Preprint for J. Non-Equilib. Thermodyn., Vol.22, No. 4, pp.

311-355 (1997)

Endoreversible Thermodynamics
Karl Heinz Homann, Josef Maximilian Burzler,
and Sven Schubert
Institut fur Physik, Technische Universitat,
D-09107 Chemnitz, Germany
Communicated by B. Andresen, Copenhagen, Denmark

Abstract
All energy transformation processes occurring in reality are irreversible and
in many cases these irreversibilities must be included in a realistic description of
such processes. Endoreversible thermodynamics is a non-equilibrium approach in
this direction by viewing a system as a network of internally reversible (endore-
versible) subsystems exchanging energy in an irreversible fashion. All irreversibili-
ties are conned to the interaction between the subsystems. In this review a general
framework for the endoreversible description of a system is presented, followed by
a discussion of the performance of such systems. Thereafter the scope of the re-
view is narrowed to time-independent stationary or cyclicly operating systems. We
present the endoreversible theory of heat engines, and give an overview over the dif-
ferent heat transfer laws used in the entropy interactions between the subsystems.
Also engine cycles dierent from the Carnot cycle and internal irreversibilities as
well as the design optimization for such systems are discussed. These aspects are
also important in the description of refrigerators and heat pumps which follows.
Then combined and staged systems comprising several subsystems and their per-
formance are reviewed and we conclude with a presentation of selected applications
of endoreversible thermodynamics.

1. Introduction
We are living in a world full of heat engines, refrigerators and other energy transforma-
tion devices. Even life can be regarded as special form of energy transformation. The
understanding of these devices and of the processes they perform have been central to
our technological development in the past and will be of equally great importance for
our future development.
Equilibriumthermodynamics as it evolved during the 19th century has provided a macro-
scopic theory for the description of these thermodynamic systems. However, it puts its
focus more on the states rather than on the processes occuring between the states. In
doing so it became a powerful theory, but the prize to be paid till today is that more
often than not processes are understood as a (quasi-static) sequence of equilibria ne-
glecting irreversibilities. This idea of reversible processes is however at variance with
the experience that in everyday live the processes occuring are not reversible, and {
more important { are not designed to be reversible, because the desired nite rates for
the energy transformation require nite and thus irreversible uxes.

1
During the years there have been a number of attempts in engineering and physics to
overcome this view, and one of the non-equilibrium thermodynamic elds, which devel-
oped during the last 20 years, has been labeled endoreversible thermodynamics [1,2].
The basic idea of the eld is to describe a non-equilibrium system as a collection of
equilibrium systems, such that all the irreversibilities occuring in a process are due to
the interaction between those subsystems. This approach is certainly supported by the
observation that in our surroundings we can indeed nd subsystems which can to a good
approximation be treated as equilibrium systems. In addition, all the power of conven-
tional equilibrium thermodynamics can be used for the description of the subsystems,
while at the same time dissipative processes are no longer neglected.
Most of the work published in the eld of endoreversible thermodynamics has been fo-
cused on determining performance predictions for heat engines and other energy trans-
formation devices which include the necessary irreversibilities caused by nite transfor-
mation rates. The approach is thus closely connected to nite time thermodynamics
[3{10], which originally put its focus on the inclusion of dissipative eects caused by
nite time or nite rate operation. In that eld the very idea of endoreversibility has
often been used [6,11{15] to describe the systems under consideration, but also other
modeling assumptions have been made [5,16]. Here, we shall conne ourselves to en-
doreversible systems in the original sense, with two narrow exceptions which will become
apparent later.

