Applied Thermal Engineering 45-46 (2012) 15e23

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

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

Oscillating flow in a stirling engine heat exchanger

M. Kuosa a, *, K. Saari a, A. Kankkunen a, T.-M. Tveit b
a
Aalto University, School of Engineering, Department of Energy Technology, P.O. Box 14400, FI-00076 Aalto, Finland
b
Single-Phase Power AS, Jacob Neumanns Vei 15, NO-1384 Asker, Norway

a r t i c l e i n f o a b s t r a c t

Article history: Three heat exchangers exist in modern Stirling engines: a heater, a cooler, and a regenerator. Here a study
Received 10 August 2011 that deals principally with tubular heaters and coolers is carried out. The calculation procedure for the
Accepted 14 March 2012 oscillating flow heat transfer is presented. Literature sources are studied to find the most suitable
Available online 28 March 2012
correlations by comparing them to each other and to the classical turbulent flow correlations encoun-
tered in the literature. The enhancement of heat transfer by means of a few circumferential slots inside
Keywords:
the tubes and the pressure losses of oscillatory flow are discussed. Non-circular cross-section conduits
Oscillating flow
with rectangular and triangular cross-sections are investigated and compared to the smooth circular
Heat exchanger
Slotted and non-circular tubes
tubes. The increment of the performance of an idealised Stirling engine with slotted heat exchanger
Stirling engine tubes is compared to the case with smooth ones. The ratio of the gain in the shaft power and pumping
losses is 2.22. The Carnot efficiency increment is 2.7%.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction statistical analysis to obtain an experimental correlation equation

for the cycle-averaged Nusselt number in terms of three similarity
Convective heat transfer equations for oscillating flow encoun- parameters: Re, Reu, and the dimensionless fluid displacement Au.
tered in Stirling engines have not been published until recent Their experimental runs were performed for air in a heated pipe for
decades. The common practice has been that the turbulent corre- the following ranges of parameters: 7 < Re<7000; 7<Reu<180. and
lations, e.g. the DittuseBoelter equation [1,2], are used since ‘no 0.06<Au<2.21. On the basis of an experimental and numerical
correlations existed up to now to calculate the heat transfer coef- study, Zhao and Cheng (1996) [8] derived Eq. (2), which consists of
ficient and friction factor in oscillating flow’ [3]. Turbulence takes only two similarity parameters, Reu and Ao. They plotted Nu=A0;85 o
time to form and settle and the picture to date is quite complex as a function of Reu (0<Reu<500), while the dimensionless oscil-
when heating and pressure variation are added. In the fully lating amplitude of fluid Ao variable got the values 8.5, 15.3, 20.4,
turbulent and fully laminar range the steady unidirectional and 34.9. De Monte et al. (1996) [9] (3) made a comprehensive
approximations are fairly good [4]. The laminar range is covered analytical study of oscillating heat transfer in Stirling machines
well in the latest papers in terms of oscillating convective heat regarding the inside and outside heat transfer of the heater, cooler,
transfer. In the literature survey several papers were found about and regenerator and the effect of heat-exchange effectiveness on
heat transfer for the oscillating pipe flow encountered in Stirling the performance of a Stirling engine. They applied the empirical
engine heat exchangers. In this study they are compared to three correlation proposed by Tang and Cheng (1993) [7] and suggested
classical unidirectional flow equations (Table 1). that the correlation was valid under the following range of condi-
As early as 1929 Richardson and Tyler [5] discovered the so- tions: 11 < Remax<10,995; 1.75<Reu<45. Walther et al. (1998) [6]
called ‘annular effect’ in an oscillatory flow in a pipe, i.e. the performed a theoretical analysis to evaluate the influence of
maximum velocity in an oscillatory flow occurs near the wall rather developing flow on the heat transfer associated with laminar
than at the centre of the pipe, as in the case with unidirectional oscillating pipe flow. Their heat transfer correlation (4) (the integral
steady flow. This applies especially at higher cycle frequencies [6]. Nusselt number describing the averaged wall heat transfer as
Tang and Cheng (1993) [7] (Eq. (1), Table 1) employed multivariate a function of the dimensionless oscillating frequency, the pipe
length, and the bulk velocity) was plotted for the range: 1000

Remax
Retrans
max ; 50
Reu
900; 0.2
L
900, where
pffiffiffiffiffiffiffiffi
Retrans
max ¼ 400 Reu .
* Corresponding author. Tel.: þ358 505381394.
E-mail addresses: maunu.kuosa@aalto.fi (M. Kuosa), kari.saari@aalto.fi (K. Saari), In this study Eqs. (1)e(4) were first tested by numerical calcu-
ari.kankkunen@aalto.fi (A. Kankkunen), tmt@sppower.no (T.-M. Tveit). lations in a tubular heat exchanger and then compared to three

