International Journal of Refrigeration 151 (2023) 1–13

Contents lists available at ScienceDirect

International Journal of Refrigeration

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

A magnetocaloric air-conditioning system prototype✩

Un prototype de système de climatisation magnétocalorique
Guilherme F. Peixer a , Maria C.R. Silva a , Anderson Lorenzoni a , Gislaine Hoffmann a ,
Diego dos Santos a , Gabriel M. do Rosário a , Elias Pagnan a , Hígor F. Teza a , Pedro M. Silva a ,
Sergio L. Dutra a , Marcelo C. Ribeiro a , Marcelo A.A. Rosa b , Allan Döring c , Bernardo P. Vieira a ,
Alan T.D. Nakashima a , Paulo A.P. Wendhausen b , Cristiano S. Teixeira c , Jaime A. Lozano a ,
Jader R. Barbosa Jr. a ,∗
a POLO - Research Laboratories for Emerging Technologies in Cooling and Thermophysics, Department of Mechanical Engineering, Federal University of Santa
Catarina (UFSC), Florianópolis, SC, 88040-900, Brazil
b
MAGMA - Magnetic Materials Group, Department of Mechanical Engineering, Federal University of Santa Catarina (UFSC), Florianópolis, SC, 88040-900, Brazil
c
Lab3M - Laboratory of Magnetism and Magnetic Materials, Department of Materials Engineering, Federal University of Santa Catarina
(UFSC), Blumenau, SC, 89036-256, Brazil

ARTICLE INFO ABSTRACT

Keywords: This paper discusses the design procedure and first results of a large-scale magnetic air-conditioning system
Magnetic refrigeration prototype demonstrated in a relevant environment compatible with the Technology Readiness Level (TRL) 6.
Air conditioning The prototype, which contains all components of a functional cooling system, was experimentally evaluated
Performance evaluation
in an especially developed calibrated calorimeter designed according to international technical standards for
System optimization
testing split-type air-conditioning units. The proposed experimental analysis is one of the first to quantify
Mots clés: the impact of external losses associated with the heat exchangers on the cooling system performance.
Réfrigération magnétique
For outdoor and indoor temperatures fixed at 35 and 27 ◦ C, respectively, the coefficient of performance
Climatisation
disregarding thermal, flow and mechanical losses beyond the limits of the active magnetic regenerator was
Évaluation des performances
Optimization du système
5.72, corresponding to a second-law efficiency of 20.7%. When these losses and other essential operating
variables, such as the heat exchanger fan power and the valve power consumption, are included, one-order-of-
magnitude reductions are observed in the performance parameters, indicating that they should not be neglected
in realistic assessments of not-in-kind cooling technologies.

1. Introduction Among several candidates, Magnetic Refrigeration (MR) stands out

as one of the most promising not-in-kind cooling technologies (Qian
The demand for refrigeration and air-conditioning systems is ex- et al., 2016). The main factors attracting the attention of industry and
pected to rise substantially in the coming decades (IEA, 2018). Main- academia toward MR are (i) the use of solid refrigerants instead of
stream cooling technologies rely on a combination of mechanical com- harmful gases; (ii) the high thermodynamic efficiency potential due
pression and expansion of a vapor and liquid–vapor phase change heat to a combination of factors such as the reversibility of the magne-
transfer. After more than a century of development, vapor compression tocaloric effect (MCE) and the recovery of magnetization work; and
has become a mature technology with efficient and compact systems (iii) recycling opportunities for the magnetocaloric materials (MCM)
and a reliable supply chain. However, many refrigerants currently
and permanent magnets (PM).
used in such systems are under phase-down schedules and are not
Most MR prototypes are based on the Active Magnetic Regener-
viable long-term options (Gaußet al., 2017). As a result, the need
ator (AMR) cycle (Barclay and Steyert, 1982), whose main practical
for more effective solutions has mobilized the scientific community to
appeal is to increase the temperature span well above the adiabatic
seek alternatives to mechanical vapor compression in terms of energy
temperature change of the solid refrigerant. In the AMR, the material is
efficiency and environmental footprint.

✩ An abridged version of this manuscript was presented at the 19th International Refrigeration and Air-Conditioning Conference at Purdue, July 10–14, West
Lafayette, IN, USA.
∗ Corresponding author.
E-mail address: [email protected] (J.R. Barbosa Jr.).

https://doi.org/10.1016/j.ijrefrig.2023.03.014
Received 17 December 2022; Received in revised form 18 February 2023; Accepted 6 March 2023
Available online 9 March 2023
0140-7007/© 2023 Elsevier Ltd and IIR. All rights reserved.
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Nomenclature HHEx hot heat exchanger

MCE magnetocaloric effect
Roman letters MCM magnetocaloric material
FC flux concentrator
𝐵 magnetic flux density, T
Gd Gadolinium
𝐶𝑂𝑃 coefficient of performance
HMS hydraulic management system
𝑐p fluid specific heat capacity, J kg−1 K−1
MC magnetic circuit
𝑚̇ f mass flow rate, kg h−1
MR magnetic refrigerator
𝑚 mass, kg
MRU magnetic refrigeration unit
𝑃 pressure, Pa
HMS hydraulic management system
𝑄̇ heat transfer rate, W
PM permanent magnet
𝑄̇ C cooling capacity, W
𝑅g inner radius of the rotor, m
𝑅i inner radius of the stator, m
𝑅Mag outer radius of the rotor, m assembled as a porous matrix subjected to alternating fluid blows syn-
𝑅o outer radius of the stator, m chronized with successive magnetization and demagnetization steps.
𝑅s outer radius of the magnet segments, m A functional magnetic refrigeration system should contain at least
𝑇 temperature, o C four main subsystems (Trevizoli et al., 2016a): (i) an AMR to generate
𝑈 overall heat transfer coefficient, W K−1 m−2 the refrigerating effect; (ii) a magnetic circuit (MC) to produce the
𝑈𝐴 overall thermal conductance, W K−1 changing magnetic field on the MCM in the AMR; (iii) a Hydraulic Man-
𝑉̇ volumetric flow rate, m3 s−1 agement System (HMS) to control the alternating fluid blows through
𝑊̇ power, W the porous matrices; and (iv) heat exchangers (HEx) to thermally
connect the fluid and the indoor and outdoor environments. Moreover,
Greek letters mechanical transmission and control systems are usually required to
𝛥 Variation, - operate these systems.
𝜀 porosity, % Many lab-scale prototypes have been designed and commissioned
𝜂 second law efficiency, % worldwide (Yu et al., 2010; Kitanovski et al., 2015; Greco et al., 2019;
Kamran et al., 2020; Kitanovski, 2020). An overview of their subsystem
𝜔 Angular speed, rad s−1
configurations, main design parameters and performance metrics is
𝜏 torque, N m
presented in Table 1. Notice that some quantities were retrieved from
𝛩 angle, o
data plots, so minor discrepancies are expected. Except for the heat
Subscripts pump of Dall’Olio et al. (2021), all systems operate as refrigerators.
Only the prototype of Nakashima et al. (2021) employs actual HEx;
amr active magnetic regenerator
the others use electric heaters and thermal baths to emulate the heat
amb ambient load and thermal reservoir temperatures. Thus, except for Nakashima
C cold et al. (2021), the values of 𝑄̇ C and 𝛥𝑇span refer to the AMR. Notice that
CE cold end 𝑄̇ C,max and 𝛥𝑇max are the maximum values reported by the authors, not
chex cold heat exchanger necessarily the no-load and zero-span values. The last column of Table 1
curie Curie temperature displays the 𝐶𝑂𝑃 at a specified operating point, not necessarily the
DSC differential scanning calorimeter highest 𝐶𝑂𝑃 .
fan fan MR is yet to achieve the performance targets required for com-
Gd gadolinium mercializaion. Since most prototypes developed so far disregarded the
heat heaters influence of HEx and other losses, discrepancies between performance
HE hot end figures are common (Greco et al., 2019; Kamran et al., 2020), and
H hot cannot be attributed only to different operating conditions and design
characteristics. More often than not, pump, motor, valve and HEx
hhex hot heat exchanger
losses are overlooked in the evaluation procedures set forth by various
La La(Fe,Mn,Si)13 Hz alloy
authors, giving rise to unnecessary skepticism about the technology.
MCM magnetocaloric material
This paper presents the design and first results of a magnetocaloric
pump pump
air-conditioning system prototype. The Magnetic Refrigeration Unit
rot rotor (MRU) was developed using a system-level multi-objective optimization
sv solenoid valve of the four main subsystems (Peixer et al., 2023): AMR, MC, HEx
sys system and HMS. The experimental tests were carried out in a calibrated
wall calorimeter wall calorimeter designed and constructed for the present project (Silva
et al., 2021). The MRU performance was evaluated in terms of the AMR
Abbreviations
and system values for 𝑄̇ C , 𝐶𝑂𝑃 and 𝜂 for controlled indoor and outdoor
AMR active magnetic regenerator temperature conditions. Although the system performance parameters
CHEx cold heat exchanger are inferior to values typical of commercial-grade air conditioners, the
HEx heat exchanger prototype achieved a cooling capacity of 490 W for a temperature span
of 16.8 ◦ C, one of the most significant operating points on record.
Furthermore, when system-level losses are disregarded, 𝐶𝑂𝑃 and 𝜂
exhibit order-of-magnitude increases, demonstrating the efficiency im-
provement potential of MR.

