Академический Документы
Профессиональный Документы
Культура Документы
Research Paper
H I G H L I G H T S
A R T I C L E I N F O A B S T R A C T
Article history: Computational models emerge as essential tools in the challenge of developing competitive thermo-
Received 6 August 2015 electric refrigerators for the domestic sector.
Accepted 8 November 2015 This paper presents a computational model for thermoelectric refrigerators that simulates the entire
Available online 1 December 2015
system under transient state, including the thermoelectric modules, heat exchangers, insulated com-
partments, and hot and cold reservoirs. Also, temperature-dependent Peltier, Seebeck, Thomson and Joule
Keywords:
effects are implemented.
Thermoelectric cooling
A prototype of a thermoelectric refrigerator has been built and tested to conduct the verication and
Thermoelectric refrigerator
Dynamic simulation model validation of the computational model. The most important outputs are predicted with deviations lower
Finite differences than 10%.
Experimental validation The effect on the outputs of temperature-independent properties has been assessed. Results indi-
cate that deviations are up to twice as high as those obtained for temperature-dependent properties, so
these simplications are invalid in the simulation of thermoelectric refrigerators under real operation.
2015 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.applthermaleng.2015.11.021
1359-4311/ 2015 Elsevier Ltd. All rights reserved.
340 A. Martinez et al./Applied Thermal Engineering 95 (2016) 339347
A common thermoelectric refrigerator is composed of thermo-
electric modules, a heat exchanger at either side of the modules and
( )
Q Thomson = I Thot T cold ( p n )NM (3)
rates that determine the performance of the thermoelectric modules normal operation temperature is around 0.025 W/mK, similar to that
can be found in Reference 28. Finally, Eq. (10) provides the current of common insulation materials. Therefore, no signicant heat ow
intensity owing through the legs once an external voltage is applied. rate is expected through this path.
This expression is obtained by including in Eq. (6) the correspond-
ing nodes of Fig. 1.
2.4. Heat exchangers
Tp1 Tp2
( p1 n1 )T1 ( p10 n10 )T10 p1 The cold-side heat exchanger comprises all the elements that
2 transfer heat from the interior air to the thermoelectric modules,
T T T T T T whereas the hot-side heat exchanger is composed of those that trans-
V + MN + n 1 n1 n 2 p10 p9 p10 + n10 n9 n10
2 2 2 fer heat from the modules to the ambient. Fig. 1 displays a nned
9
Tp( j1) Tp( j+1) Tn( j1) Tn( j+1) heat sink at either side of the modules, although heat pipes,
p( j)
2
n( j)
2
thermosyphons or simple plane plates could be mounted. The char-
I= j=2 (10)
L p 9 p1 p10 9 s acterization of these systems is complex, since convection,
+ + p( j) + 2 p conduction, contact and constriction take place, along with the pres-
S 2 2 S
MN
p j=2 p
ence of phase-change uids. That is the reason why a single node
s
L 9 9
represents each heat exchanger in the electrical analogy, and the
+ n n1
+ n10 + n( j) + 2 n
S n 2 2 j=2 S n thermal resistances (Rhe,hot; Rhe,cold) are directly model inputs. An ex-
perimental procedure is performed to calculate them, presented in
The model considers 1-d heat conduction across the legs, thus sections 4.1.3 and 4.1.4.
neglecting convective heat transfer between ceramic plates. The In parallel, specic heat and mass of each heat exchanger are the
reason is that the gap between plates in a module is so short com- model inputs that determine the thermal capacities (Che,hot; Che,cold).
pared to length and width that air between them does not ow Finally, thermal resistance Rinter is introduced to simulate the
and virtually behaves like a solid [29]. Air thermal conductivity for heat transferred directly from the hot to the cold heat exchanger,
Table 3
Calculation of the thermal resistance of the nned heat sink.
V (V) I (A) T (C) R (K/W) BR (K/W) V (V) I (A) T (C) R (K/W) BR (K/W)
30.01 3.00 8.1 0.090 0.005 10.01 1.01 13.2 1.306 0.078
30.02 3.01 8.3 0.092 0.005 10.00 1.01 13.2 1.307 0.078
30.01 3.01 8.2 0.091 0.005 9.98 1.00 13.0 1.303 0.079
344 A. Martinez et al./Applied Thermal Engineering 95 (2016) 339347
See that outputs one and two are directly connected to steady power consumption of the thermoelectric modules with 95% of de-
state, whereas the last one relates to transient performance. The test viations lying on the global condence interval [7.31; 1.99].
is conducted for 8, 16 and 24 V of voltage applied to the modules Similarly, 95% of deviations for the temperature difference
and is replicated twice. It should be remembered that the modules between the inner air and the ambient provide a global con-
are connected electrically in series and each one is able to operate dence interval of [7.54; 4.15], whereas those for the time required
at a maximum voltage of 12 V. to reduce the temperature of the interior by 10 C lie on the global
The V&V uses the uncertainty analysis presented in the Appen- condence interval of [5.58; 9.95]. Therefore, a general value of
dix and establishes that the model is accepted provided that all the 10% can be established to indicate the accuracy of the computa-
simulated values fall within the corresponding 95%-condence in- tional model.