2. Endoreversibility and endoreversible systems

2.1 Denition of the endoreversible subsystems
An endoreversible system consists of a number of subsystems which interact with each
other and with their surroundings. We choose the subsystems so as to insure that each
one undergoes only reversible processes. All the dissipation or irreversibility occurs in
the interactions between the subsystems or the surroundings. An endoreversible system
is thus dened by the properties of its subsystems and of its interactions. We call
processes of such systems endoreversible process.
For a given thermodynamic system, the choice of an appropriate subdivision into subsys-
tems depends on what relaxation processes are included in the dissipation, i.e. depending
on what is considered a reversible process. Very often the separability of relaxation time
scales inside the composite system allows to dene the subsystems in a natural way:
the relaxation processes inside each subsystem must be fast compared to the rates of
relaxation between subsystems, i.e. the interactions between subsystems. However, one
should note that slow or fast relaxation is not equivalent to strong or weak dissipation,
so once again the dening property of an endoreversible system is that all subsystems
undergo only reversible processes.
If a subsystem is for instance a spatially uniform working uid, than the requirement that
it undergoes only reversible processes means that it is always in internal thermodynamic
equilibrium. But subsystems can be also more aggregate objects, namely engines (or
more general energy transformation devices). If for instance such an engine takes in heat
at temperature TH and converts it into work and heat discharged at temperature TL ,
then endoreversibility requires its eciency  to be the Carnot eciency C = 1 ? TL =TH .
This will become more apparent in our rst example.

2
 1100.0
0.0 100.0200.0300.0400.0500.0600.0700.0800.0900.01000.0
TH TH

qH

Temperature T
K TiH
2 3
TiH

P 1 4
TL

qL
TiL=TL -3 2 7
Entropy S
1e+01

Fig. 1 : Model of the endoreversible Novicov engine with nite heat conduction K to the high
temperature heat reservoir (left). TS-diagram of a Carnot cycle with a temperature dierence
to the high temperature heat reservoir (right).

2.2 An introductory example

As a simple introductory example we consider the Novicov engine [17,18], a simplied
version of the Curzon-Ahlborn engine treated later. The Novicov engine is a continuously
operating, reversible Carnot engine with the internal temperatures TiH and TiL . It is in
direct contact with the external low temperature heat bath at TL and is coupled to an
external high temperature heat bath at TH through a nite heat conductance K (see
Figure 1). The heat baths are both considered to be innite such that the in- and out ux
of energy does not change their temperatures.
The question is now how the irreversibility introduced by the nite heat conductance
in uences the performance of the engine. Does it for instance have an eect on the
eciency of the system?
We rst note that the total heat ux through the system is limited, and thus the power
produced by the engine is limited as well. To characterize the performance of this
endoreversible system in more detail we want to determine the maximumpower available
and the eciency at the operating point of maximum power.
Due to the nite heat conductance heat is only transported to the Carnot engine, if
its high temperature TiH is lower than the bath temperature TH . The heat ux qH
transported from the heat bath to the engine is assumed to be proportional to the
temperature dierence (Newtonian heat conduction)
qH = K(TH ? TiH ): (1)
At the low temperature side the heat can be discharged to the heat bath at TL without
any resistance. Thus this interaction between heat bath and engine is characterized not
by a transfer law but by the requirement that the lower temperature TiL of the Carnot
engine is the same as the bath temperature
TL = TiL : (2)
We note that the Carnot engine is characterized by three energy uxes and two temper-
atures: qH enters the engine at temperature TiH , the heat ux to the low temperature
heat bath qL leaves the engine at temperature TiL , and P is the power delivered by the
engine. The heat baths are described by (TH ; qH) and (TL ; qL) respectively, and the heat
conduction contains the parameter K. Three of the variables (TH ; TL; K) have exter-
nally xed values. All the variables (energy uxes, temperatures, and K) are related
by the interaction between the heat bath and the engine and by the constraints coming

3
from the endoreversibility of the Carnot engine. As the engine operates continuously in
a steady state, all the energy uxes have to balance
0 = qH ? qL ? P: (3)
In addition, as the engine operates reversibly the entropy uxes to and from the engine
have to cancel
0 = TqH ? TqL : (4)
iH iL
Solving now for P we obtain
   
