* Correspondence to: Adrian Bejan, Department of Mechanical Engineering and Materials Science, Duke University,

Durham, NC 27708-0300, U.S.A.

Received 16 January 2001
Copyright 2002 John Wiley & Sons, Ltd. Accepted 4 June 2001
INTERNATIONAL JOURNAL OF ENERGY RESEARCH
Int. J. Energy Res. 2002; 26:545}565 (DOI: 10.1002/er.804)
Fundamentals of exergy analysis, entropy generation
minimization, and the generation of #ow architecture
Adrian Bejan*
Department of Mechanical Engineering and Materials Science, Duke University, Durham, NC 27708-0300, U.S.A.
SUMMARY
This paper outlines the fundamentals of the methods of exergy analysis and entropy generation minimiz-
ation (or thermodynamic optimization*the minimization of exergy destruction). The paper begins with
a review of the concept of irreversibility, entropy generation, or exergy destruction. Examples illustrate the
accounting for exergy #ows and accumulation in closed systems, open systems, heat transfer processes, and
power and refrigeration plants. The proportionality between exergy destruction and entropy generation
sends the designer in search of improved thermodynamic performance subject to "nite-size constraints and
speci"ed environmental conditions. Examples are drawn from energy storage systems for sensible heat and
latent heat, solar energy, and the generation of maximum power in a power plant model with "nite heat
transfer surface inventory. It is shown that the physical structure (geometric con"guration, topology) of the
system springs out of the process of global thermodynamic optimization subject to global constraints. This
principle generates structure not only in engineering but also in physics and biology (constructal theory).
Copyright 2002 John Wiley & Sons, Ltd.
KEY WORDS: exergy analysis; entropy generation minimization; EGM; thermodynamic optimization;
constructal theory; topology optimization; self-organization in nature; self-optimization in
nature
1. EXERGY, NOT ENERGY
The energy crisis of the 1970s and the continuing emphasis on e$ciency (conservation of fuel
resources) has led to a complete overhaul of the way in which power systems are analyzed and
improved thermodynamically. The new methodology is exergy analysis and its optimization
component is known as thermodynamic optimization, or entropy generation minimization (EGM).
This new approach is based on the simultaneous application of the "rst law and the second law in
analysis and design (Evans, 1969; Reistad, 1970; Nerescu and Radcenco, 1970; Brodyanskii, 1973;
Haywood, 1980; Ahern, 1980). In the 1990s it has become the premier method of thermodynamic
analysis in engineering education (Moran, 1989; Bejan, 1982, 1988; Feidt, 1987; Sieniutycz and
Salamon, 1990; Kotas, 1995; Moran and Shapiro, 1995; Radcenco, 1994; Shiner, 1996; Bejan et al.,
1996) and it is now sweeping every aspect of engineering practice (Stecco and Moran, 1990, 1992;
Valero and Tsatsaronis, 1992; Bejan and Mamut, 1999; Bejan et al., 1999). It is particularly well
suited for computer-assisted design and optimization (Sciubba and Melli, 1998; Sciubba,
1999a, b). In this paper we review the fundamentals of the method, its current status, and a few
examples of the ways in which the method can be used for system optimization. Extensive reviews
of the greatly expanding exergy and entropy generation "eld have been published (Bejan,
1996a, b, 1997).
To begin with, we must distinguish between exergy and energy in order to avoid any confusion
with the traditional energy-based methods of thermal system analysis and design. Energy #ows
into and out of a systemalong paths of mass #ow, heat transfer, and work (e.g. shafts, piston rods).
Energy is conserved, not destroyed: this is the statement made by the "rst law of thermodynamics.
Exergy is an entirely di!erent concept. It represents quantitatively the &useful' energy, or the
ability to do or receive work*the work content*of the great variety of streams (mass, heat,
work) that #ow through the system. The "rst attribute of the property &exergy' is that it makes it
possible to compare on a common basis di!erent interactions (inputs, outputs, work, heat).
Another bene"t is that by accounting for all the exergy streams of the system it is possible to
determine the extent to which the system destroys exergy. The destroyed exergy is proportional to
the generated entropy. Exergy is always destroyed, partially or totally: this is the statement made
by the second law of thermodynamics. The destroyed exergy, or the generated entropy is
responsible for the less-than-theoretical thermodynamic e$ciency of the system.
By performing exergy accounting in smaller and smaller subsystems, we are able to draw a map
of how the destruction of exergy is distributed over the engineering system of interest. In this way
we are able to pinpoint the components and mechanisms (processes) that destroy exergy the most.
This is a real advantage in the search for improving e$ciency (always by "nite means), because it
tells us fromthe start howto allocate engineering e!ort and resources. To the optimal allocation of
resources we return in the example of Section 4.1.
In engineering thermodynamics today, emphasis is placed on identifying the mechanisms and
systemcomponents that are responsible for thermodynamic losses, the sizes of these losses (exergy
analysis), minimizing the losses subject to the global constraints of the system (entropy generation
minimization), and minimizing the total costs associated with building and operating the energy
system (thermoeconomics). To review the fundamentals of thermoeconomics is not the objective
of this paper. A description of thermoeconomics and its applications, as an evolutionary
development beyond exergy analysis and entropy generation minimization, is available in
Bejan et al. (1996).
The method of thermodynamic optimization or entropy generation minimization (EGM) is
a "eld of activity at the interface between heat transfer, engineering thermodynamics, and #uid
mechanics. The position of the "eld is illustrated in Figure 1. The method relies on the
simultaneous application of principles of heat and mass transfer, #uid mechanics, and engineering
thermodynamics, in the pursuit of realistic models of processes, devices, and installations. By
realistic models we mean models that account for the inherent irreversibility of engineering
systems and processes.
Thermodynamic optimization may be used by itself (without cost minimization) in the
preliminary stages of design, in order to identify trends and the existence of optimization
opportunities. The optima and structural characteristics identi"ed based on thermodynamic
546 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 1. The interdisciplinary "eld covered by the method of thermodynamic optimization, or entropy
generation minimization (Bejan, 1982). The lower diagram is due to Smith (2000).
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 547
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 2. General de"nition of #ow system (open thermodynamic system) in communication
with the atmosphere (Bejan, 1988).
optimization can be made more realistic through subsequent re"nements based on global cost
minimization, by using the method of thermoeconomics. Space limitations do not permit this
method to be covered in this article, however, examples of its use are mentioned in other articles
in this issue. Modern presentations of thermoeconomics can be found in Benelmir et al. (1991,
1992, 1997), Von Spakovsky (1994), Von Spakovsky and Frangopoulos (1993, 1994), Bejan et al.
(1996), Benelmir and Feidt (1997), Tsatsaronis (1999), Valero et al. (1999), and Olsommer et al.
(1999a, b).
2. ENTROPY GENERATION, OR EXERGY DESTRUCTION
Here is why in thermodynamic optimization we must rely on heat transfer and #uid mechanics,
not just thermodynamics. Consider the most general system}environment con"guration, namely
a system that operates in the unsteady state, Figure 2. Its instantaneous inventories of mass,
energy, and entropy are M, E, and S. The system experiences the net work transfer rate =Q , heat
transfer rates (QQ
"
, QQ

