International Journal of Thermal Sciences 50 (2011) 206e213

Contents lists available at ScienceDirect

International Journal of Thermal Sciences

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

Modied theoretical model for thermoacoustic couples

Hadi Babaei, Kamran Siddiqui*
Department of Mechanical & Materials Engineering, The University of Western Ontario, London, Canada N6A 5B9

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 28 April 2010
Received in revised form
14 September 2010
Accepted 19 September 2010
Available online 2 November 2010

The theoretical model of thermoacoustic couples is modied to incorporate more realistic physical
processes that are acoustic dissipation within the stack, and the heat exchange between the stack and its
surroundings, which were not considered in the previous models. The results show good agreement
between the present and previous models at high thermal conductivity of the stack. However, at low
stack thermal conductivity, the previous models not only signicantly overestimate stack temperature
difference (DTs) but at certain stack positions they predict values of DTs greater than the theoretically
possible values. The comparison of DTs values predicted by the present and previous theoretical models
with the experimental results shows a good agreement between the present model and the experiments
while, the previous models signicantly overpredict DTs for the given range of drive ratios. It is shown
that the incorporation of realistic physical processes signicantly improves the accuracy of the theoretical model for the thermoacoustic couples.
2010 Elsevier Masson SAS. All rights reserved.

Keywords:
Thermoacoustics
Thermoacoustic couples
Stack temperature difference

1. Introduction
Heat can be transferred from a cold reservoir to a hot reservoir
by utilizing sound energy. This technology is called thermoacoustic
refrigeration and it has introduced a new vision towards environmentally-friendly refrigeration systems. Thermoacoustic refrigerators, which convert sound energy into heat energy use noble gases
as the refrigerant and thus, eliminate the use of conventional
harmful refrigerants. Furthermore, the inherent simplicity, fewer
components and dimensional exibility make thermoacoustic
refrigerators more appealing as the new generation of sustainable
refrigerators. Development of the thermoacoustic linear theory
played a major role in the recent advancement of this technology
[1,2]. However, in several cases, disagreement between the
predictions of the linear theory and experimental results have been
reported which could be due to the simplied assumptions made in
the theory. Nevertheless, this theoretical model is the only available
tool to design thermoacoustic devices.
A thermoacoustic refrigerator consists of a resonator tube,
a stack, two heat exchangers and an acoustic source (e.g. a loudspeaker) to excite a standing wave inside the tube. Several different
congurations of the stack (which is the heart of a thermoacoustic
device) have been proposed and/or used such as parallel plates, pin
array or a porous medium. The two heat exchangers are attached to
either sides of the stack and maintain the desired temperature

* Corresponding author.
E-mail address: ksiddiqui@eng.uwo.ca (K. Siddiqui).
1290-0729/$ e see front matter 2010 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ijthermalsci.2010.09.011

gradient along the stack by transferring heat to and from the

thermal reservoirs located outside the resonator tube. The stack of
a thermoacoustic refrigerator in the absence of heat exchangers is
called a thermoacoustic couple.
Thermoacoustic couples are used for the study of fundamental
thermoacoustic processes [3,4]. Schematic of a thermoacoustic
couple inside a resonator tube is shown in Fig. 1. A half-wavelength
acoustic standing wave generated by the acoustic source excites the
working gas parcels within the resonator tube. Due to the interaction of the gas parcels with the solid surface of the stack, temperature gradient is established along the stack. The stack (or couple)
temperature difference, herein after referred to as DTs, is dened as
the steady-state temperature difference between the hot-end of the
stack (located close to the pressure antinode) and the cold-end of
the stack (located away from the pressure antinode). The resonator
tube section between the acoustic source and the cold-end of the
stack is referred to as the cold-side ambient duct whereas, the
resonator tube section between the hot-end of the stack and closed
end is referred to as the hot-side ambient duct (see Fig. 1). Improved
understanding of the fundamental thermoacoustic process occurring in thermoacoustic couples is very crucial for the design and
development of efcient thermoacoustic devices which have a great
potential to serve as sustainable energy sources.
Thermoacoustic couple was rst introduced by Wheatley et al.
[3] in 1983. By applying the thermoacoustic linear theory, they
proposed a procedure and nally an equation to estimate DTs. In
their formulation, they neglected the change in the work ow of
the stack by assuming a short stack. They also assumed that the
thermal conductivity of the stack is high enough that the heat

