Energy 44 (2012) 902e910

Contents lists available at SciVerse ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Thermodynamic optimization of a Stirling engine

M.C. Campos a, J.V.C. Vargas a, *, J.C. Ordonez b
a
Programa de Pós-Graduação em Engenharia e Ciência dos Materiais, PIPE, and Departamento de Engenharia Mecânica, Universidade Federal do Paraná, UFPR, CP 19011,
81531-980 Curitiba, PR, Brazil
b
Department of Mechanical Engineering and Center for Advanced Power Systems, Florida State University, Tallahassee, FL 32310-6046, USA

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

Article history: A Stirling engine configuration consisting of two cylinders, a regenerator and a sliding disc actuating
Received 4 November 2011 mechanism (“swashplate”) is considered in this paper. A mathematical model, which combines funda-
Received in revised form mental and empirical correlations, and principles of classical thermodynamics, mass and heat transfer
9 April 2012
accounting for variable heat transfer coefficients, is developed. The proposed model is then utilized to
Accepted 30 April 2012
Available online 8 June 2012
simulate numerically the system transient and steady state response under different operating and
design conditions. A system global optimization for maximum performance in the search for optimal
parameters that lead to maximum cycle efficiency is performed with low computational time. Appro-
Keywords:
Stirling engines
priate dimensionless groups are identified and the results presented in normalized charts for general
Thermodynamic optimization application. The numerical results show that the two-way maximized system efficiency, hmax;max , occurs
Heat engines when two system characteristic parameters, the ratio between the total swept volume during the
expansion, and the total swept volume, 4, and the ratio between the heat transfer area of the hot side
heat exchanger and the total heat exchange area, y, are optimally selected, i.e., ð4; yÞopt yð0:5; 0:4Þ. The
two-way maximized cycle efficiency found with respect to the optimized parameters is sharp, in the
sense that a 225% variation of the calculated efficiency values was observed within the range of tested
configurations in this study, and “robust” (i.e., relatively insensitive) to the variation of several param-
eters, thus stressing the importance to be considered in actual design. It is also found that the twice-
maximized cycle efficiency and the total engine work output increase monotonically with the temper-
ature of the hot source, Th. As a result, the model is expected to be a useful tool for simulation, design,
and optimization of Stirling engines.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction A previous work by Curzon and Ahlborn [2] showed that the
efficiency of a Carnot engine with a finite heat transfer time
Stirling engines are classified as thermal (or heat) engines depends only upon the temperatures Th and Tc, and is given by:
within the theoretical thermodynamic framework. The Stirling sffiffiffiffiffiffi
engine thermodynamic cycle displays a theoretical thermal effi- Tc
ciency equal to the Carnot limit [1]. However, the heat transfer in hmax ¼ 1 (1)
TH
this ideal limit must occur isothermally and reversibly, which
demands infinite time, therefore, zero power is observed in the As a result, the Stirling engine is a promising alternative, taking
efficiency limit, 1Tc/Th, in which Th is the temperature of the hot into account that it is an external combustion engine with possi-
reservoir and Tc the temperature of the cold reservoir. The bilities for better combustion control and the potential to use
mechanical difficulty to accomplish the ideal volume variation in multiple fuels. Therefore, with proper design, Stirling engines are
the Stirling cycle is also responsible to separate the real from the expected to be less expensive and less polluting than Diesel engines
ideal cycle. and even gas turbines.
The expectation of achieving thermal efficiencies close to the
Carnot limit developed interest in the Stirling engine among
researchers. Several authors have published theoretical and
* Corresponding author. Dept. of Mechanical Engineering, Universidade Federal experimental studies on that possibility. Reader and Hooper [3]
do Parana, Caixa Postal 19011, Curitiba, PR CEP 81531-980, Brazil. Tel.: þ55 41 3361
presented experimental measurements and analyses of Stirling
3307.
E-mail address: [email protected] (J.V.C. Vargas). engines. Uriel and Berchowitz [4] described analytically different

0360-5442/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2012.04.060
 M.C. Campos et al. / Energy 44 (2012) 902e910 903