,
2
, QQ
L
) with n#1 temperature reservoirs (
"
,

,
2
,
L
), and mass #ow
rates (mR
'"
, mR
"'
) through any number of inlet and outlet ports. Noteworthy in this array of
interactions is the heat transfer rate between the system and the environmental (atmospheric)
temperature reservoir, QQ
"
, on which we focus shortly.
The thermodynamics of the system consists of accounting for the "rst law and the second law
(e.g. Moran and Shapiro, 1995; Bejan, 1997),
dE
dt
"
L

G"
QQ
G
!=Q #
'"
mR h!
"'
mR h (1)
548 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
SQ
"
"
dS
dt
!
L

G"
QQ
G

G
!
'"
mR s#
"'
mR s*0 (2)
where h is shorthand for the sum of speci"c enthalpy, kinetic energy, and potential energy
of a particular stream at the boundary. In Equation (2) the total entropy generation rate SQ
"
is simply a de"nition (notation) for the entire quantity on the left-hand side of the inequality
sign. We shall see that it is advantageous to decrease SQ
"
; this can be accomplished by
changing at least one of the quantities (properties, interactions) speci"ed along the system
boundary.
We select the environmental heat transfer QQ
"
as the interaction that is allowed to #oat as
SQ
"
varies. Historically, this choice was inspired (and justi"ed) by applications to power
plants and refrigeration plants, because the rejection of heat to the atmosphere was of little
consequence in the overall cost analysis of the design. Eliminating QQ
"
between Equations (1)
and (2) we obtain
=Q "!
d
dt
(E!
"
S)#
L

G

1!

"

QQ
G
#
'"
mR (h!
"
s)!
"'
mR (h!
"
s)!
"
SQ
"
(3)
The power output or input in the limit of reversible operation (SQ
"
"0) is
=Q
'`
"!
d
dt
(E!
"
S)#
L

G

1!

"