H. Babaei, K. Siddiqui / International Journal of Thermal Sciences 50 (2011) 206e213

Nomenclature
A
a
BR
cp
DR
E_ 2

DE_ 2
f
H_ 2
K
k
L
l
Pm
PA
p1
Pr
Q_
Q_
2

rh
s1
T
DT
VT
U1
u1
x

cross-sectional area (m2)

speed of sound (m/s)
blockage ratio of the stack, dened as the ratio of the
working gas volume to the total volume
isobaric heat capacity of the working gas (J kg
1 K
1)
drive ratio, dened as the ratio of the antinode
pressure amplitude to the mean pressure
work ux (W)
consumed (dissipated) work ux (W)
resonant frequency (Hz)
total energy ux (W)
thermal conductivity (W m
1 K
1)
wave number (m
1)
length (m)
half of the stack plate thickness (m)
mean pressure (Pa)
antinode pressure amplitude (Pa)
pressure amplitude (Pa)
Prandtl number
heat ux (W)
thermoacoustic heat ux (W)
hydraulic radius (m)
amplitude of the entropy per unit mass (J kg
1 K
1)
temperature (K)
temperature difference (K)
temperature gradient (K/m)
volume ow rate (m3/s)
velocity amplitude (m/s)
position, coordinate along sound-propagation
direction (m)

transferred from the cold-end of the stack to the hot-end of the

stack by the thermoacoustic process, returns via the stack plates
through conduction, and there is no heat exchange between the
stack and its surroundings i.e. they assumed that the stack or the
couple is thermally isolated.
Atchley et al. [4] slightly modied the theoretical relation of
Wheatley et al. [3] by considering the conduction back through
both the stack plates and the working gas. They also conducted
experiments to measure DTs using three different composite stacks
each made of ve plates. The stack length varied from 6.8 to
10.7 mm and the thermal conductivity varied from 0.48 to
5.8 W/m K. at a range of drive ratios from 0.17% to 1.99%. The
temperature was measured using a thermopile. They compared
experimental and theoretical values of DTs for the stack with the
thermal conductivity of 5.8 W/m K for the given drive ratios. A good
agreement was observed between the experimental results and
theoretical predictions for drive ratios less than 0.4% and the
discrepancy started at higher drive ratios. At the drive ratio of 1.99%,

Fig. 1. Schematic of a thermoacoustic couple inside a resonator tube.

x1
xc
y0

b
dk
dv
g
G
l
r
u

207

gas displacement amplitude (m)

stack center position (m)
half of the plate spacing (m)
thermal expansion coefcient (K
1)
thermal penetration depth (m)
viscous penetration depth (m)
ratio, isobaric to isochoric specic heats
normalized temperature gradient
wavelength (m)
density (kg m
3)
angular frequency (rad s
1)

Superscripts
w
complex conjugation
Subscripts
a
ambient
c
cold
ce
cold-end
con
consumed
cond
conduction
crit
critical
d
duct
g
gas
h
hot
he
hot-end
m
mean
max
maximum
n
normalized
s
stack
t
total

theoretical values up to 100% larger than experimental values were

observed.
Piccolo and Cannistraro [5] performed experimental investigations on a thermoacoustic couple. They used ve polyethylene
plates as the stack with the length of 7 cm. The thermal conductivity of the plates was about 0.11 W/m K. They measured DTs by
using two miniature thermo-resistances inserted at the centers of
stack ends. They compared their experimental results with the
values of DTs predicted by the theoretical model of Atchley et al. [4]
at three drive ratios of 0.15%, 0.27% and 0.38%. They observed good
agreement between the experimental and theoretical results at the
driver ratio of 0.15% which started to diminish with an increase in
the drive ratio. At the drive ratio of 0.38%, theoretical values up to
200% greater than experimental ones were observed.
Worlikar et al. [6] developed a 2D numerical model to predict
the steady-state temperature difference developed along a thermoacoustic couple. They used a numerical solver to incorporate the
energy equation in their calculations. Two parallel plates having the
desired thickness were considered as the stack. The boundary of
the computational domain was dened to enclose half thickness of
the plates. The surfaces of the boundary perpendicular to the
acoustic wave propagation were assumed to impose the conditions
of the idealized standing wave. Periodic conditions were applied on
the surfaces parallel to the wave propagation to avoid the formation
of mean ow. The thickness, length and plate spacing of the stack
were assumed to be 0.1905, 6.85 and 1.338 mm, respectively. The
thermal conductivity of the stack was assumed to be about
5.7 W/m K (a high thermal conductivity stack). They compared the
prediction of their numerical simulation with the experimental

