Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Journal of the Korean Physical Society, Vol. 67, No. 10, November 2015, pp.

1755∼1766

Revisiting the Difference between Traveling-wave and


Standing-wave Thermoacoustic Engines
- A Simple Analytical Model for the Standing-wave One

Kyuichi Yasui,∗ Teruyuki Kozuka,† Masaki Yasuoka and Kazumi Kato


National Institute of Advanced Industrial Science and Technology (AIST),
2266-98 Anagahora, Shimoshidami, Moriyama-ku, Nagoya 463-8560, Japan

(Received 17 July 2015, in final form 30 September 2015)

There are two major categories in a thermoacoustic prime-mover. One is the traveling-wave
type and the other is the standing-wave type. A simple analytical model of a standing-wave ther-
moacoustic prime-mover is proposed at relatively low heat-flux for a stack much shorter than the
acoustic wavelength, which approximately describes the Brayton cycle. Numerical simulations of
Rott’s equations have revealed that the work flow (acoustic power) increases by increasing of the
amplitude of the particle velocity (|U |) for the traveling-wave type and by increasing cosΦ for the
standing-wave type, where Φ is the phase difference between the particle velocity and the acoustic
pressure. In other words, the standing-wave type is a phase-dominant type while the traveling-wave
type is an amplitude-dominant one. The ratio of the absolute value of the traveling-wave com-
ponent (|U |cosΦ) to that of the standing-wave component (|U |sinΦ) of any thermoacoustic engine
roughly equals the ratio of the absolute value of the increasing rate of |U | to that of cosΦ. The
different mechanism between the traveling-wave and the standing-wave type is discussed regarding
the dependence of the energy efficiency on the acoustic impedance of a stack as well as that on
ωτα , where ω is the angular frequency of an acoustic wave and τα is the thermal relaxation time.
While the energy efficiency of the traveling-wave type at the optimal ωτα is much higher than that
of the standing-wave type, the energy efficiency of the standing-wave type is higher than that of
the traveling-wave type at much higher ωτα under a fixed temperature difference between the cold
and the hot ends of the stack.

PACS numbers: 43.35.Ud


Keywords: Thermoacoustic prime mover, Traveling-wave type, Standing-wave type
DOI: 10.3938/jkps.67.1755

I. INTRODUCTION prime-mover (engine). The thermodynamic cycle in a


traveling-wave prime-mover is widely known to be sim-
ilar to the Stirling cycle (Fig. 1(a)), which is typical in
When a sound wave propagates in a narrow tube with Stirling engines [5,6,9–11]. The thermodynamic cycle in
some temperature gradient along the tube, a small gas a standing-wave prime-mover is similar to the Brayton
parcel moves forward and backward periodically in the cycle (Fig. 1(b)), which is typical in gas turbines and
tube with the sound frequency exchanging some heat air-breathing jet engines [1,3–6,12].
with the wall of the tube [1–4]. As the instantaneous In the Brayton cycle of a standing-wave prime-mover,
local pressure temporally oscillates in a sound wave, a the pressure inside a gas parcel is kept constant (isobaric
gas parcel repeats expansion and contraction. When the process) when the gas parcel is at the right or the left side
temperature of the gas parcel is higher during its expan- of its vibration, as shown in Fig. 1(c). In other words,
sion than during its contraction at the moment of the when the particle’s velocity is nearly zero, the pressure
same volume due to the heat exchange with the wall, inside a gas parcel is kept constant. Between the two
the gas parcel does net work every cycle because the en- ends of its vibration, the gas parcel moves adiabatically
closed area in the volume-pressure diagram (Figs. 1(a) (Figs. 1(b) and (c)). With regard to the Stirling cycle
and (b)) corresponds to the net pV work done by the of a traveling-wave prime-mover, please refer to Refs. [5,
gas parcel [1–8]. In other words, the sound’s inten- 6], and [9-11].
sity increases. The system is called a thermoacoustic A thermoacoustic prime-mover is usually made of a
pipe with a short stack in it at an appropriate position
∗ E-mail: k.yasui@aist.go.jp (Figs. 2(a) and (b)). The short stack is often made of ce-
† present affiliation: Aichi Institute of Technology
-1755-
-1756- Journal of the Korean Physical Society, Vol. 67, No. 10, November 2015

Fig. 2. (Color online) A typical thermoacoustic prime-


mover (engine): (a) traveling-wave type and (b) standing-
wave type.

tensified in the narrow tubes. The frequency of sound


usually coincides with the resonance frequency of the
pipe. Thermoacoustic prime-movers are classified into
two types depending on the phase difference between the
particle’s velocity and the acoustic pressure [13]. The
traveling-wave type is defined as Φ = 0, where Φ is the
phase lead of the particle’s velocity (U) relative to the
acoustic pressure (P) as follows:
P = |P |ei(ωt+φ) , (1)
U = |U |e i(ωt+θ)
, (2)
Φ = θ − φ, (3)
with |P | and |U | being the amplitudes of P and U , re-
spectively, i the unit imaginary number, ω the angular
frequency of sound, and φ and θ the phase angles of P
and U , respectively. The actual physical quantities are
the real parts of P and U . The standing-wave type is
defined as Φ = π/2. The actual phase difference in ex-
periments is determined by the configuration of the pipe.
For example, in the experiment of Yazaki et al. [14],
Fig. 1. (Color online) A schematic representation for a gas Φ ∼ π/6 was observed at the lower temperature side of
parcel undergoing a thermoacoustic cycle. (a) The pressure- the stack in a looped pipe (Fig. 2(a)). They regarded this
volume diagram for the Stirling cycle for the traveling-wave system as a traveling-wave type because Φ was small as
engine. (b) The pressure-volume diagram for the Brayton in a true traveling-wave type (Φ = 0). When a partition
cycle for the standing-wave engine. (c) The motion of a gas was inserted in the looped pipe, they observed Φ ∼ π/2
parcel near a wall with a temperature gradient in the Brayton at the lower temperature side of a stack. They regarded
cycle of the standing-wave engine corresponding to (b). this system as a standing-wave type. In this system, the
work flow, which is the acoustic power defined by Eq. (4),
took on a small negative value at the lower temperature
ramics, with many straight narrow tubes along the stack. side of a stack and a larger positive value at the higher
When one side of a stack is cooled and the other side is temperature side. In other words, a sound was radiated
sufficiently heated, a sound wave is generated and is in- from both sides of a stack in this system. This is some-
Revisiting the Difference between Traveling-wave · · · – Kyuichi Yasui et al. -1757-

