A 2D Computational Model of an Active Magnetocaloric Regenerator

with Parallel Plates

Francisco Rodríguez-Méndez1, Bruno Chinè1, Simone Fabbrici2, Franca Albertini2, Daniela Negri3,
Francesco Cugini4, Massimo Solzi4

1. School of Materials Science and Engineering, Costa Rica Institute of Technology, Cartago, Costa Rica.
2. Institute of Materials for Electronics and Magnetism, National Research Council, Parma, Italy.
3. Independent Researcher, Fidenza, Italy.
4. Department of Mathematical, Physical and Computer Sciences, University of Parma, Italy.

Abstract

In this work, a 2-dimensional time-dependent model of a magnetic refrigeration system with an active magnetic
regenerator (AMR) is developed using COMSOL Multiphysics ®. The model consists of a regenerator with
parallel plates, hot and cold heat exchangers, and a working fluid. It uses experimental measurements of
∆𝑇𝑎𝑑 (𝐻, 𝑇) and 𝑐𝑝 (𝐻, 𝑇) as an input to account for the MCE, later providing characteristic response metrics of the
cooling system such as the temperature span, pressure difference, field fluid velocity and cooling power. With this
approach, three different magnetocaloric materials (MCM) were investigated – Ni49.6Mn34.2In16.1, and Ni50Mn35In15
Heusler compounds, and Gd – with direct and inverse MCE, where most of the important operation conditions of
the system remained fixed, for instance, fluid flow rate, working frequency, magnetic field strength, and number
and dimensions of the plates. The results indicate that Gd performs better than the other two materials in terms of
cooling capacity and temperature span, being the Ni49.6Mn34.2In16.1 alloy the one that generates the lowest
performance. Even though the Heusler compounds did not perform as well as the Gd, this model encourages
future studies where the working parameters of the magnetic refrigeration system can be varied to enhance and
optimize the performance of these MCMs and overall, the response of the thermo-magnetic device.

Keywords: magnetocaloric materials, magnetic refrigeration, computational modeling and simulation, non-
isothermal flow, heat transfer, fluid dynamics.

Introduction remains constant [9], therefore, the change in the

magnetic entropy is compensated by a change in the
Around 17% of the primary energy produced world- lattice entropy, which results in an adiabatic temper-
wide is consumed by refrigeration and air condition- ature change ∆𝑇𝑎𝑑 of the material [10] [11]. When
ing systems [1] [2], reaching a power intake of 2000 the material experiences an increase in its tempera-
TWh, approximately [3]. Furthermore, these thermal ture under the application of a magnetic field, the
management devices are based on vapor compres- MCE is called direct, but if the material cools down,
sion technologies, which still use refrigerants with a then it is defined as an inverse MCE [12].
high greenhouse effect potential, such as fluorocar- Nowadays, there are many research prototypes using
bons and carbon dioxides, contributing with the the principle of magnetic refrigeration to obtain a
7.8% emissions of greenhouse gases to the atmo- competitive thermal span and cooling power [13],
sphere [4] [5]. Therefore, as an energy-efficient and but still no commercial products are available. For
environmental-friendly alternative, magnetic refrig- this new technology to have a niche in the market,
eration is considered one of the best emergent tech- significant work must be done in the field of mate-
nologies to replace vapor compression plants [6]. rials science, specifically in the processing and man-
Compared with modern cooling systems, magnetic ufacturing of magnetocaloric materials with better
refrigeration uses a solid material as a refrigerant, properties [14] [15]. Among the most critical re-
which exhibits zero ozone depletion and global war- quirements, both the adiabatic temperature and en-
ming potential. On the other hand, it can also reach tropy changes of the material must be as large as
efficiencies up to 60% of the Carnot cycle, against possible in small magnetic field variations, as they
the 40% of the best options currently available on the are directly related to the temperature span of the
market [7], obtaining the potential ability to reduce device and the quantity of heat that can be transferred
energy consumption by up to 20%-30% [8]. in one cooling cycle [16] [17], respectively.
Magnetic refrigeration technology is based on the The benchmark material with the best properties is
magnetocaloric effect (MCE). This phenomenon gadolinium (Gd), which has been used in most of the
occurs when a material with magnetic properties is magnetocaloric devices [18] [19]. But since Gd is a
exposed to a magnetic field, resulting in a change in heavy rare-earth with low availability and high cost,
its magnetic order and a consequent decrease in the alternative rare-earth-free materials are the subject of
entropy of the magnetic contribution. Under isen- current research. Ni-Mn-based Heusler alloys are
tropic conditions, the total entropy of the material low cost, recyclable, highly available, and rare-earth

