8 19 18.3 19.2 19.4 19.7

model. In the calculation of thermal drop the maximum deviation from the experimental results is
1.2 K.
3160 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162

Table 2
Comparison between model-prototype results with a Peltier supplied voltage of 12 V an ambient temperature of
296.6 K and 60% of relative humidity
Configuration Testing characteristics Temperature (K) Model Prototype
1 Outer fins with a 12 V fan Tint 278.9 277.7
Without inner fins and without a fan Thp 311.2 308.4
Tcp 263.3 261.6
(Tamb
Tint) 17.8 19
2 Outer fins with a 4 V fan Tint 280.6 279.8
Without inner fins and without a fan Thp 316.2 312.1
Tcp 265.2 264
(Tamb
Tint) 16.1 16.9
3 Outer fins with a 12 V fan Tint 278.5 277.3
With inner fins and without a fan Thp 312.6 309.4
Tcp 265.3 264.2
(Tamb
Tint) 18.1 19.3
4 Outer fins with a 12 V fan Tint 273.7 272.7
Inner fins with a 3 V fan Thp 311.4 308.8
Tcp 266.6 266
(Tamb
Tint) 22.9 23.9

The thermal drop decreases when the Peltier voltage decreases too, as it was expected. On the
other hand and for the same Peltier voltage, thermal drop increases if the ambient tempera-
ture increases too. This effect is accurately predicted by the calculation model, because of the fact
that the Peltier electric resistance has been introduced in the model as a function of its
temperature.
Subsequently, the dissipaters thermal resistances have been changed for a fixed Peltier
supply voltage of 12 V. The fan voltage has been changed to vary the air flow, and inside the
following configurations have been studied: Flat plate (no fins) and without a fan; fins
dissipater without a fan; fins dissipater with a 3 V fan (0.5 W). The comparison between
the numerical and experimental results are shown in Table 2 for the studied configurations.
It can be shown that the accuracy of the model is high. The maximum error in thermal
drop is 1.2 K. In the same way, the present study makes it clear that dissipater thermal
resistance is very important for the performance of thermoelectric refrigerators, as it was said
in [9]. Specifically the thermal drop increases up to 40% when configuration 4 is used instead
of 2.
A comparative analysis of the experimental COP and the one obtained with the model has been
also carried out for all the cases listed in Tables 1 and 2. The results are shown in Fig. 7, where the
high accuracy of the calculation model can be observed, since all the values result in a margin of
relative error of ±7%.
The graph of Fig. 8 shows the performance of the prototype tested with the maximum power,
for different ambient temperatures. It can be observed that thermal drop stays higher than 21 C
with an average power consumption of 1 kW h/day.
 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162 3161

0.65

0.6

0.55

Calculation model COP 0.5

0.45

0.4

0.35

0.3

0.25

0.2

0.15
0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65
Experimental COP

Fig. 7. Comparison of the COP numeric data with the experimental ones.

55
50 Power Consumed
45
40 T hot Peltier
35
T hot Dissipater
30
Temperature (ºC)

25 T ambient

20
15 Thermal Drop

10 Inner
5
0
T cold Dissipater
-5 T cold Peltier

-10
-15
0 1000 2000 3000 4000
TIME (min)
T Cold dissipater Power consumed (W) T ambient
T cold peltier T hot peltier T hot dissipater
Thermal drop inner Temperature

Fig. 8. Experimental data of the prototype with a variable ambient T.

3162 D. Astrain et al. / Applied Thermal Engineering 25 (2005) 3149–3162

6. Conclusions

A computational calculation model has been developed to simulate thermoelectric refrigerators.

The accuracy of the model has been experimentally proved in different conditions, providing an
acceptable result: a maximum error in COP of ±7% and a maximum discrepancy in thermal drop
of 1.2 K.
The importance of heat dissipater thermal resistance on the performance of the thermoelectric
refrigerators has been demonstrated with both simulations and experimental data.
A thermoelectric refrigerator with an inner volume of 55 · 10
3 m3 has been designed and built.
It needs a continuous electric current (maximum 12 V) what makes it suitable for automobile
industry applications, since it offers the following advantages with respect to vapour compression:

• A more ecological system because it does not use refrigerants.

• More silent and robust since it minimizes the moving parts (it does not need a compressor).
• More precise in the control of temperatures, since it does not need to carry out start–stop
cycles, making it possible to vary the supply voltage in a progressive way.

The main disadvantage is linked to the electricity consumption. The electric energy consumed
by a thermoelectric refrigerator is higher in comparison with current compression refrigerators
(approximately the same as a vapour compression one with an inner volume of 100 · 10
3 m3).

References

[1] G. Min, D.M. Rowe, Cooling performance of integrated thermoelectric microcooler, Solid-State Electronics 43 (5)
(1999) 923–929.
[2] J.M. Gordon, K.C. Ng, H.T. Chua, A. Chakraborty, The electro-adsorption chiller: a miniaturized cooling cycle
with applications to micro-electronics, International Journal of Refrigeration 25 (8) (2002) 1025–1033.
[3] J.G Vián, D. Astrain, M. Domı́nguez, Numerical modelling and design of a thermoelectric dehumidifier, Applied
Thermal Engineering 22 (4) (2002) 407–422.
[4] J.G. Vián, D. Astrain, J.J. Aguas, Thermoelectric equipment to keep laboratory test-tube at a controlled
temperature, Journal of Thermoelectricity (3) (1999) 52–65.
[5] D.M. Rowe, CRC Handbook of Thermoelectrics, ISBN 0-8493-0146-7, 1995, pp. 19–25.
[6] D.M. Rowe, X. Bhandari, Modern Thermoelectrics, Holt, Rinehart and Winston, London, 1983, pp. 7–13.
[7] G.V. Parmelee, R.G. Huebscher, Heat transfer by forced convection along a smooth flat surface, Heat Piping Air
Conditioning 19 (8) (1947) 115.
[8] T.M. Ritzer, P.G. Lau, Economic Optimization of Heat Sink Design, 13th International Conference on
Thermoelectrics, Kansas City, MO, 1994, pp. 77–100.
[9] D. Astrain, J.G. Vián, M. Domı́nguez, Increase of COP in the thermoelectric refrigeration by the optimisation of
heat dissipation, Applied Thermal Engineering 23 (17) (2003) 2183–2200.