QQ
G
#
'"
mR (h!
"
s)!
"'
mR (h!
"
s) (4)
In engineering thermodynamics each of the terms on the right-hand side of Equation (4) is
recognized as an exergy of one type or another (see Section 3), and the calculation of =Q
'`
is known as exergy analysis. Subtracting Equation (3) from Equation (4) we arrive at the
Gouy-Stodola theorem,
=Q
'`
!=Q "
"
SQ
"
(5)
In Equation (5) =Q
'`
is "xed because all the heat and mass #ows (other than QQ
"
) are "xed.
Pure thermodynamics (e.g., exergy analysis) ends, and the method of entropy generation
minimization (EGM) begins with Equation (5). The lost power (=Q
'`
!=Q ) is always positive,
regardless of whether the system is a power producer (e.g. power plant) or a power user (e.g.
refrigeration plant). To minimize lost power when =Q
'`
is "xed is the same as maximizing power
output in a power plant, and minimizing power input in a refrigeration plant. This operation is
also equivalent to minimizing the total rate of entropy generation.
The critically new aspect of the EGM method*the aspect that makes the use of thermo-
dynamics insu$cient, and distinguishes EGM from pure exergy analysis*is the minimization of
the calculated entropy generation rate. Optimization and design (the generation of structure)
are the di!erence. To minimize the irreversibility of a proposed con"guration, the analyst must
use the relations between temperature di!erences and heat transfer rates, and between pressure
di!erences and mass #ow rates. The analyst must express the thermodynamic nonideality of
the design SQ
"
as a function of the topology and physical characteristics of the system, namely,
"nite dimensions, shapes, materials, "nite speeds, and "nite-time intervals of operation. For this
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 549
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 3. Closed system and process en route to thermal and mechanical equilibrium with the environment.
the analyst must rely on heat transfer and #uid mechanics principles, in addition to thermo-
dynamics. Only by varying one or more of the physical characteristics of the system, can the
analyst bring the design closer to the operation characterized by minimum entropy generation
subject to size and time constraints. We illustrate this technique by means of a few very basic
models in Section 4.
3. EXERGY ANALYSIS
There is a rich nomenclature and mathematical apparatus associated with de"ning and calculat-
ing the exergies of various entities. Most of these exergy names are attached to the four types of
terms shown on the right side of Equation (4). This nomenclature must be used with care,
especially now as the method is applied for the "rst time to areas where energy-based methods are
still the norm. The key feature is this: exergy is the maximum (theoretical) work that can be
extracted (or the minimum work that is required) from the entity (e.g. stream, amount of matter)
as the entity passes from a given state to one of equilibrium with the environment. As such, exergy
is a measure of the departure of the given state from the environmental state*the larger the
departure, the greater the potential for doing work.
To illustrate the calculation of exergy, consider the following examples in which the environ-
mental state is represented by the atmospheric temperature
"
and pressure P
"
. If the entity is
a closed system("xed mass, Figure 3) at an initial state represented by the energy E, entropy S and
volume <, then its exergy (expressed in joules) relative to the environment is
"E!E
"
!
"
(S!S
"
)#P
"
(<!<
"
) (6)
In this expression is known as the non-ow exergy of the given mass (Moran, 1989; Bejan, 1988),
and the subscript 0 indicates the system properties in the state of thermal and mechanical
equilibrium with the environment. The environmental state is also known as the restricted dead
550 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 4. Open system, steady state, and one stream en route to thermal and mechanical
equilibrium with the environment.
state*&dead' because once in this state the system cannot deliver any more work relative to the
environment. It is &restricted' because in this state the system is in thermal and mechanical
equilibrium with the environment, but not in chemical equilibrium. Equation (6) is general in the
sense that the internal construction and materials (e.g. single phase vs. multi phase) of the given
mass are not speci"ed. Equation (6) can be generalized further for cases where the chemical
composition of the given mass may change en route to chemical equilibrium with the environ-
ment (Moran, 1989; Bejan, 1988).
Note that the non-#ow exergy has its origin in the "rst term identi"ed on the right-hand side
of Equation (3) or Equation (4). The system &accumulates' the quantity (E!
"
S) as potential
work, and, in going from left to right in Figure 3, this work potential decreases from (E!
"
S) to
(E
"
!
"
S
"
). The di!erence, which is E!E
"
!
"
(S!S
"
), represents all the work that could be
produced during the process. From this quantity we must subtract the work fraction done by the
system against the atmosphere, P
"
(<
"
!<). The resulting expression is Equation (6).
As a second exergy calculation example, consider the steady-#ow system with a stream of mass
#ow rate mR , where the given (inlet) state is represented by the speci"c enthalpy h, entropy s, kinetic
energy
`
<`, and gravitational potential energy gz, where z is the altitude of the inlet. The speci"c
-ow exergy is expressed in J kg, and is evaluated relative to the environment (
"
, P
"
):
e
V
"(h#
`
<`#gz)!(h#
`
<`#gz)
"
!
"
(s!s
"
) (7)
As shown in Figure 4, the subscript 0 indicates the properties of the stream that reached thermal
and mechanical equilibrium with the environment. In other words, e
V
is the change in the value of
the group h#
`
<`#gz!
"
s, in going from the inlet to the outlet. The exergy #ow rate of this
stream is the product mR e
V
. More general versions of Equation (7) are available for streams that
can also exchange chemical species with the environment (Moran, 1989; Bejan, 1988).
The streams of #ow exergy were identi"ed already in the third and fourth terms of Equations
(3) and (4), where h was shorthand for the sum (h#
`
<`#gz). Continuing with this shorthand
notation, we note that the group (h!
"
s) represents the speci"c -ow availability of the stream.
The di!erence between the #ow availability at the indicated state (inlet, Figure 4) and the #ow
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 551
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
availability of the same stream at the environmental state (outlet, Figure 4) is the #ow exergy,
namely e
V
"(h!
"
s)!(h
"
!
"
s
"
), which is the same as Equation (7).
A third example is the exergy content of heat transfer. If the heat transfer rate QQ enters the
system by crossing a boundary of local temperature , then its exergy stream relative to the
environment (
"
) is
EQ
/
"QQ

1!

"

(8)
Alternative notations used for EQ
/
are EQ
O
and EQ
VO
. The heat-transfer exergy #ow rate EQ
/
is zero
when "
"
, i.e. as in the case of the heat rejected to the ambient by power and refrigeration
plants. In Figure 2 an exergy stream is associated with each of the heat inputs QQ

,
2
, QQ
L
, while the
QQ
"
stream carries no exergy EQ
/"
"QQ
"
(1!
"
/
"
)"0. The exergy #ows associated with heat
transfer are accounted for by the second term on the right-hand side of Equations (3) and (4).
It is important to stress that in Figure 2 and Equation (8)
"
is the temperature of those regions
of the environment that are su$ciently close to the system but not a!ected by the discharge. The
purpose of this modeling decision is to place inside &the system' all the irreversibilities associated
with the internal and external e!ect of the physical installation that resides inside the system. The
system comprised the installation and the surrounding regions that are a!ected (e.g. heated) by
the installation. In an actual power or refrigeration plant, the rejected heat current QQ leaves the
installation and enters the neighbouring environmental #uid (air, water) at a temperature
somewhat higher than
"
. Further down the line, the same QQ reaches the true environment at