1
free, making them a promising material for magnetic
refrigeration applications. However, many of them
exhibit low adiabatic temperature change under cy-
cling due to a large thermal hysteresis [20] [21] [22].
In the other hand, the material adiabatic temperature
change alone is not sufficient to achieve the temper-
atures required for most refrigeration applications –
∆𝑇𝑎𝑑 of Gd at 1T magnetic field is only ~3K, for
example. To expand this operating range, the princi-
ple of an active magnetic regenerator (AMR) is used.
An AMR is a structure composed of a magneto-
caloric material through which a fluid is pumped in Figure 1. Phases of the Brayton refrigeration cycle.
the opposite direction to the heat flow. This regen-
This work aims to study the performance of three dif-
erator has a double function, as it plays the role of
ferent MCMs – Ni49.6Mn34.2In16.1, and Ni50Mn35In15
both a cooling element and a heat regenerator [23].
Heusler compounds, and Gd – in an AMR cycle by
Therefore, to generate a large temperature gradient,
building a robust computational model in COMSOL
the operation of the AMR must ensure that the fluid
Multiphysics®. This approach, as shown in Fig. 2,
keeps cycling from a cold end to a hot end of the de-
uses a complete set of parallel plates as regenerator,
vice, taking the heat load and transferring it to the
cold and hot heat exchangers (CHEX and HHEX,
environment [24]. A full cycle, which is also known
respectively), and a working fluid. The model has a
as the Brayton refrigeration cycle [25], consists in
fixed geometry and operates with constant values
four different steps, as also shown in Fig. 1:
for fluid velocity, cycle period, and magnetic field,
a) Magnetization of the AMR as it is exposed to the
but changes the system initial temperature to match
magnetic field, resulting in the increment of the
the temperature of each MCM studied at which the
temperature of each piece of the MCM.
MCE is maximized (Curie temperature). The ∆𝑇𝑎𝑑
b) Fluid flow from the cold end through the heated
and heat capacities of the materials are measured
MCM to the hot end, where the heat received
experimentally rather than mean field approxima-
from the MCM is transmitted to the environment.
tions, and the fluid motion considers viscous losses.
c) Demagnetization of the AMR as the magnetic
Main metrics used to evaluate the magnetic device
field is removed, resulting in the decrease of the
performance are the maximum temperature span ob-
temperature of each piece of the MCM.
tained, the cooling capacity at no-load condition, and
d) Fluid flow from the hot end through the cooled
the coefficient of performance (COP).
MCM (where the fluid is cooled) to the cold end,
where the heat is absorbed from the surroundings
(the refrigerator chamber). AMR Modeling
Several computational studies have been carried out
to assess the performance of this AMR cycle using The model is based on the COMSOL Multiphysics ®
various magnetocaloric materials. Tusek et al. [26] Conjugate Heat Transfer module that combines the
used Gd in a packed-bed regenerator and found that Heat Transfer in Solids and Fluids and the Laminar
smaller spheres yielded larger cooling capacity due Flow physics under the Non-Isothermal Flow
to better heat transfer properties. Tomc [27] reported Multiphysics interface. The dependent variables,
the first method for optimizing the minimum number such as fluid pressure and velocity, and temperature
of layers necessary to increase the performance of an distribution across the AMR, are solved using the
AMR with parallel plates. Trevizoli et al. [28] [29] Finite Element Method. Therefore, the velocity and
developed a computational model that also uses an pressure profiles are first determined along the fluid
AMR with a packed-bed of spheres, concluding that and then they are used to solve the temperature dis-
in order to increase the precision of the simulations tribution in both the solid and fluid domains during
the irreversibility of the internal heat transfer of the the cold and hot fluid flows, respectively. This pro-
system, viscous losses of the working fluid, losses cess is repeated for several cooling cycles for the
due to the demagnetization phase, and heat losses to system to reach a steady-state condition.
the environment must be taken into account. Zhang
Governing equations of the model
et al. [30] used modeling and simulation techniques
to optimize the length of the AMR layers manufac- Assuming the fluid flow is laminar, the velocity dis-
tured with different LaFeMnSiH compounds. They tribution in the fluid is determined by solving the
found that the best performance of the regenerator Navier-Stokes momentum and continuity equations
occurred at a frequency of 0.1Hz with a layer length 𝜕𝑈
of 200mm, obtaining a simulated span temperature 𝜌𝑓 ( + (𝑈 · ∇)𝑈) − 𝜇𝑓 ∇2 𝑈 + ∇𝑝 = 0 (1)
𝜕𝑡
of 41K. Aprea et al. [2] [31] assessed six different
∇·𝑈 =0 (2)
AMRs, demonstrating that parallel plates regenera-
tors have greater efficiency, but only if the plates are where 𝜌𝑓 is the density of the fluid, 𝜇𝑓 is the dynamic
perfectly parallel, being the packed-bed of spheres viscosity of the fluid, 𝑈 is the fluid velocity field, 𝑝
the most viable option for real applications. is the pressure, and 𝑡 is time.

