(9.6.08)Temadag om Kølemetoder
Recent Developments in Room Temperature Active Magnetic Regenerative
Refrigeration
Kurt Engelbrecht, Nini Pryds, Christian Bahl
Risø DTU
(9.6.08)Temadag om Kølemetoder
Magnetic Refrigeration Background
• Alternative to vapor compression
• Magnetic refrigeration used to first break 1K temperature barrier near 1935, but used a one shot technique with a very low temperature span– Giauque awarded the Nobel Prize in chemistry in 1949
• First regenerative magnetic refrigerator proof-of-concept device built by Brown in 1976
• Possible applications include residential/commercial space cooling, refrigeration, hydrogen liquefaction/storage
(9.6.08)Temadag om Kølemetoder
Advantages of Magnetic Refrigeration
• Environmentally friendly– no ozone depleting or direct global warming associated with
the refrigerant
• Possibly more efficient than current technologies– no superheat or throttling loss– no compressor losses– nominally equivalent heat exchanger loss– relatively easy to achieve efficient part load operation
• Quiet operation
(9.6.08)Temadag om Kølemetoder
Room Temperature Magnetic Materials
The magnetocaloric effect is greatest at the Curie point of the magnetic material, which can be adjusted by alloying
200 225 250 275 300 325 3500
2
4
6
8
10
12
Temperature (K)
Adi
abat
ic T
empe
ratu
re C
hang
e (K
)0-5 Tesla
0-2 Tesla
Gadolinium (Gd) and Erbium (Er) alloy
(9.6.08)Temadag om Kølemetoder
Magnetic Refrigeration Concept
(9.6.08)Temadag om Kølemetoder
MagnetizationHot-to-Cold FlowCold-to-Hot FlowDemagnetization
Cold
ReservoirHot
Reservoir
Heat Rejection
Refrigeration
Active Magnetic Regenerative Refrigerator Cycle (AMRR)
(9.6.08)Temadag om Kølemetoder
Magnetic Refrigerator Design Considerations
• System configuration – rotary, reciprocating, etc.
• Magnetocaloric material selection
• Regenerator design
• Magnetic field
• Heat transfer fluid selection
(9.6.08)Temadag om Kølemetoder
AMRR System Configurations
• Stationary bed with an electromagnet– magnetic field is controlled by varying the current
to the electromagnet
QH QC
magnet
regenerator
displacer
(9.6.08)Temadag om Kølemetoder
AMRR System Configurations
• Reciprocating systemmagnet
regenerator
displacer
(9.6.08)Temadag om Kølemetoder
AMRR System Configurations
• Rotary system magnet
regenerator
pump
(9.6.08)Temadag om Kølemetoder
Permanent
Magnet
bed rotation
Magnetization in a Rotary Bed
Magnetic field “sweeps”through bed.
magnetized region
zero magnetic field
(9.6.08)Temadag om Kølemetoder
Choosing Magnetocaloric Materials
• Adiabatic temperature change
• Isothermal entropy change
• Thermal conductivity
• Magnetic hysteresis
• Practical issues – cost, corrosion, toxicity, manufacturability
(9.6.08)Temadag om Kølemetoder
Magnetocaloric Material Properties
• Magnetic materials generally experience a phase change from a ferromagnet to a paramagnet at the Curie Point
• Second-order magnetic phase transitions (SOMT)– transition associated with the alignment of magnetic
moments in the material– effect is nearly instantaneous (order of nanoseconds)– generally reversible magnetization
• First order-magnetic phase transitions (FOMT)– latent heat and large magnetocaloric effect associated with
transition– higher irreversibilities than second order– longer time associated with transition
(9.6.08)Temadag om Kølemetoder
Summary of Magnetocaloric Material Properties
13 (0-1.45
Tesla)30FOMT287x=0.1
6532FOMT318x=0
MnAs1-xSbx
17.128FOMT287LaFe11.7Si1.3H1.1
2727FOMT269Gd5Si3.5Ge0.5
~05.55.8SOMT294Gd