"
(i.e. it crosses the
"
boundary, Figure 2). The interaction between energy systems and their
surrounding #uids forms the subject of the environmental #ows documented in several articles in
this issue.
Exergy accounting and the spatial distribution of exergy destruction are illustrated in Figure 5.
The top "gure shows the traditional, energy-#ow analysis of a simple Rankine cycle power plant.
The heat input is QQ
'
, and the net power output is =Q
'
!=Q

"p
''
E
/'
. The fraction p
''
is the
second-law or rational e.ciency of the power plant. It is a relative measure of the combined
imperfections of the power plant. The corresponding exergy wheel for the #ows through and
around a refrigeration plant is illustrated in Figure 6.
The calculated widths of the exergy destruction streams indicate a ranking of the components
as candidates for thermodynamic optimization. The exergy #ow analysis (Figures 5 and 6,
bottom) can be performed inside each component in order to determine the particular features
(e.g. combustion, fouling, heat transfer, pressure drop) that dominate the irreversibility of that
component. Finally, the success of the thermodynamic improvements that are implemented can
only be evaluated by repeating the exergy analysis and registering the changes in exergy
destruction and second-law e$ciency.
The current literature reviewed in references shows that exergy principles are being applied to
a wide variety of thermal/chemical processes. Avoidable destructions of exergy represent the
waste of exergy sources such as oil, natural gas and coal. By devising ways to avoid the
destruction of exergy, better use can be made of fuels. Exergy analysis determines the location,
type, and true magnitude of the waste of fuel resources, and plays a central role in developing
strategies for more e!ective fuel use.
552 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 5. The conversion and partial destruction of exergy in a power plant based on the simple Rankine
cycle. Top: the traditional notation and energy-interaction diagram. Bottom: the exergy wheel diagram
(Bejan, 1988), and the de"nition of the second law e$ciency p
''
.
4. ENTROPY GENERATION MINIMIZATION
Most of the newest work and opportunities for advances are in developing strategies for the
optimal allocation (con"guration, topology) of resources. This work is known as exergy destruc-
tion minimization, irreversibility (entropy generation) minimization, or thermodynamic optim-
ization. It is often subjected to overall constraints such as "nite sizes, "nite times, material
types, and shapes. Recent reviews of the literature (Bejan, 1996a, b) show that thermodynamic
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 553
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 6. The conversion and partial destruction of exergy in a refrigeration plant based on the simple vapor
compression cycle. Top: the traditional notation and energy-interaction diagram. Bottom: the exergy wheel
diagram (Bejan, 1988), and the de"nition of the second law e$ciency p
''
.
optimization is making fast progress in cryogenics, heat transfer engineering, energy storage
systems, solar power plants, fossil-fuel power plants, and refrigeration plants (Feidt, 1998). The
examples collected in this section illustrate the opportunities for devising strategies of optimal
allocation.
4.1. Optimal allocation of heat transfer area
Heat transfer principles combined with thermodynamics shed light on why energy systems are
imperfect, and why they possess geometric structure*why their hardware is arranged in certain
554 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 7. Model of power plant with two heat transfer surfaces, and the maximization of power output
(or the minimization of entropy generation) subject to "xed heat input (Q
'
) and "xed total heat
transfer surface (C
'
#C
'
"C, constant).
amounts, and in certain ways in space. A power plant owes its irreversibility to many factors, one
of which is the transfer of heat across "nite temperature di!erences. This e!ect has been isolated
in Figure 7. The power plant is the vertical segment marked between the high temperature
'
and
the ambient temperature
'
. The heat input Q
'
("xed) and the rejected heat Q
'
must be driven by
temperature di!erences: the temperature gaps
'
!
''
and
''
!
'
account for some of the
space occupied by the power plant. Heat transfer surfaces reside in these spaces. The rest of the
space is reserved for the rest of the power plant: for simplicity, this inner space is assumed to be
irreversibility free (endoreversible),
S
"
"
Q
'

''
!
Q
'

''
"0 (9)
The entire irreversibility of this power plant model is concentrated in the spaces occupied by the
two temperature gaps. The simplest heat transfer model for these spaces is the proportionality
between heat current and temperature di!erence,
Q
'
"C
'
(
'
!
''
) (10)
Q
'
"C
'
(
''
!
'
) (11)
Each thermal conductance (C
'
, C
'
) is proportional to its area for heat transfer, e.g. C
'
";
'
A
'
and C
'
";
'
A
'
, where each heat transfer area (A
'
, A
'
) is multiplied by its corresponding overall
heat transfer coe$cient (;
'
, ;
'
). When ;
'
and ;
'
are equal (or of the same order of magnitude),
the total area constraint (A
'
#A
'
"A, constant) is represented adequately by the total conduc-
tance constraint (Bejan, 1988, 1997)
C
'
#C
'
"C (12)
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 555
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
where C is "xed. More general constraints, valid for unequal heat transfer coe$cients, can also be
used (Bejan, 1996a, b).
The analytical model is completed by the "rst law, written for the power plant as a closed
system operating in steady state or in an integral number of cycles, ="Q
'
!Q
'
. Combining
this with the preceding relations, we obtain the power output as a function of the conductance
allocation fraction x"C
'
/C (in heat exchanger terminology this dimensionless parameter is also
known as C*),
=
Q
'
"1!