2
 Ni₄₉.₆Mn₃₄.₂In₁₆.₁ (heating)
5
Ni₄₉.₆Mn₃₄.₂In₁₆.₁ (cooling)
4 Ni₅₀Mn₃₅In₁₅
Gd
3

∆T (K)
1

-1

Figure 2. AMR model and boundary conditions used. -2

280 285 290 295 300 305 310 315 320 325 330
Then, the temperature distribution in the solid regen-
T (K)
erator and the heat exchangers domains is calculated
by solving the equation Figure 3. ∆𝑇𝑎𝑑 for each magnetocaloric material used
𝜕𝑇𝑠 under a magnetic field change of 1.8T.
𝜌𝑠 𝑐𝑝,𝑠 − 𝑘𝑠 ∇2 𝑇𝑠 = 𝑄̇𝑀𝐶𝐸 + 𝑄̇𝐻𝑇 (3)
𝜕𝑡 Ni₄₉.₆Mn₃₄.₂In₁₆.₁ (0T)
4450
where 𝑐𝑝,𝑠 is the specific heat capacity of the solid, Ni₄₉.₆Mn₃₄.₂In₁₆.₁ (1.8T)
𝑇𝑠 is the solid temperature, and 𝑘𝑠 is the solid thermal 3700 Ni₅₀Mn₃₅In₁₅ (0T)
Ni₅₀Mn₃₅In₁₅ (1.8T)
conductivity. Here, the MCE is introduced in the 2950 Gd (0T)
term 𝑄̇𝑀𝐶𝐸 as a heat source in the energy equation Gd (1.8T)
2200
and defined as a function of the ∆𝑇𝑎𝑑 (𝐻, 𝑇) with
respect to time, and the 𝑐𝑝 (𝐻, 𝑇) of the MCM. Is it 1450
550
important to note that both ∆𝑇𝑎𝑑 (𝐻, 𝑇) and 𝑐𝑝 (𝐻, 𝑇)
cp (J/kg.K)

