Research Article – JTEN – 2016 – 90

Journal of Thermal Engineering http://eds.yildiz.edu.tr/journal-of-thermal-engineering/Articles

Yildiz Technical University Press, Istanbul, Turkey Manuscript Received May 12, 2016; Accepted June 29, 2016
Vol. 2, No. 6, Special Issue 5, pp. 962-970, December, 2016.

This paper was recommended for publication in revised form by Regional Editor Hafız Muhammed Ali

INVESTIGATION OF THE THERMAL PERFORMANCE OF CRYOGENIC REGENERATOR

AS A POROUS STRUCTURE

Muhammad Ali Kamran

Muhammad Shad Department of Mechanical Engineering
Department of Mechanical Engineering University of Engineering and Technology
University of Engineering and Technology Peshawar, Pakistan
Peshawar, Pakistan

Keywords: cryocooler, regenerator, CFD, hydrodynamic parameters, in-efficiency

* Corresponding author: M. Shad, Phone: +923045754703
E-mail: [email protected]

ABSTRACT INTRODUCTION

An efficient cryocooler is one of the essential requirements for Cryocooler is an integral part of thermal imaging
cooling of Infrared (IR) sensors to low temperatures in high systems used in military operations, which cools down the
resolution night vision systems. A regenerator is an imperative Infrared (IR) sensors to maintain high accuracy of these sensors.
component of cryocooler which has a significant effect on Essentially, the cryocooler performance depends on the
cooling performance of a cryocooler. In this research work a efficiency of a regenerator. A regenerator consists of an array
Computational Fluid Dynamic (CFD) methodology based on of porous matrix material which is exposed to oscillating flow
thermal equilibrium modelling approach has been implemented of a working fluid. Hot and cold fluids pass through the
to analyze the thermal performance of a regenerator under regenerator and exchange heat with matrix material
different design conditions. The regenerator was modeled as a periodically. An efficient regenerator for cryocooler is a
porous media with time varying boundary conditions in the
challenge to design, and has been a focus of special
commercial software package FLUENT by incorporating the
consideration for the last decade or so due to complex flow
effects of temperature dependent physical properties of both,
behavior inside the porous matrix material. Recent studies on
matrix material and working fluid. Simulations were conducted
at different cyclic flow velocities and effect of these variations regenerator design approaches have shown Computational
on temperature swing, pressure drop, in-efficiency, capacity Fluid Dynamics (CFD) methodology to be the best amongst the
ratio and number of heat transfer units of the regenerator have available options, since it allows performance analysis under
been studied for design of better regenerators. The results show periodic flow conditions at the least cost.
that any increment in the velocity of flow, increase the Many cryogenic engineers and scientists around the
temperature swing, pressure drop and in-efficiency while globe have modelled and analyzed regenerator as a porous
decrease the capacity ratio and number of heat transfer units of structure [1-3]. Suzuki and Muralidhar [4] using a local thermal
the regenerator with fixed geometry. In this research work, it is non-equilibrium model investigated pulsating flow inside a
also concluded that the thermal performance of regenerator porous media and indicated that at low Reynolds number,
strongly depend upon the temperature dependent physical effectiveness/efficiency of regenerator increases with an
properties of both matrix material and working fluid. increase of frequency but decreases with increase of frequency
at high Reynolds number. Tao et al. [5] analyzed different
regenerators for pulse tube cryocooler and concluded that
cooling performance of cryocooler enhances by increasing the
specific heat and density of regenerator matrix material but
decreases with an increase of thermal conductivity. Conrad et
al. [6] numerically modeled the regenerator as a two