1359-4311/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2012.03.023
16 M. Kuosa et al. / Applied Thermal Engineering 45-46 (2012) 15e23

Nomenclature VSE swept volume of piston

v mean velocity of flow
A cross-sectional area of moving element or flow x fluid or piston displacement, local coordinate
Ao, AR, Au dimensionless amplitude of fluid displacement or xmax amplitude of fluid displacement
stroke
a1, a2 inside and outside width of rectangular duct Greek symbols
b1, b2 inside and outside height of rectangular duct a heat transfer coefficient
c outer leg of triangular duct D difference
Cf Darcy friction factor dw wall thickness
d inner diameter of circular tube εl entrance effect correction factor
dh hydraulic diameter of duct or tube hc Carnot efficiency
f function in the correlation formula given by Petukhov l Fanning friction factor
and Popov lw conductance of wall
G total conductance L relative amplitude of fluid displacement, critical length
L tube length ratio
Nu Nusselt number m dynamic viscosity
P shaft power r density
PF pumping losses sw shear force at duct wall
Pr Prandtl number F heat flow
Re Reynolds number 4 phase angle, drag coefficient induced by sudden
qm mass flow rate changes of flow by geometry
qv volume flow rate u oscillation frequency
Rem Re average over half a cycle
Reu kinetic Reynolds number (¼ rud2/m) subcripts
T temperature i inner
TH heat source temperature o outer
TL coolant temperature C cooler
um cross-sectional mean velocity E heater
umax amplitude of the cross-sectional mean velocity DE dead volume
u mean velocity of unidirectional flow F friction
VE momental expansion volume SE swept volume
VDE dead volume of expansion space x local

classical correlations (5)e(7) by Dittus and Boelter [10], Petukhov heat transfer of oscillating flow by narrow boundary layers on the
and Popov [11], and Gnielinski [12], respectively, meant for unidi- walls of heater and cooler tubes. In addition to the circular tubes,
rectional turbulent developed flow but applied to the oscillating circumferentially slotted and non-circular rectangular and trian-
pipe flow conditions. Eq. (7) of Gnielinski [12] covers a large Re gular tubes are studied to improve the performance of the Stirling
range: 3000 < Re < 106. engine that is being studied.
The reciprocating flow in Stirling engines changes its direction
constantly. During one cycle the laminar, transitional, and turbulent 2. Assumptions and input values
forms of flow modes occur. In this paper a high-frequency (8-Hz)
low heat source temperature engine is considered. The review of In order to calculate the heat transfer in the heat exchangers the
these equations is performed in order to find tools to enhance the following assumptions are made according to de Monte et al.

Table 1
Oscillating and classical unidirectional heat transfer equations.

 2
Au xmax p$um d Re
Nu ¼ 0:494 þ 0:0777$ $Re0:7  0:00162$Re0:4 Re0:8
u Au ¼ ¼ ¼ (1)
1 þ Au L u$L L Reu

xmax
Nu ¼ 0:02$A0:85
o $Re0:58
u A0 ¼ (2)
d
 2
AR 1 dh Remax
Nu ¼ 0:494 þ 0:0777$ $Re0:7  0:00162$Re0:4 $ð4$Reu Þ0:8 AR ¼ (3)
1 þ AR 4 L Reu
 
 Re 0:951 0:703 1 L Reu
Nu ¼ 5:78 þ 0:00918$Re0:969 $L0:367 þ 0:178  L $Re0:526 L ¼ (4)
u Remax  u 2 d Remax

Nu ¼ 0:023$Re0:8 Pr 0:4 (5)

ðf =2ÞRe Pr
Nu ¼   f ¼ ð3:64logRe  3:28Þ2
  f 1=2   (6)
ð1 þ 13:6$f Þ þ 11:7 þ 1:8$Pr 1=3 Pr 2=3  1
2