are dependent on the applied magnetic field and the 450

instant temperature of the regenerator. The relation 350

that describes this effect is calculated by 250
∆𝑇𝑎𝑑 (𝐻, 𝑇) 150
𝜌𝑠 𝑐𝑝 (𝐻, 𝑇) = 𝑄̇𝑀𝐶𝐸 (4) 280 285 290 295 300 305 310 315 320 325
𝑑𝑡
T (K)
∆𝑇𝑎𝑑 (𝐻, 𝑇) and 𝑐𝑝 (𝐻, 𝑇) parameters are introduced
in the model by using interpolated functions based Figure 4. Specific heat values, for each magnetocaloric
on experimental data for each of MCM studied, as material used under a magnetic field of 0T and 1.8T.
illustrated in Figs. 3 and 4, where the Ni50Mn35In15
∆𝑇𝑎𝑑 data was measured in [32], and its 𝑐𝑝 values are Boundary conditions and design parameters of
based on measurements found in [33]. the AMR model
Besides, it is worth noting that unlike Ni50Mn35In15 The regenerator in the model is made of the MCM,
and Gd, which manifest conventional MCE at their the cold and hot heat exchangers are built of copper,
Curie transition temperature, the Ni49.6Mn34.2In16.1 and the heat transfer fluid is pure water. As shown in
Heusler alloy exhibit a thermal hysteresis caused by Fig. 2, external boundaries have adiabatic conditions
the first-order ferromagnetic-to-paramagnetic transi- imposed on them, except for the outward boundaries
tion that takes place at its martensitic phase transfor- of CHEX and HHEX, which have a prescribed heat
mation. This characteristic not only affects the flux. The solid domains and the fluid are assumed to
material ∆𝑇𝑎𝑑 cycling behavior, as it is shown in the be in perfect contact following the boundary condi-
Fig. 3, where one curve describes the MCM heating tion given in the relation
phase and a different one the cooling process, but 𝜕𝑇𝑓 𝜕𝑇𝑠
also displaying an inverse MCE near 300K. (𝑘𝑓 )| = (𝑘𝑠 )| (6)
𝜕𝑦 𝑦=𝐻𝑓𝑙 𝜕𝑦 𝑦=𝐻𝑓𝑙
Subsequently, the heat transfer for incompressible
fluids with convective terms, with the 𝑄̇𝐻𝑇 term Also, in each fluid-solid interface, the “no-slip wall”
linking the solid and fluid domains, is calculated by condition has been imposed.
the next equation The geometrical configuration implemented in the
model is summarized in Table 1, where a total of 26
𝜕𝑇𝑓 parallel plates were built into the regenerator. For the
𝜌𝑓 𝑐𝑝,𝑓 ( + (𝑈 · ∇)𝑇𝑓 ) − 𝑘𝑓 ∇2 𝑇𝑓 = −𝑄̇𝐻𝑇 (5)
𝜕𝑡 Brayton refrigeration cycle described in the intro-
duction, the time steps for the four stages are sym-
where 𝑐𝑝,𝑓 is the specific heat capacity of the fluid, metrically distributed. This parameter along others
𝑇𝑓 is the fluid temperature, and 𝑘𝑓 is the fluid thermal of equally importance, such as the strength of the
conductivity. The velocity field calculated in Eqs. magnetic field, the cycle period, the fluid velocity in
(1) and (2) are used in equation (5) to determine the the fluid flows phases, the initial prime temperature
convective heat transfer. for each MCM, and the heat transfer coefficients for

3
Table 1. Geometrical configuration of the model. Table 2. Initial and working conditions of the model.
Part Dimensions (mm2) Parameter Value
Regenerator plates 80 x 0.25 Magnetic field strength 1.8T
Cold heat exchanger 50 x 10 Total cycle period 2s
Hot heat exchanger 50 x 5 Fluid flows step time 1s
Fluid channel 230 x 14 (de)magnetization step time 1s
Heat exchanger-regenerator gap 10 Fluid velocity 0.024 m/s
Gap between plates 0.01 Heat flux CHEX (hC) 0 W/(m2.K)
Heat flux HHEX (hH) 10E3 W/(m2.K)
the outward boundaries of the heat exchangers are Ni49.6Mn34.2In16.1 initial temperature 298K
provided in Table 2. For the latter parameter, perfect Ni50Mn35In15 initial temperature 316K
thermal contact between the hot heat exchanger and Gd initial temperature 292K
the outside environment is assumed by making the
value of the heat transfer coefficient in the outward where 𝜂𝑝𝑢𝑚𝑝 is the pump efficiency, assumed to be
boundary of the hot end very large. In the other hand, around 0.8 for most devices.
to assess the performance of the regenerator at max- Finally, the COP of the AMR is calculated by
imum cooling capacity, the heat transfer coefficient 𝑞𝑐′
at the CHEX is set to 0 W/(m2.K), i.e. no cooling- 𝐶𝑂𝑃 = − ′ (13)
𝑤𝑡𝑜𝑡
load is considered in this simulation.