962
 Research Article

dimensional porous media to measure the hydrodynamic

Figure 1. Regenerator mesh geometry

significance of these changing properties on the performance of

parameters like viscous and inertial resistance for different
a regenerator.
mesh geometries in axial and radial directions for both steady
and oscillatory flow. They obtained the Darcy's permeability The work presented in this paper has used a CFD
assisted approach to analyze the performance of a regenerator
and Forchheimer's inertial coefficients experimentally and
for fixed geometry, operating and hydrodynamics parameters
compared it with CFD simulation results and concluded that the
Darcy's permeability and Forchheimer's inertial coefficients for different mesh porosity and mass flow rates of the working
fluid. Variable properties for both the matrix material and
were usually unaffected to pressure and frequency. Landrum et
al. [7] experimentally investigated the effects of average fluid working fluid were incorporated in the working model. Details
of the numerical methodology is discussed next.
pressure on hydrodynamic parameters across different
regenerator fillers under steady state conditions. In their study,
MATERIAL AND METHOD
hydrodynamic parameters of the porous media were iteratively
adjusted in FLUENT simulation to match with experimental The regenerator was modeled as a porous media using
results. Axial pressure drop at different supply pressures was the commercially available software ANSYS Fluent, which
also measured. A direct correlation between axial pressure solved the volume averaged equations of mass, momentum and
drops and mass flow rates was reported. Nair and Krishnakumar energy detailed below.
[8] numerically investigated heat transfer in a wire meshes
regenerator by using 2-D local thermal equilibrium porous Conservation of mass
media model. Pathak et al. [10] performed experimental and
CFD based study for porous model under steady and oscillatory (ερ) + ∇. (ερv) = 0 (1)
flow conditions to determine the Darcy's permeability and
Forchheimer's coefficients. These coefficients represent the Momentum equation
viscous losses and inertial losses in the flow respectively. The Porous media in Fluent is modelled by merging
work discussed above was directed towards understanding of Darcy’s and Forchheimer’s term in the force expression of the
the hydrodynamic properties of the porous media used in momentum equation.
regenerators. Very little attention was given to the heat transfer
issues which define the performance of a regenerator. Recently,
F=− v + C ρ|v|v (2)
analysis of the regenerator performance for heat transfer has
also attracted the interest of researchers. Costa et al. [12] used
the thermal non-equilibrium porous media model to analyze the The first term on the right hand side of the expression
performance of a Stirling engine regenerator, and concluded is the Darcy’s term, which represents the pressure drop while
that the adopted model can be used with a high confidence level the second expression models Forchheimer’sterm. The inertial
for design of Stirling engine regenerators. resistance factor C and the permeability α for porous media
It is clear from the above that CFD methodology is the required for solving this equation are provided by the user
most popular technique towards analysis of regenerators. depending on the mesh geometry. Values of viscous and inertial
However, few studies have been directed towards heat transfer resistance are 2.35 x 1010 1/m2 and 47000 1/m respectively for
aspects of regenerators. The studies that have been conducted 325 SS mesh size considered in the present work [7].The
have assumed the physical properties to remain constant over volume average momentum equation is:
the working cycle of a regenerator. However, properties like
viscosity, density, specific heat and thermal conductivity of (ερ v) + ∇. (ερvv) = −ε∇ p + ∇. (ετ) + − v +
both the matrix material and working fluid vary with a change
in temperature. Therefore, there is a need to incorporate C ρ|v|v (3)
variable properties in the simulations to determine the effect/

963
 Research Article

Conservation of Energy appropriate DEFINE_PROFILE user defined function (UDF)

was selected from the FLUENT user manual [13], compiled and
The porous media volume averaged energy equation assigned at both ends of the regenerator. A Stirling cryocooler
accounts for the fluid and porous medium interactions. In this generally operates at a frequency of 45Hz, which was used to
research work local thermal equilibrium has been assumed calculate the time period of the sine function employed.
between the porous media and the working Fluid [13]. Temperature of the working fluid at inlet of the regenerator
FLUENT process a single energy equation by considering local during heating period was set at 300K, while inlet temperature
thermal equilibrium between solid and fluid. This assumption for the cooling cycle was kept at 80K. Second order upwind
become true for the conditions in which the variation in discretized governing equations were solved by employing the
temperature between the two phases is not large and no internal Semi Implicit Method for Pressure Linked Equations
heat generated source available. This means that the fluid and (SIMPLE) algorithm with PRESTO discretization scheme for
matrix were assumed to be at the same temperature throughout pressure interpolation. Convergence criteria for the field
the regenerator. Thus a single energy equation was solved by residuals was kept at 10-6. Table 1 summarizes the regenerator
FLUENT. parameters used.