2
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Table 1
Overview of magnetocaloric cooling and heat pumping prototypes.
Author Configuration Design Parameter Performance Metric
MC MCM HMS HEx 𝑚MCM B f 𝑚̇ f 𝑄̇ C,max [W] 𝛥𝑇max [K] COP[-]
(poles) [kg] [T] [Hz] [kg h−1 ] (𝛥𝑇span [K]) (𝑄̇ C [W]) (𝑄̇ C [W]/𝛥𝑇span
[K])
Bahl et al. HB – MV – – 1.24 – – – – –
(2011) (4) (peak)
Engelbrecht HB 0.25–0.8 mm MV – 2.8 1.24 1–2 200–700 1010 25.4 1.8
et al. (2012) (4) Gd spheres (peak) (0) (0) (400/8.9)
Lozano et al. HB 0.25–0.8 mm MV – 2.8 1.24 1.5 200–600 400 16 1.51
(2013) (4) Gd spheres (peak) (10) (200) (400/7.1)
Bahl et al. HB 0.25–0.8 mm MV – 2.8 1.24 1.5–2.25 200–600 1010 25.4 –
(2014) (4) Gd spheres (peak) (0) (0)
Eriksen et al. RS 0.3–0.6 mm MV – 1.7 1.13 0.75 180 360 25 3.1
(2015) (2) Gd and GdY (peak) (2) (0) (82.5/15.5)
spheres
Fortkamp RS 0.3–0.6 mm MV – 1.7 1.13 1 80–185 110 19 5.1
et al. (2018) (2) Gd and GdY (peak) (12) (50) (80/10.5)
spheres
Dall’Olio RS 0.4–0.63 mm SV – 3.4 1.44 0.5–1.2 500–1280 950 15.8 6.7
et al. (2021) (2) La(Fe,Mn,Si)13 Hy (mean) (5.6) (90) (340/10.3)
particles
Masche et al. RS 0.4–0.63 mm SV – 3.4 1.44 0.5–1.2 400–1300 815 15.8 5.7
(2021) (2) La(Fe,Mn,Si)13 Hy (mean) (5.6) (20) (288/10.3)
particles
Masche et al. RS 0.35–0.71 mm SV – 3.83 1.44 0.5–1.7 420–1650 1000 20.8 15.9
(2022) (2) Gd spheres (mean) (8) (0) (444/7.3)
Tura and HB 0.3 mm PP – 0.11 1.47 2–4 22–96 50 29 1.6
Rowe (2011) (2) Gd spheres (peak) (10) (0) (–/2.5)
Arnold et al. HB 0.5 mm PP – 0.65 1.25 0.5–0.8 25–118 96 33 –
(2014) (2) Gd spheres (mean) (0) (0)
Christiaanse HB 0.3–0.4 mm PP – 0.022 1.47 1 10 2 8 –
et al. (2018) (2) MnFePSi (peak) (6) (0)
particles
Govindappa HB 0.3–0.425 mm PP – 0.056 1.47 1 14 12.5 11.5 –
et al. (2018) (2) MnFeP1−x Asx (peak) (2.5) (0)
particles
Jacobs et al. HB 0.177–0.246 mm MV – 1.52 1.44 4 750–1290 3042 18 1.6
(2014) (–) La(Fe,Mn,Si)13 H (peak) (0) (0) (2000/12)
particles
Lionte et al. RS 0.4 mm MV – 5.5 1.34 1.12 420 900 21.4 9.7
(2020) (4) Gd and Gd-Er (peak) (18.35) (768) (900/18.35)
plates
Tušek et al. RS 0.3 mm MV – 0.6 0.98 0.25–4 – – – –
(2010) (4) Gd plates (mean)
Huang et al. CC 0.4–0.8 mm SV – 1.18 0.875 0.5–1.7 70–260 162.4 11.6 1.85
(2019) (2) Gd spheres (mean) (0) (0) (102.4/0)
Aprea et al. DU 0.4–0.5 mm MV – 1.2 1.25 0.36–1.79 420 – 13.5 –
(2014) (2) Gd spheres (peak) (0)
Aprea et al. DU 0.4–0.5 mm MV – 1.2 1.25 0.26–1.38 300–420 200 11.9 2.5
(2016) (2) Gd spheres (peak) (1.2) (0) (200/5.9)
Trevizoli HB 0.55 mm PP – 0.195 1.69 0.25–1 5–103 53.7 30 4.6
et al. (2016b) (2) Gd spheres (peak) (0) (0) (11/5)
Trevizoli HB 0.8 mm PP – 0.195 1.69 0.25–1 5–103 33 31 3.1
et al. (2017) (2) Gd spheres, (peak) (0) (0) (6.6/3)
plates and pins
Nakashima HB 0.55 mm MV – 0.195 1.69 0.25–0.5 10–100 32 10 2.3
et al. (2018) (2) Gd spheres (peak) (0) (22) (25/5)
Lozano et al. RS 0.425–0.6 mm MV – 1.7 0.892 0.4–1.4 150–200 150 12 0.54
(2016) (2) Gd spheres (mean) (0) (0) (80.4/7.1)
Capovilla RS 0.425–0.6 mm MV – 1.7 0.892 0.4–0.8 100–200 104 8 2.46
et al. (2016) (2) Gd spheres (mean) (4) (48) (72/4)
Nakashima RS 0.4 mm SV Tube-fin 2.24 0.98 0.5–1 125–225 28 14.2 0.37
et al. (2021) (2) Gd and GdY (mean) (14.2) (28) (24/12.5)
Spheres

Legend: HB: Halbach array, RT: Rotor-stator, CC: Custom C-shaped magnet, DU: Double-U magnet, MV: Mechanical valves, SV: Solenoid valves, PP: Piston pump.

3
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Fig. 1. The Magnetic Refrigeration Unit (MRU) and its main components.

Fig. 2. Photograph of the MRU inside the calorimeter.