Performance evaluation metrics Simulation Results and Discussion

The assessment of the performance of the system is MCMs assessment
determined by the temperature span achieved for the
device, and the cooling capacity. ∆𝑇𝑠𝑝𝑎𝑛 is the temp- Fig. 5 illustrates the development of the ∆𝑇 over time
erature difference between the hot and cold heat ex- for (a) Ni49.6Mn34.2In16.1, (b) Ni50Mn35In15, and (c) Gd
changers. Then, for the cooling capacity, the amount materials, respectively. For each simulation, 600
of cooling per cycle absorbed is determined by the cycles were performed, which is equivalent to 1200s
integration of the heat flux through the outward with a time step of 0.1s. As expected, the Gd is
boundary of the CHEX as follows capable of build a larger AMR ∆𝑇 than the Heusler
compounds studied, due to its bigger ∆𝑇𝑎𝑑 . It is also
𝜏 𝐿𝐻𝐸𝑋
𝑞𝑐′ = ∫ ∫ 𝑞𝑐′′̇ 𝑑𝑥𝑑𝑡 (7) noticeable that, for the total cycles simulated, the Gd
0 0 and Ni50Mn35In15 do not arrive to a steady-state con-
where 𝜏 is the cycle period and 𝐿𝐻𝐸𝑋 is the length of dition, where the ∆𝑇 between two consecutive cy-
the heat exchanger. The corresponding refrigeration cles is equal or less than 0.01K, meaning we need to
capacity per cycle is now calculated by the relation increase the study number of cycles until the above
condition is fulfilled. Meanwhile, Ni49.6Mn34.2In16.1
𝑞𝑐′ reached a steady-state in about 200s.
𝑞𝑐′̇ = (8)
𝜏 Here, the behavior of Ni49.6Mn34.2In16.1 alloy is worth
Derived from the cooling capacity, we can also cal- a closer look. As it displays an inverse MCE under
culate the coefficient of performance (COP) of the 300K along with a thermal hysteresis, the hot part of
system. First, the rejected heat per cycle is deter- the system is uncapable of building a positive temp-
mined by the integration of the heat flux through the erature difference with the surroundings. This means
outer boundary of the HHEX as follows that the device is no longer a cooling system, but a
𝜏 𝐿𝐻𝐸𝑋 heat pumping one, since the temperature difference
𝑞𝑟′ = ∫ ∫ 𝑞𝑟′′̇ 𝑑𝑥𝑑𝑡 (9) between the hot side of the AMR and the surround-
0 0 dings now allows heat to enter the regenerator. To
Then, the total work done by the AMR is the sum of avoid this situation, better understanding of the in-
the work required to magnetize the regenerator and verse MCE and the thermal hysteresis of the material
the work required to displace the fluid is needed, together with a study of another operation
′ ′ ′ conditions, for example, dual magnet configuration
𝑤𝑡𝑜𝑡 = 𝑤𝑚𝑎𝑔 + 𝑤𝑝𝑢𝑚𝑝 (10)
with half the regenerator magnetized at half a cycle
The work required to magnetize and demagnetize the with the other half demagnetized, and different
regenerator can be calculated with the equation initial conditions, such as the working frequency, or
′
𝑤𝑚𝑎𝑔 = 𝑞𝑟′ + 𝑞𝑐′ (11) the fluid volume displaced in one cycle.
Fig. 6 shows the 2D temperature evolution across the
The pressure drop in the regenerator is then needed entire magnetic device during the (a) cold and (b) hot
to calculate the pumping power required to move the blow steps of the Brayton cycle described. For con-
working fluid through the AMR. Generally, the rela- venience, the images were taken using the Gd as the
tionship between these two variables is described by model material. In the cold blow step of the cooling
𝑈𝑓 ∇𝑝 cycle, the heat generated in the AMR due to the MCE
𝑤̇ 𝑝𝑢𝑚𝑝 = (12) is transferred into the working fluid and moved
𝜂𝑝𝑢𝑚𝑝
towards the HHEX, where it will heat up the hot side