208

H. Babaei, K. Siddiqui / International Journal of Thermal Sciences 50 (2011) 206e213

results of Atchley et al. [4]. They reported good agreement between

their numerical results and experimental measurements [4] as well
as linear theory at low drive ratios. At moderate drive ratios, their
results showed good agreement with experiments [4] but deviated
from linear theory. For the low range of drive ratios, they also found
linear variation of temperature within the stack as predicted by
linear theory. However, for moderate range of drive ratios, they
predicted non-uniform temperature prole deviated from the
linear theory. They also showed that the heat exchange between
the gas and solid mainly occurs at the ends of the stack.
Cao et al. [7] conducted a two-dimensional numerical study to
investigate thermoacoustic phenomenon beyond Rotts [8,9] onedimensional analysis which assumed no heat exchange between gas
and solid. They numerically solved the compressible NaviereStokes,
continuity, energy and state equations to simulate and analyze the
total energy ux density near and within a parallel-plate thermoacoustic couple. They dened a numerical domain which
enclosed half of the plate spacing between two parallel plates with
one surface of a plate inside the domain. They considered zero
longitudinal temperature gradient along the solid plate (DTs / 0)
by assuming innite thermal conductivity (Ks / N) and heat
capacity of the plate. They showed that the energy ows out and in
at the cold-end and hot-end of the plate, respectively, in the direction perpendicular to the acoustic wave propagation. They observed
that this energy ow which is due to heat conduction was sharply
peaked at the extreme ends of the plate. They did not provide any
physical explanation for this behavior. It was showed that this
energy exchange occurred within a region whose length was
comparable to the gas displacement amplitude. Like Hoer [10],
they concluded that the heat exchangers in thermoacoustic devices
should be shorter than gas displacement amplitude. They also
showed that the energy is transported mainly along the acoustic
wave propagation and the heat exchange between gas and solid is
negligible in the central part of the plate. They showed that more
energy ows in the domain through the vertical boundary close to
the cold-end of the plate than ows out of the domain through the
vertical boundary close to the hot-end of the plate. They attributed
this trend to the energy dissipation due to viscous and thermal
effects near the plate.
Mozurkewich [11] considered a pore of a thermally isolated thermoacoustic stack and analytically calculated the three-dimensional
time-averaged temperature distribution within the pore. They
considered both innite and nite thermal conductivities for the solid
plate. In the former case (Ks / N), the temperature gradient along the
solid is zero (DTs / 0). In the case of the nite thermal conductivity,
a temperature gradient was developed along the solid plate (DTs s 0).
They demonstrated that the heat pumped by the thermoacoustic effect
reduces in the presence of a temperature gradient (nite thermal
conductivity). They predicted a linear prole for the axial time-average
temperatures of gas and solid in the central region of the pore.
Assuming a low thermal conductivity solid wall, they argued that the
solid plate temperature gradient approaches zero over a distance
smaller than the displacement amplitude and gas temperature
gradient approaches a constant value which is about the critical
temperature gradient over a distance comparable to the displacement
amplitude.
Mozurkewich [12] developed a one-dimensional model to
predict the transverse heat exchange in a thermoacoustic pore.
They reported some new features by considering a parallel-pore
heat exchanger located next to one end of the stack. They assumed
no direct heat conduction between the walls of the heat exchanger
and stack. Finite and innite thermal conductivity was assumed for
the stack and heat exchanger, respectively. They found that the gas
temperature can be nonuniform within the heat exchanger. They
also demonstrated that there is a net heat exchange between the

heat exchanger and the stack even when no temperature difference

is assumed between the heat exchanger and the adjacent wall of
the stack.
Ishikawa and Mee [13] applied a 2D NaviereStokes solver to
model the energy transfer in a thermoacoustic couple maintained
at a constant temperature. They demonstrated that heat transfer
occurs over a distance approximately equal to the displacement
amplitude from the ends of the couple. They showed that when the
plate spacing is in the order of thermal penetration depth or bigger,
heat transfer occurs even in the shortest examined stack whose
length was comparable to the acoustic displacement amplitude.
Zoontjens et al. [14] developed a numerical model that takes
into account the inuence of the thickness of the solid plate in
thermoacoustic couples. They found that the ow structure and
distribution of energy density at the stack ends is highly inuenced
by the thickness of the stack plates.
Fundamental understanding of the thermoacoustic phenomenon is essential to improve the thermoacoustic model. Thermoacoustic couple can be considered as a primary thermoacoustic
device where the thermoacoustic phenomenon can be examined
[3]. As the above literature review shows, there is a large
disagreement between the experimental and theoretical values of
DTs, which is observed at drive ratios as low as 0.4%. In all these
cases, theoretical values overestimated DTs compared to the
experimentally measured values. This discrepancy could be due to
the simplistic assumptions made in the previous studies. For
example, in the previous models, it was assumed that the change in
stack work ow is small and thus ignored [3,4]. The change in the
stack work ow implies acoustic dissipation within the stack,
which was neglected in the previous theoretical models for thermoacoustic couples. All real thermoacoustic systems have acoustic
dissipation within the stack. Similarly, the stack was considered to
be thermally isolated. That is, the heat transfer between the stack
and its surroundings was neglected.
This was based on the assumption that the thermal conductivity
of the stack is large enough that the heat transferred by the thermoacoustic process is transferred back via conduction through the
stack plates [3] or conduction through the stack plates and the
working gas [4]. However, it has been recommended in the literature to keep the thermal conductivity of the stack plate lower for
the higher performance of the stack [15]. Thus, the assumption of
isolated stack is not valid in this situation. The present study is
aimed at improving the previous theoretical models by incorporating more realistic assumptions.