2. Magnetic refrigeration unit design The prototype was designed by a multidisciplinary team comprising
experts in thermal, hydraulic and magnetic systems, and specialists
2.1. Overview in instrumentation, controls and material sciences. A Set-Based Con-
current Engineering convergence strategy was adopted to design the
prototype in a Lean Product Development environment. In the design
A rendered image of the MRU and its main components, excluding process, four main subsystems were considered: (i) AMR, (ii) MC, (iii)
the cold heat exchanger (CHEx), is depicted in Fig. 1. A photograph of HEx, and (iv) HMS. Since a segregated subsystem design does not guar-
the prototype inside the calorimeter is shown in Fig. 2. An explanatory antee optimal performance, a system-level integrated multi-objective
video of the MRU and the calorimeter can be found in: https://youtu. optimization was adopted, using low-computational cost models for
be/5lVMstvAjgk. each subsystem. The results provided design guidelines for all four

4
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Fig. 4. Comparison between experimental data and numerical results for the modulus
of 𝐵 at the center of the magnetic gap.

and 𝑅Mag , are 185 and 338 mm, respectively. The outer radius of the
PM segments, 𝑅s , is 315 mm, and the rotor length is 190 mm.
Fig. 3. Cross-section view (first quadrant) of the MC-AMR assembly. Black regions: The MC dimensions reported above were determined from a combi-
soft magnetic material; white regions: air; light gray regions: hard magnetic material; nation of subsystem (component)-level and system-level optimization
Dark gray region at the bottom left: stator/shaft; arrows: remanence directions.
procedures. At the subsystem level, the design was based on the nu-
merical method and optimization proposed by Fortkamp et al. (2023),
which consists of minimizing the deviation between a target (i.e.,
subsystems in integration events. Next, each subsystem was optimized idealized) trapezoidal magnetic waveform and a numerical profile cal-
individually and the guidelines were updated. The design cycle was culated adjusting the remanence directions of the PM segments. To
repeated until a final system configuration was achieved, culminating reduce the mass of hard magnetic material, the external radii of the
in a patent application (Lozano et al., 2021). PM segment layers were decreased independently until the calculated
magnetic field reached 99% of its original value. Similarly, the size of
2.2. Magnetic circuit subsystem the air (void) segment in the FC was gradually enlarged until either the
saturation of the soft magnetic material or the magnetic flux density in
The MC consists of a rotor-stator configuration, known for some the
the low magnetic field region extrapolated pre-determined threshold
highest cooling capacities in the literature (Trevizoli et al., 2016a). In
values.
this configuration, two concentric cylinders are separated by the AMR
At the system level, Machine Learning was employed to reduce the
(air) gap. The cross-section area of the first quadrant of the assembly is
overall cost and power consumption of the system by using a set of
presented in Fig. 3, where the segmentation of the PMs and remanence
optimized MC geometries to train a Ridge Regression coupled with
directions are merely illustrative.
Polynomial Feature pre-processing routines (Fortkamp et al., 2021).
The inner cylinder (stator) is made from 0.5-mm hollow circular
The main MC-related system variables are the mass and cost of the
laminated sheets of E-145 electrical steel. This design reduces eddy cur-
hard and soft ferromagnetic materials, and the magnetic flux density,
rent losses, which would otherwise appear due to the time-dependent
𝐵, profile generated in the AMR gap.
nature of the magnetic field and the electrical conductivity of the stator
Fig. 4 compares the experimental and numerical modulus of 𝐵 at
material (Lozano et al., 2016; Nakashima et al., 2021; Dall’Olio et al.,
the center of the AMR gap as function of the angular position, 𝛩.
2021). These losses increase the system power consumption and reduce
The numerical simulations were performed in COMSOL Multiphysics
the cooling capacity as they are dissipated as heat in the stator next to
the AMRs. The laminated sheets are stacked around a fixed shaft and and the data were acquired using a gaussmeter (LakeShore 425) and
tied in by two flanges. The stator length is 190 mm, and its inner and a transversal probe (MMT-6J04-VF) (Trevizoli et al., 2015; Fortkamp
outer radii, 𝑅i and 𝑅o , are 34 and 134 mm, respectively. et al., 2023). A good agreement is observed at low fields and in
The rotor is made from soft ferromagnetic material and PM seg- the regions with the steepest flux density variation. The maximum
ments. A servo motor (WEG-SWA 56-4-8, driver WEG-SCA06) coupled deviations (below 5%) take place around the peak. Moreover, the low-
with a pulley and belt system rotate it at a specified frequency. Since field range covers a wide region of the magnetized volume and is very
the MC has two magnetic poles, the AMR operation frequency is twice close to 0 T, an important requirement of AMR operation. Even though
that of the rotor. The PMs are all grade N50M sintered Nd-Fe-B, with a the high-field magnetic region does not display a plateau as wide as
minimum remanence of 1.41 T. Each pole comprises 32 epoxy-glued other systems (Lozano et al., 2016), high values of 𝐵 were achieved,
PM segment assembled from eight different segment geometries. To with a peak at 1.15 T.
decrease the PM mass, the number of segments in each PM layer is
reduced as one moves away from the center of the magnetic pole (see 2.3. Active magnetic regenerator subsystem
Fig. 3). Approximately 12.5 L of PM are used in the MC. The soft
ferromagnetic material part of the rotor, the flux concentrator (FC), The AMRs consist of porous beds of MCM housed in stainless steel
is made from 1- and 5-mm thick laminated sheets of S235JR. Unlike casings positioned within the air gap between the rotor and the stator,
the stator, lamination is intended to facilitate manufacturing instead of as shown in Figs. 3 and 5. The casings, shaped as sectors of a cylindrical
preventing eddy currents. The FC has two void regions: one to position shell, are attached to the stator and remain static during operation.
the PM segments (light gray regions) and another to reduce the overall They were designed to withstand cyclical mechanical loads with a
mass of the system (white region). The inner and outer rotor radii, 𝑅g minimum thickness to maximize the internal volume available for the

5
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Fig. 5. AMR assembled in the radial gap between the rotor and stator.