'
/
'
1!(Q
'
/
'
C)

1
x
#
1
1!x
(13)
The minimization of entropy generation (S
"
) is equivalent to the maximization of power output
(=), cf. Equation (5). The =expression (13) can be maximized with respect to x, and the result is
x
'
"1/2 (Bejan, 1988), or
C
''
"C
' '
(14)
In conclusion, there is an optimal way to allocate the constrained hardware (C) to the two ends
of the power plant, that is, if the maximization of power output subject to "xed heat input (Q
'
)
and "xed size (C) is the purpose. Equation (14) also holds for refrigerating machines modeled in
the same way (Bejan, 1996a, b).
The maximization of =is shown graphically in Figure 7. Small conductances strangle the #ow
of heat, and demand large temperature di!erences. The power output is small when the
temperature di!erence across the innermost (reversible) compartment is small. The "rst and third
frames of Figure 7 show that when the two conductances are highly dissimilar in size, large
temperature gaps are present, and the power output is small. The best irreversible performance is
somewhere in the middle, where the conductances are comparable in size.
4.2. Optimal latent-heat storage temperature
Here is an example of how the type of energy-storage material can be chosen based on
thermodynamic optimization. A simple way to perform the thermodynamic optimization of the
latent heat storage process was proposed by Lim et al. (1992), Figure 8. The hot stream of
initial temperature
`
comes in contact with a single phase-change material through a "nite
thermal conductance UA, assumed known, where A is the heat transfer area between the melting
material and the stream, and ; is the overall heat transfer coe$cient based on A. The
phase-change material (solid or liquid) is at the melting point

. The stream is well mixed

at the temperature
"'
, which is also the temperature of the stream exhausted into the
atmosphere (
"
).
The steady operation of the installation modelled in Figure 8 accounts for the unsteady (cyclic)
operation in which every in"nitesimally short storage (melting) stroke is followed by a short
energy retrieval stroke: mR is stopped, and the recently melted phase-change material is solidi"ed
to its original state by the cooling e!ect provided by the heat engine positioned between

and

"
. In this way, the steady-state equivalent model of Figure 8 represents the complete cycle*that
is, storage followed by retrieval.
556 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 8. The generation of power using a phase-change material and a hot stream that is ultimately
discharged into the ambient (Lim et al., 1992).
The steady cooling e!ect of the power plant can be expressed in two ways:
QQ

";A(
"'
!

) (15)
QQ

"mR c
"
(
`
!
"'
) (16)
By eliminating
"'
between these two equations we obtain
QQ

"mR c
"
N
'"
1#N
'"
(
`
!

) (17)
in which N
'"
is the number of heat transfer units of the heat exchanger surface,
N
'"
"
;A
mR c
"
(18)
Of interest is the minimization of entropy generation, or the calculation of the maximumrate of
exergy (useful work, =Q in Figure 8) that can be extracted from the phase-change material. For
this, we model as reversible the cycle executed by the working #uid between

and
"
=Q "QQ

1!

"

(19)
and, after combining with Equation (17), we obtain
=Q "mR c
"
N
'"
1#N
'"
(
`
!

1!

"

(20)
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 557
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 9. Model of a solar thermal power plant with collector-ambient heat loss and a heat exchanger
between the collector and a Carnot energy conversion cycle (Bejan, 1982).
By maximizing =Q with respect to

*that is, with respect to the type of phase-change material,

we obtain the optimal melting and solidi"cation temperature:

'
"(
`

"
)` (21)
The maximum power output that corresponds to this optimal choice of phase-change
material is
=Q
`
"mR c
"

`
N
'"
1#N
'"

1!

"

`
(22)
The same results, Equations (21) and (22), could have been obtained by minimizing the total rate
of entropy generation. One way to improve the power output of the single-element installation of
Figure 8 is by placing the exhaust in contact with one or more phase-change elements of lower
temperatures. This method is illustrated in Lim et al. (1992).
4.3. Optimal solar collector temperature
The optimization of solar energy conversion has been studied under the banners of two
fundamental problems. One is concerned with establishing the theoretical limits of converting
thermal radiation into work, or calculating the exergy content of radiation. The other problem
deals with the delivery of maximumpower from a solar collector of "xed size (Bejan, 1982, DeVos,
1992). This problem has also been solved in many subsequent applications (Bejan, 1996a), which
are united by a characteristic and important design feature: the collector operating temperature
can be optimized.
This optimization opportunity is illustrated in Figure 9. A power plant is driven by a solar
collector with convective heat loss to the ambient. The heat loss is assumed to be proportional to
the collector-ambient temperature di!erence, QQ
"
"(;A)

!
"
). The internal heat exchanger
between the collector and the hot end of the power cycle (the user) is modelled similarly,
QQ "(;A)
'
(

!
"
). There is an optimal coupling between the collector and the power cycle
such that the power output is maximum. This design is represented by the optimal collector
558 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 10. Two sources of irreversibility in the heating (charging) stroke of a
sensible-heat storage process (Bejan, 1982).
temperature (Bejan, 1982)

'

"
"
0`
`
#R0
`
1#R
(23)
where R"(;A)

/(;A)
'
and 0
`
"
`
/
"
is the uppermost (&stagnation') temperature of the
collector. This optimum has its origin in the trade-o! between the Carnot e$ciency of the
reversible part of the power plant (1!
"
/
"
) and the heat loss to the ambient, QQ
"
. The power
output is the product QQ (1!
"
/
"
). When