times used as another definition of the standing-wave In the present study, a simple analytical model
type [15]. However, according to the previous definition of a thermoacoustic prime-mover is proposed for the
(Φ = π/2), the work flow at the lower temperature side standing-wave type at a relatively low heat-flux for a
is zero: stack that is much shorter than an acoustic wavelength,
 which provides new insight into the differences in na-
A1 ω 1
I= P U dt = A1 Re[P U ] ture between the standing-wave and the traveling-wave
2π 2 types. The numerical result based on the simple model is
1 1 compared with the result of numerical simulations of the
= A1 |P ||U | cos(φ − θ) = A1 |P ||U | cos Φ, (4)
2 2 momentum and the continuity equations under Rott’s
where A1 is the area of the cross section of a stack in acoustic approximation [2,25–30]. Analytically, the sim-
which the working gas moves. ple model is shown to satisfy the momentum equation.
As seen in the experiment of Yazaki et al. [14], Φ is Furthermore, the simple analytical model approximately
neither 0 nor π/2 in actual experiments. Thus, Yazaki describes the Brayton cycle in Figs. 1(b) and (c). In
et al. [16] expressed the arbitrary value of Φ as the su- addition, the different mechanism of an increase in the
perposition of the traveling-wave type (Φ = 0) and the work flow between the traveling-wave and standing-wave
standing-wave type (Φ = π/2) as follows. thermoacoustic engines based on the numerical simula-
tions of Rott’s equations is discussed. The meanings of
U = |U |ei(ωt+θ) = |U |ei(ωt+φ+Φ) the traveling-wave and the standing-wave components of
any thermoacoustic engine are also discussed. Further-
= |U | cos Φei(ωt+φ) + |U | sin Φei(ωt+φ+π/2) , (5)
more, different mechanisms for the dependence of the en-
where |U | cos Φ is the traveling-wave component and ergy efficiency on the acoustic impedance of a stack and
|U | sin Φ is the standing-wave component. This kind of on the radius of a narrow tube in a stack are discussed
decomposition has also been used by Hasegawa et al. between the traveling-wave and the standing-wave types
[17]. by using numerical simulations of Rott’s equations. The
The different characteristics of the traveling-wave and energy efficiency is compared between the traveling-wave
the standing-wave components have already been dis- and the standing-wave types under various conditions.
cussed in the literature [9–11,18–24]. According to Tom- Although the difference between the traveling-wave and
inaga and Yazaki [21–24], the energy conversion per unit standing-wave types has already been discussed [2,5,9–
volume, which is defined as the divergence of the work 11,18–24,31–39], many of the above points have not yet
flux (work flow per unit area of cross section) (W = been addressed.
div( AI1 )), is separately expressed for the traveling-wave
and the standing-wave components as follows:
W = Wp + WS + Wk , (6) II. MODEL
dTm
Wp = 0.5Re(χα )β |P ||U | cos Φ, (7)
dx
dTm Although the position of a stack in a pipe is very im-
WS = −0.5Im(χα )β |P ||U | sin Φ, (8) portant in real thermoacoustic systems [2, 12, 14, 21, 31,
dx 32, 40–47], we restrict the discussion in the present pa-
Wk = 0.5Im(χα )(KT − KS )ω|P |2 , (9) per to the processes inside a stack. A stack is assumed

2J1 [(i − 1) ωτα ] to consist of straight narrow tubes.The proposed sim-
χα = √ √ , (10)
(i − 1) ωτα J0 [(i − 1) ωτα ] ple analytical model for a thermoacoustic prime-mover
of the standing-wave type is a simple superposition of
τα = r2 /(2α), (11)
the traveling-wave and the reflected wave in a narrow
where Wp and WS are the traveling-wave and the tube of a stack with a constant phase difference of π/2
standing-wave components of the energy conversion, re- between the acoustic pressure and the particle’s velocity
spectively, Wk is the dissipation due to thermal conduc- for each wave. The model is simply expressed as follows:
tion (Wk < 0), χα is the thermoacoustic function for
a tube with a circular cross section given by Eq. (10), P = Aei(ωt−kx) + Bei(ωt+kx) , (12)
β is the thermal expansion coefficient, Tm is the time- iπ/2
Ce
averaged temperature at the position, x is the distance U= (Aei(ωt−kx) − Bei(ωt+kx) ), (13)
ρm c0
along the narrow tube, KT and KS are the isothermal
and the adiabatic compressibilities of the working gas, where A and B are the amplitudes of the acoustic pres-
respectively, ω is the angular frequency of sound, τα is sure (real number) of the travelling wave and the re-
the thermal relaxation time defined by Eq. (11), J0 and flected wave, respectively, k is the wave number (k =
J1 are the zeroth- and the first-order Bessel functions, ω/c0 ), ρm is the time-averaged density of air (1.3 kg/m3 ),
respectively, r is the radius of the narrow tube, and α is c0 is the sound velocity (340 m/s), and C is a non-
the thermal diffusivity of the gas. dimensional constant that depends on the shape of a
-1758- Journal of the Korean Physical Society, Vol. 67, No. 10, November 2015

narrow tube in a stack as follows: where ν and α are the kinematic viscosity and the ther-
ωτν mal diffusivity of the working gas, respectively. χα and
C= (for circular cylinder), (14a) χν are thermoacoustic functions that allow us to describe
4
2 the three-dimensional phenomena in a tube by using one-
C = ωτν (for parallel plates), (14b) dimensional equations. For a tube with a circular cross
3
section, the thermoacoustic functions can be analytically
where
expressed by using two quantities: the thermal relax-
τν = r2 /(2ν), (15) ation time τα and the viscous relaxation time τν which
are defined by Eqs. (11) and (15), respectively. The ther-
r is the radius of the narrow tube (for parallel plates, the moacoustic functions for a tube with a circular cross sec-
distance between the plates is 2r), and ν is the kinematic tion are given by Eqs. (10) and (21):
viscosity of the working gas.