(ερ E + (1 − ε)ρ E ) + ∇. v(ρ E + p) = Table 1. Geometric parameters of Regenerator

∇. K ∇T + (τ . v) (4)
Symbol Definition Value
where K = ε K + (1 − ε) K Th Hot End Temperature 300 K

A 2-D axi-symmetric geometry of the regenerator was Tc Cold End Temperature 80 K

meshed in ICEMCFD (Figure 1). The simulations were L Length of regenerator 0.050 m
conducted for an incompressible viscous working fluid using
pressure based transient solver with absolute velocity D Diameter of regenerator 0.010 m
formulations and considering laminar flow. Helium was used as Ac Cross section area of regenerator 0.000079 m2
the working fluid while porous matrix material was taken to be
stainless steel. Aff Free flow area of regenerator 0.000055 m2
To incorporate the temperature dependent properties Am Matrix material area 0.000024 m2
of the matrix material (thermal conductivity and specific heat)
and working fluid (viscosity, thermal conductivity and specific Ac Convictive heat transfer area 0.134694 m2
heat) in the simulations, National Institute of Standards and dw Diameter of mesh screen wire 0.000035 m
Technology (NIST) [15-16] data was used. This was then fed
into Fluent using the best fit Polynomial curve. As the flow of α Porosity of matrix material 0.7
working fluid in the regenerator is cyclic, the simulations must Dh Hydraulic diameter of regenerator 0.000082 m
be conducted under varying boundary conditions. The heating
(flow of fluid from hot end to cold end) and cooling operations λ Heating/Cooling flow period 0.011 sec
(flow of fluid from cold end to hot end) are periodic in nature, f Operating Frequency 45 Hertz
and form one complete cycle. Both modes are operational for SS Stainless steel matrix material
half the time period of the complete cycle, since the end of
heating period marks the start of the cooling period and vice He Helium working Fluid
versa with an instantaneous switch from heating to cooling. To
realize this, velocity inlet boundary condition was applied at
both ends of the regenerator using the Sine function. An

964
 965

Since the simulations were conducted under varying boundary

conditions, 5 inner iterations per time step were performed for
all the results presented with a time step of 0.1ms. This number
was chosen to be the minimum required to attain ‘convergence’
of the solution and was identified by comparing the simulation
results for different inner iterations. A comparison of
temperature profiles for simulations conducted using 5 and 10
iterations shown in Figure 2 shows that the temperature at inlet
and outlet of the regenerator for both 5 and 10 iterations per
time step remain same at all times. This confirms that increasing
the inner iteration beyond 5 would only result in added
computational overhead. Velocity of the working fluid in the
regenerator was varied in the range of 20m/s to 50m/s. For fixed
regenerator geometry considered in the present work, this
equates to a mass flow rate variation from 0.266g/s to 0.631g/s.
The results obtained from simulation are discussed next.

RESULTS AND DISCUSSION

Before discussing the results, the nomenclature of
terms used in this section are defined for clarity. The heat
transfer performance of the regenerator is most appropriately
expressed by efficiency of regenerator given by [3]

Figure 2. Temperature variation with time at inlet and out of

regenerator for 5 and 10 iterations per time step

Figure 3. Temperature distribution along regenerator for heating and cooling period

965
 966

NTU = (6)
( )

Matrix capacity ratio (CR) measures the thermal

capacity of the matrix material relative to the minimum heat
capacity of working fluid. CR is the ratio of thermal capacity of
the matrix to the minimum flow stream thermal capacity [3].