Table 2 Table 3
Compound and total masses of MCM in the AMRs and their respective porosities. Curie temperatures of the AMR layers.
𝑚MCM [g] 𝑚Gd [g] 𝑚La [g] 𝜖 [-] Layer Material 𝑇curie [◦ C] 𝑇DSC [◦ C] Mass fraction [%]
AMR 1 1309.32 320.88 988.44 0.417 1 Gd 17.0 17.0 24.0
AMR 2 1304.29 318.86 985.43 0.419 2 La(Fe,Mn,Si)13 Hy 19.3 17.5 7.6
AMR 3 1275.83 297.9 977.93 0.431 3 La(Fe,Mn,Si)13 Hy 22.2 20.9 7.6
AMR 4 1294.01 306.83 987.18 0.423 4 La(Fe,Mn,Si)13 Hy 24.6 24.4 7.6
AMR 5 1280.84 304.62 976.22 0.429 5 La(Fe,Mn,Si)13 Hy 27.9 27.4 7.6
AMR 6 1296.14 314.29 981.85 0.422 6 La(Fe,Mn,Si)13 Hy 30.0 30.0 7.6
AMR 7 1297.4 320.5 976.9 0.422 7 La(Fe,Mn,Si)13 Hy 32.6 33.5 7.6
AMR 8 1289.91 305.98 983.93 0.425 8 La(Fe,Mn,Si)13 Hy 34.6 36.5 7.6
9 La(Fe,Mn,Si)13 Hy 36.4 37.8 7.6
10 La(Fe,Mn,Si)13 Hy 39.0 39.3 7.6
11 La(Fe,Mn,Si)13 Hy 41.3 41.4 7.6
MCM (Peixer et al., 2017). Each casing has two ends, one hot and one
cold, and each end has two fluid ports, one inlet and one outlet, which
operate alternatively letting the fluid flow along the porous bed in a
direction specified by the stage of the AMR cycle. Although the system The number of AMR layers and their respective transition tem-
was designed to operate with a maximum of sixteen beds, only eight peratures were determined by the (limited) solid refrigerant avail-
were utilized in the present study due to a limited MCM availability. ability (particularly the La-based alloys) and the system temperature
requirements. Thus, one Gd layer and ten La(Fe,Mn,Si)13 Hz layers
The porous media in the AMRs consisted of packed beds of Gd
were used. The Curie temperature, 𝑇curie , and mass fraction of each
and La(Fe,Mn,Si)13 Hz arranged in layers with different magnetic tran-
layer of the optimal AMR configuration were defined via numerical
sition temperatures. While the Gd particles (supplied by the Baotou
simulation (Vieira, 2020), as shown in Table 3. In addition to the desired
Research Institute of Rare Earths, China) are largely spherical, the
𝑇curie , Table 3 also presents the Curie temperatures of the samples sup-
La(Fe,Mn,Si)13 Hz particles (commercialized as Calorivac-HS2 by Vacu-
plied by the manufacturer, 𝑇DSC , measured by a Differential Scanning
umschmelze GmbH & Co) resemble slightly elongated cylinders (Vieira
Calorimeter (Netzsch DSC21400A-1128-L) (Rosa et al., 2022). Notice
et al., 2021). The mean diameters of the Gd and La(Fe,Mn,Si)13 Hz par-
that the mass fractions are nominal values based on the simulations.
ticles, 400 and 650 μm, are dictated by their respective manufacturing
Again, due to the characteristics of the AMR beds, it is impossible to
processes. The mean height and width of the internal cross-section area
match the target compositions due to small inter-layer variations.
of the AMR casing (i.e., the area occupied by the porous matrix) are 45 The main consequence of the above-mentioned inter-AMR mass
and 57 mm, respectively. The mean AMR bed length is 130 mm. The differences is the flow imbalance between the regenerators, which is
masses of Gd, 𝑚Gd , and La(Fe,Mn,Si)13 Hz , 𝑚La , in each bed, and the a deterrent to the overall performance. As a result, fine-tuning the
resulting bed porosity, 𝜖, are presented in Table 2. opening and closing times of the solenoid valves (see Section 2.4) is
The difference between the mass composition of each regenerator is required to adjust the outlet temperature of the regenerators. Therefore,
due to a combination of factors, including manufacturing imprecisions the primary metric used to assess the performance of each regenerator
of the regenerator parts, assembly of such parts and geometric charac- pair is the outlet temperature of the cold end during the hot blow.
teristics of the magnetocaloric material. While minor regenerator casing Although individual AMR pressure drop and permeability evaluations
dimension variations were anticipated, more significant dimensional were done prior to the experiments reported here (Teza et al., 2021),
deviations occurred in the purpose-built wire meshes that separate any correction aimed at attenuating deviations between regenerator
the different regenerator layers. In addition, the substantial particle beds is performed on the spot using the hot-blow cold-end temperature
shape and size heterogeneity pointed out by Rosa et al. (2022) – as a metric.
the La(Fe,Mn,Si)13 Hz particles are more cylindrical than spherical – The AMR was also designed combining system- and subsystem-level
combined with the fact that the assembly of each regenerator is a procedures. At the subsystem-level, several experimental tests (Teza
manual process contributed significantly to the differences between the et al., 2021) were performed to validate numerical models (Vieira
masses of the AMR beds. et al., 2021) and evaluate the mechanical and chemical stability of

6
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Fig. 6. Hydraulic diagram of the magnetocaloric refrigeration unit (MRU).

the MCM (Rosa et al., 2022). At the system-level, lumped-element the same pair of solenoid valves and arranged at diametrically opposite
models (Nakashima et al., 2022) were integrated into the optimization positions in the air gap. To assure the continuous fluid flow thought
scheme, enabling faster and accurate performance predictions. In ad- the HEx, AMR set 1 operates simultaneously opposite to set 3, and
dition to 𝑄̇ C,amr and 𝛥𝑇amr , the most relevant AMR parameters to the set 2 is opposite to set 4. Hence, as the system blow fraction is fixed
system performance are the mass and cost of MCM, the fluid pumping at 25%, at any given instant, as one set of AMR beds goes through
power and the magnetization power. the hot blow, the others experience the cold blow, magnetization and
demagnetization phases, respectively.
2.4. Hydraulic management subsystem
At the subsystem level, the HMS was designed focusing on achieving
Fig. 6 presents the MRU hydraulic diagram. The thermal fluid is a an adequate hydraulic profile whilst minimizing the power consump-
2% vol. solution of ENTEK FNE (MacDermid Enthone GMbH, Germany) tion. Although mechanical valve systems were designed and evalu-
in deionized water. A gear pump (Thebe P11 7br) moves the fluid ated in previous works (Santos et al., 2021a), commercially available
through the circuit. The oscillatory hydraulic profile in the AMR is set solenoid valves (Asco SC8210-112/220V-NC) were chosen to control
by eight solenoid valves (one pair of solenoid valves for each pair of the flow through the AMR beds. Not only do the solenoid valves present
AMRs), sixteen check-valves (two for each AMR bed) and four fluid similar transient responses, they are also much more flexible regarding
manifolds. Two heat exchangers, one hot and one cold, two filters and changes in operating conditions, robustness, simplicity and low power
a liquid tank are the remaining components of the liquid line. consumption (6 W of nominal power consumption per valve) (Santos
The gear pump is positioned upstream to the HHEx to facilitate et al., 2021b). The HMS parameters relevant to the system optimization
the dissipation in the latter of the heat generated in the former. Next, are the nominal power consumption of the solenoid valves and the
the fluid flows to the high pressure hot manifold and is distributed to pumping power through the components.
the solenoid valves and AMR beds. The fluid is collected in the high-
pressure cold manifold and direct to the CHEx. Then, the fluid is once
again distributed to the AMRs via the low-pressure cold manifold and, 2.5. Data acquisition and instrumentation
after flowing through the beds, it is collected by the low-pressure hot
manifold and directed to the liquid tank and pump. Each AMR bed
Fig. 6 indicates the temperature and pressure measurements along
has one inlet and one outlet ports on each side (hot and cold). Check
the hydraulic circuit. The mass flow rate is measured at the inlets
valves are placed at the ports of the cold side to assure a proper flow
distribution through the AMRs. of CHEx and HHEx. The rotor is equipped with torque and power
Magnetic and hydraulic profiles are synchronized via a control logic meters, which are also used to measure the fan power at the CHEx
employing the signal of an encoder coupled to the rotor. Additionally, and HHEx and the pumping power at the hydraulic pump. An encoder
the signal of eight Hall Effect sensors placed either at or between four and eight Hall-effect sensors measure the rotor’s angular position and
of the eight AMR beds are used to fine tune the actuation of solenoid magnetic flux density. Data acquisition and control were implemented
valves. According to Fig. 6, the AMR beds are divided into four sets (1, in LabView using National Instruments hardware (My-Rio 19000 and
2, 3 and 4) and into two other sets (A and B), which are controlled by cDAQ-9179). The expanded uncertainties are presented in Table 4.

7
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Fig. 7. Schematic representation of the MRU and the calorimeter.

Table 4 Table 6
Experimental uncertainties of the measuring instruments installed in the MRU. Reconditioning equipment employed in the calorimeter.
Measurement Uncertainty Equipment Quantity Specifications
Fluid temperature ±0.11 ◦C Electric heaters 8 1750 W heating power (each)
Air temperature ±0.11 ◦ C Immersion heaters 4 1000 W heating power (each)
Fluid pressure ±0.01 bar Water reservoir 2 30 L stainless steel tanks
Torque ±0.25 N m Axial fans 4 EBM-Papst S4E350-AN19-43
Power consumption ±3.5 W Fan coil unit 1 30000 Btu h−1 cooling capacity
Mass flow rate (CHEx) ±6.8 kg h−1 Chiller 1 Connected to the fan coil unit
Mass flow rate (HHEx) ±1%