T iL
P = qH 1 ? T = K(TH ? TiH ) 1 ? T : T L (5)
iH iH
For given temperatures of the heat baths and given K, the ow of heat through the
engine and the power produced by the inner Carnot engine will depend only on the
operating temperatures of the Carnot engine. As TH and TL are xed, the only control
to in uence the overall performance of the endoreversible engine is TiH , and we nd the
power P as a function of TiH only. Equation (5) alone is thus characterizing the entire
endoreversible Novicov engine with Newtonian heat conduction.
The maximum power is determined by dierentiation with respect to TiH
 
dP TH
0 = dT = K T 2 ? 1 ; T L (6)
iH iH
p
from which we nd TiH = TH TL . Operating with this temperature the maximum
power is p p 2
Pmax = K TH ? TL (7)
and the eciency in terms of the bath temperatures is
r
(Pmax ) = 1 ? TTiL = 1 ? TTL : (8)
iH H

The reader should note the remarkable fact that this eciency does not depend on the
size of the heat conductance K. Also note that this eciency is not a bound for heat
engines operating not at the maximum power point.
This simple example has shown how with a relatively modest eort new and interesting
results can be obtained for the performance of heat engines operating out of equilibrium.
The basic quantities to describe the system were the energy and entropy uxes, the
temperatures and the heat conductance. The underlying reason for the simplicity of our
derivation is the basic assumption of endoreversibility, which allows to make use of the
Gibbs equation for the internally reversible working uid leading to the coupled balance
equations for energy and entropy.

3. Endoreversible systems { a formal description

In this section we want to generalize the above example towards a general and more
formal description of endoreversible systems. Basically, an endoreversible system is a
network of reversible subsystems exchanging energy. Setting up the mathematical de-
scription of an endoreversible system is quite easy, usually a number of balance equations

4
and transport equations have to be combined. In that the task is similar to the descrip-
tion of an electrical network, where dierent elements are connected and the Kirchho
rules are used to determine the relation between the currents and voltages.
For an endoreversible system things are a bit more complicated, for energy is not the
only exchanged quantity. In each interaction between subsystems energy is accompanied
by another quantity, be it for instance entropy, momentum or a particle ux. This leads
to the special structure of endoreversible systems, which we will present below.

3.1 Thermodynamics of equilibrium systems

We start our description by considering the basic building block of endoreversible sys-
tems, a thermal equilibrium system. It is described by its state variables, but as usual
for equilibrium systems, one has some freedom in the choice of these state variables. As
an example consider an ideal gas, which can be described for instance by its volume and
pressure, or by its volume and entropy. To simplify the description of the theory we
start with a special choice of variables, namely the energy picture. In the energy picture
the energy E is considered to be a function of all the other extensive thermodynamic
variables of a system.
In the following, Xi will denote the extensive thermodynamic variables of subsystem i,
for instance the volume Vi , or particle number Ni , all of which are counted by . We
note that there may be thermodynamic variables like the surface of the subsystem, which
are not truly extensive, but for the sake of simplicity are here called 'extensive' too. The
entropy Si of subsystem i is a well dened state variable due to endoreversibility and
belongs to the set of extensive variables. Thus the state of the subsystem is uniquely
described by the set of its extensities fXi g. We then have
Ei = Ei(Xi ): (9)
Note that (9) denes the properties of subsystem i, i.e. specifying Ei as a function of
the Xi determines what the thermal behavior of that subsystem is. The energy E is not
conned to be the internal energy, it can in addition include the translational kinetic
energy, the rotational kinetic energy, or the potential energy in one or more external
elds. Also all elastic energy could be included. In each case the proper 'extensive' (or
more common 'dynamical') variables must be included in the set of extensities. Naturally
the energy then becomes also a function of these variables. Even pure mechanical systems
are within the scope of this description, which is very convenient for instance in the
combined treatment of dynamic and thermodynamic systems.
Due to endoreversibility all the standard equilibrium relations hold within a subsystem.
We obtain the respective conjugate intensive variables Yi from (9):
@Ei :
Yi = @X (10)

i
The Gibbs equation becomes X
dEi = Yi dXi ; (11)

and in each subsystem the extensive and the intensive variables are related via the
equations of state (10).
Due to the Gibbs equation each in ux of an extensity Ji = X_ i into the system carries
an accompanying in ux of energy Ii [19]:
Ii = Yi Ji: (12)

5
For instance any heat ux q is carried by an entropy ux q=T, or an angular momentum
ux M (torque) carries an energy ux !M, where ! is the angular velocity.