Nu ¼
ðf =8Þ$ðRe  1000Þ$Pr
  3000 < Re < 106 f ¼ ð0:79$lnRe  1:64Þ2 (7)
1 þ 12:7$ðf =8Þ$ Pr 2=3  1 104 < Re < 5,106
 M. Kuosa et al. / Applied Thermal Engineering 45-46 (2012) 15e23 17

where 4 is the crank angle. By substituting V ¼ x$A and AE ¼ ASE ¼

ADE ¼ A, we get

xmax $A
xE $A ¼ ð1  cosðutÞÞ þ xDE $A (9)
2
TH TL
Differentiating Eq. (9) with respect to time gives the cross-
sectional mean velocity um

xmax $u  xmax $u
um ¼ $sin ut ¼ $sin4 (10)
2 2
The phase angle 4 ¼ ut of the cross-sectional mean velocity
Fig. 1. Alpha-type Stirling engine. ranges from 0 to 2p. The amplitude of the cross-sectional mean
velocity umax (Eq. (11)) is obtained when 4 ¼ p/2. It is related to the
amplitude of the fluid displacement xmax by the piston [8] (stroke,
bore: 0.15 m, 0.3 m) and is the distance between the top and the
(1996) [9]: 1) the working fluid temperatures in the heater and
bottom dead ends.
cooler are uniform and time-independent; 2) the ideal gas equation
applies; 3) the cycle gas density in the heat exchangers is constant; xmax $u
4) an ‘alpha’-type engine is considered; 5) the machine is working umax ¼ (11)
2
in cyclic steady-state operation, characterised by sinusoidal laws of
Eq. (10) is used as the hydrodynamic boundary condition for the
motion of the piston. Input values are presented in the context of
calculation of the heat transfer characteristics in the oscillating pipe
the calculations.
flow. The mean velocity um is illustrated in Fig. 2 as a function of the
The alpha-type engine is one of three main types of Stirling
crank angle (where umax is 3.75 m/s).
engine arrangements. The alpha-type engine uses two pistons that
By means of um and umax we can calculate Rem, Remax and qm m
mutually compress, expand, and move the working gas between
and qm max. These parameters are encountered in the literature
hot and cold spaces (Fig. 1). See, for instance, Martini 1983 [13] for
correlations (1)e(4) of the oscillating pipe flow.
a more detailed description of the arrangements. The tubular heat
Fig. 2 presents the cross-sectional mean velocity for piston. For
exchangers in the engine that are studied are realised by counter-
the heater tubes that were studied (helium as working medium:
flow configuration.
150  C, 120 bar) the corresponding cross-sectional mean velocity
amplitude umax is 14 m/s and Remax ¼ 47,423 at a frequency of 8 Hz
3. Piston motion and dimensionless numbers
because of the different face areas between the piston and the
bundle of 670 tubes. According to de Monte et al. (1996) [9],
The time-averaged velocity in a Stirling engine is zero, or at least
very low, since in a Stirling engine we do not have a pulsating flow, 2
but an oscillating flow, meaning that the stationary component of Rem ¼ Remax (12)
p
the flow rate is zero.
The Schmidt theory is one of the isothermal calculation The Re (¼Rem) linked to the Eqs. (1) And (3)e(7) is based on the
methods for Stirling engines [14]. For the alpha-type engine the cross-sectional mean velocity um. It is a kind of average over half
momental expansion volume VE in Eq. (8) is function of a swept a cycle (Eq. (12)).
volume of the piston VSE and the dead volume of expansion space
VDE [14]. 4. Calculation procedure

VSE Eq. (2) by Zhao and Cheng (1990) [8] is used in demonstrating
VE ¼ ð1  cos4Þ þ VDE (8)
2 the calculation sequence. um is calculated from the Eq. (10), while u

5
4
3
2
um / m/s

1
0
-1 0 45 90 135 180 225 270 315 360

-2
-3
-4
-5
o
φ/
Fig. 2. Cross-sectional mean velocity um as a function of the piston crank angle.
18 M. Kuosa et al. / Applied Thermal Engineering 45-46 (2012) 15e23

1,2

0,8
Nu / Ao0.85
0,6

0,4

0,2

0
0 200 400 600 800 1000 1200
Reω / -
Fig. 3. Nu=A0;85
o as a function of Reu.