Table 5 The indoor chamber is equipped with temperature and humidity re-
Dimensions of the calorimeter.
conditioning systems, whereas only the temperature is reconditioned in
Dimensions Width Length Height
the outdoor chamber. In the indoor chamber reconditioning system, dry
[mm] [mm] [mm]
electric heaters and electric heaters immersed in water tanks control
External - calorimeter 2200 4860 2640 the temperature and the humidity, respectively. Axial fans homogenize
Internal - indoor chamber 2060 2380 2490
the air in the chamber. The outdoor chamber reconditioning system
Internal - outdoor chamber 2060 2190 2500
consists of a fan coil unit coupled with a chiller and tangential fans. The
reconditioning equipment of both chambers is summarized in Table 6.
Nine thermocouples, eight humidity transducers, and one pres-
2.6. Calorimeter and heat exchanger sure transducer installed in each chamber provide data for calculating
their respective wet-bulb temperatures. Seven thermocouples and one
A calibrated calorimeter was developed to test the MRU according thermo-hygrometer measure the temperature and the humidity of the
to international (ISO-5151, 2017) and Brazilian (Inmetro, 2011, 2012, air surrounding the calorimeter. Electric power is measured by two
2013) standards. Although some specifications were not fully met power transducers with 1000- and 2000-W full scales. The condensate
due to laboratory space limitations, most parameters, particularly the mass is measured by a calibrated scale. Temperatures are also moni-
instrument tolerances, were within the margins deemed adequate for tored in the immersion heater and condensate reservoir. The expanded
characterizing the system in a relevant environment. uncertainties of the calorimeter instrumentation are shown in Table 7.
The calorimeter consists of two chambers separated by an insulated Heat leaks through the insulated walls must be considered during
wall to emulate the outdoor (warm) and indoor (cold) conditions. the calorimeter calibration. Here, the calibration involved the use
Table 5 shows the calorimeter external dimensions and the internal of electric heaters to produce a steady-state thermal load inside the
dimensions of the indoor and outdoor chambers. The walls are insu- chambers with a temperature difference of at least 11 ◦ C relative to
lated with polyisocyanurate, which has a thermal conductivity between the outside. The overall thermal conductance of the calorimeter walls,
0.02 and 0.03 W m−1 K−1 ; the wall separating the indoor and outdoor 𝑈 𝐴, is calculated from the wall heat transfer rate and the temper-
chambers has a thickness of 150 mm, whereas the thickness of the ature difference. Since the heat transfer area is known, the overall
remaining walls is 70 mm. The MRU and calorimeter are schematically heat transfer coefficient, 𝑈 , can be computed as well. This procedure
represented in Fig. 7. is repeated three times: first heating the indoor chamber only, then

8
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Table 7 The AMR and system cooling capacities, 𝑄̇ C,amr and 𝑄̇ C,sys , are
Experimental uncertainties of the measuring instruments installed in the calorimeter.
calculated, respectively, as:
Measurement Equipment Uncertainty
𝑄̇ C,amr = 𝑚̇ f 𝑐p 𝛥𝑇C,amr (3)
Water temperature Thermocouple ±0.25 ◦ C
Air dry bulb temperature Thermocouple ±0.25 ◦ C
Air dry bulb temperature Thermo-hygrometer ±0.12 ◦ C 𝑄̇ C,sys = 𝑊̇ fan + 𝑊̇ heat + 𝑄̇ wall (4)
Air relative humidity Humidity transducers ±2.8%
Air relative humidity Thermo-hygrometer ±2% where 𝑐p is assumed equal to that of pure water. 𝛥𝑇C,amr is the
Electric Power Power transducers (1000 W) ±3.6 W temperature difference between the fluid entering and exiting the cold
Electric Power Power transducers (2000 W) ±3.6 W end of the AMR, measured in the manifolds on the cold side. 𝑊̇ fan
Pressure Pressure Transducers ±2 kPa represents the power supplied to the fans, 𝑊̇ heat is the power to the
Mass Scale ±0.1 g heaters and 𝑄̇ wall is the heat gain through the walls of the indoor
chamber.
Table 8 The AMR and system coefficients of performance, 𝐶𝑂𝑃amr and
Calorimeter calibration results. 𝐶𝑂𝑃Sys , are calculated as:
Wall 𝑈 𝐴 [W K−1 ] 𝑈 [W m−2 K−1 ] 𝑄̇ C,amr
Dividing wall 1.7 ± 0.1 0.313 ± 0.016 𝐶𝑂𝑃amr = (5)
𝛥𝑃amr 𝑉̇ l + 𝜏𝜔
Indoor chamber walls 13.9 ± 0.1 0.482 ± 0.005
Outdoor chamber walls 17.7 ± 0.1 0.779 ± 0.005 𝑄̇ C,sys
𝐶𝑂𝑃sys = (6)
𝑊̇ pump + 𝑊̇ rot + 𝑊̇ CHEx + 𝑊̇ HHEx + 𝑊̇ SV
where 𝛥𝑃amr is the AMR pressure drop (measured at the manifolds,
heating the outdoor chamber only, and finally heating both chambers disregarding the tubing, adapters and casing losses), 𝜏 is the torque
simultaneously. The result are shown in Table 8. (disregarding friction in the bearings and parasitic losses such as eddy
Due to their favorable thermal-hydraulic characteristics, currents) and 𝜔 is the rotor angular speed. 𝑊̇ rot is the power required
fan-supplied multi-circuited herringbone-wave tube-fin HEx were used to drive the rotor, 𝑊̇ pump is the power consumption of the hydraulic
to connect the system to the indoor and outdoor chambers. The cou- pump, 𝑊̇ CHEx and 𝑊̇ HHEx are the fan powers of each HEx, and 𝑊̇ SV is
pling between the AMR, HEx and thermal reservoirs creates two char- the power required to actuate the solenoid valves.
acteristic temperature spans: the AMR temperature span, 𝛥𝑇amr , and the Lastly, the AMR second-law efficiency, 𝜂amr , and the system second-
system temperature span, 𝛥𝑇sys . Due to the finite sizes of the HEx, the law efficiency, 𝜂sys , are calculated as:
former is always larger than the latter. However, most MR prototypes ( )
𝑇HE − 𝑇CE
did not employ actual HEx; instead, Joule-effect heaters were used 𝜂amr = 𝐶𝑂𝑃amr (7)
𝑇CE
to generate the thermal loads (Kitanovski et al., 2015). As a result, ( )
𝑇H − 𝑇C
external losses are rarely quantified in the MR literature. 𝜂sys = 𝐶𝑂𝑃sys (8)
At the subsystem level, the 𝜀 − 𝑁𝑇 𝑈 method was used to design the 𝑇C
HEx using empirical Colburn-𝑗 and friction factor relationships (Peixer 3. Results and discussion
et al., 2022). At the system level, the liquid- and air-side pumping
power contributions in the HHEx and CHEx were included in the overall The MRU performance was evaluated at five different flow rates
system power consumption. Moreover, the thermal coupling between between 300 and 700 kg h−1 . For each 𝑚̇ f , the system was tested at two
the heat transfer fluid and the air in the indoor and outdoor chambers operating points defined by a thermal load (set by the electric heaters
was realized via energy balances in both HEx. The HEx dimensions in the indoor chamber) and a temperature span, with 𝑇H always fixed
were as follows: 515 mm (height), 550 mm (width) and 245 mm at 35 ◦ C. At the first operating point (Case I), international standards
(length), with an eight fins-per-inch density and tube ID and wall were followed (ISO-5151, 2017; Inmetro, 2011, 2013, 2012), and 𝑇C
thickness of 11.9 mm and 0.4 mm, respectively. Forced air flow through was fixed at 27 ◦ C. The second operating point (Case II) was set by
the fin-tube arrays was supplied by EBM-Papst S4E350-AN19-43 axial turning off the indoor chamber heaters. Notice that the performance of
fans. first-order MCM AMRs is strongly affected by 𝑇H , so fine tuning this pa-
rameter is usually adopted to enhance the system performance (Masche
2.7. Test parameters and performance metrics et al., 2021). However, such a strategy was not pursued in the present
work.
The MRU is evaluated in terms of 𝛥𝑇 , 𝑄̇ C , 𝐶𝑂𝑃 and 𝜂 associated As mentioned above, at any given time, the mass flow rate is split
with the AMR and overall system. Since most MR prototypes are not between the beds A and B of a given AMR pair (Fig. 6). Hence, the
equipped with actual HEx, the AMR-based parameters are better suited system mass flow rate is approximately twice that of one AMR bed.
for comparisons with other MR systems, while the system parameters Here, the blow fraction was set at 25%, and the operating frequency
can be compared with other cooling technologies. The temperature was kept fixed at 0.6 Hz. Although the system can perform at higher
spans 𝛥𝑇amr and 𝛥𝑇sys are defined as: frequencies with sixteen AMR beds (full load), at the present half-
load condition (eight beds), magnetic cogging is still significant, which
𝛥𝑇amr = 𝑇HE − 𝑇CE (1) increases both the torque and the power consumption.