3.2 Endoreversible systems

With these denitions we can now give a more precise description of an endoreversible
system:
Endoreversible systems are networks of reversible subsystems connected by interactions
in which energy is exchanged between them or with the surroundings.
Each subsystem i is characterized by a number of contact points (or contacts), through
which the subsystem receives or discards energy. Through each contact the energy is
transported by an extensity (a carrier) Xi , for instance entropy or volume. The contacts
for the same extensity in one subsystem are numbered by r.
Each
;r
functions assigned to it (Yi;r ; Ji;r ; Ii;r ). Here
contact has three (time-dependent) ;r
Ii ;r is the energy ux into the system, Ji is the associated ux of the carrier Xi and
Yi is the corresponding thermodynamic intensity for that contact.
To make the notation easier we combine the indices (i; ; r) into a single one , which
counts all dierent contacts in the composite network:
Y = Yi;r ; etc. (13)
Endoreversibility guarantees that the energy and extensity in ux at each contact are
always related by
I = Y J ; (14)
in other words, the intensity puts an energy value on the in ux of an extensity.
We distinguish between two dierent types of subsystems, reservoirs and engines.

3.2.1 Reservoirs
A reservoir is a thermodynamic system in equilibrium, characterized by either
a) given intensities Yi. This is the case for innite reservoirs where the in ux of an
extensity does not change the value of the intensity. In the introductory example
both heat baths were of this type.
b) its extensities Xi and its energy function Ei(Xi ). Then the intensities are known
Yi = @Ei =@Xi , and due to its internal equilibrium they are uniform throughout
the subsystem and thus the contact intensities Yi;1 = Yi;2 = : : :  Yi are equal
for all r. From the balance equations for the extensities and the energy one nds
X
X_ i = Ji;r (15)
r
and X X X X
E_ i = Ii;r = Yi;r Ji;r = Yi Ji;r ; (16)
;r ;r  r
where we have assumed that the extensities are neither destroyed nor produced
within a subsystem.

6
3.2.2 Engines
An engine is a reversible subsystem, for which the contact variables are related by special
balance requirements for the extensities and the energy. For an engine operating in a
steady state one requires X
0 = Ji;r (17)
r
and X X
0= Ii;r = Yi;r Ji;r ; (18)
;r ;r
while for cyclic engines with cycle time ttot
Z ttot X
0= dt Ji;r (19)
0 r
and Z ttot X Z ttot X
0= dt Ii;r = dt Yi;r Ji;r (20)
0 ;r 0 ;r
holds. Note that for the endoreversible engine one does not need to know the equations of
state for its working uid. In the introductory example equations (3) and (4) correspond
to equations (18) and (17), respectively.

3.2.3 The interactions

The interactions describe, how the contacts of the subsystems exchange energy. The
contact points are connected by the interactions such that each contact belongs to one
specic interaction. An interaction
is characterized by the set of contacts which belong
to it, and by the specic extensity X  , through which the contacts exchange energy. If
the interaction is reversible, only contacts for the exchanged extensity are needed. If the
interaction is irreversible, entropy is produced, and then at least one additional contact
is needed in which the produced entropy can be deposited.
Some of the extensities (like angular momentum) and energy are conserved quantities
by nature, while others (like the particle number of chemical species) are not. In a (com-
plete) interaction all the conserved quantities must balance to zero. We shall see later
however, that often it suces to consider only partial interactions. A specic interaction

can be either reversible or irreversible, and can be either of the two following cases:
a) All the contact intensities ;  2
obey Y = Y . In the above discussed Novicov
engine the interaction between the engine and the lower heat bath is of this type.
b) The interaction is dened by a transport law which gives either the ux of the
extensity
J = J (fY! g; fXig; zm ) (21)
or the corresponding ux of the energy
I = I (fY! g; fXig; zm ) (22)
at each of the involved contacts as functions of the intensities, the extensities (for
reservoirs) and of additional external parameters zm , which are counted by m.
These parameters are mentioned here explicitly, as they will be used as 'controls'
to adjust the uxes in optimizing the performance of endoreversible systems.