In the simple analytical model (Eqs. (12) and (13)), 2J1 [(i − 1) ωτν ]
the thermoacoustic effect is simply expressed by the χν = √ √ , (21)
(i − 1) ωτν J0 [(i − 1) ωτν ]
phase factor (eiπ/2 = i). We should note that A and
B are real numbers. At the lower temperature side of a where J0 and J1 are the zeroth- and the first-order Bessel
stack (x = 0), the phase lead of the particle’s velocity functions, respectively. For sufficiently small ωτν , the
(U) relative to the acoustic pressure (P) is Φ = π/2 due thermoacoustic function χν is approximated as follows
to the phase factor (eiπ/2 ), which satisfies the condition [21]:
for a standing wave. In this model, the attenuation of
acoustic waves is neglected [48]. Although similar discus- 1
sions have been reported [19,24,49–54], no report on the χν = 1 − i ωτν (for circular cylinder) for ωτν  π,
4
simple analytical model of Eqs. (12) and (13) has been (22a)
published. 2 π
Firstly, we discuss the difference between the simple χν = 1 − i ωτν (for parallel plates) for ωτν  ,
3 2
analytical model and the normal acoustic waves without (22b)
any thermoacoustic effect. For normal waves (Eqs. (16)
and (17)), there is no phase difference between the acous- Using the approximation (Eqs. (22a) or (22b)), the
tic pressure and the particle’s velocity for the traveling simple analytical model (Eqs. (12) and (13)) satisfies
or the reflected wave (For the sum of the waves, there is Eq. (18) because k = ω/c0 .
a phase difference) [55]: Finally, we show that the simple analytical model ap-
proximately describes the Brayton cycle in Figs. 1(b)
P = Aei(ωt−kx+α) + Bei(ωt+kx+α) , (16)
and (c). Firstly, we show that the isobaric processes (1
1 → 2 and 3 → 4 in Figs. 1(b) and (c)) are approximately
U= (Aei(ωt−kx+α) − Bei(ωt+kx+α) ), (17)
ρm c0 described by Eqs. (12) and (13). Here, we restrict the
where A, B, α and β are real constants. discussion to the condition kx  1, which means that
Next, we show that the simple analytical model the vibration amplitude of a gas parcel in Fig. 1(b) is
(Eqs. (12) and (13)) satisfies the momentum equation un- very small (very low acoustic intensity) and that the tube
der Rott’s acoustic approximation. The momentum and (stack) is very short compared to an acoustic wavelength.
the continuity equations in a narrow tube under Rott’s Then, the real parts of Eqs. (12) and (13) are crudely
acoustic approximation are given, respectively [2]. approximated respectively, as follows, where one should
note that the real parts of P and U are the real physical
dP iωρm
=− U, (18) quantities:
dx 1 − χν
dU iω[1 + (γ − 1)χα ] Re(P ) = (A + B) cos ωt, (23)
=− P 1
dx γPm Re(U ) = (B − A) sin ωt, (24)
χα − χν 1 dTm ρm c0
+ U, (19)
(1 − χν )(1 − σ) Tm dx where Re(P ) and Re(U ) mean the real parts of P and
where P and U are the acoustic pressure and the parti- U , respectively. In the processes of 1 → 2 and 3 → 4 in
cle’s velocity as defined in Eqs. (1) and (2), respectively, Fig. 1(c), the particle’s velocity is nearly zero (Re(U ) ∼
and x is the position along a straight tube with its ori- 0). This implies that sinωt ∼ 0. From Eq. (23),
gin at the lower temperature side of a stack (the left
end of a stack). ρm , Pm , γ, σ are the mean density, d(Re(P ))
= −ω(A + B) sin ωt. (25)
the mean pressure, the ratio of specific heats, and the dt
Prandtl number of the working gas, respectively. The
Prandtl number is defined by Thus, when the particle’s velocity is nearly zero, the
time derivative of the acoustic pressure is also nearly zero
σ = ν/α, (20) (acoustic pressure is nearly constant). This implies that
Revisiting the Difference between Traveling-wave · · · – Kyuichi Yasui et al. -1759-

Eqs. (12) and (13) approximately describe the isobaric end of a stack and n = N to the right end of the stack),
process in Figs. 1(b) and (c). and Mb,n is a matrix defined by
Next, the adiabatic processes (2 → 3 and 4 → 1 in
Figs. 1(b) and (c)) are considered. For the adiabatic pro- Mb,n =
 ωρm,n 
cess, Re(P )V γ = constant holds, where V is the volume 0 i(1−χν,n )
of a gas parcel. Thus, the time derivative of Re(P )V γ is Δx × ω[1+(γ−1)χα,n ] (χα,n −χν,n ) 
1 dTm  , (32)
zero: iγPm (1−χν,n )(1−σ) Tm,n dx n

d(Re(P )) γ dV where Δx is the length of a short  narrow tube, ρm,n ,


V + Re(P )γV γ−1 = 0. (26) χν,n , χα,n , Tm,n , and dTm /dxn are the mean density
dt dt
(time-averaged density), the thermoacoustic functions,
The volume of the gas parcel is approximately given by
the mean temperature (time-averaged temperature), and
∂(Re(ξ)) the gradient of the mean temperature at the nth short
V = (Δx)S , (27) tube, respectively. With Eq. (31), the acoustic pressure
∂x
and the particle’s velocity at any point in a stack can be
where (Δx) is the initial length of a gas parcel in the calculated from those at the left end of the stack (n = 0
direction of vibration (x), S is the cross section of the or x = 0):
gas parcel, and ξ is the displacement of the gas parcel in a  
complex number (the real part is the real displacement). P (xn )
= (E + Mb,n−1 )(E + Mb,n−2 )
The displacement of the gas parcel ξ is approximately U (xn )
given by  
P (x0 )
 · · · (E + Mb,1 )(E + Mb,0 ) , (33)
i U (x0 )
ξ = U dt = − U. (28)
ω
where E is the unit matrix. The mean temperature
Inserting Eq. (28) into Eq. (27) and using Eqs. (12) and (time-averaged temperature) and its gradient in the ma-
(13), we obtain trix Mb,n in Eq. (32) are calculated according to the
method of Ueda [26]. When negligible thermal conduc-
c c tion occurs between a tube and its surroundings, the
V = −(Δx)S Re(iP ) = (Δx)S (A+B) sin ωt,
ρm c20 ρm c20 total enthalpy flow (Hf low ) along a narrow tube is con-
(29) served:

where the same crude approximation has been made as dHf low
= 0. (34)
in Eqs. (23) and (24). Inserting Eqs. (23), (25) and (29) dx
into Eq. (26) approximately yields
The total enthalpy flow is given by [2,23]
2 γ  