2. Theoretical background
By linearizing momentum, continuity and energy ow equations, simplied linear model of thermoacoustic was developed
[2,6]. Consider a stack of parallel plates with x axis along the
direction of the acoustic wave propagation and y axis perpendicular
to the plane of the stack. Let the plate thickness be equal to 2l and
the plate spacing be equal to 2y0. The existing thermoacoustic
model is based on the following assumptions [2,6].
The half spacing between the stack plates is assumed to be
greater than the thermal penetration depth (dk). This assumption is
called the boundary-layer approximation which is a good approximation in thermoacoustic devices where typically, dk  y0  2dk
[2,6].
The length of the stack is considered to be signicantly less than
the wavelength of the acoustic wave (i.e. Ls  l). It is called the
short stack approximation. With this approximation and assuming
standing wave phasing between pressure and velocity, the velocity
and pressure can be expressed as [2,6],

H. Babaei, K. Siddiqui / International Journal of Thermal Sciences 50 (2011) 206e213

p1 PA coskx;

u1

l
y0




PA
sinkx
rm a

(1)

Finally, it is assumed that the stack is short enough that p1 and

u1 could be regarded as independent of x along the stack and the
temperature difference along the stack is less than the mean
temperature of the stack (i.e. DT  Tm). So the thermophysical
properties of the gas are assumed to be independent of x within the
stack. Thus, p1, u1 and thermophysical properties are evaluated at
the stack midpoint [2,6].
The total power ow (i.e. total energy ux) through the stack
H_ 2;s , and the acoustic power (work ux) consumed by the
refrigerator stack DE_ 2;s are expressed in the simplied linear form
by the following equations [2,6].

#
"
p


p dv 
Ag dk jp1 jju1 j
1 Pr Pr
_
p
1 Pr

G
H2;s
4 rh 1 PrL
y0
1 Pr



Ag Kg Asolid Ksolid VT
(2)

DE_ 2;s

Ag Ls

4 rh

"

g
1 p1 j2 dk u
G

p
1
gpm
1 Pr L

rm ju1 j2 dv u

(3)

L
2

where, L 1
dv =rh dv =2rh2 , and G is dened as the ratio of

the temperature gradient along the stack (VT) to the critical
temperature gradient (VTcrit). The critical temperature gradient is
dened as the stack temperature gradient when the temperature
change of the gas parcel due to adiabatic compression or expansion
is the same as the temperature change along the stack plate, and is
expressed as [2],

VTcrit

Tm bup1
rm cp u1

(4)

If VT VTcrit i.e. G 1, no heat ow between the gas parcel and

stack plate would occur [2]. For a device to operate as a refrigerator
i.e. to transfer heat from the cold-end to the hot-end of a stack, the
temperature change of the gas parcels due to adiabatic compression or expansion must be greater than the corresponding
temperature change of the stack plate. Thus, for a thermoacoustic
refrigerator, G  1 [2].

209

Considering a short stack and neglecting the change in the work

ow of the stack, previous studies proposed an equation to estimate DTs [3,4]. They assumed sufciently high thermal conductivity
for the stack. Therefore, they neglected heat transfer between the
stack and its surroundings and assumed that the thermoacoustic
heat ow Q_ 2;s is conducted back to the cold-end of the stack by
either the solid plates or both the solid plates and working gas [3,4].
Applying Eq. (6), it can be shown that their approach and
assumptions imply,

H_ 2;s 0

(7)

In the present study, the simplied linear thermoacoustic

theory (Eqs. (2) and (3)) has been used along with a more realistic
approach to predict DTs. In this approach, we considered the
acoustic dissipation inside the stack and the heat exchange
between the stack and the surroundings, which were neglected in
the previous models. This will allow to predict DTs in a situation
which is closer to that in the real systems. The present approach is
based on applying the energy balance on a control surface
enclosing the cold-end of a thermoacoustic couple. The inuence of
energy dissipation within the stack on energy pattern was
demonstrated through the numerical simulation of Cao et al. [7].
Acoustic power consumed by the stack and the heat exchange
between the stack and its surroundings have been incorporated
through this energy balance. Fig. 2(a) shows the schematic of
a thermoacoustic couple inside a resonator tube. The control
surface is outlined with the dashed line in the gure. The energy
uxes crossing the boundaries of the control surface are also shown
in the gure.
It is assumed that the dissipated acoustic power in the cold-side
ambient duct leaves the device. The acoustic energy ow into the
stack DE_ 2;sum is the sum of the acoustic power consumed by the
stack DE_ 2;s and by the hot-side ambient duct DE_ 2;ad . Neglecting
the acoustic power dissipated in the hot-side ambient duct, the
required acoustic power is equal to the acoustic power consumed
by the stack. It is also assumed that there is no lateral heat transfer
in the stack.
The term Q_ cond;ce represents the heat exchange between the
cold-end of the stack and the cold-side ambient duct. This term is

3. Theoretical modeling
The acoustic approximation to the total power owing in the x
direction can be written as [15],

1 h ~ i 1
H_ 2 x Re p1 U
1 rm Tm
2
2
dTm
Asolid Ksolid
dx

h
i

~ 1 dA
Ag Kg
Re s1 u
(5)

On the right-hand side, the rst term is the acoustic power, the
second term is Tm times the entropy ux, and the last term is the
heat conduction through the stack and the working gas [15].
Eq. (5) indicates that the summation of time-averaged longitudinal work ow E_ 2;s , thermoacoustic heat ow Q_ 2;s , and
conduction heat ow Q_ cond;s is the time-averaged longitudinal
total energy ow at any point in the stack. That is,

H_ 2;s E_ 2;s Q_ 2;s Q_ cond;s

(6)

Fig. 2. (a) Schematic of a thermoacoustic couple inside a resonator tube. The control
surface is outlined with a dashed line. (b) The theoretical variations of stack and gas
temperatures, and the temperatures at different locations inside and outside the stack.
Note that, only one stack plate is shown in the gure to illustrate the temperature
variations.