terval for the mean, formed by the three experimental values. Coecient of operation (COP) can be calculated with Eq. (13),
Equation (A1) is used to obtain these condence intervals. becoming 0.33 0.05, 0.19 0.02 and 0.09 0.01 for the experi-
mental tests with 8, 16 and 24 V respectively. The corresponding
5. Results simulated values are 0.34, 0.17 and 0.09. As expected all these are
extremely low compared to vapour compression systems, and de-
5.1. System validation crease with applied voltage [2,3,5].
Table 4 shows the experimental values of the three outputs, the (T amb T air )
R ins
corresponding averages and overall uncertainties, and the simu- COP = (13)
lated values provided by the computational model. Deviations W + W fan
between simulated and experimental values are provided by Eq. (12).
Again, Eq. (A1) is used to obtain the condence intervals for the
5.2. Thermoelectric module validation: inuence of
deviations.
temperature-dependent parameters
exp-sim
Deviation = 100 (12) Additionally, a specic test for V&V of the thermoelectric module
sim
has been conducted, since this is the most complex element in the
Firstly, the V&V is successfully achieved, since the nine con- thermoelectric refrigerator, and its performance might be masked
dence intervals include the corresponding simulated value. As in the V&V of the global system.
expected, increasing voltages lead to higher electric power con- All the nodes, resistances and capacities are removed from the
sumptions of the thermoelectric modules and higher air-to- model, except for those representing the thermoelectric module.
ambient temperature differences. Likewise, less time is required to Its hot side temperature is set to 50 C; then 4, 8 and 12 V are sup-
lower the interior temperature by 10 C. plied. Simulated values of current intensity (I), temperature difference
Secondly, regarding the electric power consumption of the ther- between ends (Tm), and heat ow rate extracted by the modules
moelectric modules, condence intervals for the deviations between (Q ext ) are compared to those provided by the manufacturer [30]
simulated and experimental values are 2.78 4.53, 0.00 1.99 and under two scenarios: Q ext = 0 , and T = 0.
0.74 1.59, for 8, 16 and 24 V, respectively, as the last columns in Table 5 provides the results, where deviations are calculated
Table 4 indicate. The lowest value is 7.31, which corresponds to the with Eq. (12). Since the manufacturer does not provide con-
lowest value obtained for 8 V, whereas the highest value is 1.99, dence intervals for these outputs, the highest deviation determines
which corresponds to the highest value obtained for 16 V. There- whether or not the model predicts with enough accuracy the per-
fore, the computational model can be said to simulate the electric formance of the thermoelectric module. Since all deviations fall
Table 4
Results of the verication and validation.
Mean O Mean O
Table 5
Results of the verication and validation for the thermoelectric module Marlow RC12-6L.
I (A) Tm (C) Q ext (W) I (A) Tm (C) Q ext (W) I (%) Tm (%) Q ext (%)
4V Q ext 1.00 30.0 - 0.99 29.2 - 1.0 2.7
Tm = 0 1.60 25.0 1.59 - 25.5 0.6 2.0
8V Q ext 2.40 52.0 2.26 50.6 - 6.2 2.8
Tm = 0 3.10 44.0 3.15 44.3 1.6 - 0.7
12 V Q ext 3.70 67.0 3.76 62.8 - 1.6 6.7 -
Tm = 0 4.50 55.0 4.65 56.0 3.2 - 1.8
A. Martinez et al./Applied Thermal Engineering 95 (2016) 339347 345
Table 6
Results of the verication and validation for the thermoelectric module Marlow RC12-6L with constant properties.
I (A) Tm (C) Q ext (W) I (A) Tm (C) Q ext (W) I (%) Tm (%) Q ext (%)
4V Q ext 1.00 30.0 1.1 29.3 9.1 2.4
Tm = 0 1.60 25.0 1.73 24.5 7.5 2.0
8V Q
ext 2.40 52.0 2.36 51.0 1.7 2.0
Tm = 0 3.10 44.0 3.46 42.2 10.4 4.3
12 V Q
ext 3.70 67.0 3.80 64.6 2.6 3.7
Tm = 0 4.50 55.0 5.19 52.9 13.3 4.0
This paper goes in line with this objective and presents a com-
putational model for thermoelectric refrigerators that outperforms
any other model in this eld. It simulates the thermoelectric modules,
heat exchangers, insulated compartments, and hot and cold
reservoirs, including the temperature-dependent Peltier, Seebeck,
Thomson and Joule effects. Furthermore, it simulates the entire
system under transient state, thus standing out as a valid design
and optimization tool for thermoelectric refrigerators.