and the stroke xmax are constants of 50.3 1/s (w8 Hz) and 0.15 m 5. Heat transfer
respectively. The phase (crank) angle of the piston 4 and um vary
according to Fig. 2. Rem also varies but Reu has a constant value of Pressurised hot water (TH) on the shell side is used as a heat
1022. The thermophysical properties of helium at a pressure of source, as shown in Fig. 1. For the heat transfer calculations
120 bar [15] are used to calculate the density r, the dynamic a constant temperature of 150  C and a pressure of 120 bar inside
viscosity m, the Prandtl number Pr, and the conductivity l that the heater tubes were selected for the working medium, helium.
appear in the dimensionless numbers. By using the mean velocity The Nu numbers of correlations (1)e(7) on the helium side are
amplitude of 14 m/s encountered in the heated tubes (umax), we can plotted in Fig. 4 as a function of Reu.
calculate the amplitude of the fluid displacement in the tubes Fig. 4 indicates that when the frequency of the oscillation is
(xmax) from Eq. (11). The dimensionless oscillating amplitude of small some of both correlation groups (oscillating and unidirec-
fluid Ao in the pipes is calculated through the pipe diameter d and tional turbulent flow) have almost the same slopes and the graphs
xmax. Finally Eq. (2) gives the mean value of 52.6 W/(m2K) for Nu. are almost convergent. As the Reu increases the slope of the laminar
According to Fig. 2, we see that um receives its maximum value oscillating correlations starts to decrease gradually.
when the phase angle is 90 . By changing the frequency 0 < f
8 Hz Eq. (2) from Zhao and Cheng (1996) [8] gives the second largest
(0 < u
50 1/s) and keeping 4 as a constant (p/2), we can plot Nu values of the oscillating graphs. It is a conservative estimate of
Nu=A0;85
o as a function of Reu [8], as presented in Fig. 3. the heat transfer rate for a reciprocating flow in a pipe of infinite
The plotted Nu=A0;85
o as a function of Reu is well compatible with length. Here only two dimensionless parameters, Ao and Reu, are
Zhao and Cheng 1996 [8] but it is extended here to the range included. The other oscillation flow Eqs. (1), (3), and (4), which
0 < Reu
1022, while it was originally presented for the include the classical Re number as a third non-dimensional term
range w 0.500. (commonly used in the unidirectional pipe flow) give smaller Nu
In this paper we will plot the graphs for Nu as a function of Reu. values.

Zhao & Cheng Tang & Cheng mod. Walther et al. mod. Walther et al.
de Monte et al. Dittus & Boelter Gnielinski Petukhov & Popov

80

60
Nu / -

40

20

0
0 200 400 600 800 1000 1200
Reω / -
Fig. 4. Nu numbers of correlations studied as a function of Reu.
 M. Kuosa et al. / Applied Thermal Engineering 45-46 (2012) 15e23 19

Fig. 5. The 90 elbow (left) and the open-end 90 edge (right).

Fig. 7. The heater tube with three circumferential slots.

By comparing the non-dimensional oscillating amplitudes AR

and Au in the sources de Monte et al. (1996) [9] and Tang and Cheng
(1993) [7], it is evident that Re ¼ Remax. The modified (mod) According to Mills (1995) [12], turbulent flows with simply
diagrams are arrived at by this assumption. Under this circum- defined hydrodynamics at a pipe entrance are seldom encountered
stance the highest oscillating Nu values were plotted from Eq. (1). in engineering practice. Most often there is a sharp 90 edge,
Otherwise, we did not get the positive Nu values from that a bend, or an elbow (Fig. 5).
equation. The corresponding hydrodynamic entrance lengths vary from
We did not find the oscillating heat transfer correlations for the about 10 to 15 diameters, where no large-scale eddies are present,
turbulent flow. However, we can conclude that unidirectional to about 30e40 diameters, where there are large-scale eddies [12].
turbulent correlations give slightly higher dimensionless heat The effect of entrance configuration on the average heat transfer for
transfer coefficients at the higher frequencies (Reu numbers) than the turbulent tube flow is presented in the Fig. 6.
the calculated coefficients from the oscillating equations of Zhao It is visible in the Fig. 6 that the open-ended 90 edge and 90
and Cheng (2) and the modified Tang and Cheng (1). elbow give the largest eddies and the most long-range average Nu
Ahn and Ibrahim (1992) [16] conducted two-dimensional enhancement. The observation that ‘the edges improve the inside
analysis of oscillating gas flow in laminar, transition and turbu- heat transfer’ in the tubular heat exchanger was studied more
lent regimes inside Stirling engine heat exchangers. Numerical data carefully.
were compared to the experimental results of other investigators It is suggested that a few sharp edges inside the pipe could
with good accuracy in laminar and turbulent flow regimes. They produce a greater local Nu number and a narrower laminar
reviewed several criteria for the critical Reynolds number for boundary layer along the entire length compared to the smooth
oscillating pipe flow. The onset of the turbulence and frictional pipe. The axial slots (grooves) are used e.g. in the annular ducts of
losses in an oscillatory turbulent pipe flow is discussed in another electric machines on the stator and/or rotor surfaces in order to
source, Zhao and Cheng (1996) [17]. The oscillating flow is in the enhance the heat transfer [19].
pffiffiffiffiffiffiffiffi
turbulent region when Ao Reu  961 [17]. At the 8-Hz frequency The heater tube length (170 mm) was divided into four similar
that was studied we get 2966 for such a product (of Ao ¼ 93 and parts by means of three circumferential slots (Fig. 7).
Reu ¼ 1017) and the flow is turbulent. The depth and width of one slot opening are assumed to be
2e3 mm and 7.5 mm, respectively. In this case the effective heat
6. Heat transfer enhancement transfer length is still 170 mm, even if the total pipe length with
slots is increased to 192.25 mm.
In this work several traditional and some up-to-date heat A factor and numerical fit was made for the open-ended 90
transfer enhancement methods, e.g. regarding the laminar edge entrance effect. Eq. (2) from Zhao and Cheng (1996) [8] was
convective heat transfer problem as a conduction problem [18], multiplied by this factor. The effects in the entrance and after the
were gone through. However, here only the effect of circumferen- sequential slots in the straight plain tube were assumed to be
tial slots in a circular pipe and the enhancement of heat transfer by identical. In addition this was compared to the factor εl in Eq. (13) of
non-circular square and triangular ducts are reported. Sukomel et al. (1987) [20] (Fig. 8).