( ) ( )
CR = = = (7)
( ) ( )

Figure 3 shows the temperature profile along the

regenerator length for heating and cooling period at a fluid
velocity of 50m/s. Recall that the simulations were set such that
the hot end temperature was fixed during the heating period
while the cold end temperature was fixed for the cooling period.
Figure 4. Temperature swing variation with velocity It is evident from the Figure that during heating period, the
temperature of fluid does not reach the desired temperature at
( ) ( )
= = (5) cold end (fixed for the cooling period). Similarly, during
( , ) ( )
reverse flow, the temperature of fluid does not reach the fixed
, , ,

However, in the literature of cryogenics, the temperature at hot end during cooling period. Actually, there is
regenerator performance is more often described in terms of a difference in temperature during the heating and cooling
inefficiency, and the same convention will be used here. period at all axial locations. This difference in temperature is
referred to as the temperature swing of matrix material, and is
Inefficiency = 1 – efficiency undesirable.
Another non dimensional parameter of importance is
the Number of Heat transfer units (NTU), which expresses the
non-dimensional size of the regenerator given by [3].

Figure 5. Temperature contours for heating period

966
 967

Figure 6. Temperature contours for cooling period

Figure 8. Pressure drop variation with mass flow rate

Figure 7. Comparison of Temperature swing

967
 968

Table 2. Computed flow parameters from the simulation results

Values
Parameters Constant
Temperature dependent properties
properties
Velocity (m/s) 20 30 40 50 40
Mass Flowrate (g/s) 0.27 0.38 0.51 0.64 0.51
Reynold's number 138.9 202.81 273.95 352.34 273.95
CR 312.5 217.04 162.67 131.76 -
NTU 320 229.1 205.64 194.21 -
Pressure drop (KPa) 55.4 88.44 144.7 154.9 -
Temperature at inlet (K) 298.6 297.4 296.7 295.5 298.9
Temperature change in cooling period (K) 218.6 217.4 216.7 215.5 218.9
Temperature at outlet (K) 81.3 82.1 84.2 85.5 80.8
Temperature change in heating period (K) 217.8 217.9 215.8 214.5 219.2
In-efficiency (%) cooling period 0.636 1.182 1.500 2.045 0.50
In-efficiency (%) heating period 0.591 0.955 1.909 2.500 0.36
Difference in In-efficiency 0.045 0.227 -0.409 -0.455 0.14
Average In-efficiency (%) 0.614 1.068 1.705 2.273 0.43

The effect of fluid velocity on the temperature swing temperature swing values change considerably when the
is shown in Figure 4, which compares the temperature swing more realistic variable properties model is used. Hence, if
at mid-length of the regenerator. The temperature swing CFD methodology is to be used for designing of regenerators,
increases from 1.787 K to 4.721 K when the velocity is the simulations must consider variations of properties with
increased from 20 m/s to 50 m/s, which shows that the matrix temperature.
material was not able to respond to higher mass flow rates as Figure 8 shows the variation of pressure drop across
efficiently at lower CRs. the regenerator matrix. The pressure drop across the
A further elaboration of this effect has been regenerator increases with increase of mass flowrate
presented in Figures 5 and 6 which compare the temperature (velocity) of fluid. A comparison with Landrum et al. [6]
variations along the regenerator at velocities of 30m/s and shows that the CFD methodology adopted was able to predict
40m/s during heating and cooling periods. A closer look at the pressure drops in the porous structure with a high level of
the temperature contours of heating and cooling period for precision. As mass flowrate varied from 0.3g/s to 0.6g/s,
the same velocity will show that the temperature change of pressure drop across the regenerator increased from 47KPa to
the working fluid is different during heating and cooling 136 KPa.
periods. For instance, at 30 m/s, the fluid undergoes a The results suggest that for fixed regenerator
temperature change of 217.9K (from 300K to 82.1K) while geometry, increasing the velocity of the working fluid
during the cooling period, a change in temperature of 217.4K increases the inefficiency of the regenerator. This is due to a
was observed (from 80K to 297.4K. this shows that the decrease in CR and NTU with increasing mass flow rates.
heating cycle was more efficient than the cooling cycle. However, Figure 9 shows that increase in mass flow rate
However, at 40m/s the heating cycle was less efficient than decreases the CR more significantly compared to the NTU at
the cooling cycle since the temperature changes were noted higher mass flow rates. Thus the inefficiency can be reduced
to be 215.8K and 216.7K in heating and cooling period by adjusting the CRs if high mass flow rates are desired. This
respectively. The cooling cycle continued to be more efficient can be done by adjusting the porosity of the matrix material.
than the heating cycle at higher velocities but the average Since the pressure drop increases across the regenerator with
inefficiency of regenerator is still increasing with higher increase of velocity, decreasing the porosity will result in
velocities. A summary of the temperatures noted at all further pressure drops. The results of simulations for different
velocities tested is presented in Table 2. porosities of matrix material are shown in Figure 10 where
To understand the effect of varying physical the difference between the pressures values at inlet and outlet
properties on the simulation results, a comparison of the is equivalent to the pressure drop for that regenerator. It is
temperature swing for fixed and variable properties at a evident that when the porosity of the regenerator increases,
velocity of 40m/s is provided in Figure 7. It is evident that the the pressure drop decreases. The porosity of regenerator