(K)(K)(J/kg-K)(K)HysteresisΔTad-ΔsMTypeTCurieMaterial
(9.6.08)Temadag om Kølemetoder
AMRR Loss Mechanisms
• Heat transfer / regeneration losses
• Pumping / viscous dissipation losses
• Axial conduction
• Eddy current heating
• Heat exchanger and motor losses– nominally equivalent to vapor compression
(9.6.08)Temadag om Kølemetoder
Regenerator Design
• Packed regenerator– packed sphere, packed powder, etc.– relatively easy to achieve high surface area– fluid flow profile is generally well-distributed– high pressure drop
• Flat plate regenerator– low pressure drop– requires small dimensions to achieve high heat
transfer performance– theoretically best regenerator performance– performance highly dependent on geometry
(9.6.08)Temadag om Kølemetoder
Prototype AMRR Performance Summary
540GdSiGeGa
100Gd5Si2Ge2
spheres (0.2mm)230Gd~200 1.4 (P)Nanjing University
spheres (0.25-0.355mm)250Gd & GdEr
spheres (0.25-0.355mm)1427Gd & GdEr
spheres (0.43-0.5mm)1415Gd331.5 (P)
0600
spheres (0.15-0.3mm)38100Gd~6005 (S)Astronautics
24402 (S)
spheres (0.3 mm)26100Gd4844 (S)Chubu Electri / Tokyo University
47Gd252 (S)(2004)
spheres (0.2mm)200Gd492 (S)
crushed part. (0.4mm)500Gd, GdTb & GdEr742 (S)University of Victoria (2006)
KWcm3Tesla
Regen typedTQCRegen materialVregenμoHMaxSystem
(9.6.08)Temadag om Kølemetoder
Promising AMRR Innovations
• High performance magnetocaloric materials with tunable Curie Temperatures
• Layered regenerators
• Rotary systems
(9.6.08)Temadag om Kølemetoder
Efficiency Dependence on Regenerator Volume –Packed Sphere Regenerator8.76 kW Space Conditioning Application
5 10 15 20 25 30 352.75
3
3.25
3.5
3.75
4
4.25
4.5
4.75
Regenerator volume, L
Coe
ffici
ent o
f per
form
ance
layered bed
non-layered bed
vapor compression
B
A
(9.6.08)Temadag om Kølemetoder
Practical AMRR Concerns
• High fluid mass flow– pressure lower than vapor compression systems
• System size
• Materials processing
• Material oxidation
• Cost
(9.6.08)Temadag om Kølemetoder
AMRR Design Consideration
,~ v refAMRR
vc f mc
hmm c T
ΔΔ
&
&
• The mass flow rate of fluid in an AMRR is much higher than an equivalent vapor compression system
• no phase change in the fluid
• For a practical system, the fluid flow rate is ~20 times the refrigerant flow rate for vapor comp
• Requires larger connecting piping to reduce pressure drop
for equal refrigeration capacity
(9.6.08)Temadag om Kølemetoder
Magnetic Refrigeration Research at Risø DTU
Partnership between:
Duration: 4 yearsEnding date: 31.12.2010
Funding: 20 Mkr5 Ph.D. students, 3 Postdocs
ChallengesDemonstrate cost-effective systems at commercially relevant temperature spans with high efficiency and environmentally friendly materials
(9.6.08)Temadag om Kølemetoder
Test Machine at Risø DTU
• Parallel plate regenerator of Gd– plate thickness 0,9mm
• Maximum magnetic field ~1.1 Tesla
• Uses a permanent magnet
• Temperature span of 9 K
• Operational since Nov 2007
(9.6.08)Temadag om Kølemetoder
Future Work at Risø
• Currently designing next generation prototype AMRR
• Project goal: 100W cooling at 40K temperature span
• System should be practical and cost effective
• Prototype will be built in 2009
(9.6.08)Temadag om Kølemetoder
Conclusions
• Improvements have been made to prototype AMRR systems recently, but the technology is still not mature
• Systems using a layered regenerator beds have been shown higher performance over a relatively large temperature span
• New magnetocaloric materials have potential to significantly improve AMRR performance
• Gd remains
(9.6.08)Temadag om Kølemetoder
Modeling Efforts