4
 Figure 6. Temperature gradient across the AMR during
(a) hot blow and (b) cold blow phases.
key parameter that directly determines the efficacy
of the AMR, further study on finding the optimal
value of this variable is needed, since if the fluid ve-
locity is too slow, the cooling effect of the MCM is
Figure 5. ∆𝑇 behavior over time at the hot and cold ends not fully utilized, but if it is too high, the CHEX will
of the AMR for (a) Ni49.6Mn34.2In16.1, (b) Ni50Mn35In15, absorb heat coming from the HHEX as well as the
and (c) Gd materials.
outside environment. In the simulations performed,
of the system. The other way around, during the hot with a fluid velocity inside the channels near 0.08m/s
blow phase, after the regenerator is demagnetized, the amount of fluid displaced through the channels
the temperature of the fluid is reduced by the colder in each cycle is 30% of the fluid inside the AMR.
AMR. Then, the cooled fluid is transferred towards Obtained ∆𝑇𝑠𝑝𝑎𝑛 , cooling capacity, and COP values
the CHEX, reducing the temperature of the cold side are summarized in Table 3. The first thing to note is
of the system. the COP negative value of Ni49.6Mn34.2In16.1, which
is explained with the fact exposed before, the AMR
Performance Evaluation working with this martensitic Heusler compound
Three metrics are used to evaluate the MCMs studied functions as a heat pumping device instead of a
in the simulations: first, the ∆𝑇𝑠𝑝𝑎𝑛 between the cold cooling system. Besides this behavior, its cooling
and hot heat exchangers of the system; and second capacity and ∆𝑇𝑠𝑝𝑎𝑛 are the lowest if compared with
and third, the cooling power and the coefficient of the values achieved by the Ni50Mn35In15 and the Gd.
performance COP at no-load. Now, the Gd performs better in terms of ∆𝑇𝑠𝑝𝑎𝑛 and
For the latter two parameters, as stated in Eq. 12, the cooling power calculated, but the Ni50Mn35In15 alloy
pressure losses along the AMR and the fluid velocity does not fall far from it, in fact, its numbers follow
are needed. Fig. 7 show the (a) change in pressure
across the regenerator, (b) the velocity profile during Table 3. Performance parameters for all MCMs.
the hot and cold fluid blows, and (c) the 2D velocity Parameter Value
distribution in the AMR. Pressure losses remain low Ni49.6Mn34.2In16.1
– ~3360Pa – due to the geometry of the AMR and Temperature span 4.5K
the perfect alignment between the parallel plates. In Cooling capacity 151.05W
a real regenerator, ensuring that each plate is perfect- Coefficient of performance -1.05
ly aligned is a difficult task to accomplish, leading to Ni50Mn35In15
major pressure losses, vortex generation in the fluid, Temperature span 14.89K
and temperature maldistribution. Cooling capacity 847.81W
Velocity profiles display a huge change in its value Coefficient of performance 1.80
when the fluid enters the regenerator channels. This Gadolinium
behavior obeys Bernoulli’s principle as the area is Temperature span 17.25K
drastically reduced coupled with the change in the Cooling capacity 974.78W
pressure. As the fluid velocity along the channel is a Coefficient of performance 0.79

5
 Conclusions

This work presents a two-dimensional model of an

active magnetocaloric regenerator built with parallel
plates made of magnetocaloric materials. The model
approach utilizes three different MCMs as refriger-
ants and calculates its performance response in terms
of ∆𝑇𝑠𝑝𝑎𝑛 , cooling capacity, and COP achieved at no
load. The simulations show that the martensitic com-
pound Ni49.6Mn34.2In16.1 has an undesirable behavior
as it acts as a pumping heat device due to its intrinsic
inverse MCE and thermal hysteresis, also demon-
strating the lowest ∆𝑇𝑠𝑝𝑎𝑛 and cooling power if com-
pared with the other materials. Proper modifications
to the operative conditions must be considered to
exploit this material in a magnetocaloric machine.
Gd is able to generate the largest ∆𝑇𝑠𝑝𝑎𝑛 and cooling
power but followed closely by Ni50Mn35In15, with a
∆𝑇𝑠𝑝𝑎𝑛 of about 15K and a cooling capacity of ap-
proximately 847.81W. Also, Ni50Mn35In15 has the
highest efficiency (COP), which makes it a viable
option with great cost effectiveness, but for applica-
tions working near 316K. However, the results ob-
tained encourage future analysis where different
parameters of the model can be varied with the pur-
pose of finding the optimal working conditions for
every magnetocaloric material studied.