A prototype of thermoelectric refrigerator has been built and
tested to obtain experimental data for verication and validation
of the computational model. The corresponding statistical analy-
sis indicates that the model predicts the most important parameters
with deviations lower than 10%. Specically, 95% of the devia-
tions between experimental and simulated values of the electric
power consumption of the thermoelectric modules come within the
interval [7.31%; 1.99%]; likewise, the interval [7.54%; 4.15%] in-
cludes 95% of the deviations between experimental and simulated
Fig. 4. Absolute value of deviations for temperature-dependent properties (cross) values of the temperature difference between the interior air
and constant properties (dot). and the ambient; additionally, the interval [5.58%; 9.95%] does so
for the time required to lower the temperature of the inner air by
10 C.
The model has been used to assess the effect on the outputs of
within 10% the highest being 6.7% the V&V of the model is suc- considering temperature-independent properties. It has been dem-
cessfully achieved. onstrated that deviations are even twice as high as those obtained
Furthermore, Table 6 presents a similar test but introduces the for temperature-dependent properties, so these simplications are
thermoelectric properties as temperature-independent. Seebeck co- invalid in the simulation of thermoelectric refrigerators under real
ecient, thermal conductivity and electrical resistivity are set operation.
respectively at 200 V/K, 1.54 W/mK and 0.01 mm, provided by
the corresponding equations in Table 1 at 300 K. Thomson effect dis- Acknowledgements
appears, as Eq. (5) indicates.
When absolute values of all deviations are plotted together, as The authors would like to thank the Spanish Ministry of Economy
Fig. 4 shows, one can see that deviations generally increase when and Competitiveness and FEDER Funds (DPI2014-53158-R) for sup-
constant properties are considered, being up to twice as high as porting this work.
those obtained for variable (temperature-dependent) properties.
Furthermore, this fact gains relevance as the supplied voltage in- Appendix
creases and large temperature gradients appear in the legs. In this
regard, it must be noted that the most used control system for the Every output experimental variable Y, depending on X1, X2, ,
interior temperature of a thermoelectric refrigerator is the on/off Xn uncorrelated input variables, is provided along with its overall
[5,17]. This controller introduces a pattern for the voltage sup- uncertainty (OY) to form the corresponding condence interval, as
plied to the modules, alternating periods at maximum voltage shown in Eq. (A1). The overall uncertainty is a combination of the
and periods at zero voltage. Therefore, temperature-independent systematic uncertainty (BY) and the random uncertainty (PY), the
properties could lead to incorrect predictions when simulating former being calculated with Eq. (A2), whereas the latter is pro-
the performance of real thermoelectric refrigerators under normal vided by Eq. (A3). In this equation, t represents the coverage factor
operation. corresponding to the t-distribution for N replications, whereas S Y
stands for the standard deviation for the average, calculated with
Eq. (A4).
6. Conclusions
Y O Y = Y B Y2 + PY2 (A1)
Up to now, the thermoelectric application with the most signif-
icant impact on the civil market is the thermoelectric refrigerator, 2 2 2
exhibiting high reliability and robustness for low-power demand. Y 2 Y 2 Y 2
B Y2 = B + B ++ B (A2)
The challenge now is the development of thermoelectric refrigera- X 1 X1 X 2 X2 X n Xn
tors able to compete with vapour-compression ones in the domestic
PY2 = (tS Y )
2
sector. (A3)
346 A. Martinez et al./Applied Thermal Engineering 95 (2016) 339347
[32] Sunon KDE1208PTS1-6. <http://datasheet.octopart.com/KDE1208PTS1-6 [38] D. Astrain, J. Vian, Study and optimization of the heat dissipater of a
-Sunon-Fans-datasheet-7265344.pdf>, n.d.. thermoelectric refrigerator, J. Enhanc. Heat Transf. 12 (2005) 159170,
[33] Ahlborn T190-0. <http://www.ahlborn.com/getle.php?1966.pdf>, n.d.. doi:10.1615/JEnhHeatTransf.v12.i2.20.
[34] Ahlborn Almemo 5690-1M09. <http://www.ahlborn.com/getle.php?1523.pdf>, [39] J. Vin, D. Astrain, Development of a hybrid refrigerator combining
n.d.. thermoelectric and vapor compression technologies, Appl. Therm. Eng. 29 (2009)
[35] R.G. Sargent, Verication and validation of simulation models, J. Simul. (2013) 33193327, doi:10.1016/j.applthermaleng.2009.05.006.
1224, doi:10.1109/WSC.2009.5429327. [40] A. Martnez, D. Astrain, A. Rodrguez, Experimental and analytical study on
[36] H. Coleman, W. Steele, Experimentation, Validation, and Uncertainty Analysis thermoelectric self-cooling of devices, Energy 36 (2011) 52505260,
for Engineers, 3rd ed., John Wiley & Sons, Hoboken, NJ, 2009. doi:10.1016/j.energy.2011.06.029.
[37] LairdTech Thermoelectric Handbook. <http://www.lairdtech.com/brandworld/
library/THR-BRO-Thermal%20Handbook%200110.pdf>, n.d..