210

H. Babaei, K. Siddiqui / International Journal of Thermal Sciences 50 (2011) 206e213

quantied through the schematic of a gas parcel oscillation in the

vicinity of a stack plate as shown in Fig. 2(b). Assuming that the
stack cold-end (i.e. the control surface) located at xce in Fig. 2(b) is at
a temperature of Tce and that in the vicinity of the control surface
i.e. xce Dx, the cold-side ambient duct is at the mean temperature
(Tm) which is greater than Tce. Heat transfer from the cold-side
ambient duct to the control surface is through conduction due to
the temperature gradient of Tm
Tce =Dx. Assuming a linear
temperature variation in the stack, we can write, Tm
Tce DTs =2.
Assuming a linear prole for the temperature is not fully realistic
due to the acoustic dissipation which serves as a heat source inside
the stack. An increase in the drive ratio causes a raise in acoustic
dissipation and more deviation from the linear assumption. As
mentioned earlier, Worlikar et al. [6] numerically demonstrated
that the linear variation of temperature within the stack is a valid
assumption at low drive ratios (as predicted by linear theory).
However, they predicted nonuniform temperature prole at
moderate drive ratios. Nevertheless, a linear temperature prole is
assumed in this study.
Cao et al. [7] and Ishikawa and Mee [13] numerically showed that
the net heat exchange between solid and gas in a thermoacoustic
couple occurs over a distance approximately equal to the gas
displacement amplitude from the ends of the couple. Swift [2] also
argued that the optimum length of a heat exchanger should be equal
to the peak to peak gas displacement amplitude at the heat exchanger
location. Hence, it is reasonable to assume that the temperature
difference of Tm
Tce exists within a peak to peak gas displacement
from the cold-end. So we can write Dx 2x1,ce Therefore, the heat ux
entering the stack through the control surface from the stack
surroundings via conduction, can be estimated by,

DTs
Tm
Tce
Kg A
Q_ cond;ce Kg A
Dx
4x1;ce

(8)

where, A is the cross-sectional area of the resonator. Applying the

energy balance on the control surface shown in Fig. 2, we have,

H_ 2;s DE_ 2;s Q_ cond;ce

(9)

Substitute H_ 2;s , DE_ 2;s and Q_ cond;ce from Eqs. (2), (3) and (8),
respectively, into Eq. (9), and solving for DTs, we have,

It is worth to mention that such a criterion was not considered in

the previous studies.
4. Theoretical results
The values of DTs predicted from the present model are
compared with that from the model developed by Wheatley et al.
[3] and its modied version presented by Atchley et al. [4]. The
drive ratio, blockage ratio, normalized stack spacing and the length
of the stack are assumed as DR 1.2%, BR 97%, y0/dk 2.9 and
Ls 3 cm, respectively. These parameters are chosen as they match
with the parameters for the experimental study conducted to
validate the present model (discussed in the Section 5). To investigate the effects of the stack thermal conductivity on the theoretical predictions, DTs predicted from these models is shown in
Fig. 3 for different stack thermal conductivities that vary from 40 to
0.04 W/m K. The values of DTs,max are also shown in Fig. 3 which
bounds the physically possible values of DTs.
Fig. 3(a) shows that for high thermal conductivity of the stack
(40 W/m K), the results from all three models agreed well in the
region of xcn < 0.4. As xcn increases, the difference between the
present and previous models starts to increase. The plot also shows
that DTs predicted from the present model becomes zero before the
velocity antinode (xcn 1.57). This is simply due to the way the
present model is formulated. That is, the energy balance is applied
to the cold-end of the stack while xcn is measured from the center of
the stack. The gure also shows that DTs predicted from all models
is less than DTs,max.
Fig. 3(b) shows the predicted values of DTs for the stack thermal
conductivity of 4 W/m K. The results show disagreement between
the predictions from the three models. The models of Wheatley
et al. [3] and Atchley et al. [4] show some disagreement for
0.1 < xcn < 0.8. However, the values of DTs from their models are
signicantly higher than that from the present model for all values
of xcn. The plot also shows that the values of DTs predicted by the
models of Wheatley et al. [3] and Atchley et al. [4] exceeds DTs,max
for xcn > 0.7 and xcn > 0.8, respectively. This implies that the values
of DTs predicted by the models of Wheatley et al. [3] and Atchley
et al. [4] in this range are physically impossible.

pi


2
p
cn Pr
1 Pr
dvn
2Lsn BRg
1cos2 xcn sin x
BRL
DTs
h sin2x 
 2L BRg
1cos2 x
A K A
p
Kg A
sn
cn
g g
Ag Pm adkn DR2
cn
solid Ksolid
p