Acknowledgements
Figure 7. AMR (a) pressure losses, (b) fluid velocity in
channels during hot and cold blow phases, and (c) 2D The first two authors gratefully acknowledge the
velocity profile during the fluid flows. financial aid provided by the Vicerrectoría de Inves-
closely the overall capabilities of Gd, with a ∆𝑇𝑠𝑝𝑎𝑛 tigación y Extensión of the Instituto Tecnológico de
and cooling power difference between them of about Costa Rica, through the project 5402-1351-2301.
2.3K and 125W, respectively. For COP metric, the
Ni50Mn35In15 Heusler compound has a higher value References
than the Gd. This means that Ni50Mn35In15 displays a
higher ratio of useful cooling provided to work re- [1] D. Coulomb, J. L. Dupont and A. Pichard, "The
quired, which can be translated to the factor that this Role of Refrigeration in the Global Economy,"
magnetocaloric material has lower energy consump- 29th Informatory Note on Refrigeration
Technologies, Technical Report, International
tion and the highest efficiency of all three MCMs an-
Institute of Refrigeration, Paris, France, 2015.
alyzed here, but still falling behind if compared with
most air conditioners (which have COP values of 2.3 [2] C. Aprea, A. Greco, A. Maiorino and C. Masselli,
"Analyzing the energetic performances of AMR
to 3.5, approximately), for example.
regenerator working with different magnetocaloric
Nevertheless, the performance responses achieved materials: Investigations and viewpoints,"
by the Ni50Mn35In15 have the potential to make this International Journal of Heat and Technology, vol.
material a preferable alternative over Gd considering 35, no. 1, pp. S383-S390, 2017.
the abundancy and cost of its constituent elements. [3] International Energy Agency, "The Future of
One drawback we must consider with this Heusler Cooling," 2018. [Online]. Available:
compound is that the metrics obtained were calcu- http://www.iea.org/reports/the-future-of-cooling.
lated at its prime temperature (i.e., its Curie transi- [Accessed July 2023].
tion temperature), which is around 316K, meanwhile [4] D. Coulomb, J.-L. Dupont and V. Morlet, "The
Gd works near room temperature ~292K. Therefore, Impact of the Refrigeration Sector on Climate
the applications where this material will find its Change," 35th Informatory Note on Refrigeration
usefulness fall into industrial processes and perhaps Technologies, Technical Report, International
electrical/electronic devices. Another possibility is Institute of Refrigeration, Paris, France, 2016.
to carry out a proper tuning of the elemental compo- [5] Y. Hwang, "Harmonization of the Life Cycle
sition of this Heusler compound, with the purpose of Climate Performance Methodology," 32nd
shifting its Curie transition temperature closer to Informatory Note on Refrigeration Technologies,
room temperature, while preserving its magneto- International Institute of Refrigeration, Paris,
France, 2016.
caloric behavior and performance response.