open end 90 edge 90 elbow T-confluence

90 round bend 180 round bend calming section
3,5
3
2,5
Nux / Nu

2
1,5
1
0,5
0
0 20 40 60 80 100
x/d
Fig. 6. The effect of entrance configuration on the average heat transfer in the turbulent tube flow.
20 M. Kuosa et al. / Applied Thermal Engineering 45-46 (2012) 15e23

Zhao & Cheng x EpsL, no slots x EpsL (Sukomel et al.)

90 edge, no slots open end 90 edge
3,5
3
Nux / Nu 2,5
2
1,5
1
0,5
0
0 5 10 15 20 25 30 35
x/d
Fig. 8. Local Nux enhancement divided by cycle-averaged Nusselt number Nu.

Nux ¼ 0:022Re0:8 Pr 0:43 εl 7. Pressure losses

(13)
εl ¼ 1:38ðx=dÞ0:12
The internal flow pressure loss of the smooth and slotted tubes
In Fig. 8 we can see the mean heat transfer from Zhao and Cheng was studied and the loss of the smooth tubes of a non-circular
(2) and the entrance effect without and with circumferential slots. cross-section (rectangular and triangular) was compared to the
The effect of the factor εl [20] with slots compared to the mean heat loss in the circular tube.
transfer looks promising. It is selected to account for the change in Two different friction factor definitions are in common use in
the heat transfer coefficient in the initial section of the tube and the literature [21], the Darcy friction factor Cf and the Fanning
after the slots. friction factor l:
The performance of the selected non-circular cross-sections of
rectangular and triangular tubes was investigated and compared to 2sw 1
Cf ¼ ¼ l (17)
that of the circular tube. Eqs. (14)e(16) for the total conductance G ru 4
were derived for the circular, rectangular, and equilateral triangular
The losses are calculated as follows:
tubes.
  2
L rv
l
¼
1
þ
1
þ
1
(14) Dp ¼ l þ S4 (18)
G ai pdi lw di þ do ao pdo dh 2
p
dw 2 The drag coefficient 4 includes all sudden changes in the flow as
l is the pipe length, lw and dw the conductance and thickness of the a result of the geometry. dh is the hydraulic diameter.
wall. For the circular pipe di and do are the inner and the outer pipe Eq. (18) applies all kinds of flow patterns, e.g. laminar and
diameters and ai and ao the corresponding heat transfer coeffi- turbulent. The drag coefficient l depends on the Reynolds number
cients at the surfaces. for the flow within the duct. The hydraulic diameter dh allows Eq.
(18) to be applied to any given cross-section [22]. The plot of l is
l 1 1 1 presented in the source [22] for tubes of various cross-sections as
¼ þ þ (15) a function of the Reynolds number.
G 2ai ða1 þ a2 Þ a1 þ a2 þ b1 þ b2 2ao ðb1 þ b2 Þ
2lw The friction coefficient Cf of unidirectional flow can be calcu-
b1  a1
lated from the correlation (19) for laminar and from the correlation
For the rectangular duct a1 and a2 (and b1 and b2) are the inside (20) for the turbulent flow [23].
(and outside) width and height of the duct.
64
l 1 1 1 Cf ¼ ; Re < 2000 (19)
¼  þ  þ (16) Re
G 2d w lw dw 3ao c
3ai c  3 c
tan30+ dw tan30+ Cf ¼ 0:079Re4 ;
1
Re < 105 (20)