actuation mechanisms used in Stirling engines and presented et al. [14] performed the pre-sizing of the thermal storage and the
mathematical models for the thermodynamic simulation of the solar receiver area of a thermodynamic micro power station and
cycle considering variable heat transfer coefficients and the pres- searched for possible optimum temperature levels. However, all
sure drop in tubular heat exchangers of circular cross-section. such analyses were performed with steady state models.
A finite time thermodynamic analysis of the Stirling cycle was Two studies were found in the literature [15,16] that utilized
presented by Ladas [5] with a sinusoidal volume variation and so dynamic models which included thermal losses to optimize the
called Beta configuration. The mass and energy balance equations performance and design parameters of the Stirling engine, after
were nondimensionalized and the maximum efficiency obtained experimental validation against data obtained from the General
was 14%. Motor GPU-3 Stirling engine prototype, showing significant engine
Finite time thermodynamics, endoreversible and simplified efficiency increase with respect to the original configuration, i.e.,
models have been used for Stirling engine optimization. For from 39% to 51% increase. Although the study was dimensional and
example: Blank and Wu [6] studied the power output and thermal specific to the GPU-3 Stirling engine, it shows the importance of
efficiency of a finite time, optimized, extra-terrestrial, solar-radiant using dynamic models for engine optimization in order to produce
Stirling heat engine, in which the heat source and sink were assumed more realistic results. For example, Karabulut [17] analyzed a free
to have infinite heat-capacity rates, obtaining expressions for piston Stirling engine with a dynamic model, which made possible
optimum power and efficiency at optimum power; heat pumps to show that the closed cycle performs a stable operation within
based on the reversed Stirling cycle were also studied [7] using a small range of the hot end temperature and damping coefficient
a simplified model, which allowed a first optimization of real gas of the piston motion that could be partially enlarged by inverting
cycles, showing that efficiencies much higher than those achievable the engine into an open-cycle engine.
with an ideal gas, and similar to those of vaporecompression cycles The bibliographic review shows that there is a lack of studies on
can be obtained; a finite time thermodynamic optimization of the thermodynamic optimization of Stirling engines with dynamic,
a Stirling engine was performed by Popescu et al. [8], considering an dimensionless and more elaborated mathematical models. There-
endo- and exo-irreversible cycle, i.e., accounting for general irre- fore, this paper’s objective is to present a dimensionless dynamic
versibilities, finding optimum operating conditions leading to mathematical model to simulate the thermodynamic behavior of
maximum power output in good agreement with experimental a Stirling engine in the transient regime as a function of geometrical
data; Erbay and Yavuz [9] studied the Stirling heat engine operating and operating parameters relevant to the engine design. Appro-
in a closed regenerative thermodynamic cycle finding the maximum priate dimensionless groups are introduced in order to present
power density and efficiency, in addition to the compression ratio at normalized simulation results for general engineering application.
maximum power density; Bhattacharyya and Blank [10] reported Acknowledging the finite availability of space in any engineering
major theoretical considerations concerning the design of an project, a constraint accounting for the total volume occupied by
endoreversible Stirling cycle with ideal regeneration; Senft [11] used the engine is imposed. The mathematical model is then utilized to
the classic Schmidt thermodynamic model (isothermal model) for optimize some engine operating and design parameters for
Stirling engines and revisited the problem of identifying optimal maximum cycle efficiency.
engine geometry; Rogdakis et al. [12] performed the optimization of
stable operation of the Stirling cycle in the free piston configuration, 2. Mathematical model
and for simplicity reasons used the Schmidt analysis; a thermoeco-
nomic optimization of an irreversible Stirling cryogenic refrigerator Fig. 1 depicts a schematic diagram of the Stirling engine
cycle was presented by Tyagi et al. [13] finding that the effect of configuration considered in this work, consisting of: two cylinders,
regenerative effectiveness is more than those of the other parame- the hot and cold side heat exchangers, the regenerator and the
ters on all the performance parameters of the cycle, and Mathieu sliding disc actuating mechanism (“swashplate”). One possible

Fig. 1. Schematic representation of the sliding disk mechanism (“swashplate”) and cylinders-regenerator unit consisting of a hot space, a cold space and a regenerator.
904 M.C. Campos et al. / Energy 44 (2012) 902e910

arrangement is a four-cylinder swashplate-drive double-acting Vs2

engine (e.g., the Ford-Philips 4-215 engine [4]), which is the one V2 ¼ Vc þ Vd2 þ ½1 þ cosðutÞ (3)
2
considered in this work, i.e., each piston is responsible for the
where Vh and Vc represent the volume occupied by the hot and cold
compression process in one side and the expansion process in the
heat exchangers respectively, Vd-1 and Vd-2 represent the dead
other side.
volumes for expansion and compression, and Vs-1 and Vs-2 the total
The proposed thermodynamic model considers three control
swept volumes for expansion and compression for each cylinder.
volumes: hot space (CV-1), cold space (CV-2), and the regenerator
The phase angle between sets of cylinders, a, is shown in Fig. 2a,
(CV-3). The model is dynamic, therefore in the analysis, tempera-
and the total number of cylinders, results directly from its value. For
tures in the three control volumes vary with respect to time, and
example, for the swashplate mechanism considered in this work,
volume in CV-1 and CV-2 also vary in time due to piston-
with a ¼ 90 , the result is an engine with 4 cylinders and 4 modules
displacement, but CV-3 undergoes constant volume processes
as shown schematically in Fig. 2b.
since the regenerator volume is constant. The basic assumptions
In Equations (2) and (3) the term (Vh þ Vd-1) represents the total
are: uniform temperatures in the control volumes, negligible
dead expansion volume and (Vc þ Vd2) the total dead compression
pressure drops across the regenerators and heat exchangers, and
volume.
the working fluid is treated as an ideal gas. The idea is to model the
The time derivatives of the expansion and compression volumes
engine as an assembly of modules, like the one shown in Fig. 1, with
for the sliding disk mechanism are given by:
identical thermodynamic performance. It is therefore sufficient to
analyze one module and obtain the total engine power (or work) dV1 V
multiplying the power of one module by the engine’s number of ¼  s1 u sinðut þ aÞ (4)
dt 2
modules.
The pistons are driven by the swashplate mechanism illustrated
in Fig. 1, resulting in a pure sinusoidal reciprocating motion having dV2 V
¼  s2 u sinðutÞ (5)
a phase difference between the adjacent pistons, given by the angle dt 2
between pistons, a. Therefore, the total expansion volume, V1, and
the total compression volume, V2, vary in time, and are calculated The volumes Vh and Vc are functions of the heat exchanger
by [4]: geometry as follows:

Vs1 N1 $L1 $p$D21

V1 ¼ Vh þ Vd1 þ ½1 þ cosðut þ aÞ (2) Vh ¼ (6)
2 4

α = 90 o

b Compression Space
Cold Heat Exchanger
Module

4 1

Expansion Space

1 2 3 4

Hot Heat Exchanger

Regenerator

Fig. 2. (a) Illustration of the phase angle between piston units, a, and (b) The schematic representation of the components of one module in the double-acting swashplate Stirling
engine considered in this work.
 M.C. Campos et al. / Energy 44 (2012) 902e910 905

N2 $L2 $p$D22 The temperature TR is found using the regenerator effectiveness,

Vc ¼ (7) ε, which is also a function of the flow direction and the tempera-
4
tures T1, T2 and Tm, in the following way:
where N1, L1 and D1 are the number, length, and diameter of the
tubes in the hot side heat exchanger, respectively, and N2, L2 and D2 TR ¼ a½T2 þ εðTm  T2 Þ þ b½T1 þ εðTm  T1 Þ (17)
are the number, length and diameter of the tubes in the cold side
heat exchanger. The conditional temperatures, Tc1 and Tc2 are, therefore given
Assuming uniform internal pressure, p, and neglecting the by:
volume occupied by the gas inside the regenerator, the mass and
energy conservation equations for the expansion and compression Tc1 ¼ a$TR þ b$T1 (18)
volumes state that:
Tc2 ¼ a$T2 þ b$TR (19)
dm1
_
¼ m (8)
dt Equation (14) is rewritten, solving for dTm/dt:

dðm1 T1 Þ dV dTm cp ε dm1

cv ¼ Q_ 1  p 1 þ cp mT
_ c1 (9) ¼ ½aðT2  Tm Þ þ bðTm  T1 Þ (20)
dt dt dt mR cR dt

The heat transfer rates in the expansion and compression

dm1 dm2
þ ¼ 0 (10) volumes are obtained as follows:
dt dt
Q_ 1 ¼ h1 A1 $ðTh  T1 Þ (21)
dðm2 T2 Þ dV
cv ¼ Q_ 2  p 2 þ cp mT
_ c2 (11)
dt dt
Q_ 2 ¼ h2 A2 $ðTc  T2 Þ (22)
where Tc1 and Tc2 are conditional temperatures that depend on
the gas flow direction inside the engine and on the heat exchange in where h1 and h2 are the heat transfer convection coefficients, A1
the regenerator; m1 and T1 are the mass and temperature in the and A2 the heat transfer areas in the expansion and compression
expansion volume, respectively; m2 and T2 are the equivalent volumes, respectively. The heat transfer areas A1 and A2, are given
quantities in the compression volume; cp and cv are the working by:
fluid specific heat at constant pressure and at constant volume,
respectively, and Q_ 1 and Q_ 2 are the heat transfer rates in the hot A1 ¼ N1 $L1 $p$D1 (23)
and cold side heat exchangers, respectively.
Combining the differential forms of the equations of state for each A2 ¼ N2 $L2 $p$D2 (24)
control volume, i.e., dðmi Ti Þ=dt ¼ 1=RðpðdVi =dtÞ þ Vi ðdp=dtÞÞ for
i ¼ 1, 2, referring to CV-1 and CV-2, respectively, with Equations The convection heat transfer coefficients vary during the engine
(9)e(11), it is possible to obtain differential equations to compute the operating cycle due to the variation in the gas mass flow rates and
internal pressure and the mass in the expansion volume as follows: they are calculated using Colburn’s correlation [18] as follows:

 _   
Q_
4=5
Q1 1 dV1 1 dV2 0:023 ReD Pr1=3 k
ðg  1Þ þ 2 þ pg þ h ¼ (25)
dp Tc1 Tc2 Tc1 dt Tc2 dt D
¼ (12)
dt V1 V2
þ where k is the gas thermal conductivity; Pr is the Prandtl number,
Tc1 Tc2
and ReD is the Reynolds number based on the tube internal
    diameter.
dm1 V1 dp p dV1 Q_ 1 In order to proceed with the optimization of engine parameters
¼ þ  (13)
dt RgTc1 dt RTc1 dt cp Tc1 for maximum efficiency, a volume constraint is introduced, to
characterize the engine’s finite space availability:
where R is the gas constant and g the working fluid specific heats
ratio cp/cv.
V * ¼ VC þ Vh þ Vd1 þ Vd2 þ VS1 þ VS2 (26)
The heat transfer rate between the regenerator matrix and the
gas is computed as follows: where V*
is a reference volume, kept fixed during the optimization
procedure.
dTm dm1
Q_ R ¼ mR cR ¼ cp ½aðT2  TR Þ þ bðTR  T1 Þ (14) In order to nondimensionalize the mathematical model, the
dt dt following scales are used:
where the coefficients a and b alternate between 0 and 1 according
to the flow direction. TR is the temperature of the gas leaving the 2:p V*
t* ¼ D* ¼ Dp A* ¼
regenerator; Tm the temperature of the regenerator matrix mR the u D*
(27)
matrix mass, and cR the regenerator specific heat. mt cv mt RTc mR cR
h* ¼ p* ¼ T * ¼ Tc M* ¼
The values of the coefficients a and b are given by: t * A* V* mt cp

dm1 The dimensionless variables are therefore defined by:

ða ¼ 1Þ and ðb ¼ 0Þ for >0 (15)
dt
~t ¼ t T
T~ i ¼ *i ~ ¼
p
p
and t* T p*
(28)
dm1 mi ~ ¼ V1 ~ ¼ V2
ða ¼ 0Þ and ðb ¼ 1Þ for <0 (16) ~i ¼
m V 1 V 2
dt mt V* V*
906 M.C. Campos et al. / Energy 44 (2012) 902e910

Additionally, the following dimensionless parameters for the  

4 V~ þV
~c
h
engine are used: ~ ¼
D (38)
1 ~
½y þ ð1  yÞwA
~ ¼ hi D
h ~ ¼ Di ~ ¼ Ai
A
i i i
h* D* A*
(29)
Using Eucken’s correlation, k ¼ m(cp þ 5.R/4) [19] for the gas
~ ¼ Vi ~
V
L
Li ¼ i*
T
T~ h ¼ h
i thermal conductivity and Red ¼ m _ L D=ðm VÞ for the Reynolds
V* D Tc
number, as a function of the mass flow rate, tube length, tube
The index i, in Equations (28) and (29), refers to the points in diameter and volume occupied by the working fluid in the heat
s1, s2, d1, d2, c, h, 1, 2, m, R, c1, c2 in Fig. 1, representing exchanger, it is possible to obtain expressions for h ~ $A
~ and
1 1
the expansion swept volume, the compression swept volume, the ~ ~
h2 $A2 as functions of the dimensionless mass flow rate, dm ~ 1 =d~t,
dead expansion volume, the dead compression volume, cold side the gas properties and the dimensionless parameters previously
heat exchanger, hot side heat exchanger, expansion space, defined:
compression space, regenerator matrix, fluid in the regenerator,
conditional temperature in the expansion space, and conditional
" ! !#  !4=5
~ 1=5 dm 
~ ¼ g ð9g  5Þ mV * t *
temperature during compression, respectively, in the applicable 1=3 2=3
Z14 yA  ~ 1
~ $A
h  
variables. 1 1 ~ ~
 dt 
27:39 mt D2p D1
Substituting the dimensionless variables into Equations (12),
(13) and (20) we obtain: (39)

   
~ $A
h ~ $ T~  T~ ~ $A
h ~ $ 1  T~ ! ~ ~
!
1 1 h 1 2 2 2 1 dV 1 dV
þ ~g
þp 1
þ 2
~
dp T~ c1 T~ c2 ~ ~
T c1 dt ~
T c2 d~t
¼ (30)
d~t ~
V
1
~
V
þ 2
T~ c1 ~
T
c2

( !" #)  !4=5
" # ~ 1=5 d m 
~ ¼g
1=3 ð9g 5Þ2=3 mV * t *
dm~1 ~ ~   ~ $A Z24 ð1yÞA  ~ 1
~ dp dV ~ $A
~ $ T~  T~ 1 h  
¼ 1 ~ þ gp
~ 1h ~  d~t 
V (31) 2 2
27:39 mt D2p wD
d~t d~t ~
1 1 h 1
dt g$T c1
1