< (sin ωt) >= , (30) A1 χα − χν


γ+1 Hf low = Re P U 1 −
2 (1 + σ)(1 − χν )
where <> means the time-averaged value. For air, γ = 2
1.4 and < (sin ωt)2 > = 0.58. This crudely holds because A1 ρm cp |U | dTm
+ Im[χα + σχν ]
< (sin ωt)2 > is slightly larger than 0.5 as the isobaric 2ω(1 − σ 2 )|1 − χν |2 dx
processes (1 → 2 and 3 → 4) with sin ωt ∼ 0 should be 2
dTm
excluded in the time average. We should note that the − Ai κi (35)
actual thermodynamic cycle in a standing-wave thermoa- i=1
dx
coustic engine deviates considerably from the Brayton
cycle [56]. where A1 is the area of the cross section of a stack where
For a comparison with the results from the simple an- the working gas moves, A2 is that occupied by the solid
alytical model, we performed numerical simulations of material of the stack, the overbar denotes the complex
Rott’s equations (Eqs. (18) and (19)) for a thermoacous- conjugate of the value, cp is the specific heat of the work-
tic prime-mover by using the transfer matrix method of ing gas at constant pressure, and κ1 and κ2 are the ther-
Ueda [25, 26]. In the transfer matrix method of Ueda mal conductivity of the working gas and that of the solid
[25,26,41,57,58], Eqs. (18) and (19) can be expressed by material of the stack, respectively.
using a matrix as In the present numerical simulations, the thermal con-
      ductivity of air (κ1 ) is calculated as a function of temper-
Pn+1 Pn Pn ature and that of a stack is assumed to be κ2 = 1.7 W/m
− = Mb,n , (31) K. From Eq. (35), the temperature gradient is given by
Un+1 Un Un

χα−χν
where a narrow tube is divided into N short tubes, Pn
ΔTm Hf low − A21 Re P U 1 − (1+σ)(1−χν)
and Un are the acoustic pressure and the particle’s veloc- = A ρ c |U |2 2 . (36)
ity at the nth short tube (n = 0 corresponds to the left Δx 1 m p
2 2 Im[χα + σχν ] − Ai κi
2ω(1−σ )|1−χν | i=1
-1760- Journal of the Korean Physical Society, Vol. 67, No. 10, November 2015

Fig. 3. (Color online) The results of numerical simulations of Rott’s equations for the traveling-wave-type thermoacoustic
prime-mover at 41 Hz as a function of the position (x) along a stack. The working gas is air at atmospheric pressure. The
radius (r) of a narrow tube in a stack is 0.2 mm, and the length of the stack is 100 mm. The porosity of the stack is 0.82.
The pressure amplitude (|P |) of an acoustic wave at the lower temperature side of the stack (x = 0) is 3.4 kPa. The acoustic
impedance (P/U ) at x = 0 is assumed to be 9ρm c0 , where c0 is the sound velocity (ρm = 1.3 kg/m3 , c0 = 340 m/s) [26]. ωτα
= 0.244 at x = 0. Both θ and φ are 0 at x = 0. Hf lux is determined as Hf lux = −2.244iC (= −3494 W/m2 ) to reproduce the
temperature difference between the cold and the hot ends of a stack observed in the experiment of Ueda et al. [45] under the
experimental conditions which differ slightly from those in this figure: (a) temperature (Tm ), (b) total enthalpy flux (Hf lux =
Hf low /A1 ), work flux (I/A1 ) and heat flux (Q/A1 ), (c) amplitudes of the acoustic pressure (|P |) and the particle velocity (|U |),
(d) phase angles of the acoustic pressure (φ) and the particle velocity (θ), and (e) rate of change of each component of the work
flux (|P |, |U |, and cos(θ − φ)).

The temperature at any point in a stack is obtained obtained by using the Carnot efficiency (ηCarnot ) [2,3]:
by the spatial integration of Eq. (36).
The heat flow (Q) is calculated by using the work flow TC
ηCarnot = 1 − , (39)
(I) calculated from Eq. (4) as follows: TH

where TC and TH are the temperatures at the lower- and


Q = Hf low − I. (37)
the higher-temperature sides of a stack, respectively.
The energy efficiency (η) of a prime-mover is calculated
by
III. RESULTS AND DISCUSSION
ΔI
η= , (38)
QH Firstly, numerical simulations of the momentum and
the continuity equations under Rott’s acoustic approxi-
where ΔI = IH − IC , with IH and IC being the work mation (Eqs. (18), (19) or Eqs. (32), (33)) are performed
flows at the higher- and the lower-temperature sides of for the traveling-wave-type thermoacoustic prime-mover
a stack, respectively, and QH being the heat flow at the at 41 Hz under conditions similar to those in the exper-
higher temperature side of a stack. The upper limit of iment by Ueda et al. [45] (Fig. 3). The working gas is
the energy efficiency of a thermoacoustic prime-mover is air at atmospheric pressure. The physical characteristics
Revisiting the Difference between Traveling-wave · · · – Kyuichi Yasui et al. -1761-

Fig. 4. (Color online) The results of numerical simulations of Rott’s equations for the standing-wave type prime-mover at 41
Hz as a function of the position (x) along a stack. The radius (r) of a narrow tube in a stack is 0.7 mm, and the length of the
stack is 35 mm. θ = π/2 and φ = 0 at x = 0. TC = 291 K and TH = 443 K (at x = 35 mm). Hf lux = −5000 W/m2 . ωτα =
2.05 and ωτν = 2.86 at x = 17.5 mm. The other conditions are the same as those in Fig. 3: (a) Work flux (I/A1 ), (b) |P | and
|U |, (c) φ and θ and (d) rate of change of each component of the work flux (|P |, |U |, and cos(θ − φ)).