 1 Pr Pr

8gBRg
1Tm Lsn cotxcn 1 Pr 1 PrL
4x1;ce
Ls
L

Ag Pm a
2 sin2xcn
d
8g kn DR 1PrL

The temperature difference of the couple (DTs) estimated by the

model of Eq. (9) takes into consideration both the acoustic power
dissipated in the stack and the heat exchange between the stack
and its surroundings, which were not considered previously.
Furthermore, as mentioned earlier, a thermoacoustic couple is
similar in conguration to a thermoacoustic refrigerator except for
the heat exchangers. Thus, it must hold the condition G  1 i.e. the
temperature gradient along the couple should always be less than
the critical temperature gradient. If Ls is the length of the stack then,

DTs  VTcrit  Ls

(11)

Thus, the maximum possible DTs attained by a thermoacoustic

couple is,

DTs;max VTcrit  Ls

(12)

(10)

Fig. 3(c) shows the values of DTs for the stack thermal conductivity
of 0.04 W/m K. The plot shows that the model of Wheatley et al. [3]
predicts signicantly high values of DTs and except for xcn < 0.065,
their model predicts physically impossible values of the stack
temperature difference. The model of Atchley et al. [4] predicted
relatively small magnitudes of DTs however, it exceeds DTs,max for
xcn > 0.3. The values of DTs predicted by the present model are
lower than that predicted by the models of Wheatley et al. [3] and
Atchley et al. [4] and remains less than DTs,max for the entire range
of xcn.
Fig. 3 shows that when the stack thermal conductivity Ks is
large, the values of DTs predicted by all three models are relatively
close. This could be due to the reason that when Ks is large, the
conduction through the stack plates is dominant compared to the
conduction through the gas within the stack and conduction heat

H. Babaei, K. Siddiqui / International Journal of Thermal Sciences 50 (2011) 206e213

4
3.5
3

Ts(C)

2.5
2
1.5
1
0.5
0

0.2

0.4

0.6

0.8

1.2

1.4

1.6

cn

16
14
12

T (C)

10
8

211

energy balance, DTs predicted by their model is the largest. Atchley

et al. [4] added conduction through the gas inside the stack in their
model but ignored heat transfer between the stack and the
surroundings. The values of DTs predicted by their model are lower
than that of Wheatley et al. [3] but higher than the present model.
The present model accounts for the conduction inside the stack
through plates and gas, and also the heat exchange between the
stack and its surroundings. Note that the energy balance at the
cold-end implicitly accounts for the heat ux from the hot-end of
the stack to the hot-end ambient duct. This heat exchange result in
a decrease in DTs.
Another reason for the higher values of DTs predicted by the
models of Wheatley et al. [3] and Atchley et al. [4] is the acoustic
dissipation whose magnitude increases from the pressure antinode
to the velocity antinode. Both of these models did not consider
acoustic dissipation in the stack. Acoustic dissipation is the
conversion of acoustic energy into heat, which acts as a heat source
within the stack and contributes to the lowering of DTs.
Furthermore, at lower values of Ks, in a certain range of xcn, the
models of Wheatley et al. [3] and Atchley et al. [4] predicted values
of DTs which are higher than DTs,maxwhich implies that for stacks
with lower thermal conductivity (recommended for real applications) and for a range of stack positions, the models of Wheatley
et al. [3] and Atchley et al. [4] predicted physically impossible
values of DTs. Thus, these models are not reliable under these
conditions. However, the present model that incorporates real
physical processes, predicts physically possible values of DTs for the
entire range of Ks at all stack positions.

5. Experimental validation

80

Theoretical values predicting the stack temperature difference

have overestimated experimental ones in all previous studies. For
example, Atchley et al. [4] reported up to 100% higher values of
theoretical DTs at the drive ratio of 1.99% for the stack thermal
conductivity of about 5.8 W/m K. Piccolo and Cannistraro [5]
reported 200% higher values of theoretical DTs at the drive ratio
of 0.38% for the stack thermal conductivity of about 0.11 W/m K.
These observations are consistent with the above theoretical
analysis which shows that the previous models predict higher
values of DTs as the stack thermal conductivity decreases. Thus,
a good validation of a theoretical model would be the one that is
based on the low stack thermal conductivity. Therefore, to validate
the present model, we conducted experiments with a stack of low
thermal conductivity.

60

5.1. Experimental set-up

40

The experiments were performed in a Plexiglas resonator tube

121 cm in length with the inner cross-sectional area of
11.5  11.5 cm2. The thickness of the Plexiglas sheets is 1.1 cm
which holds the rigid wall assumption. Air at atmospheric pressure
was used as the working gas. To excite a half wave-length acoustic
standing wave in the tube, a 200 W loudspeaker (model PR 65
NEO) was used. A function generator (Agilent 33120A) was used to
generate sinusoidal waves at a frequency 146 Hz. A 220 W amplier (Pioneer SA-1270) was used to amplify the output signal of the
function generator. The accuracy of the frequency and amplitude of
the output signal were 1 mHz and 0.1 mV, respectively. The
amplied signal was used to drive the loudspeaker. A microphone
(model 377A10 PCB Piezotronics) was used to monitor the acoustic
pressure. The open circuit sensitivity of the microphone was
1.94 mV/Pa at 251.2 Hz. The microphone was placed in a drilled
hole at the closed end of the tube. The frequency response of the
microphone was nearly at between 20 and 1000 Hz. The acoustic