(40)
dT~ m ε h    i dm~1
¼ * a T~ 2  T~ m þ b T~ m  T~ 1 (32) ~ , per cycle,
~ and heat transfer, Q
The dimensionless network, W,
d~t M d~t 1
are found by:
The dimensionless temperatures T~ R , T~ c1 and T~ c2 are defined !
as follows: dW~ ~
dV dV~
~ 1 2
¼ p þ (41)
h  i h  i ~
dt ~
dt d~t
T~ R ¼ a T~ 2 þ ε T~ m  T~ 2 þ b T~ 1 þ ε T~ m  T~ 1 (33)
~
dQ  
1 ~ $A
¼ h ~ T~  T~
T~ c1 ¼ a$T~ R þ b$T~ 1 (34) ~
dt
1 1 h 1 (42)

~ and Q ~ at the beginning of the cycle,

T~ c2 ¼ a$T~ 2 þ b$T~ R (35) Assigning a zero value to W 1
the total dimensionless work W ~
~ t and heat transfer Q
1t during the
engine cycle are represented by the values of W ~ and Q ~ respec-
Several parameters, related to the volume distribution are 1
tively, when ~t ¼ 1 (end of the cycle).
needed to study and optimize the engine: s, 4, j, x, and b. This last ~
~ t and Q
The cycle efficiency, h, is therefore defined using W 1t :
parameter is introduced to study the influence in the engine
performance of the relation between the reference volume and the
piston diameter. Additionally, the following parameters related to ~t
W
the heat exchangers geometry are used: y, Z1, Z2, and w. The
h ¼ ðg  1Þ ~ (43)
Q 1t
nomenclature fully presents their definitions and mathematical
expressions.
It is possible to show that several parameters defined in the 3. Numerical method and error analysis
previous paragraph have the following relationships:
Equations (30)e(32) define a system of three ordinary differ-
ential equations that was solved numerically with an adaptive time
y$ð1  xÞ
w ¼ (36) step fourth-fifth order Runge-Kutta method [20], together with
xð1  yÞ
Equations (33)e(35), (39), (40) in the dimensionless time interval
ð0  ~t  1Þ. As a result, the thermodynamic properties of the gas
 
~ þV~c during an operating cycle were obtained. Such dimensionless time
V h ¼ jð1  sÞ (37)
interval is the time required to complete an engine cycle.
 M.C. Campos et al. / Energy 44 (2012) 902e910 907

For each set of geometric and operating parameters, the system

a 12
is solved iteratively for correct boundary conditions, i.e., requiring ψ = 0. 9 x = 0.9 p* = 10
that the unknowns p ~, m~ 1 and T~ m computed at the end of the cycle
ε = 0.95 y = 0.5 M* = 8
ð~t ¼ 1Þ do not differ from their values at the beginning of the cycle 10 ~ ~
since the working fluid is in the same thermodynamic state. This is ~
p Th = 4 A = 25 Z1 = 600
done by computing a cycle relative error and satisfying the Z 2 = 400 ϕ = 0.5 σ = 0.9
following convergence criterion [21]: 8
 