(
'
the Carnot factor is too small. When

'
'
the heat input QQ drawn from the collector is too small, because the heat loss to the
ambient QQ
"
is large.
Corresponding optimal couplings have been identi"ed for solar-driven power plants of many
power-cycle designs, extraterrestrial power plants, and refrigeration systems driven by solar
power (Bejan, 1996a). These optima are obtained based on the maximization of power output, or
the minimization of the total rate of entropy generation.
4.4. Optimal sensible-heat storage time interval
The opportunity for minimizing the destruction of exergy during exergy storage becomes evident
if we examine the system shown in Figure 10. The storage system (the left side of the "gure)
contains a batch of liquid (m, c). The liquid is held in an insulated vessel. The hot-gas stream
mR enters the system through one port and is gradually cooled as it #ows through a heat exchanger
immersed in the liquid bath. The spent gas is discharged directly into the atmosphere. As time
passes, the bath temperature and the gas outlet temperature
"'
approach the hot-gas inlet
temperature,
`
.
If we model the hot gas (steam, products of combustion) as an ideal gas with constant speci"c
heat c
"
, the temperature (history) of the storage system is expressed in closed form by the
equations
(t)!
"

`
!
"
"1!exp (!y0) (24)
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 559
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565

"'
(t)!
"

`
!
"
"1!y exp (!y0) (25)
where y and the dimensionless time 0 are de"ned as
y"1!exp (!N
'"
), N
'"
"
hM
'
A
'
mR c
"
, 0"
m c
"
mc
t (26)}(28)
In these equations, A
'
is the total heat-exchanger surface separating the stream from the liquid
bath, and hM
'
is the mean heat transfer coe$cient (in time and space) based on A
'
. Built into the
model is the assumption that the liquid bath is well mixed, i.e., that the liquid temperature () is
a function of the time (t) only. As expected, both and
"'
approach
`
asymptotically. They
approach
`
faster when N
'"
is higher.
Turning our attention to the irreversibility of the energy-storage process, Figure 10 shows that
the irreversibility is due to two distinct parts of the apparatus. First, there is the "nite-A(t)
irreversibility associated with the heat transfer between the hot stream and the cold liquid bath.
Second, the stream exhausted into the atmosphere is eventually cooled down to
"
, again by heat
transfer across a "nite A' (t). Neglected in the present model is the irreversibility due to the
pressure drop across the heat exchanger traveled by the stream mR .
The combined e!ect of the competing irreversibilities noted in Figure 10 is a characteristic of all
sensible-heat storage systems. Because of it, only a fraction of the exergy content of the hot stream
can be stored in the liquid bath. In order to see this, consider the instantaneous rate of entropy
generation in the overall system delineated in Figure 10,
SQ
"
"mR c
"
ln

"

`
#
QQ
"

"
#
d
dt
(mc ln ) (29)
where QQ
"
"mR c
"
(
"'
!
"
). Important is the entropy generated during the entire charging-time
interval 0!t, which, using Equations (24)}(29), can be put in dimensionless form as
1
mc
R
"
SQ
"
dt"0

ln

"

`
#t

#ln(1#tp
'
)!tp
'
(30)
where the "rst-law e$ciency p
'
is shorthand for the right-hand side of Equation (24), and where
t"(
`
!
"
)/
"
.
Multiplied by
"
, the entropy-generation integral j
R
"
SQ
"
dt calculated above represents the
destroyed exergy*the bite taken by irreversibilities out of the total exergy supply brought into
the system by the hot stream
E
V
"tEQ
V
"tmR c
"
ln [
`
!
"
!
"
ln (
`
/
"
)] (31)
On this basis, we de"ne the entropy-generation number N
`
as the ratio of the lost exergy divided
by the total exergy invested during the time interval 0!t:
N
`
(0, t, N
'"
)"