2
0

0.2

0.4

0.6

0.8

1.2

1.4

1.6

cn

c 160
140
120

Ts(C)

100

20
0

0.2

0.4

0.6

0.8

1.2

1.4

1.6

cn

Fig. 3. Comparison of the values of DTs predicted by the present and previous models
at stack thermal conductivity of (a) Ks 40 W/m K, (b) Ks 4 W/m K, (c) Ks 0.04 W/
m K. The values of physically possible maximum stack temperature difference (DTs,max)
is also presented. Solid line, present model; dashed line, model of Atchley et al. [4];
dash-dotted line, model of Wheatley et al. [3]; dotted line, DTs,max. Theoretical calculations for all models were conducted at, DR 1.2%, BR 97%, y0/dk 2.9 and Ls 3 cm.

transfer between the stack and the surroundings. However, as Ks

decreases, the conduction through the gas inside and outside the
stack becomes more signicant. The model of Wheatley et al. [3]
ignored conduction through the gas and thus, to satisfy the

H. Babaei, K. Siddiqui / International Journal of Thermal Sciences 50 (2011) 206e213

Fig. 4(aec) shows the comparison of the experimental DTs with

that obtained from the present model and the models of Wheatley
et al. [3] and Atchley et al. [4] at drive ratios of 0.45%, 0.82% and
1.2%, respectively.
Excellent agreement between the results from the present
model and experiments are observed at the drive ratio of 0.45%.
With an increase in the drive ratio, some differences are observed
between the results from the present model and the experiments.
This difference at the higher drive ratios could be due to the
nonlinear effects and the assumption that the temperature variation within the stack is linear [6,8]. However, these differences are
signicantly small compared to that the differences between the
experimental results and predictions from the models of Wheatley
et al. [3] and Atchley et al. [4]. For example, at the drive of ratio of
0.82% (Fig. 4(b)), the values predicted by the present model are on
average 85% higher than the experimental results whereas, the
values of DTs from the models of Wheatley et al. [3] and Atchley
et al. [4] are 1900% and 700% higher than the experimental results.
At the drive ratio of 1.2%, these differences are about 100%, 1200%
and 600% for the present model, Wheatley et al.s [3] model and
Atchley et al.s [4] model, respectively. Note that at the lowest drive
ratio (0.45%), the values of DTs from the models of Wheatley et al.
[3] and Atchley et al. [4] are 2700% and 600% higher than the
experimental results.
Furthermore, at all given drive ratios, the trends of DTs from
experiments and the present model are very similar e.g. both have
the maximum DTs at the same normalized stack position of all given
drive ratios. Whereas, the trends of DTs from the models of
Wheatley et al. [3] and Atchley et al. [4] are signicantly different
from the experimental trends at all given driver ratios. The good
agreement between the values of DTs from the present model and
experiment validates the given model and conrms that the
present model with the associated modications can predict DTs
with much better accuracy than the previous models, over a wide
range of thermal conductivities that are used in the practical
applications.

60

50

Ts(C)

40

30

20

10

0.2

0.4

0.6

0.2

0.4

0.6

0.8

1.2

1.4

1.6

0.8

1.2

1.4

1.6

xcn

110
100
90
80
70

Ts(C)

5.2. Experimental results

60
50
40
30
20
10
0

cn

160
140
120
100

pressure was recorded via an oscilloscope (model DSO3062A).

Type-K thermocouples with the diameter of 0.25 mm and the
accuracy of 0.1  C were used to measure the temperature. The
temperature data were acquired via a 16-channel data acquisition
card (PCI-6036E, National Instruments) using the LabView data
acquisition software at a sampling rate of 2 Hz. Reticulated
Vitreous Carbon (RVC) with 20 pores per linear inch was used as
the stack material whose thermal conductivity is about 0.04 W/m K
[16]. The average pore size was about 0.05 inch or 1.27 mm
(2y0 1.27 mm). Thus, the ratio of the half of the plate spacing to
thermal penetration depth (y0/dk) was estimated to be 2.9 which
falls very well into the boundary-layer approximation [2]. The
porosity of the material was measured and found to be 97%. Based
on the material porosity, the plate thickness (2l) was estimated to
be equal to 0.0196 mm. Similar approach was used by Adeff et al.
[16]. The length of the stack was 3 cm. Temperatures were
measured at the centers of the two end-planes of the stack (i.e. at
the hot and cold-ends of the stack). The measurements were performed at eight different stack positions (xcn) along the resonator
tube between the closed end and the center of the tube. That is,
between the pressure antinode and velocity antinode. The data
were acquired at six drive ratios (ranging from 0.18% to 1.2%) at
each position. One thermocouple was used to monitor and record
the bulk temperature in the resonator tube. The bulk temperature
of the room was also monitored by a digital thermometer placed
outside the tube.