ujþ1  uj 
εcycle ¼  tol (44) 6
uj

where u ¼ p ~ 1 or T~ m , tol is a convergence tolerance limit (e.g.,

~; m
4
0.01 or 1%), and j  0 is the iteration counter. All three unknowns
need to be tested according to the criterion of Equation (44) to
ensure convergence. 2
The process for finding correct boundary conditions starts from
initial values estimated at the beginning of the cycle in the first
iteration, which are replaced by the calculated values at the end of 0
the first iteration, to start the second iteration and so on, until 0 0.2 0.4 0.6 0.8 1
~
convergence is attained. In this work, the selected point to start the Vt
process computing the cycle relative error according to Equation
(44) was the beginning of compression, with estimated initial b 12
ψ = 0 .9 x = 0. 9
~0 ¼ 1, T~ m;0 ¼ 1 and m
values at j ¼ 0, p ~ 1;0 evaluated with the ideal
p* = 10 ε = 0.95
gases equation of state using p ~0 ¼ 1, T~ m;0 ¼ 1.
In the simulation, H2 was used as the working fluid, and its
10 y = 0.5 M* = 8
~
p ~ ~
properties were taken from the technical literature [22]: Th = 4 A = 25
cp ¼ 14.510 J/(kg K), R ¼ 4.124 J/(kg K), m ¼ 1,25 105 kg/(m s). 8 Z1 = 600 Z 2 = 400
Additionally, the following values were used: Tc ¼ 300 K, ϕ = 0.5 σ = 0.9
u ¼ 10 p rad/s ¼ 3000 rpm, a ¼ 90 and Dp ¼ 0.1 m.
Regarding the optimization of parameters, first a physical 6
investigation is conducted to select the appropriate parameters to
be optimized, which in the case of the present study was the pair
4
ð4; yÞ. Therefore two levels of optimization were carried out in this
study for maximum system efficiency. The ranges of variation of
each parameter to be optimized were: 0:25  4  0:7 and 0:2  y  2
0:7. The selected discretization in all ranges for the efficiency
maximization was the coarsest set for which the optimal value of
each parameter did not change as the sets became finer, while the 0
0 90 180 270 360
relative error was kept below 1% in all cases [21].
θ
4. Results
~V
Fig. 3. (a) The p ~ t diagram of the double-acting swashplate Stirling engine, and (b)
The p~  q diagram of the double-acting swashplate Stirling engine.
Initially, it was investigated the existence of optimal values for
the volume parameters, s, 4 and x, using the following values for
the remaining parameters: p* ¼ 10 bar, b ¼ 2, j ¼ 0,9, A ~ ¼ 25 ,
y ¼ 0.5, Z1 ¼ Z2 ¼ 400, T~ h ¼ 3, M* ¼ 8 and x ¼ 0.5. Notice that once,
b and Dp, are given, the reference volume V* can be computed. The
total mass of gas can also be obtained from the specification of p*,
using Equation (25).
In order to indicate the validity of the assumptions, numerical
procedure and engine characteristics, the analysis starts by pre-
senting a p ~V ~ t diagram, which is shown in Fig. 3a. The compres-
sion and expansion processes are clearly devised in the graphic,
showing that a positive network is achieved. For the engine
configuration simulated in Fig. 3, the efficiency was h ¼ 24:25 %.
Fig. 3b shows a p ~  q diagram, i.e., the behavior of the dimension-
less indicated pressure with respect to crank angle during one
engine cycle. Both graphs corroborate the expected trends for
a Stirling engine, obtained with the simulations conducted with the
mathematical model introduced in this paper.
As the percentage of the reference volume occupied by the total
swept volume increases, s (0,80, 0,85 and 0,90), it is observed an
increase in the cycle efficiency as shown in Fig. 4. This analysis
illustrates the variation of the cycle efficiency against the parameter
4, that represents the fraction of the expansion swept volume with Fig. 4. Efficiency behavior with respect to 4, and varying s.
908 M.C. Campos et al. / Energy 44 (2012) 902e910

Fig. 5. Behavior of the efficiency h with respect to 4, and varying x. Fig. 7. Maximized efficiency, hmax, as a function of M*, and varying ε.

respect to the total swept volume. With an equal allocation of due to the increase in the flow velocity inside the heat exchanger
volume to the heat exchangers V ~ and V ~ c , expressed by the
h tubes (which increases the heat transfer coefficient) and, the
parameter x ¼ 0.5, it is observed a maximum efficiency corre- resulting diameter reduction, due to the fixed heat exchange area
sponding to 4 z 0.5, i.e., for equal swept volumes during ~ fixed.
constraint, i.e., for y and A
compression and expansion. The optimal value of 4 decreases Fig. 7 illustrates the effect of the regenerator effectiveness, ε, on
slightly for smaller values of s. the maximized efficiency which increases as ε increases, which is
Setting s ¼ 0.9, the search for higher efficiencies continues by expected since more energy is captured for work conversion. The
varying the parameter x, i.e., by varying the volume distribution graph also illustrates by means of the parameter M*, that the mass
between the heat exchangers. increase in the regenerator is unnecessary above M* y 1.
The simulation results are illustrated in Fig. 5. Higher effi- Figs. (4)e(7) characterize the optimal value 4opt y 0.5 as
ciency values are obtained for larger values of x, i.e., when more “robust’’, taking into consideration that it is insensitive to the
volume is allocated to the hot end heat exchanger. It is important variation of design parameters. This is valuable information from
to note that x ¼ ðV ~ =V~ þV ~ c Þ, therefore increasing x does not
h h the point of view of engine design.
imply in an increase of the hot side dead volume, which is Fig. 8 illustrates a second geometric optimization opportunity
known to be a negatively affecting factor in the performance of for the cycle. It shows that the cycle efficiency can be maximized
Stirling engines. Increasing x means that more of the total one more time, now with respect to the parameter y, which
available volume is allocated to the hot heat exchanger, therefore represents the percentage of heat transfer area of the hot side heat
increasing the heat exchanger capability of collecting the avail- exchanger with respect to the total available heat transfer area. The
able heat supply. graph also illustrates the variation of the parameter p*, which is
Once that the existence of a maximum efficiency corresponding proportional to the total mass of working fluid used by the engine.
to an optimal value of 4opt y 0.5 was demonstrated, the model was It is found that as the total mass of working fluid is reduced, the
used to study the effect of the parameters Z1 and Z2, that represent twice-maximized efficiency increases. It is observed that the
the length to diameter aspect ratio of the heat exchanger tubes. The optimal value yopt y 0.4, is practically insensitive to variations of p*.
results, shown in Fig. 6, indicate that larger values of these This finding is remarkable, as it characterizes the double optimi-
parameters increase the engine efficiency. This trend is justified zation as “robust”, i.e., independent of the main design parameters.