• Developed a 1D AMRR model– used model to determine limits of AMRR technology
and investigate control techniquesK. L. Engelbrecht, G. F. Nellis, and S. A. Klein, 2006, Predicting the Performance
of an Active Magnetic Regenerator Refrigerator used for Space Cooling and Refrigeration, HVAC&R Journal 12(4):1077-1095
K. L. Engelbrecht, G. F. Nellis, and S. A. Klein, 2006, Modeling the Transient Behavior of an Active Magnetic Regenerative Refrigerator, Proceedings of the 11th International Refrigeration and Air Conditioning Conference at Purdue,
West Lafayette, IN
• Used the model to predict the performance of a rotary AMRR prototype
(9.6.08)Temadag om Kølemetoder
Model Verification – Rotary Input Data
Modeling parameters - geometry
Parameter Value Parameter Valueregenerator type packed sphere maximum applied field 1.5 Teslaregenerator material Gd sphere size for packing 0.425-0.5 mmregenerator beds 6 porosity 0.362regenerator height 6.6 mm motor efficiency N/Sregenerator width 15.9 mm dwell ratio 1/3regenerator length 2.025 in magnet arc 120°
regenerator volume 33 cm3 heat transfer fluid90% water/10% ethylene glycol
(9.6.08)Temadag om Kølemetoder
Uncertainty in Model Inputs
x
( )m t&
heat transfer fluid properties:
cold end, fluid enters at TC
magnetic regenerator material properties:
( ) ( ) ( ) ( ), , , , , , , ,rr o r o r
o T
sk T x T x H c T x H xH
μ μ ρμ
⎛ ⎞∂⎜ ⎟∂⎝ ⎠
( ) ( ) ( ), , ,f f f fc T k T Tρ μ
( ),oH x tμ
hot end, fluid enters at TH
Tf (x,t)
Tr (x,t)
regenerator bed configurationAc, L, as, dh, ε, Nu(Re, Pr), f(Re), keff(Re, Pr,..)
(9.6.08)Temadag om Kølemetoder
Model Results for New Nusselt Number and Reduced Magnetic Field
(9.6.08)Temadag om Kølemetoder
Modeling Error vs. Temperature Span
(9.6.08)Temadag om Kølemetoder
Important AMRR Design/Modeling Considerations
• Effective magnetic field in the regenerator material is different than the field in the gap of the magnet
– suggests that the permeability of prospective magnetocaloric materials should be considered along with other properties
• Heat transfer in the regenerator – model agreement improved using a Nu correlation developed for a
liquid rather than primarily for a gas
(9.6.08)Temadag om Kølemetoder
FLUID PUMP
COLD HX
MAGNETOCALORICWHEEL
PERMANENTMAGNET
DRIVE MOTOR
Example of a Prototype Rotary AMRR
(9.6.08)Temadag om Kølemetoder
Regenerator Flow Unbalance
• for some flow configurations, it is possible that flow through each regenerator may be dependent on flow direction
fluid flow direction
regenerator bed
(9.6.08)Temadag om Kølemetoder
Model Sensitivity to Flow Unbalance
5.3633.7739.130.67417.48
1.4320.9322.360.34411.87
6.0538.0444.090.67111.56
3.1221.4924.610.35113.15
0.1467.6667.800.8140.24
0.6534.7335.380.3540.83
1.9451.2653.200.6710.32
1.0634.6635.720.3410.21
WWWLiter/minHzK
Cooling Power Loss
5% Unbalance
No Unbalance
Fluid Flow Rate
Cycle FrequencyΔTCase
(9.6.08)Temadag om Kølemetoder
Thermal Properties of the Regenerator Housing
• The regenerator housing is made of G-10, k ~ 0.8 W/m-K, α ~ 4e-7
6.6mm
15.9mm
(9.6.08)Temadag om Kølemetoder
Reduced Magnetic Field in the Regenerator
• in a magnetically permeable object, the magnetic field will be less than the magnetic field in free space
– actual magnetic field depends on the shape of the material and the permeability as a function of magnetic field
• For a sphere of constant permeability in a uniform magnetic field the magnetic field in the material is
00 0
0
3
2effH Hμ μ μ
μ
=+
permeability of regenerator material
• It is important to determine the actual effective magnetic fieldin the regenerator