6
[6] A. Kitanovski, U. Plaznik, U. Tomc and A. cycle that exploits thermal hysteresis," Nat Mater,
Poredos, "Present and future caloric refrigeration vol. 17, no. 10, pp. 929-934, 2018.
and heatpump technologies," International Journal [21] F. Scheibel, T. Gottschall, A. Taubel, M. Fries, K.
of Refrigeration, vol. 57, pp. 288-298, 2015. P. Skokov, A. Terwey and W. Keune, "Hysteresis
[7] E. Brück, O. Tegus, O. Tegus, D. Thanh and K. Design of Magnetocaloric Materials—From Basic
Buschow, "Magnetocaloric refrigeration near room Mechanisms to Applications," Energy Technology,
temperature," Journal of Magnetism and Magnetic vol. 6, no. 8, pp. 1397-1428, 2018.
Materials, vol. 310, no. 2, pp. 2793-2799, 2007. [22] L. F. Cohen, "Contributions to Hysteresis in
[8] D. Velázquez, C. Estepa, E. Palacios and R. Magnetocaloric Materials," physica status solidi
Burriel, "A comprehensive study of a versatile (b), vol. 255, no. 2, 2017.
magnetic refrigeration demostrator," International [23] J. A. Barclay and W. A. Steyert, "Active magnetic
Journal or Refrigeration, vol. 63, pp. 14-24, 2016. regenerator". ENERGY United States, Department
[9] S. Choi, U. Han, H. Cho and H. Lee, "Review: of US Department of Energy Patent US4332135A,
Recent advances in household refrigerator cycle 27 01 1981.
technologies," Applied Thermal Engineering, no. [24] H. Ucar, D. Paudyal and O. Boyraz, "Using
132, pp. 560-574, 2018. numerical methods to screen magnetocaloric
[10] V. K. Pecharsky and K. A. G. Jr, "Magnetocaloric materials in an active magnetic regenerative cycle,"
effect and magnetic refrigeration," Journal of International Journal of Refrigeration, vol. 120,
Magnetism and Magnetic Materials, vol. 1, no. 3, pp. 50-57, 2020.
pp. 44-56, 1999. [25] J. Tusek, A. Kitanovski, S. Zupan, I. Prebil and A.
[11] A. Tishin and Y. Spichkin, The Magnetocaloric Poredos, "A comprehensive experimental analysis
Effect and its Applications, Boca Raton: CRC of gadolinium active magnetic regenerators,"
Press, 2003. Applied Thermal Engineering, vol. 53, pp. 57-66,
[12] T. Gottschall, "On the magnetocaloric properties of 2013.
Heusler compounds: Reversible, time- and size- [26] J. Tušek, A. Kitanovski, I. Prebil and A. Poredoš,
dependent effects of the martensitic phase "Dynamic operation of an active magnetic
transition," TUprints - TU Darmstadt publication regenerator (AMR): Numerical optimization of a
service, Darmstadt, 2016. packed-bed AMR," International Journal of
[13] A. Greco, C. Aprea, A. Maiorino and C. Masselli, Refrigeration, vol. 34, no. 6, pp. 1507-1517, 2011.
"A review of the state of the art of solid-state [27] U. Tomc, "Improvements to the heat transfer of an
caloric cooling processes at room-temperature active magnetic regenerator," University of
before 2019," International Journal of Ljubljana, Ljubljana, 2016.
Refrigeration, vol. 106, p. 66–88, 2019. [28] P. V. Trevizoli, A. T. Nakashima, G. F. Peixer and
[14] T. Gottschall, K. P. Skokov, M. Fries, A. Taubel, I. J. R. B. Jr., "Performance evaluation of an active
Radulov, F. Scheibel, D. Benke, S. Riegg and O. magnetic regenerator for cooling applications –
Gutfleisch, "Making a Cool Choice: The Materials part I: Experimental analysis and thermodynamic
Library of Magnetic Refrigeration," Advanced performance," International Journal of
Energy Materials, vol. 9, no. 34, 2019. Refrigeratio, vol. 72, pp. 192-205, 2016.
[15] A. Kitanovski, "Energy Applications of [29] P. V. Trevizoli, A. T. Nakashima, "Performance
Magnetocaloric Materials," Advanced Energy evaluation of an active magnetic regenerator for
Materials, vol. 10, no. 10, 2020. cooling applications – part II: Mathematical
[16] J. Romero Gómez, R. Ferreiro Garcia, A. De modeling and thermal losses," International Journal
Miguel Catoira and M. Romero Gómez, of Refrigeration, vol. 72, pp. 206-217, 2016.
"Magnetocaloric effect: A review of the [30] M. Zhang, O. Abdelaziz, A. M. Momen and A.
thermodynamic cycles in magnetic refrigeration," Abu-Heiba, "A numerical analysis of a
Renewable and Sustainable Energy Reviews, magnetocaloric refrigerator with a 16-layer
Elsevier, vol. 17, no. C, pp. 74-82, 2013. regenerator," Scientific Reports, vol. 7, 2017.
[17] K. P. Skokov, A. Y. Karpenkov, D. Y. Karpenkov [31] C. Aprea, A. Greco, A. Maiorino and C. Masselli,
and O. Gutfleisch, "The maximal cooling power of "Energy performances and numerical investigation
magnetic and thermoelectric refrigerators with of solid-state magnetocaloric materials used as
La(FeCoSi)13 alloys," Journal of Applied Physics, refrigerant in an active magnetic regenerator,"
vol. 113, 2013. Thermal Science and Engineering Progress, vol. 6,
[18] M. Balli, O. Sari, C. Mahmed, C. Besson, and J. pp. 370-379, 2018.
Forchelet, "A pre-industrial magnetic cooling [32] S. Singh, L. Caron, S. W. D’Souza, T. Fichtner, G.
system for room temperature application," Applied Porcari, S. Fabbrici, C. Shekhar, S. Chadov, M.
Energy, vol. 98, pp. 556-561, 2012. Solzi and C. Felser, "Large Magnetization and
[19] K. Engelbrecht, D. Eriksen, C. Bahl, R. Bjørk, J. Reversible Magnetocaloric Effect at the Second-
Geyti, J. Lozano, K. Nielsen, F. Saxild, Smith and Order Magnetic Transition in Heusler Materials,"
N. Pryds, "Experimental results for a novel rotary Advanced Materials, vol. 28, pp. 3321-3325, 2016.
active magnetic regenerator," International Journal [33] A. G. Condal, "Giant caloric and multicaloric
of Refrigeration, vol. 35, pp. 1498-1505, 2012. effects in magnetic alloys," Department de Física
[20] T. Gottschall, A. Gràcia-Condal, M. Fries, A. de la Matèria Condensada, Universitat de
Taubel, L. Pfeuffer, L. Mañosa, A. Planes, K. P. Barcelona, 2021.
Skokov and O. Gutfleisch, "A multicaloric cooling