𝛥𝑇sys = 𝑇H − 𝑇C (2) 3.1. Temperature pull down

where 𝑇HE is the fluid temperature in the outlet manifold at the hot Fig. 8 presents the temperature pull-down curves (MR and calorime-
side; 𝑇CE is measured at the corresponding spot at the cold end. 𝛥𝑇sys ter) for an AMR mass flow rate of 600 kg h−1 (Case II). The values of 𝑇C ,
is the difference between the outdoor chamber temperature, 𝑇H , itself 𝑇H , 𝑇HE , 𝑇CE , and the ambient temperature, 𝑇amb , are shown for a period
an average of local temperatures measured in the outdoor chamber, and of more than 5 h. The results demonstrate that temperatures around
the indoor chamber temperature, 𝑇C , also an average of indoor chamber 22 ◦ C can be reached, as expected for air-conditioning applications.
local temperatures. The thermal fluid stabilizes at around 2 h, whereas the indoor air takes

9
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Fig. 8. Transient temperature profiles for an AMR mass flow rate of 600 kg h−1 and a frequency of 0.6 Hz (Case II).

more than 5 h. The HEx (external) irreversibilities are clearly visible, in the tubing connecting the AMR and the CHEx, the fluid flow losses
and the difference between 𝛥𝑇sys and 𝛥𝑇amr poses a large penalty on the in the manifold and hydraulic components and the CHEx effectiveness
system, especially since achieving large temperature spans is among are strongly influenced by the mass flow rate and, therefore, affect
the main obstacles of MR. Temperature differences of around 2.8 ◦ C the system’s cooling capacity. Additionally, there is the effect of the
and 1.2 ◦ C are required in the HHEx and CHEx, respectively, reducing AMR temperature span, which has not been kept fixed in any of the
the span produced by the AMR. The difference is higher in the HHEx assessments performed.
due to a larger heat rejection rate, which is also influenced by the heat Fig. 10 evaluates the system and AMR thermodynamic efficiencies
dissipation in the hydraulic pump. in terms of the 𝐶𝑂𝑃 for Cases I and II. Firstly, in absolute terms, in
addition to the cooling capacity and temperature span values specified
3.2. Steady-state results above, the power consumption and efficiency obtained in the present
tests fall short of the requirements for practical applications of such
Fig. 9 presents system and AMR operating points (Cases I and systems. For instance, conventional vapor compression systems can de-
II) for various mass flow rates. Although Fig. 9(a) reveals a superior velop values of 𝐶𝑂𝑃sys and 𝜂sys above 2.5 and 10%, respectively (Silva
system’s response at 600 kg h−1 , at the current development stage, the et al., 2021), requiring magnetocaloric systems to attain at least similar
system is still incapable of producing a suitable operating point for figures. However, when the influence of the HEx and other non-AMR
air-conditioning applications, which for the present prototype corre- power consumption factors are excluded from the analysis, a twofold
sponds to a system cooling capacity of 9000 BTU h−1 (2637 W) at a increase of the 𝐶𝑂𝑃 is observed, demonstrating a significant perfor-
standardized system temperature span where the outdoor and indoor mance improvement potential. For Case I, the maximal 𝐶𝑂𝑃sys and
temperatures are fixed at 35 and 27 ◦ C (ISO-5151, 2017). However, 𝐶𝑂𝑃amr are 0.30 (500 kg h−1 ) and 5.72 (300 kg h−1 ). For Case II, the
maximal 𝐶𝑂𝑃sys and 𝐶𝑂𝑃amr are 0.16 (500 kg h−1 ) and 5.57 (300 kg
considering the AMR performance only (i.e., excluding the losses be-
h−1 ). Again, for completeness, the ranges associated with the variables
yond its limits), Fig. 9(b), the system achieved some of the largest
in Eq. (6) are 𝑊̇ pump : 290 W to 810 W, 𝑊̇ rot : 150 to 210 W, 𝑊̇ CHEx :
cooling capacities and temperature spans on record. The maximum
125 to 130 W, 𝑊̇ HHEx : 185 to 190 W and 𝑊̇ sv : 7 to 8.5 W. Hence, it
values for 𝑄̇ C,sys and 𝛥𝑇sys are around 317 W and 12.6 ◦ C, for 600 kg
is evident the efficiency of some components, e.g., pump and motors,
h−1 . As for the AMR, the maximum 𝑄̇ C,amr and 𝛥𝑇amr are around 490 W
are determinants of the overall system performance. Improving the
and 16.8 ◦ C, also for 600 kg h−1 .
individual efficiencies of the ancillary components is the subject of
When comparing both figures, the remarkable influence of the HEx
future studies.
on the system’s performance becomes apparent. First, the temperature
Fig. 11 assesses the second-law thermodynamic efficiency, 𝜂, for
span is reduced by approximately 3 or 4 K due to the finite HEx thermal
Cases I and II. As a consequence of the results presented in Fig. 10,
conductances. Other irreversibility sources are the dead volume and
the second-law efficiency in this preliminary assessment is still insuf-
fluid mixing effects that decrease the cooling capacity at the cold end of ficient to enable the practical application of magnetic cooling in air
the AMR. 𝑄̇ C,sys is also reduced by thermal leaks in the tubing between conditioners. However, again, if the influence of non-AMR power con-
the manifolds and the indoor calorimeter chamber. Hence, it becomes sumption elements is disregarded, a twofold increase in 𝜂 is obtained,
clear that, to evaluate MR prototypes properly, one must consider the which might be enough for some practical applications. For Case I, the
influence of the HEx and the cold fluid transport on the cooling capacity maximum values of 𝜂sys and 𝜂amr are 0.8% (500 kg h−1 ) and 20.7%
and temperature span. For completeness, the ranges associated with the (300 kg h−1 ). For Case II, the maximum values of 𝜂sys and 𝜂amr are
variable in Eq. (4) are 𝑊̇ fan : 125 to 130 W, 𝑊̇ heat : 40 to 170 W and 𝑄̇ wall : 0.69% (500 kg h−1 ) and 22.6% (300 kg h−1 ).
0 to 20 W.
As can be seen from Fig. 9, the flow rates responsible for the 4. Conclusions
maximal cooling capacity are different from the AMR and system
standpoints. Several reasons are involved in that behavior, as complex The design and assembly of a large-scale magnetic air-conditioning
and intricate trade-offs connect the AMR, cold heat exchanger and low- prototype and a purpose-built calibrated calorimeter were presented to-
temperature thermal reservoir. For instance, the parasitic heat gains gether with preliminary results and a detailed performance assessment.

10
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Fig. 9. Operating points (Cases I and II) for various mass flow rates: (a) system and (b) AMR.

Fig. 10. Results for the AMR and system 𝐶𝑂𝑃 as a function of the mass flow rate for (a) Case I and (b) Case II.

Fig. 11. Results for the AMR and system 𝜂 as a function of the mass flow rate for (a) Case I and (b) Case II.