"
E
V

R
"
S
"
dt"1!
tp
'
!ln (1#tp
'
)
0 [t!ln (1#t)]
(32)
560 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
The entropy-generation number takes values in the range 0}1, the N
`
"0 limit representing the
elusive case of reversible operation. Note the relation N
`
"1!p
''
, where p
''
is the second-law
e$ciency of the installation during the charging process.
Charts of the N
`
(0, t, N
'"
) surface show that N
`
decreases steadily as the heat-exchanger size
(N
'"
) increases (Bejan, 1982). This e!ect is expected. Less expected is the fact that N
`
goes through
a minimum as the dimensionless time 0 increases. For example, the optimal time for minimum
N
`
can be calculated analytically in the limit t;1, where Equation (32) becomes
N
1
"1![1!exp (!y0)]`/0 (33)
The solution of the equation cN
`
/c0"0 is
0
'
"1.256[1!exp (!N
'"
)] (34)
In other words, for the common range of N
'"
values (1}10), the optimal dimensionless charging
time is consistently a number of order 1. This conclusion continues to hold as t takes values
greater than 1.
Away from the optimal charging time (i.e. when 0P0 or 0PR), the entropy-generation
number N
`
approaches unity. In the short-time limit (0;0
'
), the entire exergy content
of the hot stream is destroyed by heat transfer to the liquid bath, which was initially at
environmental temperature
"
. In the long-time limit (0<0
'
), the external irreversibility takes
over: the used stream exits the heat exchanger as hot as it enters (
"'
"
`
), and its exergy
content is destroyed entirely by the heat transfer (or mixing) with the
"
atmosphere. The
traditional ("rst-law) rule of thumb of increasing the time of communication between heat source
and storage material is counterproductive from the point of view of avoiding the destruction of
exergy.
5. STRUCTURE SPRINGS OUT OF THERMODYNAMIC OPTIMIZATION
The summarizing conclusion that unites the examples reviewed in Section 4 is that the physical
result of global optimization of thermodynamic performance is structure (con"guration, topo-
logy, geometry, architecture, pattern). The examples covered structure in space (Section 4.1),
temperature (Sections 4.2 and 4.3) and time (Section 4.4) This structure-generating principle
deserves to be pursued further, in increasingly more complex system con"gurations. The genera-
tion of structure in engineering has been named constructal method; the thought that the same
principle accounts for the generation of shape and structure in natural #ow systems is constructal
theory (Bejan, 2000).
The principle of organizing structure for the purpose of extracting and using maximum exergy
from a hot stream is particularly relevant to the integrative conceptual design of energy #ow
systems for aircraft. The same principle applies to systems in which all the functions are driven
by the exergy drawn fromthe limited fuel installed on board: ships, automobiles, military vehicles,
environmental-control suits, portable power tools, etc. Additional support for this view is
provided by the record on powered #ight, engineered and natural. Figure 11 shows the cruising
speeds of insects, birds and airplanes, next to the theoretical speed obtained by minimizing
the power (exergy rate) destroyed during #ight (Bejan, 2000). The speeds of the aircraft
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 561
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Figure 11. Cruising speeds of insects, birds and airplanes, and the theoretical speed (<&M') for
minimum rate of exergy destruction (Bejan, 2000).
compiled in this "gure refer to the optimal cruising conditions*the speed for minimal power
consumption*not the maximal speed of the aircraft.
The performance record of the natural and engineered #ow systems (e.g., Figure 11) suggests
that the constructal principle is important not only in engineering but also in physics and biology
in general. In this theoretical framework the airplane emerges as a physical extension of man, in
the same way that the body of the #ying animal (e.g., bat, bird) developed its own well adapted
extensions. All such extensions are discrete marks on a continuous time axis that points
toward the better and the more complex. This theoretical line of inquiry is explored in a new book
(Bejan, 2000).
ACKNOWLEDGEMENTS
I want to thank my colleagues Michel Feidt (University Henri PoincareH , Nancy), Enrico Sciubba (University
of Rome I, La Sapienza) and Richard A. Smith (Air Force Research Laboratory) for their constructive
contributions to the revised form of this article.
NOMENCLATURE
A "area (m`)
C "thermal conductance (W K)
c, c
.
"speci"c heat (J kg K)
562 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
E "energy (J)
EQ
/
"exergy transfer rate (W)
e
`
"speci"c #ow exergy (J kg)
E
`
"#ow exergy (J)
EQ
`
"#ow exergy rate (W)
EGM "entropy generation minimization
g "gravitational acceleration (m s`)
h "speci"c enthalpy (J kg)
hM "overall heat transfer coe$cient (Wm` K)
m, M "mass (kg)
mR "mass #ow rate (kg s)
P "pressure (Pa)
Q "heat transfer (J)
QQ "heat transfer rate (W)
R "ratio of thermal conductances, Equation (23)
s "speci"c entropy (J kg K)
S "entropy (J K)
SQ
"
"entropy generation rate (WK)
t "time (s)
"temperature (K)
; "overall heat transfer coe$cient (Wm` K)
< "velocity (ms)
< "volume (m`)
=Q "power (W)
x "conductance allocation fraction
y "dimensionless group, Equation (26)
z "elevation (m)
Greek symbols
p
'
""rst law e$ciency
p
''
"second law e$ciency
0 "dimensionless time, Equation (28)
0 "temperature ratio, Equation (23)
"speci"c non#ow exergy (J kg)
"non-#ow exergy (J)
t "dimensionless temperature di!erence, Equation (30)
Subscripts
b "bath
c "collector
C "Carnot, reversible
H "high
L "low
m "melting
max "maximum
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 563
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
opt "optimum
out "outlet
p "pump
rev "reversible
t "turbine
u "user
0 "environment
REFERENCES
Ahern, JE. 1980. he Exergy Method of Energy Systems Analysis. Wiley: New York.
Bejan A. 1982. Entropy Generation through Heat and Fluid Flow. Wiley: New York.