Fig. 6. Behavior of the maximized efficiency, hmax, with respect to Z2, and varying Z1. Fig. 8. Efficiency h as a function of y for variable p*.
 M.C. Campos et al. / Energy 44 (2012) 902e910 909

[7] Angelino G, Invernizzi C. Potential performance of real gas Stirling cycle heat
pumps. International Journal of Refrigeration 1996;19(6):390e9.
[8] Popescu G, Radcenco V, Costea M, Feidt M. Finite-time thermodynamics
optimisation of an endo- and exo-irreversible Stirling motor. Revue Generale
de Thermique 1996;35(418e19):656e61.
[9] Erbay LB, Yavuz H. Analysis of the Stirling heat engine at maximum power
conditions. Energy 1997;22(7):645e50.
[10] Bhattacharyya S, Blank DA. Design considerations for a power optimized
regenerative endoreversible Stirling cycle. International Journal of Energy
Research 2000;24(6):539e47.
[11] Senft JR. Optimum Stirling engine geometry. International Journal of Energy
Research 2002;26(12):1087e101.
[12] Rogdakis ED, Bormpilas NA, Koniakos IK. A thermodynamic study for the
optimization of stable operation of free piston Stirling engines. Energy
Conversion and Management 2004;45(4):575e93.
[13] Tyagi SK, Lin G, Kaushik SC, Chen J. Thermoeconomic optimization of an
irreversible Stirling cryogenic refrigerator cycle. International Journal of
Refrigeration 2004;27(8):924e31.
[14] Mathieu A, Feidt M, Rochelle P, Grosu L, Gualino D, Grappe B. Preliminary
sizing and optimization of a micro solar power plant by a parametric sensi-
tivity study. Environmental Engineering and Management Journal 2010;
9(10):1381e7.
~ t as functions of T~ .
Fig. 9. Efficiency h and work W [15] Timoumi Y, Tlili I, Ben Nasrallah S. Design and performance optimization of
h
GPU-3 Stirling engines. Energy 2008;33(7):1100e14.
Finally, the effect of varying the temperature of the hot source, [16] Timoumi Y, Tlili I, Ben Nasrallah S. Performance optimization of Stirling
engines. Renewable Energy 2008;33(9):2134e44.
Th, in the twice-maximized cycle efficiency is considered. The [17] Karabulut H. Dynamic analysis of a free piston Stirling engine working with
results in Fig. 9 indicate an increase in the efficiency with the closed and open thermodynamic cycles. Renewable Energy 2011;36(6):1704e9.
increase in the temperature of the hot source. Fig. 9 also illustrates [18] Bejan A. Convection heat transfer. 2nd ed. New York: John Wiley & Sons; 1995.
the engine total work, W ~ t , that corresponds to the twice-maxi- [19] Reid RC, Prausnitz JM, Poling BE. Properties of gases and liquids. 4th ed.
Boston: McGrawHill; 1987.
mized efficiency, as a function of T~ h . [20] Kincaid D, Cheney W. Numerical analysis. 1st ed. Belmont-CA: Wadsworth;
1991.
[21] Editorial. Journal of heat transfer editorial policy statement on numerical
5. Conclusions accuracy. ASME Journal of Heat Transfer 1994;116:797e8.
[22] Bejan A. Heat transfer. 1st ed. New York: John Wiley & Sons; 1993.
In this paper a mathematical model to simulate the operation of
Stirling engines is introduced. A thermodynamic optimization of Nomenclature
the cycle for maximum thermal efficiency is conducted. The
numerical results show that there is an optimal engine geometry A: total heat transfer area, A1 þ A2
that maximizes the thermal efficiency. The results are reported A1: heat transfer area in the expansion volume
A2: heat transfer area in the compression volume
using dimensionless variables, in normalized charts, which are a, b: coefficients, Equations (15) and (16)
general for the engine configuration considered, i.e., with a sliding cp: working fluid specific heat at constant pressure
disk mechanism (“swashplate”). cR: regenerator specific heat
cv: working fluid specific heat at constant volume
The study identifies the location of the thermodynamic optima D1: diameter of the tubes in the hot side heat exchanger
in terms of two important design parameters, 4 and y, and their D2: diameter of the tubes in the cold side heat exchanger
sensitivity to other engine parameters. The thermodynamic optima h1: heat transfer convection coefficient in the expansion volume
h2: heat transfer convection coefficient in the compression volume
are sharp, in the sense that a 225% variation of the calculated effi-
k: gas thermal conductivity
ciency values was observed within the range of tested configura- L1: length of the tubes in the hot side heat exchanger
tions in this study, and “robust” (i.e., relatively insensitive) to the L2: length of the tubes in the cold side heat exchanger
variation of several parameters, e.g., p* (or mt), T~ h , M*, ε, Z1, Z2, x and m1: mass in the expansion volume
m2: mass in the compression volume
s. It is therefore reasonable to state that this conclusion is of great mR: regenerator matrix mass
importance for engine design. mt: total working fluid (gas) mass
N1: number of tubes in the hot side heat exchanger
N2: number of tubes in the cold side heat exchanger
Acknowledgements Pr: Prandtl number
p: internal pressure
Q_ 1 : heat transfer rate in the hot side heat exchanger
The authors acknowledge with gratitude the support of the
Q_ 2 : heat transfer rate in the cold side heat exchanger
Center for Advanced Power Systems at Florida State University, Q_ R : heat transfer rate in the regenerator
AFOSR (Award No. FA9550-06-1-0527), the Brazilian National Q~ : dimensionless heat transfer