The performance results demonstrated that there is room for improving Nevertheless, compared to other MR systems, the present prototype
the MR prototype in terms of cooling capacity, power consumption reached one of the most significant operating points, with a cooling
and efficiency before any attempt to commercialize the technology. capacity of 490 W at an AMR temperature span of 16.8 ◦ C. Moreover,

11
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

this operating condition was sustained in a relevant environment with- Fortkamp, F.P., Cattelan, L.F.P., Peixer, G.F., Lozano, J.A., Barbosa, Jr., J.R., 2021.
out any fine tuning the outdoor temperature, a sensitive performance Using machine learning to design a large scale magnetic circuit. In: Proceedings
in Nineth IIR International Conference on Caloric Cooling, Thermag IX. Maryland,
parameter in first-order AMRs.
USA.
Further improvements, such as (i) increasing the operating fre- Fortkamp, F., Eriksen, D., Engelbrecht, K., Bahl, C., Lozano, J., Barbosa, Jr., J.R., 2018.
quency by reducing the torque required to move the rotor by employing Experimental investigation of different fluid flow profiles in a rotary multi-bed
a continuous array of sixteen (full load) instead of eight (half load) active magnetic regenerator device. Int. J. Refrig. 91, 46–54.
Fortkamp, F.P., Nakashima, A.T.D., dos Santos, V.M.A., Lozano, J.A., Barbosa Jr., J.R.,
AMR beds; (ii) reducing parasitic heat gains in the tubing and hydraulic
2023. Computationally-efficient optimization of the remanence angles of permanent
components at the cold side (between the AMR and the CHEx); and magnet circuits for magnetic refrigeration. J. Magn. Magn. Mater. 569, 170429.
(iii) optimizing hydraulic components to reduce flow losses (e.g., man- Gauß, R., Homm, G., Gutfleisch, O., 2017. The resource basis of magnetic refrigeration.
ifolds and check valves), are expected to enhance the cooling capacity, J. Ind. Ecol. 21, 1291–1300.
Govindappa, P., Trevizoli, P., Niknia, I., Christiaanse, T., Teyber, R., Rowe, A., 2018.
temperature span, power consumption, and efficiency.
Experimental characterization of multilayer active magnetic regenerators using first
External irreversibilities have also been carefully evaluated. Their order materials: Multiple points of equilibrium. J. Appl. Phys. 124, 134901.
highly detrimental impact must be considered when assessing the Greco, A., Aprea, C., Maiorino, A., Masselli, C., 2019. A review of the state of the
performance of not-in-kind technologies. Here, a twofold increase in art of solid-state caloric cooling processes at room-temperature before 2019. Int.
the performance metrics was encountered when evaluating the system J. Refrig. 106, 66–88.
Huang, B., Lai, J., Zeng, D., Zheng, Z., Harrison, B., Oort, A., van Dijk, N., Brück, E.,
from the AMR and system standpoints. From an AMR perspective, the 2019. Development of an experimental rotary magnetic refrigerator prototype. Int.
maximal 𝐶𝑂𝑃amr and 𝜂amr were 5.72 and 22.6%, whereas from a system J. Refrig. 104, 42–50.
perspective, the maximal 𝐶𝑂𝑃sys and 𝜂sys were 0.3 and 0.8%. The IEA, 2018. The Future of Cooling. Technical Report, IEA, Paris, License: CC BY 4.0.
differences between them reveal the importance of considering thermal, Inmetro, 2011. Portaria n.◦ 007 - Requisitos de avaliação da conformidade para
condicionadores de ar. Technical Report, INMETRO.
fluid flow and mechanical losses in addition to the AMR performance Inmetro, 2012. Portaria n.◦ 643 - requisitos de avaliação da conformidade para
to avoid raising misleading expectations about the technology. condicionadores de ar. Technical Report, INMETRO.
Inmetro, 2013. Portaria n.◦ 410 - requisitos de avaliação da conformidade para
Declaration of competing interest condicionadores de ar. Technical Report, INMETRO.
ISO-5151, 2017. Non-Ducted Air Conditioners and Heat Pumps - Testing and Rating
for Performance. Technical Report, ISO.
The authors declare that they have no known competing finan- Jacobs, S., Auringer, J., Boeder, A., Chell, J., Komorowski, L., Leonard, J., Russek, S.,
cial interests or personal relationships that could have appeared to Zimm, C., 2014. The performance of a large-scale rotary magnetic refrigerator. Int.
influence the work reported in this paper. J. Refrig. 37, 84–91.
Kamran, M.S., Ahmad, H.O., Wang, H.S., 2020. Review on the developments of active
magnetic regenerator refrigerators – Evaluated by performance. Renew. Sustain.
Acknowledgments Energy Rev. 133, 110247.
Kitanovski, A., 2020. Energy applications of magnetocaloric materials. Adv. Energy
Mater. 10 (10), 1903741.
The authors appreciate the support from CODEMGE and EMBRAPII
Kitanovski, A., Tusek, J., Tomc, U., Plaznik, U., Ozbolt, M., Poredos, A., 2015. Magne-
through Grant No. 201813442 (‘‘Development of an Air Conditioner tocaloric Energy Conversion: From Theory to Applications. Springer International
Operated by a Magnetic Cooling Unit’’ Project). Additional funding Publishing.
was provided by the National Institutes of Science and Technology Lionte, S., Risser, M., Muller, C., 2020. A 15 kW magnetocaloric proof-of-concept unit:
Initial development and first experimental results. Int. J. Refrig. 122.
(INCT) Program (CNPq Grant No. 404023/2019-3; FAPESC Grant No.
Lozano, J.A., Capovilla, M.S., Trevizoli, P.V., Engelbrecht, K., Bahl, C.R., Bar-
2019TR0846). bosa, Jr., J.R., 2016. Development of a novel rotary magnetic refrigerator. Int.
J. Refrig. 68, 187–197.
References Lozano, J., Engelbrecht, K., Bahl, C., Nielsen, K., Eriksen, D., Olsen, U., Bar-
bosa, Jr., J.R., Smith, A., Prata, A., Pryds, N., 2013. Performance analysis of a
rotary active magnetic refrigerator. Appl. Energy 111, 669–680.
Aprea, C., Cardillo, G., Greco, A., Maiorino, A., Masselli, C., 2016. A rotary permanent
Lozano, J.A., Peixer, G.F., Lorenzoni, A.M., Hoffmann, G., Silva, M.C.R., Dutra, S.L.,
magnet magnetic refrigerator based on AMR cycle. Appl. Therm. Eng. 101,
Ribeiro, M., dos Santos, D., Rosário, G.M., Cattelan, L.F.P., Teza, H.F., Pagnan, E.,
699–703.
Fortkamp, F.P., Vieira, B.P., Nakashima, A.T.D., Barbosa Jr., J.R., 2021. Unidade
Aprea, C., Greco, A., Maiorino, A., Mastrullo, R., Tura, A., 2014. Initial experimental
magnetocalórica. Brazilian Patent No. BR 10 2021 023316. Instituto Nacional de
results from a rotary permanent magnet magnetic refrigerator. Int. J. Refrig. 43,
Propriedade Industrial.
111–122.
Masche, M., Liang, J., Dall’Olio, S., Engelbrecht, K., Bahl, C., 2021. Performance
Arnold, D., Tura, A., Ruebsaat-Trott, A., Rowe, A., 2014. Design improvements of a
analysis of a high-efficiency multi-bed active magnetic regenerator device. Appl.
permanent magnet active magnetic refrigerator. Int. J. Refrig. 37, 99–105.
Therm. Eng. 199, 117569.
Bahl, C., Engelbrecht, K., Bjørk, R., Eriksen, D., Smith, A., Nielsen, K., Pryds, N., 2011. Masche, M., Liang, J., Engelbrecht, K., Bahl, C., 2022. Performance assessment of a
Design concepts for a continuously rotating active magnetic regenerator. Int. J. rotary active magnetic regenerator prototype using gadolinium. Appl. Therm. Eng.
Refrig. 34, 1792–1796. 204, 117947.
Bahl, C., Engelbrecht, K., Eriksen, D., Lozano, J., Bjørk, R., Geyti, J., Nielsen, K., Nakashima, A.T., Dutra, S.L., Trevizoli, P.V., Barbosa, Jr., J.R., 2018. Influence of the
Smith, A., Pryds, N., 2014. Development and experimental results from a 1 kW flow rate waveform and mass imbalance on the performance of active magnetic
prototype AMR. Int. J. Refrig. 37, 78–83. regenerators. Part I: Experimental analysis. Int. J. Refrig. 93, 236–248.
Barclay, J., Steyert, W., 1982. Active magnetic regenerator. US-Patent 4332135. Nakashima, A.T., Fortkamp, F.P., de Sá, N.M., dos Santos, V.M., Hoffmann, G.,
Capovilla, M.S., Lozano, J.A., Trevizoli, P.V., Barbosa Jr., J.R., 2016. Performance Peixer, G.F., Dutra, S.L., Ribeiro, M.C., Lozano, J.A., Barbosa Jr., J.R., 2021. A
evaluation of a magnetic refrigeration system. Sci. Technol. Built Environ. 22 (5), magnetic wine cooler prototype. Int. J. Refrig. 122, 110–121.
534–543. Nakashima, A.T.D., Peixer, G.F., Lozano, J.A., Barbosa Jr., J.R., 2022. A lumped-element
Christiaanse, T.V., Trevizoli, P.V., Misra, S., Carroll, C., van Asten, D., Zhang, L., magnetic refrigerator model. Appl. Therm. Eng. 204, 117918.
Teyber, R., Govindappa, P., Niknia, I., Rowe, A., 2018. Experimental study of 2- Peixer, G.F., Dutra, S.L., Calomeno, R.S., de Sá, N.M., Lang, G.B., Lozano, J.A., Barbosa
layer regenerators using Mn–Fe–Si–P materials. J. Phys. D: Appl. Phys. 51 (10), Jr., J.R., 2022. Influence of heat exchanger design on the thermal performance of
105002. a domestic wine cooler driven by a magnetic refrigeration system. An. Acad. Bras.
Dall’Olio, S., Masche, M., Liang, J., Insinga, A., Eriksen, D., Bjørk, R., Nielsen, K., Cienc. 94, e20200563.
Barcza, A., Vieyra, H., Beek, N.V., Bez, H.N., Engelbrecht, K., Bahl, C., 2021. Novel Peixer, G., Lozano, J.A., Barbosa Jr., J.R., 2017. Performance evaluation of AMRs using
design of a high efficiency multi-bed active magnetic regenerator heat pump. Int. different casings. In: Book of Abstracts for the Danish Days on Caloric Materials
J. Refrig. 132, 243–254. and Devices. Roskilde, Denmark.
Engelbrecht, K., Eriksen, D., Bahl, C., Bjørk, R., Geyti, J., Lozano, J., Nielsen, K., Peixer, G.F., Nakashima, A.T.D., Lozano, J.A., Barbosa, Jr., J.R., 2023. System-level
Saxild, F., Smith, A., Pryds, N., 2012. Experimental results for a novel rotary active multi-objective optimization of a magnetic air conditioner through coupling of
magnetic regenerator. Int. J. Refrig. 35, 1498–1505. artificial neural networks and genetic algorithms. Appl. Therm. Eng. in review.
Eriksen, D., Engelbrecht, K., Bahl, C., Bjørk, R., Nielsen, K., Insinga, A., Pryds, N., 2015. Qian, S., Nasuta, D., Rhoads, A., Wang, Y., Geng, Y., Hwang, Y., Radermacher, R.,
Design and experimental tests of a rotary active magnetic regenerator prototype. Takeuchi, I., 2016. Not-in-kind cooling technologies: A quantitative comparison of
Int. J. Refrig. 58, 14–21. refrigerants and system performance. Int. J. Refrig. 62, 177–192.