Bejan A. 1988. Advanced Engineering hermodynamics. Wiley: New York.
Bejan A. 1996a. Entropy Generation Minimization. CRC Press: Boca Raton.
Bejan A. 1996b. Entropy generation minimization: the new thermodynamics of "nite-size devices and "nite-time
processes. Journal of Applied Physics 79:1191}1218.
Bejan A. 1997. Advanced Engineering hermodynamics (2nd edn). Wiley: New York.
Bejan A. 2000. Shape and Structure, from Engineering to Nature. Cambridge University Press: Cambridge, UK.
Bejan A, Mamut E. (eds). 1999. hermodynamic Optimization of Complex Energy Systems. Kluwer Academic Publishers:
Dordrecht, The Netherlands.
Bejan A, Tsatsaronis G, Moran M. 1996. hermal Design and Optimization. Wiley: New York.
Bejan A, Vadasz P, KroK ger DG. (eds). 1999. Energy and the Environment. Kluwer Academic Publishers: Dordrecht, The
Netherlands.
Benelmir R, Feidt M. 1997. Thermoeconomics and "nite size thermodynamics for the optimization of a heat pump,
International Journal of Energy Environment Economics 5:129}133.
Benelmir R, Evans RB, Von Spakovsky MR. 1991. Thermoeconomic analysis and design of a cogeneration system.
International Journal of Energy Environment Economics. 1:71}80.
Benelmir R, Evans RB, Von Spakovsky MR. 1992. High degree decentralization for the optimum thermoeconomic design
of a combined cycle. International Journal of Energy Environment Economics 2:155}164.
Benelmir R, Lallemand M, Lallemand A, Von Spakovsky MR. 1997. Exergetic and economic optimization of a heat pump
cycle. International Journal of Energy Environment Economics 5:135}149.
Brodyanskii VM. 1973. Exergy Method of hermodynamic Analysis. Energiia: Moskow.
De Vos A. 1992. Endoreversible hermodynamics of Solar Energy Conversion. Oxford University Press: Oxford, UK.
Evans RB. 1969. A proof that exergy is the only consistent measure of potential work (for chemical systems). Ph.D. hesis.
Dartmouth College: Hanover, New Hampshire.
Feidt M. 1987. hermodynamique et Optimisation E! nergetique des Syste& mes et Procede& s, Paris, Technique et Documenta-
tion, Lavoisier.
Feidt ML. 1998. Thermodynamics and the optimization of reverse cycle machines. In hermodynamic Optimization of
Complex Energy Systems, Bejan, A. and Mamut, E. (eds). Kluwer Academic Publishers: Dordrecht, The Netherlands.
385}402.
Haywood RW. 1980. Equilibrium hermodynamics. Wiley: New York.
Kotas TJ. 1995. he Exergy Method of hermal Plant Analysis. Krieger: Melbourne, FL.
Lim JS, Bejan A, Kim JH. 1992. Thermodynamic optimization of phase-change energy storage using two or more
materials. Journal Energy Resources echnology 114:84}90.
Moran MJ. 1989. Availability Analysis: A Guide to E.cient Energy ;se (2nd edn). ASME Press: New York.
Moran MJ, Shapiro HN. 1995. Fundamentals of Engineering hermodynamics (3rd edn). Wiley: New York.
Nerescu I, Radcenco V. 1970. Exergy Analysis of hermal Processes. Editura Tehnica: Bucharest.
Olsommer B, Favrat D, Von Spakovsky MR. 1999a. An approach for the time-dependent thermoeconomic modeling and
optimization of energy system synthesis, design and operation (Part I: Methodology and results). International Journal
of Applied hermodynamics 2(3):97}114.
Olsommer B, Favrat D, Von Spakovsky MR. 1999b. An approach for the time-dependent thermoeconomic modeling and
optimization of energy system synthesis, design and operation (Part II: Reliability and availability). International
Journal of Applied hermodynamics 2(4):177}186.
Radcenco V. 1994. Generalized hermodynamics. Editura Technica: Bucharest.
Reistad GM. 1970. Availability: concepts and applications. Ph.D. hesis, University of Wisconsin, Madison.
564 A. BEJAN
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565
Sciubba E. 1999a. Optimisation of turbomachinery components by constrained minimisation of the local entropy
production rate. In hermodynamic Optimization of Complex Energy Systems, Bejan A, Mamut E. (eds). Kluwer
Academic Publishers: Dordrecht, The Netherlands.
Sciubba E. 1999b. Allocation of "nite energetic resources via an exergetic costing method. In Bejan A, Mamut E. (eds).
hermodynamic Optimization of Complex Energy Systems. Kluwer Academic Publishers: Dordrecht, The Netherlands,
151}162.
Sciubba E, Melli R. 1998. Arti,cial Intelligence in hermal Systems Design: Concepts and Applications. Nova Science: New
York.
Shiner JS. ed. 1996. Entropy and Entropy Generation. Kluwer Academic Publishers: Dordrecht.
Sieniutycz S, Salamon P. (eds). 1990. Finite-ime hermodynamics and hermoeconomics. Taylor and Francis: New York.
Smith RA. 2000. Personal communication, May 18.
Stecco SS, Moran MJ. (eds). 1990. A Future for Energy. Pergamon: Oxford, UK.
Stecco SS, Moran MJ. (eds). 1992. Energy for the ransition Age. Nova Science: New York.
Tsatsaronis G. 1999. Design optimization using exergoeconomics. In hermodynamic Optimization of Complex Energy
Systems, Bejan A, Mamut E. (eds). Kluwer Academic Publishers: Dordrecht, The Netherlands, 101}116.
Valero A, Tsatsaronis G. (eds). 1992. ECOS'92, Proceedings of the International Symposium on E.ciency, Costs,
Optimization and Simulation of Energy Systems, Zaragoza, Spain. ASME Press: New York.
Valero LA, Correas L, Serra L. 1999. In On-line thermoeconomic diagnosis of thermal power plants. Bejan A, Mamut E.
(eds). hermodynamic Optimization of Complex Energy Systems. Dordrecht, Kluwer Academic Publishers: The Nether-
lands. 117}136.
Von Spakovsky MR. 1994. Application of engineering functional analysis to the analysis and optimization of the CGAM
Problem. Energy2he International Journal 19 (special issue): 343}364.
Von Spakovsky MR, Frangopoulos CA. 1993. The environomic analysis and optimization of energy systems (Parts I and
II). Proceedings of the International Conference on Energy Systems and Ecology. ENSEC 193, Vol. 1. ASME: New York,
123}144.
Von Spakovsky MR, Frangopoulos CA. 1994. The evironomic analysis and optimization of a gas turbine cycle with
cogeneration. hermodynamics and the Design, Analysis and Improvement of Energy Systems Vol. AES 33, ASME:
New York.
EXERGY ANALYSIS AND ENTROPY GENERATION MINIMIZATION 565
Copyright 2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:545}565