of the gas, such as the mean density, viscosity, thermal is considerably decreased (Fig. 3(c) and (e)), which has
conductivity, ratio of specific heats, and specific heat at already been pointed out by Babaei and Siddiqui [59].
constant pressure, are calculated as functions of temper- The decrease in |P | is due to the viscous damping as the
ature. The total enthalpy flux (Hf lux : total enthalpy amplitude of the particle’s velocity becomes too large
flow per unit area of cross section) is determined to re- (Eq. (18) and Fig. 3(c)). This means that a stack effec-
produce the experimental condition of the temperature tively works only within a critical length of 77.2 mm in
at the higher temperature side of a stack as Hf lux = this case. The energy efficiency at the critical length is
−2.244iC , where iC is the work flux (work flow per unit 0.23, and its ratio to the Carnot efficiency is 36.9%.
area of cross section) at the lower temperature side of a In Fig. 4, the results of numerical simulations of the
stack. The resultant energy fluxes are shown in Fig. 3(b). momentum and the continuity equations under Rott’s
The work flux increases with increasing distance up to acoustic approximation for the standing-wave-type ther-
77.2 mm due to the conversion of heat into acoustic en- moacoustic prime-mover are shown as functions of the
ergy by the thermoacoustic prime-mover. The increase position along a stack for 35 mm. In this case, the in-
in the work flux is solely due to the increase of the ampli- crease in the work flux is mostly due to the increase in
tude of the particle’s velocity (|U |) in the second term on cos(θ − φ) (Fig. 4(d)), which has never been pointed out
the right side of Eq. (19) (Fig. 3(e)), which has already before in the literature. If the heat flux is much higher
been pointed out by Ceperley [10] and others. How- than that in the case of Fig. 4, the increase in the work
ever, the work flux begins to decrease above x = 77.2 flux is not only due to the increase in cos(θ − φ) but also
mm because the amplitude of the acoustic pressure (|P |) due to the increase in |U | except near the lower temper-
-1762- Journal of the Korean Physical Society, Vol. 67, No. 10, November 2015

Fig. 5. (Color online) The results of numerical calculations of the simple analytical model (Eqs. (12) and (13)) as a function
of position in the direction of wave propagation. The pressure amplitude of the traveling wave is A = 2.66 kPa, and that of the
reflected wave is B = 2.09 kPa with C = ωτ4ν = 0.715. The horizontal axis is for 35 mm as in Fig. 4: (a) work flux, (b) |P | and
|U |,(c) φ and θ, and (d) rate of change of each component of the work flux (|P |, |U |, and cos(θ − φ)).

ature side of a stack (x = 0) where the increase in the ature difference between the cold and the hot ends of a
work flux is solely due to the increase in cos(θ − φ) in stack) than in the case of Fig. 4, however, the deviation
any case for the standing-wave type. from the simple analytical model becomes considerable
Now, the numerical result based on the proposed sim- due to the increase in |U | caused by the thermoacoustic
ple analytical model (Eqs. (12), (13)) for the standing- effect.
wave type is compared with the above result. In order to For the traveling-wave type (Fig. 3), the simple an-
reproduce the condition of Fig. 4, A = 2.66 × 103 Pa and alytical model does not reproduce the results because
B = 2.09 × 103 Pa are assumed in Fig. 5 with C = ωτ4ν the increase in |U | caused by the thermoacoustic effect
= 0.715. The results in Fig. 5 are qualitatively similar is not taken into account in the analytical model. This
to those in Fig. 4 although the quantitative agreement confirms that the mechanism of the increase in the work
is rather poor. For normal waves without any thermoa- flux is completely different between the traveling-wave
coustic effect (Eqs. (16) and (17)), the work flux is con- and the standing-wave types, at least at a relatively low
stant and does not increase (or decrease) with distance. heat-flux.
In contrary, the simple model with a constant phase dif- What is the situation when a prime mover is neither
ference of π/2 between the acoustic pressure and the par- pure a traveling-wave type nor a pure standing-wave
ticle’s velocity for each wave results in an increase in the type? In Fig. 6, the results of the numerical simulations
work flux with distance. This means that at relatively of the momentum and the continuity equations under
low heat-flux, the thermoacoustic effect in the standing- Rott’s acoustic approximation are shown as functions of
wave prime-mover is solely due to the phase-difference the initial phase of U (θ at x = 0), which is equivalent to
factor (eiπ/2 ). With a higher heat-flux (larger temper- the initial phase difference between U and P (Φ at x =
Revisiting the Difference between Traveling-wave · · · – Kyuichi Yasui et al. -1763-

Fig. 7. (Color online) The results of numerical simulations


of Rott’s equations for a thermoacoustic prime-mover with its
energy efficiency relative to the Carnot efficiency as a function
of ωτα at the center of a stack (at x = 17.5 mm) when the
radius (r) of a narrow tube in a stack is varied. The length of
a stack is 35 mm. Hf lux has been determined for each case
to fix TH = 600 K and TC = 291 K. θ = 0 (θ = π/2) at x =
0 for the traveling-wave type (standing-wave type), and φ =
0 at x = 0 for the both types. The other conditions are the
same as those in Fig. 3 for both types.