For the triangular equilateral duct c is the outer leg of the The friction coefficient of a cyclically turbulent oscillatory flow
triangle and dw the wall thickness. was measured by Zhao and Cheng in 1996 (Eq. (21)) [17].
It was found out that the smooth non-circular square and !
triangular tubes have greater free flow and exterior areas than the 1 76:6
Cf ¼ þ 0:40624 (21)
circular tube with an equivalent hydraulic diameter [4]. Hence, for Ao Re1:2
u
the equivalent heat flow magnitude the required tube lengths of
square and triangular ducts were 21% and 39% shorter than the which is valid in the range of 81
Reu
540 and 54.4
Ao
113.5.
length of the circular tube. However, the dead volume caused by The laminar and turbulent drag coefficients for unidirectional
the heat exchanger remains constant. tube flow were plotted when l ¼ 4Cf as a function of Reu. This curve
 M. Kuosa et al. / Applied Thermal Engineering 45-46 (2012) 15e23 21

unidirectional Zhao & Cheng 1996

0,1

0,08

λ/- 0,06

0,04

0,02

0
0 200 400 600 800 1000 1200
Reω / -
Fig. 9. The laminar and turbulent drag coefficients for unidirectional tube flow compared to the drag coefficient of a cyclically turbulent oscillatory flow.

is compared to the friction coefficient graph of the cyclically Friction coefficients and Nusselt numbers for fully developed
turbulent oscillatory flow in Fig. 9. laminar flow are presented for a large variety of tube shapes in the
The oscillatory Eq. (21) for the turbulent flow gives slightly source Rosenhow et al. (1985) [24]. In all cases the hydrodynamic
lower values for the friction coefficient than the unidirectional diameter is used as the characteristic length in the Reynolds and
friction Eq. (20) (Fig. 9). In the studied heat exchanger that was Nusselt numbers. The friction CfRe and Nu values of the regular
studied Ao ¼ 93.3 and Reu ¼ 1016.8 (w 8 Hz) and the flow is polygonal tubes decrease in a gently sloping manner as the number
turbulent, after Zhao and Cheng [17]. of sides decreases.
The oscillating pressure loss for the circumferentially slotted Bejan and Lorente (2006) suggest that the round tube shape is
circular tube (3 slots, 2 mm depth) is 8.17 times higher than for the the best regarding the laminar flow resistance. However, nearly
smooth tube at a frequency of 8 Hz. The friction loss towards one round shapes such as hexagonal and square tubes perform almost
tube is small (w1 W), however. as well after the constructal law. Even if the duct cross-section is
The pressure losses of unidirectional and oscillating flow for the imperfect e that is, with features such as sharp corners, which
smooth circular and non-circular tubes are presented in the Fig. 10. concentrate fluid friction e its performance is already as good as it
It is visible that unidirectional flow in the tube retains larger losses can be [25].
than the oscillating case. The length of the circular tube is 0.17 m
and the inner diameter is 0.006 m. 8. Performance with slotted tubes
The square and triangular tubes correspondingly have only
marginally lower losses (as a result of the diminished pipe lengths The thermodynamic performance of an idealised Stirling engine
that give equivalent heat transfer) compared to the smooth circular was studied by comparing the heat transfer between circum-
tube. ferentially slotted and smooth circular tubes in the heat

unidir osc osc square osc triangular

400

300
Δp / Pa

200

100

0
0 200 400 600 800 1000 1200
Reω / -
Fig. 10. Pressure losses in one smooth tube: unidirectional and oscillating flow in the circular tube and the oscillating flow in non-circular square and triangular tubes.
22 M. Kuosa et al. / Applied Thermal Engineering 45-46 (2012) 15e23

exchangers. The identical heater and cooler by the geometry for the heater (and conductance GC and GE), we got a TE increase of 2.5  C,
engine that was studied were roughly dimensioned. The effect of a TC fall of 2.3  C, and Dh w 0.0272. It means a 2.72% increment in
the slots was considered from two points of view. the efficiency of the Stirling engine by using these circumferentially
First, the difference in the shaft power was calculated according slotted tubes instead of smooth ones.
to Eq. (22) when comparing slotted and smooth tubes in the heater.
9. Conclusions
DP ¼ h$DFE (22)