968
 969

matrix material was changed from 0.5 to 0.9 causing a swing, pressure drop and in-efficiency of regenerator but
reduction in pressure drops from 130.8KPa to 93.9KPa with decreases the CR more significantly compared to the NTU.
40m/s flow velocity. Therefore the inefficiency of regenerator can be reduced by
adjusting the CRs if high mass flow rates are preferred. When
the effects of temperature dependent physical properties of
both matrix material and working fluid are incorporated in
the simulation, the temperature swing and inefficiency of
regenerator significantly increased from the values of
temperature swings and inefficiency of regenerator with
constant physical properties, thus thermal performance of
regenerator also strongly depend upon the temperature
dependent physical properties of both matrix material and
working fluid. The thermal performance of a fixed geometry
regenerator can be augmented by adjusting the velocity/mass
flowrate of working fluid and selection of matrix material.
The computational results acquired from this investigation
can be use with great confidence to design and optimize an
efficient regenerator for cryocooler applications.

NOMENCLATURE
Figure 9. Variation of dimensionless number with mass
flowrate Ac regenerator cross section area(m2)
Am convictive heat transfer area(m2)
C inertial resistance factor
Cm heat capacity of matrix material (J/K)
Cmin minimum heat capacity of working fluid (J/K)
Cpf Specific Heat of Working fluid(J/Kg-K)
Cpm specific heat of matrix material(J/Kg-K)
D regenerator diameter (m)
Dh regenerator hydraulic diameter (m)
dw wire diameter (m)
Ef total fluid energy (J/kg)
Es total solid energy (J/kg)
f frequency (Hz)
h heat transfer coefficient (W/m2 K)

Figure 10. Pressure amplitude variation at inlet and outlet Kf thermal conductivity of working fluid(W/m K)
of regenerator with porosity Ks thermal conductivity of solid matrix (W/m2 K)
L length of regenerator matrix (m)
CONCLUSIONS M mass of regenerator 9Kg)
In this research study, a cryogenic regenerator as a ṁ mass flowrate (Kg/s)
porous structure is simulated in FLUENT under periodic flow
p static pressure (Pa)
conditions for constant and temperature dependent physical
properties of matrix material and working fluid in order to Pr Prandtl number
analyze the thermal performance of regenerator. For the fixed Re Reynolds number
regenerator geometry, the effects of velocity/mass flowrate
rh regenerator hydraulic radius (m)
were investigated on the performance of regenerator. The
results showed that any increment in the velocity/mass T temperaure of working fluid (K)
flowrate of the working fluid lead to increase the temperature t time (s)

969
 970

Tin temperure at inlet (K) [7] T. J. Conrad, E. C. Landrum, S. M. Ghiaasiaan and C.