T (C)

212

80
60
40
20
0

0.2

0.4

0.6

0.8

xcn

1.2

1.4

1.6

Fig. 4. Comparison of the values of DTs from theoretical models and experiments at
drive ratios of (a) 0.45%, (b) 0.82% and (c) 1.20%. Solid line, present model; dashed line,
model of Atchley et al. [4]; dash-dotted line, model of Wheatley et al. [3]; , experimental data. The other conditions for the theoretical calculations and experiments
were, Ks 0.04 W/m K, BR 97%, y0/dk 2.9 and Ls 3 cm.

However, we would like to emphasis that the theoretical model

predicting DTs can be further improved by incorporating more
realistic assumptions such as the transverse heat conduction in the
stack and working gas, the heat transport due to second-order

H. Babaei, K. Siddiqui / International Journal of Thermal Sciences 50 (2011) 206e213

movement of acoustic streaming and the heat leakage through the

duct walls.

213

that from the experiments. These results show that the present
model predicts DTs with much more accuracy than the previous
models.

6. Conclusion
References
Previous theoretical models used to study the performance of
thermoacoustic couples were based on very simplied assumptions
that could hold only for stacks with very high thermal conductivity
which is not recommended in practical applications. In the present
study, we have modied the theoretical model of a thermoacoustic
couple by incorporating more realistic physical processes that are
consistent with the practical applications. The two major distinguishing features of the present model are the consideration of
acoustic dissipation within the stack, and the heat exchange
between the stack and its surroundings. The comparison of the
stack temperature difference (DTs) predicted by the present and
previous models shows that when the stack thermal conductivity is
low, the previous models not only signicantly overestimate DTs
but at certain stack positions they predict physically impossible
values DTs i.e. values greater than the theoretically possible values
of DTs. The present and previous theoretical models were also
compared with the experimental data acquired for low thermal
conductivity stack at three drive ratios. The comparison of the
experimental data with previous models shows that the previous
models overpredict DTs by 600% to over 2000% for the given range
of drive ratios with the theoretical proles signicantly different
from the experimental ones. The comparison of experimental data
with the present model shows excellent agreement at the drive
ratio of 0.45%. At higher drive ratios, the difference between the
present model and experimental data reached up to 100% however,
the proles of DTs predicted by the present model are identical to

[1] N. Rott, Thermoacoustics, Adv. Appl. Mech. 20 (1980) 135e175.

[2] G.W. Swift, Thermoacoustic engines, J. Acoust. Soc. Am. 84 (1988) 1145e1179.
[3] J. Wheatley, T. Hoer, G.W. Swift, A. Migliori, An intrinsically irreversible
thermoacoustic heat engine, J. Acoust. Soc. Am. 74 (1983) 153e170.
[4] A.A. Atchley, T.J. Hoer, M.L. Muzzerall, M.D. Kite, Acoustically generated
temperature gradients in short plates, J. Acoust. Soc. Am. 88 (1990) 251e263.
[5] A. Piccolo, G. Cannistraro, Convective heat transport along a thermoacoustic
couple in the transient regime, Int. J. Therm. Sci. 41 (2002) 1067e1075.
[6] A.S. Worlikar, O.M. Knio, R. Klein, Numerical simulation of a thermoacoustic
refrigerator-stratied ow around the stack, J. Comput. Phys. 144 (1998)
299e324.
[7] N. Cao, J.R. Olson, G.W. Swift, S. Chen, Energy ux density in a thermoacoustic
couple, J. Acoust. Soc. Am. 99 (1996) 3456e3464.
[8] N. Rott, Damped and thermally driven acoustic oscillation in wide and narrow
tubes, Z. Angew. Math. Phys. 20 (1969) 230e243.
[9] N. Rott, Thermally driven acoustic oscillation, part III: second order heat ux,
Z. Angew. Math. Phys. 26 (1975) 43e49.
[10] T.J. Hoer, Effective heat transfer between a thermoacoustic heat exchanger
and stack, J. Acoust. Soc. Am. 94 (1993) 1772.
[11] G. Mozurkewich, Time-average temperature distribution in a thermoacoustic
stack, J. Acoust. Soc. Am. 103 (1998) 380e388.
[12] G. Mozurkewich, A model for transverse heat transfer in thermoacoustics,
J. Acoust. Soc. Am. 103 (1998) 3318e3326.
[13] H. Ishikawa, D.J. Mee, Numerical investigations of ow and energy elds near
a thermoacoustic couple, J. Acoust. Soc. Am. 111 (2002) 831e839.
[14] L. Zoontjens, C.Q. Howard, A.C. Zander, B.S. Cazzolato, Numerical study of ow
and energy elds in thermoacoustic couples of non-zero thickness, Int.
J. Therm. Sci. 8 (2009) 733e746.
[15] G.W. Swift, Thermoacoustics: A Unifying Perspective for Some Engines and
Refrigerators. The Acoustical Society of America, NY: Melville, 2002.
[16] J.A. Adeff, T.J. Hoer, A.A. Atchley, Measurements with reticulated vitreous
carbon stacks in thermoacoustic prime movers and refrigerators, J. Acoust.
Soc. Am. 104 (1998) 32e38.