where DP is the difference in the shaft power when using slotted Earlier fully laminar and turbulent steady unidirectional equa-
and smooth tubes, DFE the corresponding heat flow difference in tions were used as an approximation for convective heat transfer in
the heater and h the efficiency (¼ 0.22) of the selected Stirling a Stirling engine. Hitherto the literature has dealt with the laminar
engine. In this first study the logarithmic temperature difference oscillating correlations. In this study the existing laminar equations
DTE between helium and water in the heater was supposed to are compared and extended to the turbulent range. It is noticed that
remain constant. The heater heat flow FE was increased as a result some of them give almost the same Nu values as turbulent unidi-
of the rise in the conductance GE and the inside heat transfer rectional equations.
coefficient ai (Eqs. (14) and (23)) while using the circumferential The salient parameters are the amplitude of the cross-sectional
slots compared to the smooth case. mean velocity, the kinetic Reynolds number, and the dimensionless
oscillating amplitude. The equations based on these two latest
FE ¼ GE DTE (23) parameters give Nu values that are almost as high as the traditional
equations in the turbulent range.
The FE values were calculated for the slotted and smooth heat Heat transfer enhancement is studied with a few circumferen-
exchanger tubes used in the Eq. (22). tial slots inside the smooth tubes. The effect of edges in the
The corresponding (oscillating flow) pressure losses were entrance and the sequential slots is assumed to be identical. In
calculated by Eqs. (17), (18), (21) and Eq. (24) addition, non-circular rectangular and triangular tubes are inves-
DPF ¼ qV $Dp (24) tigated. The square and triangular tubes have greater free flow and
exterior areas than the circular tube with the same hydraulic
for a heater consisting of 670 tubes. The drag coefficients 4 was diameter. Hence the duct lengths are also shorter. However, the
taken from the source [26]. As a result we got the power ratio dead volume in the heat exchanger remains constant.
(Eq. (25)) The friction coefficient of oscillating flow is lower than in the
unidirectional approximation. The frictional losses in slotted tubes
DP 1336; 8 W
¼ ¼ 2:22 (25) are greater but with square and triangular tubes they are margin-
DPF 603 W ally lower than with smooth circular ones.
The increment of the performance in the idealised Stirling
Here the factor εl (Eq. (13)) [20] together with Eq. (2) from Zhao
engine with slotted heat exchanger tubes in oscillating flow is
and Cheng [8] was used to calculate the (arithmetic) mean heat
compared to smooth tubes. The ratio of the gain in the shaft power
transfer coefficients for the slotted pipes (3 slots, depth 2 mm) and
and pumping losses is 2.22 compared to smooth tubes. The Carnot
the difference DFE and DP.
efficiency increment is 2.72%.
From Eq. (25) we can see, on the basis of previous assumptions,
With these simplified examinations it can be concluded that
that the gain in the shaft power of the Stirling engine compared to
methods (few edges or circumferential slots) that form narrower
the increase in the pumping losses is 2.22 when these circum-
boundary layers along the length of the tube but do not signifi-
ferentially slotted tubes are used. The same ratio is 1.94 for the
cantly increase the dead volume or pressure loss are recommended
tubes consisting of 3 slots with a depth of 3 mm.
in the design of the heater and cooler of the Stirling engine.
We also considered the efficiency increment. In this case we
kept the cooler and heater heat flows FC and FE and the shell-side
(water) temperatures constant (Fslotted pipe ¼ Fsmooth pipe). The Acknowledgements
subscript C refers to the cooler. As a result of the slots in the tube
both GC and GE increase and the logarithmic DTC and DTE are This work is supported by the Single-Phase Power AS. Thanks
reduced compared to the smooth tube (Eq. (23)). Hence the new are also given to the research milieu at Aalto University.
working fluid temperatures TE and TC can be solved for the helium
in the slotted heater and cooler and the Carnot efficiency (Eq. References
(26)) [27].
[1] G. Walker, Stirling Engines, Clarendon Press, Oxford, 1980.
T [2] H. Carlsen, Stirling Engines for Biomass e What Is the Problem? 12th Stirling
hc ¼ 1  C (26) Engine Conference, 12th ISEC, 7e9 September Durham, 2005, 12p.
TE [3] T. Colmant, M. Deledicque, O. Squilbin, S. Bonnet, M. Alaphilippe, P. Stouffs,
Quasi steady flow modelling of a small Stirling engine: comparison between
The increment in the efficiency of the Stirling engine can be calculated and measured instantaneous temperatures, 11th Stirling Engine
calculated as Conference, 11th ISEC, Rome pp. 9e23 (November 2003).
[4] A. Altman, Stirling engine heat transfer e a snappy primer heat exchanger
hc slotted  hc smooth design you can understand, Ó 2000 Alan Altman all rights reserved.
Dhc ¼ (27) [5] E.G. Richardson, E. Tyler, The transverse velocity gradient near the mouths of
hc smooth pipes in which an alternating or continuous flow of air is established, Proc.
Phys. Soc. London 42 (1929) 1e15.
where hc smooth is the solved Carnot efficiency with a smooth heater [6] C. Walther, H.-D. Kühl, T. Pfeffer, S. Schulzs, Influence of developing flow on
the heat transfer in laminar oscillating pipe flow, Forschingenieurwes 64
and cooler tubes and hc slotted is the calculated efficiency with (1998) 55e63 Ó Springer-Verlag 1998.
slotted tubes. [7] X. Tang, P. Cheng, Correlations of the cycle averaged Nusselt number in
By using the heat transfer enhancement by the factor εl [20] a periodically reversing pipe flow, Int. Comm. Heat Mass Transfer 20 (1993)
161e172.
together with Eq. (2) from Zhao and Cheng [8] to calculate the [8] T. Zhao, P. Cheng, Oscillatory heat transfer in a pipe subjected to a laminar
working fluid heat transfer coefficients in the slotted cooler and the reciprocating flow, ASME J. Heat Transfer 118 (1996) 592e598.
 M. Kuosa et al. / Applied Thermal Engineering 45-46 (2012) 15e23 23