Council of Scientific and Technological Development, CNPq (project R: gas constant

ReD: Reynolds number based on internal tube diameter
554151/2010-3-CNPq), and the Engineering and Materials Science t: time
Graduate Program (PIPE) of the Federal University of Parana. tol: cycle convergence tolerance
T1: temperature in the expansion volume
T2: temperature in the compression volume
References Tc: temperature of the cold reservoir,
Tc1: conditional temperature that depends upon the gas flow direction and the heat
[1] Moran MJ, Shapiro HN, Boettner DD, Bailey MD. Fundamentals of engineering exchange in the regenerator
thermodynamics. 7th ed. New York: John Wiley & Sons; 2011. Tc2: conditional temperature that depends upon the gas flow direction and the heat
[2] Curzon FL, Ahlborn B. Efficiency of a Carnot engine at maximum power exchange in the regenerator
output. American Journal of Physics 1975;43:22e4. Th: temperature of the hot reservoir
[3] Reader GT, Hooper C. Stirling engines. 1st ed. London: E. & F. N. Spon; 1983. Tm: temperature of the regenerator matrix
[4] Urieli I, Berchowitz DM. Stirling cycle engine analysis. 1st ed. Bristol: Adam TR: temperature of the gas leaving the regenerator
Hilger Ltd; 1984. u: general unknown
[5] Ladas HG, Ibrahim OM. Finite-time view of the Stirling engine. Energy 1994; V1: total expansion volume
19(8):837e43. V2: total compression volume
[6] Blank DA, Wu C. Power optimization of an extraterrestrial solar-radiant Stir- Vc: volume occupied by the cold heat exchanger
ling heat engine. Energy 1995;20(6):523e30. Vd1: expansion dead volume
910 M.C. Campos et al. / Energy 44 (2012) 902e910

Vd2: compression dead volume j: ratio between the volume allocated to the heat exchangers, ðV~ h þ V~ c Þ, and the
~
total dead volume, ðV ~ ~ ~
Vh: volume occupied by the hot heat exchanger d1 þ V d2 þ V h þ V c Þ
Vs1: total swept volume for expansion u: angular speed (rad/s)
Vs2: total swept volume for compression
w: diameter ratio D ~
~ =D Subscripts
1 2
~ dimensionless cycle network
W: i used to represent the following:
x: ratio between the hot side heat exchanger volume, V ~ , and the total volume s1: expansion swept volume
h
~ þV
allocated to the heat exchangers, ðV ~cÞ s2: compression swept volume
h
y: ratio between the heat transfer area of the hot side heat exchanger, A~ , and the d1: dead expansion volume
1
~ þA
total heat exchange area, ðA ~ Þ ¼ A~ d2: dead compression volume
1 2
Z1: length to diameter ratio for the tubes of the hot side heat exchanger c: cold side heat exchanger
Z2: length to diameter ratio for the tubes of the cold side heat exchanger h: hot side heat exchanger
1: expansion space
Greek letters 2: compression space
a: phase angle m: regenerator matrix
b: ratio between the reference volume V* and p.D3p R: fluid in the regenerator
g: specific heats ratio cp/cv c1: conditional temperature in the expansion space
ε: regenerator effectiveness c2: conditional temperature in the compression space
εcycle : cycle relative error max: maximized (max, max ¼ twice-maximized)
h: cycle efficiency t: total
q: crank angle, ut  180=p
s: fraction of the reference volume, V*, occupied by the total engine swept volume Superscripts
4: ratio between the total swept volume during the expansion, V ~ *: reference value (scale)
s1 , and the total
swept volume, ðV ~ ~
s1 þ V s2 Þ w: dimensionless variable