standing-wave type (Φ = π/2), the increase in the work


flow at x = 0 is due to the increases in |U | and cos(θ −φ),
respectively (Fig. 6(b)). The ratio of the absolute val-
ues for the increasing rates of |U | and | cos(θ − φ)| at
x = 0 nearly equals the ratio of the absolute values for
the traveling-wave and the standing-wave components in
Eq. (5) (Fig. 6(b)), which has never before been reported
in the literature. When the increase (or decrease) in the
Fig. 6. (Color online) The results of numerical simulations work flow is mainly caused by the increase (or decrease)
of Rott’s equations for a thermoacoustic prime-mover as a in |U | (equivalent to the case of a larger absolute value
function of θ at x = 0 (equivalent to Φ at x = 0) with Hf lux of the traveling-wave component), the prime mover is
= −2.5 × 104 W/m2 . The other conditions are the same as called an amplitude-dominant type (Fig. 6). When the
those in Fig. 4. ωτα = 2.99 at x = 0. The region for the
increase (or decrease) in the work flow is mainly caused
amplitude-dominant type (traveling-wave type) and that for
the phase-dominant type (standing-wave type) are shown in by the increase (or decrease) in cos(θ − φ) (equivalent
the figures, as well as that for both mechanisms: (a) energy to the case of larger absolute value of the standing-wave
efficiency relative to the Carnot efficiency, and (b) ratio of component), it is called a phase-dominant type.
the absolute values of the rates of change at x = 0 (|P |, |U |, In Fig. 7, the results of the numerical simulations for
and cos(θ − φ)) and the ratio of |cos(θ − φ)| (traveling-wave various radii of a narrow tube in a stack are shown for
component) to |sin(θ − φ)| (standing-wave component). a fixed temperature difference between the cold and the
hot ends of a stack (TC = 291 K and TH = 600 K). The
horizontal axis is expressed by using the corresponding
0). Such a plot has already been published by Kang et al. ωτα at the center of a stack (at x = 17.5 mm). The en-
[20]. The energy efficiency is non-zero between −π/4 and ergy efficiency relative to the Carnot efficiency is a max-
3π/5 for Φ at x = 0 under the condition of Fig. 6. The imum at ωτα = 0.12 and 1.5 for the traveling-wave and
highest energy efficiency relative to the Carnot efficiency the standing-wave types, respectively, under the condi-
is at π/4 under the condition of a constant heat flux. tion of Fig. 7, which is nearly consistent with the predic-
For the pure traveling-wave type (Φ = 0) and the pure tion based on Eqs. (7) and (8) that the optimal condition
-1764- Journal of the Korean Physical Society, Vol. 67, No. 10, November 2015

is ωτα  1 and ωτα = 3 for the traveling-wave and the


standing-wave types, respectively [19, 29, 60]. The opti-
mal radii of a narrow tube are 0.2 mm and 0.7 mm for
the traveling-wave and the standing-wave types, respec-
tively, under the condition of Fig. 7. While the energy
efficiency for the traveling-wave type at the optimal ωτα
is higher than that for the standing-wave type, the en-
ergy efficiency for the standing-wave type is higher than
that for the traveling-wave type at relatively large ωτα .
As the mechanisms for the increase in the work flow
are completely different between the traveling-wave and
the standing-wave types, the reason for the dependence
of the energy efficiency on ωτα is also different: For the
traveling-wave type, the viscous damping of an acoustic
wave is greater for smaller ωτα due to the smaller radius
of a tube resulting in larger decrease in |P | [52,61]. For
the standing-wave type, on the other hand, the increase
in cos(θ − φ) (the decrease in θ) is suppressed for values
of ωτα slightly smaller than the optimal value because
the phase shift of the particle’s velocity is suppressed Fig. 8. (Color online) The results of numerical simulations
as the heat exchange between the wall and a gas parcel of Rott’s equations for a thermoacoustic prime-mover with its
becomes faster [21]. For larger ωτα , on the other hand, energy efficiency relative to the Carnot efficiency as a function
the reason for the decrease in the energy efficiency is the of the absolute value of the acoustic impedance (P/U ) at the
same for the two types; that is, the increasing rate of |U | lower temperature side of a stack (x = 0). The length of a
decreases because the heat exchange between the wall of stack is 35 mm. Hf lux has been determined for each case to
a tube and a gas parcel takes place in closer proximity fix TH = 600 K and TC = 291 K. θ = 0 (θ = π/2) at x = 0
to the wall. One should note that |U | is increased by the for the traveling-wave type (standing-wave type), and φ = 0
at x = 0 for both types. The radius (r) of a narrow tube in
thermoacoustic effect even for the standing-wave type at
a stack is 0.2 mm (ωτα = 0.12 at x = 17.5 mm) and 0.7 mm
larger x than 0 because Φ deviates from π/2. (ωτα = 1.5 at x = 17. 5 mm) for the traveling-wave type and
Next, the dependence of the energy efficiency on the the standing-wave type, respectively. The other conditions
acoustic impedance (P/U ) at x = 0 is discussed when the are the same as those in Fig. 3 for both types.
temperature difference between the cold and hot ends of
a stack is fixed (TC = 291 K and TH = 600 K) (Fig. 8).
For a plane traveling wave, the acoustic impedance al- smaller value of the acoustic impedance. For a larger
ways equals ρm c0 , where ρm is the mean density of the value of acoustic impedance, cos(θ − φ) decreases more
gas and c0 is the sound velocity (ρm = 1.3 kg/m3 , c0 for the traveling-wave type because θ decreases more due
= 340 m/s). When a reflected wave occurs, it takes an to the smaller |U |. On the other hand, for the standing-
absolute value larger than ρm c0 [55]. A numerical study wave type, |U | decreases as the thermoacoustic effect is
of the acoustic impedance of a stack has been reported suppressed because the first term on the right side of
by Worlikar and Knio [62]. Eq. (19) overwhelms the second term as U is a smaller
For the traveling-wave type, a non-zero energy effi- imaginary number.
ciency is seen for the range from 1.3ρm c0 to 660ρm c0 in Next, the energy efficiency is compared between the
the acoustic impedance at x = 0 under the condition of traveling-wave and the standing-wave types for various
Fig. 8. For the standing-wave type, it is from ρm c0 to acoustic-pressure amplitudes at the lower temperature
49ρm c0 in the absolute value of the acoustic impedance side of a stack (at x = 0) with a fixed temperature dif-
at x = 0. The optimal acoustic impedances are 7ρm c0 ference between the two sides of a stack (TC = 291 K and
and 11ρm c0 in absolute value for the traveling-wave and TH = 600 K) when the radius of a narrow tube (ωτα ) is
the standing-wave types, respectively. The reasons for optimal for each type (Fig. 9). The energy efficiency of
the dependence of the energy efficiency on the acoustic the traveling-wave type becomes much higher than that
impedance are different between the traveling-wave and of the standing-wave type as the acoustic pressure am-
the standing-wave types. For the traveling-wave type, plitude at x = 0 increases. When the initial acoustic
a lower acoustic impedance results in a larger |U | and pressure is 0, the energy efficiency is 0 for both types.
stronger viscous damping of an acoustic wave (decrease The initial start-up of the thermoacoustic oscillations is
in |P |). On the other hand, for the standing-wave type, due to a nonlinear effect that is neglected in Eqs. (18)
viscous damping is suppressed by the phase difference of and (19) [6,63–77].
π/2 between the particle’s velocity and the acoustic pres-
sure according to Eq. (18). Instead, the rate of increase
of cos(θ − φ) decreases because φ decreases more for a
Revisiting the Difference between Traveling-wave · · · – Kyuichi Yasui et al. -1765-

wave type is much higher than that for the standing-


wave type at the optimal condition of ωτα , the energy
efficiency for the standing-wave type is higher than that
for the traveling-wave type at much higher ωτα . The
mechanism of the dependence of the energy efficiency on
ωτα as well as on the acoustic impedance of a stack is dif-
ferent between the traveling-wave and the standing-wave
types.

