Active Magnetic Regeneration - Magnetocaloric Energy Conversion: From Theory to Applications - page 138

It can be seen from Fig. 4.23 that the maximum cooling power per mass of
magnetocaloric material increases with the frequency of operation for all the ana-
lysed frequencies, except for the Carnot-like AMR cycle, where the peak speci

c
cooling power is obtained at a frequency of around 2 Hz (this is also strongly
related to the AMR s geometry, as explained in Sect. 4.4 ). Between the Brayton-

like and Ericsson-like AMR cycles there is small difference in the cooling power,
whereas the Carnot-like AMR cycle provides much less cooling power. The highest
COP can be obtained with an Ericsson-like cycle, and the lowest COP can be
obtained with a Carnot-like AMR cycle.
It is important to note that the permanent magnet assembly (in addition to the

'

fl nes the type of thermodynamic cycle that can be performed.

The mass of the magnet assembly for the Carnot-like cycle can be much smaller
than the mass of the magnet assembly for the Brayton-like or Ericsson-like cycle
(see [ 10 ] for details). Since the magnet assembly represents the major costs of a
magnetic refrigerator [ 15 ], it also makes sense to represent results as a function of
the magnet mass, as was presented by Kitanovski et al. [ 10 ].
Based on the results of the

uid fl ow regime) de

rst analysis, we can conclude that the considered
Carnot-like AMR cycle, compared to the Brayton-like or the Ericsson-like cycles,
operates with a much lower speci

ciency. This is mostly
due to the increased irreversible heat transfer losses related to the non-ideal
regeneration between the neighbouring material particles in the regenerator. Small
increments in the material

c cooling power and ef

s temperature due to quasi-isothermal magnetization will
also require a very small

'

temperature difference between the magnetocaloric

material and the working fl uid, which are unfortunately strongly limited by the heat

transfer coef cient and the available heat transfer area. Therefore, the selection of

an appropriate cycle can drastically in

uence the regeneration process.
In both the Carnot-like and Ericsson-like AMR cycles, the performance is not as
high as would be expected from just studying the T

fl

s diagram. Therefore, the

-

application of a T s diagram is not the right method to study the AMR cycle ' s

-

performance, especially because it does not account for the AMR s regeneration

process and the corresponding heat transfer losses between the working

'

fl uid and

the magnetocaloric material.
In the second analysis, which was performed by Plaznik et al. [ 9 ], three different
thermodynamic cycles were analysed: the Brayton-like, the Ericsson-like and the
Hybrid Brayton Ericsson-like AMR cycles. The aim of the investigation was the

-

same as in the rst analysis, i.e. to investigate the performance of a magnetic

cooling device with an AMR. However, in this particular case, the in fl uence of the

magnetization pro ned by the magnet assembly) on the performance of the

AMR was investigated. Furthermore, two different types of AMRs were evaluated
in the simulation (Fig. 4.24 ). The parallel-plate and packed-bed AMR, respectively.
In both AMRs, gadolinium was used as the magnetocaloric material and water was
considered as the working

le (de

fl uid. In all cases, the magnetic eld change was con-

sidered to be 1 T.