S. Kirkconnel , "Anisotropic hydrodynamic
Tout temperaure at outlet (K)
parameters of regenerator materials suitable for
Th, in temperaure at inlet during heating period (K) miniature cryocoolers," in International cryocooler
Tc,in temperature at outlet during cooling period (K) conference, 2009.
V inlet/outlet velocity magnitude (m/s) [8] A. R. Nair and K. Krishnakumar, "Numerical
modeling of single blow transient testing of a wire
α permeability (m2)
screen mesh heat exchanger," in 10th National
μ fluid dynamic viscosity (Pa s) conference on technological trends, 2009.
ρf fluid density (kg/m3) [9] R. Gheisari, A. Jafarian and M. A. Ansari, "Analytical
ρs solid density (kg/m3) investigation of compressible oscillating flow in a
porous media: A Second-order successive
ε regenerator matrix porosity
approximation technique," Int. J. Refrigeration, vol.
ρ density of working fluid (kg/m3) 35, pp. 1789 -1799, 2012.
[10] M. G. Pathak, T. I. Mulcahey and S. M. Ghiaasiaan,
gradient operator "Conjugate heat transfer during oscillatory laminar
τ stress tensor (Pa) flow in porous media," International Journal of Heat
λ Heating/cooling period (s) and Mass Transfer, vol. 66, p. 23–30, 2013.
ϵ efficiency of regenerator [11] S. C. Costa, H. Barrutia, J. A. Esnaola and M. Tutar ,
"Numerical study of the heat transfer in wound woven
wire matrix of a Stirling regenerator," Energy
Conversion and Managment, vol. 79, p. 255–64, 2014.
REFERENCES
[12] S. C. Costa , H. Barrutia, J. A. Esnaola and M. Tutar,
[1] W. M. Kays and A. L. London, Compact Heat "The thermal non-equilibrium porous media modelling
Exchangers, 2nd ed., New York: McGraw Hill, 1964. for CFD study of woven wire matrix of a Stirling
[2] P. H. Chen, Z. C. Chang and B. J. Huang, "Effect of regenerator," Energy Conversion and Managment,
oversize in wire screen matrix to the matrix holding vol. 89, p. 473–483, 2015.
tube on regenerator thermal performance," [13] A. FLUENT, " Fluent 6.3 User's Guide," 2006, pp. 7-
Cryogenics, vol. 36, pp. 365-372, 1996. 109.
[3] R. A. Ackermann, Cryogenic Regenerative Heat [14] T. R. Ashwin, G. S. V. L. Narasimham and S. Jacob,
Exchangers, New York: Plenum Press, 1997. "Comparative Numerical Study of Pulse Tube
[4] K. Muralidhar and K. Suzuki, "Analysis of flow and Refrigerators," in International cryocooler
heat transfer in a regenerator mesh using a non-Darcy conference, 2009.
thermal non-equilabrium model," Int J Heat Mass [15] NIST, "Cryogenics Technologies Group.," [Online].
Transfer, vol. 44, pp. 2493-2504, 2001. Available:
[5] Y. B. Tao, Y. W. Liu, F. Gao, X. Y. Chen and Y. L. http://cryogenics.nist.gov/MPropsMAY/material%20
He, "Numerical analysis on pressure drop and heat properties.htm. [Accessed 20 12 2015].
transfer performance of mesh regenerators used in [16] NIST, "Thermophysical Properties of Fluid Systems,"
cryocoolers," Cryogenics, vol. 49, p. 497–503, 2009. [Online]. Available:
[6] E. C. Landrum, T. J. Conrad, S. M. Ghiaasiaan and C. http://webbook.nist.gov/chemistry/fluid/. [Accessed
S. Kirkconnell, "Effect of pressure on hydrodynamic 20 12 2015].
parameters of several PTR regenerator fillers in Axial
steady flow," in International cryocooler conference,
2009.

970