ACKNOWLEDGMENTS

The authors would like to thank Yuki Ueda of Tokyo


University of Agriculture and Technology for making the
Fortran program for the transfer matrix method open
to public. We would like to thank Shinya Hasegawa of
Tokai University and Shin-ichi Sakamoto of University
of Shiga Prefecture for useful discussions. We would
Fig. 9. (Color online) The results of numerical simulations
also like to thank Nobumasa Sugimoto of Kansai Uni-
of Rott’s equations for a thermoacoustic prime-mover with its versity and Tatsuo Inoue of Genesis Research Institute
energy efficiency relative to the Carnot efficiency as a function for useful comments. This study was partly supported
of the acoustic pressure amplitude at the lower temperature by the Advanced Low Carbon Technology Research and
side of a stack. Hf lux has been determined for each case to Development Program (ALCA) from Japan Science and
fix TH = 600 K and TC = 291 K. The other conditions are Technology Agency.
the same as those in Fig. 8.

IV. CONCLUSION REFERENCES

[1] G. W. Swift, J. Acoust. Soc. Am. 84, 1145 (1988).


A simple analytical model of a standing-wave ther-
[2] G. W. Swift, Thermoacoustics: A unifying perspective
moacoustic prime-mover is proposed at a relatively low for some engines and refrigerators (Acoustical Society
heat-flux for a much shorter stack than an acoustic wave- of America, New York, 2002).
length, which approximately describes the Brayton cycle [3] J. C. Wheatley, G. W. Swift and A. Migliori, Los Alamos
in Figs. 1(b) and (c). The result of the numerical calcula- Science 14, 2 (1986).
tion based on the simple model qualitatively agrees with [4] J. Wheatley, T. Hofler, G. W. Swift and A. Migliori, Am.
that of the numerical simulations of the momentum and J. Phys. 53, 147 (1985).
the continuity equations (Eqs. (18) and (19)) derived un- [5] S. Klein and G. Nellis, Thermodynamics (Cambridge
der Rott’s acoustic approximation for the standing-wave University Press, New York, 2012).
type at relatively low heat-flux. For the traveling-wave [6] N. Sugimoto, Nagare 33, 181 (2014).
type, on the other hand, the simple model does not repro- [7] N. Sugimoto, D. Shimizu and Y. Kimura, Phys. Fluids
20, 024103 (2008).
duce the results of Rott’s equations. It confirms that the
[8] A. S. Abduljalil, Z. Yu and A. J. Jaworski, J. Power
mechanism of the traveling-wave type is different from Energy 226, 822 (2012).
that of the standing-wave type. For the traveling-wave [9] P. H. Ceperley, J. Acoust. Soc. Am. 66, 1508 (1979).
type, the work flow (acoustic power) is increased with [10] P. H. Ceperley, J. Acoust. Soc. Am. 77, 1239 (1985).
direct increases in the amplitude of the particle velocity [11] J. A. de Jong, Ph.D. thesis, University of Twente, The
(|U |) by the thermoacoustic effect as has already been Netherlands, 2015.
reported [10]. For the standing-wave type, on the other [12] F. Wu, L. Chen, D. Li, G. Ding, C. Zhang and X. Kan,
hand, the work flow increases with increasing cos(θ − φ), Appl. Energy 86, 1119 (2009).
which has not been reported until now in the literature. [13] Y. Ueda, T. Biwa, U. Mizutani and T. Yazaki, J. Acoust.
Thus, the standing-wave type could be called a phase- Soc. Am. 115, 1134 (2004).
[14] T. Yazaki, A. Iwata, T. Maekawa and A. Tominaga,
dominant type while the traveling-wave type could be
Phys. Rev. Lett. 81, 3128 (1998).
called an amplitude-dominant type. The ratio of the ab- [15] T. Biwa, Y. Ueda, T. Yazaki and U. Mizutani, Europhys.
solute value of the traveling-wave to the standing-wave Lett. 60, 363 (2002).
component of any prime mover nearly equals the ratio [16] T. Yazaki, T. Biwa and A. Tominaga, Appl. Phys. Lett.
of the absolute value of the increasing rate of |U | to that 80, 157 (2002).
of cos(θ − φ), which has not been reported in literature [17] S. Hasegawa, T. Yamaguchi and Y. Oshinoya, Appl.
until now. While the energy efficiency for the traveling- Thermal Enginrng. 58, 394 (2013).
-1766- Journal of the Korean Physical Society, Vol. 67, No. 10, November 2015

[18] R. Raspet, H. E. Bass and J. Kordomenos, J. Acoust. Acoust. 65, 1 (2004).