[9] F. de Monte, G. Galli, F. Marcotullio, An analytical oscillating-flow thermal anal- [18] Z.Y. Guy, D.Y. Li, B.X. Wang, A novel concept for convective heat transfer
ysis of the heat exchangers and regenerator in Stirling machines, 0-7803-3547-3- enhancement, Int. J. Heat Mass Transfer 41 (14) (1998) 2221e2225.
7/16 $4.00 Ó 1996 IEEE, downloaded on June 10, 2010 (from IEEE Xplore). [19] M. Kuosa, P. Sallinen, A. Reunanen, J. Backman, J. Larjola, L. Koskelainen,
[10] P. Holman, Heat Transfer, ninth ed. McGraw-Hill Education, 2002, interna- Numerical and experimental modelling of gas flow and heat transfer in the air
tional edition. gap of an electric machine, part II: grooved surfaces, J. Thermal Sci. 14 (1)
[11] X. Wang, N. Zhang, Numerical analysis of heat transfer in pulsating turbulent (March 2005) 48e55.
flow in a pipe, Int. J. Heat Mass Transfer 48 (2005) 3957e3970. [20] V.P. Isachenko, V.A. Osipova, A.S. Sukomel, Heat Transfer, Mir Publishers,
[12] A.F. Mills, Basic Heat and Mass Transfer (1995). Moscow, 1987.
[13] W.R. Martini, Stirling Engine Design Manual, National Aeronautics and Space [21] F.M. White, Viscous Fluid Flow international edition, McGraw-Hill Inc., 1991.
Administration, 1983, 1-4102-1604-7. [22] VDI Heat Atlas, VDI-Verlag GmbH, Düsseldorf, 1993.
[14] K. Hitara, Schmidt theory for Stirling engines, tentative version on January 20, [23] N.-E. Fagerholm, Termodynamiikka 479, Otakustantamo, Jyväskylä Finland,
1997. 1986, (in Finnish).
[15] V. Arp, R.D. McCarty, D.G. Friend, Thermophysical Properties of Helium-4 from [24] W.M. Rohsenov, J.P. Hartnett, E.N. Ganic, Handbook of Heat Transfer Funda-
0.8 to 1500 K with Pressures to 2000 MPa (September 1998) NIST Technical mentals, second ed. McGraw-Hill Book Company, 1985.
Note 1334 (revised). [25] A. Bejan, S. Lorente, Constructal theory of generation of configuration in
[16] K.H. Ahn, M.B. Ibrahim, Laminar/turbulent oscillating flow in circular pipes, nature and engineering, J. Appl. Phys. 100 (2006) 041301.
Int. J. Heat Fluid Flow 13 (4) (1992). [26] V. W. Kotiaho, M. J. Lampinen, A. Seppälä;, Termodynamiikan ja lämmön-
[17] T.S. Zhao, P. Cheng, Experimental studies on the onset of turbulence and siirto-opin taulukoita, raportti 136, Espoo Finland 2004 ((in Finnish)).
frictional losses in an oscillatory turbulent pipe flow, Int. J. Heat Fluid Flow 17 [27] M.J. Moran, H.N. Shapiro, Fundamentals of Engineering Thermodynamics,
(1996) 356e362. second ed. Wiley, New York, USA, 1992.