12
G.F. Peixer et al. International Journal of Refrigeration 151 (2023) 1–13

Rosa, M.A., Boeck, P., Döring, A.M., Vieira, B.P., Schafer, D., Plá Cid, C.C., Lozano, J.A., Trevizoli, P.V., Christiaanse, T.V., Govindappa, P., Niknia, I., Teyber, R., Bar-
Barbosa Jr., J.R., Wendhausen, P.A.P., da Silva Teixeira, C., 2022. Magnetocaloric bosa, Jr., J.R., Rowe, A., 2016a. Magnetic heat pumps: An overview of design
La(Fe,Mn,Si)13 H𝑧 particles and their chemical stability in heat transfer fluids principles and challenges. Sci. Technol. Built Environ. 22, 507–519.
employed in magnetic refrigeration. J. Magn. Magn. Mater. 563, 169875. Trevizoli, P.V., Lozano, J.A., Peixer, G.F., Barbosa Jr., J.R., 2015. Design of nested
Santos, D.D., Dutra, S.L., Ribeiro, M.C., Lorenzoni, A.M., Rosário, G.M.D., Silva, M.C.R., halbach cylinder arrays for magnetic refrigeration applications. J. Magn. Magn.
Hoffmann, G., Peixer, G.F., Lozano, J.A., Barbosa Jr., J.R., 2021a. Designing a hy- Mater. 395, 109–122.
draulic management system for a large-scale magnetic refrigerator. In: Proceedings Trevizoli, P.V., Nakashima, A.T., Peixer, G.F., Barbosa, J.R., 2016b. Performance
in Nineth IIR International Conference on Caloric Cooling, Thermag IX. Maryland, evaluation of an active magnetic regenerator for cooling applications – Part I:
USA. Experimental analysis and thermodynamic performance. Int. J. Refrig. 72, 192–205.
Santos, D., Silva, M., Hoffmann, G., Lorenzoni, A., Peixer, G.F., Lozano, J.A., Barbosa Trevizoli, P.V., Nakashima, A.T., Peixer, G.F., Barbosa, Jr., J.R., 2017. Performance
Jr., J.R., 2021b. Experimental evaluation of the fluid flow management system of a assessment of different porous matrix geometries for active magnetic regenerators.
magnetic air conditioner. In: Proceedings of the 26th ABCM International Congress Appl. Energy 187, 847–861.
of Mechanical Engineering. Florianópolis, Brazil, pp. COB2021–1836. Tura, A., Rowe, A., 2011. Permanent magnet magnetic refrigerator design and
Silva, M.C.R., Rosário, G.S., Dutra, S.L., Capovilla, M., Lima, R., Lorenzoni, A., dos experimental characterization. Int. J. Refrig. 34, 628–639.
Santos, D., Hoffmann, G., Peixer, G.F., Lozano, J.A., Barbosa Jr., J.R., 2021. Tušek, J., Zupan, S., Šarlah, A., Prebil, I., Poredoš, A., 2010. Development of a rotary
Development and characterization of a calibrated calorimeter to evaluate air magnetic refrigerator. Int. J. Refrig. 33 (2), 294–300.
conditioning systems. In: Proceedings of the 26th ABCM International Congress Vieira, B.P., 2020. Modeling and Optimization of Active Magnetic Regenerators Using
of Mechanical Engineering. Florianópolis, Brazil, pp. COB2021–1992. La-Fe-Si Based Alloys. (Master’s thesis). Universidade Federal de Santa Catarina,
Teza, H.F., Vieira, B., de Sá, N.M., Nakashima, A., Peixer, G.F., Hoffmann, G., Florianópolis, Brazil.
Lozano, J., Barbosa Jr., J.R., 2021. Characterization of the thermo-hydraulic Vieira, B.P., Bez, H.N., Kuepferling, M., Rosa, M.A., Schafer, D., Plá Cid, C.C.,
performance of a La(Fe,Mn,Si)13Hy regenerator. In: Proceedings of the 26th Vieyra, H.A., Basso, V., Lozano, J.A., Barbosa Jr., J.R., 2021. Magnetocaloric
ABCM International Congress of Mechanical Engineering. Florianópolis, Brazil, pp. properties of spheroidal La(Fe,Mn,Si)13 H𝑦 granules and their performance in
COB2021–1683. epoxy-bonded active magnetic regenerators. Appl. Therm. Eng. 183, 116185.
Yu, B., Liu, M., Egolf, P.W., Kitanovski, A., 2010. A review of magnetic refrigerator and
heat pump prototypes built before the year 2010. Int. J. Refrig. 33, 1029–1060.

13