Soc. Am. 94, 2232 (1993). [49] H. Tijdeman, J. Sound Vib. 39, 1 (1975).
[19] R. Raspet, J. Brewster and H. E. Bass, J. Acoust. Soc. [50] A. A. Atchley, H. E. Bass, T. J. Hofler and H. T. Lin, J.
Am. 103, 2395 (1998). Acoust. Soc. Am. 91, 734 (1992).
[20] H. Kang, G. Zhou and Q. Li, Cryogenics 50, 450 (2010). [51] A. A. Atchley, J. Acoust. Soc. Am. 92, 2907 (1992).
[21] A. Tominaga, Fundamental Thermoacoustics (Uchida [52] T. Yazaki, Y. Tashiro and T. Biwa, Proc. Roy. Soc. A
Rokakuho, Tokyo, 1998). 463, 2855 (2007).
[22] A. Tominaga J. Cryog. Soc. Jpn 25, 132 (1990). [53] B. Lihoreau, P. Lotton, M. Bruneau and V. Gusev, Acta
[23] A. Tominaga, Cryogenics 35, 427 (1995). Acustica united with Acustica 88, 986 (2002).
[24] T. Yazaki, Nagare 24, 395 (2005). [54] I. Nowak, S. Rulik, W. Wroblewski, G. Nowak and J.
[25] Y. Ueda, J. Phys. Soc. Jpn. 80, 034403 (2011). Szwedowicz, Int. J. Heat Mass Transfer 77, 369 (2014).
[26] Y. Ueda, J. Cryog. Supercond. Soc. Jpn. 47, 3 (2012). [55] L. E. Kinsler, A. R. Frey, A. B. Coppens and J. V.
[27] N. Rott, Zeit. Angew. Mathemat. Phys. (ZAMP) 20, 230 Sanders, Fundamentals of Acoustics, 3rd ed. (John Wiley
(1969). & Sons, New York, 1982).
[28] N. Rott, J. Appl. Mathemat. Phys. (ZAMP) 24, 54 [56] H. Luck and Ch. Trepp, Cryogenics 32, 690 (1992).
(1973). [57] Y. Ueda and C. Kato, J. Acoust. Soc. Am. 124, 851
[29] N. Rott, J. Appl. Mathemat. Phys. (ZAMP) 26, 43 (2008).
(1975). [58] K. Kuroda, S. Sakamoto, K. Shibata, Y. Nakano, T.
[30] N. Rott and G. Zouzoulas, J. Appl Mathemat Phys Tsuchiya and Y. Watanabe, Jpn. J. Appl. Phys. 51,
(ZAMP) 27, 197 (1976). 07GE01 (2012).
[31] F. C. Bannwart, G. Penelet, P. Lotton and J. P. Dal- [59] H. Babaei and K. Siddiqui, Energy Conv. Manage. 49,
mont, J. Acoust. Soc. Am. 133, 2650 (2013). 3585 (2008).
[32] T. Biwa, Y. Tashiro, U. Mizutani, M. Kozuka and T. [60] T. Inoue, Refrigeration 69, 1246 (1994).
Yazaki, Phys. Rev. E 69, 066304 (2004). [61] T. Biwa, J. Phys. Conf. Ser. 400, 052001 (2012).
[33] H. Chaitou and P. Nika, Energy Convers. Manage. 55, [62] A. S. Worlikar and O. M. Knio, Acustica 85, 480 (1999).
71 (2012). [63] O. Hireche, C. Weisman, D. Baltean-Carles, P. L. Quere
[34] H. Kang, Q. Li and G. Zhou, Cryogenics 49, 112 (2009). and L. Bauwens, J. Acoust. Soc. Am. 128, 3438 (2010).
[35] H. Kang, Q. Li and G. Zhou, Energy Convers. Manage. [64] M. Watanabe, A. Prosperetti and H. Yuan, J. Acoust.
50, 2098 (2009). Soc. Am. 102, 3484 (1997).
[36] G. Poignand, A. Podkovskiy, G. Penelet, P. Lotton and [65] H. Yuan, S. Karpov and A. Prosperetti, J. Acoust. Soc.
M. Bruneau, Acta Acustica united with Acustica 99, 898 Am. 102, 3497 (1997).
(2013). [66] S. Karpov and A. Prosperetti, J. Acoust. Soc. Am. 107,
[37] M. Guedra and G. Penelet, Acta Acustica united with 3130 (2000).
Acustica 98, 232 (2012). [67] S. Karpov and A. Prosperetti, J. Acoust. Soc. Am. 112,
[38] Z. J. Hu, Z. Y. Li, Q. Li and Q. Li, Energy Convers. 1431 (2002).
Manage. 51, 802 (2010). [68] M. F. Hamilton, Y. A. Ilinski and E. A. Zabolotskaya, J.
[39] M. Nouh, O. Aldraihem and A. Baz, Proc. SPIE 8688, Acoust. Soc. Am. 111, 2076 (2002).
86880L (2013). [69] N. Sugimoto and K. Tsujimoto, J. Fluid Mech. 456, 377
[40] H. Hatori, T. Biwa and T. Yazaki, J. Appl. Phys. 111, (2002).
074905 (2012). [70] N. Sugimoto and M. Yoshida, Phys. Fluids 19, 074101
[41] K. Sahashi, S. Sakamoto, K. Kuroda, and Y. Watanabe, (2007).
Jpn. J. Appl. Phys. 51, 07GE02 (2012). [71] D. Shimizu and N. Sugimoto, J. Phys. Soc. Jpn. 78,
[42] M. E. H. Tijani and S. Spoelstra, J. Appl. Phys. 110, 094401 (2009).
093519 (2011). [72] N. Sugimoto, J. Fluid Mech. 658, 89 (2010).
[43] B. J. Andersen and O. G. Symko, J. Acoust. Soc. Am. [73] G. Y. Yu, E. C. Luo, W. Dai and J. Y. Hu, J. Appl.
125, 787 (2009). Phys. 102, 074901 (2007).
[44] Y. Tsuji, S. Sakamoto, T. Ishino, Y. Watanabe and J. [74] J. A. L.Nijeholt, M. E. H. Tijani and S. Spoelstra, J.
Senda, Jpn. J. Appl. Phys. 47, 4231 (2008). Acoust. Soc. Am. 118, 2265 (2005).
[45] Y. Ueda, T. Biwa, U. Mizutani and T. Yazaki, Appl. [75] W. Polifke, A. Poncet, C. O. Paschereit and K. Dobbel-
Phys. Lett. 81, 5252 (2002). ing, J. Sound Vib. 245, 483 (2001).
[46] S. Backhaus and G. W. Swift, J. Acoust. Soc. Am. 107, [76] K. K. Durga, R. I. Sujith and M. Tyagi, Acta Acustica
3148 (2000). united with Acustica 90, 1131 (2004).
[47] K. Tourkov and L. Schaefer, Energy Conv. Managem. [77] K. Wang, D. M. Sun, J. Zhang, J. Zou, K. Wu, L. M. Qiu
95, 94 (2015). and Z. Y. Huang, Intern. J. Thermal Sci. 94, 61 (2015).
[48] M. E. H. Tijani, S. Spoelstra and P. W. Bach, Appl.

You might also like