(9.6.08)Temadag om Kølemetoder
Reduction in Cooling Power at Reduced Heat Transfer
25.4%39.10.674.017.48
17.1%22.40.344.011.87
19.6%44.10.671.011.56
10.8%24.60.351.013.15
2.2%67.80.814.00.248.3%35.40.354.00.830.3%53.20.671.00.321.0%35.70.341.00.21
WLiter/minHzK
Cooling Power Loss
Predicted Cooling Power
Fluid Flow Rate
Cycle FrequencyΔTCase
(9.6.08)Temadag om Kølemetoder
Determining Actual Magnetic Field
(9.6.08)Temadag om Kølemetoder
Governing Equations
( ) ( ) ( ) ( ) ( ) ( )
2
2
f f f ff f f s c f r f f f c f
h
rdisp c
Nu Re ,Pr k Tm t c T T a A T T c T A T
x d tenthalpy flow capacity ofheat transfer to
entrained fluidregenerator material
Tk Ax
axialdispersion
ρ ε∂ ∂⎡ ⎤ ⎡ ⎤+ − +⎣ ⎦ ⎣ ⎦∂ ∂
∂−
∂
&144424443 144424443144444424444443
1 2
( )f
m tpx
viscousdissipation
ρ∂
=∂
&
4 43 14243
( ) ( ) ( ) ( ) ( )2
21 1f f f f r rs c f r c o eff c r c
h
Nu Re ,Pr k T M T ua A T T A H k A Ad t x t
axialmagnetic work energy storedheat transfer from fluid conductionin matrix
ε μ ρ ε∂ ∂ ∂− + − + = −
∂ ∂ ∂144424443 14243 1442443144444424444443
Fluid:
Regenerator:
(9.6.08)Temadag om Kølemetoder
Modeling Efforts
• Developed a 1D AMRR model flexible design tool capable of predicting the performance of an AMRR system
• Verify the model experimentally
• Make the model publicly available
(9.6.08)Temadag om Kølemetoder
Representative Experimental Cases
xxx8
xxx7
xxx6
xxx5
xxx4
xxx3
xxx2
xxx1
highlow highlowhighlow
Fluid Flow Rate Cycle FrequencyTemperature SpanCase
(9.6.08)Temadag om Kølemetoder
Model Sensitivity to Heat Transfer Coefficient in the Regenerator
(9.6.08)Temadag om Kølemetoder
Modeling Experimental Data – Cycle Configuration
(9.6.08)Temadag om Kølemetoder
bed rotation
mass flow inmass flow out
Cold to Hot Flow in a Rotary Bed
flow activated as bed moves overcold check-valve
flow activated as bed moves overhot check-valve
(9.6.08)Temadag om Kølemetoder
Prototype AMRR Performance
540GdSiGeGa
100Gd5Si2Ge2
spheres (0.2mm)230Gd~200 1.4 (P)Nanjing University
spheres (0.25-0.355mm)250Gd & GdEr
spheres (0.25-0.355mm)1427Gd & GdEr
spheres (0.43-0.5mm)1415Gd331.5 (P)Astronautics
0600
spheres (0.15-0.3mm)38100Gd~6005 (S)Astronautics
24402 (S)Toshiba
spheres (0.3 mm)26100Gd4844 (S)Chubu Electric/
47Gd252 (S)
spheres (0.2mm)200Gd492 (S)Victoria
crushed part. (0.4mm)500Gd, GdTb & GdEr742 (S)University of
KWcm3Tesla
Regen typedTQCRegen materialVregenμoHMaxSystem
(9.6.08)Temadag om Kølemetoder
1 1 1
Efficiency Dependence on Aspect Ratio
Volume of magnetic material is held constant while the length and diameter of the regenerator bed are varied
Axial conduction losses begin to dominate as L/d
decreases
regenerator
bed
Pump losses begin to dominate as L/d increases
flow
flow
(9.6.08)Temadag om Kølemetoder
Numerical Verification: Schumann Solution
• Solid and fluid in a regenerator are initially at a uniform temperature of 0 K and fluid at temperature TH enters at t=0
• Analytical solution for transient temperature profiles of the fluid and solid exists
• Shitzer, A. and M. Levy, 1983, Transient Behavior of a Rock-Bed Thermal Storage SytemSubjected to Varibable Inlet Air Temperatures: Analysis and Experimentation, Journal of Solar Engineering, vol 105: p. 200-206
• Solution assumes constant material properties, no axial conduction, and uniform heat transfer between fluid and solid
• The temperature profiles predicted by the numerical model can beverified against this analytical solution
(9.6.08)Temadag om Kølemetoder
fluid temp
Model Verification – Single Blow
NTU=50
regenerator temp