7
To explain the case b we consider three examples:
Heat conduction between two heat baths.
In the above discussed Novicov engine the interaction of the engine to the upper heat
bath is of this type. Here two entropy contacts are coupled such that the heat ux (1) is
a function of the two bath temperatures, and K is an external parameter. The entropy
uxes do not balance as the transport is irreversible.
Particle exchange.
Let us consider a conserved particle ow J from one reservoir (contact 1) to another
(contact 2) due to a dierence in the chemical potentials  :
J = ?J1 = J2 = h(1 ? 2 ) : (23)
If we assume that the energy ux taken from contact 1 but not deposited in 2

I = I1 ? I2 = J(1 ? 2 ) (24)
is dissipated to heat and put into an entropy contact 3 we have
I3 = J(1 ? 2) (25)
and the entropy ux created is
J3 = J(1T? 2) : (26)
3

Energy transfer by angular momentum ux with friction.

Let us assume a drive shaft connects two angular momentum (work) reservoirs, trans-
porting energy from contact 1 to 2. Let us further assume that the bearing acts with a
frictional torque (contact 3) M3 = MFric = ! proportional to the angular velocity !
on the shaft. The dissipated energy goes as heat into the bearing. Then the interaction
is modeled with two contacts for the bearing, an angular momentum contact 3 and an
entropy contact 4. In a stationary state the torques M on the drive shaft (the angular
momentum uxes) have to balance
M1 + M2 + ! = 0 ; (27)
from which the contact intensities !1 = !2 = ! = ?(M1 + M2 )= are obtained. The
dissipated energy ux is put as heat into contact 4:
I4 = !M3 = !2 (28)
carried by the created entropy ux
2
J4 = !
T : (29)
4
The angular momentum ux M3 does not carry any energy into contact 3 as that contact
was assumed to have angular velocity zero.
The interactions can of course be much more complicated, however the three examples
already showed the importance of the interactions from a theoretical point of view: They
act as constraints on the contact variables.

8
3.3 The characterization of endoreversible systems
Collecting the dierent elements introduced above, we nd that an endoreversible system
is described by its contact variables and the extensities of the reservoirs. If the system
does not contain nite capacity reservoirs, then the extensities of the reservoirs can be
excluded from the description. Usually some of the contact variables and extensities
will be given, for instance the temperatures of some heat bath, while others remain
undetermined. They are however not completely free, as all of them are related by
constraints due to
{ the Gibbs relation at each contact
{ the balance equations in the reservoirs
{ the balance equations in the engines
{ the interactions.
Thus an endoreversible system is completely characterized by this set of algebraic and
ordinary dierential equations relating its contact variables and reservoir extensities.

3.3.1 Simplifying the description by excluding contact variables

The characterization of an endoreversible system can often be simplied by excluding
some of the contact variables from the description. Naturally it depends on the analysis
intended, which variables can be excluded. For instance in the above example of the
drive shaft connecting to angular momentum contacts, one might not be interested in
the dissipated energy and the angular momentum going into the bearing. Then it suces
to consider only the two contacts 1 and 2. Note that in this case of a partial interaction
the equal intensities !1 = !2 do not indicate a reversible contact, and that the energy
is no longer conserved in such an interaction.
Also some of the balance equations for the carriers Xi in the subsystems might be savely
ignored. The reason is that without an eect on the energy balance, contacts for that
carrier with zero intensities can be added, which correct the balance equation. As an
example consider the power output of the Novicov engine example. This energy ux
is certainly associated with a carrier, for instance a ux of angular momentum (in the
shaft connecting the engine to an electric generator or some other user). So in principle
the engine in the example should have a balance equation for the angular momentum,
however this balance equation can always be made correct with no eect on the energetic
balance by adding an angular momentum contact with intensity (angular velocity) zero.
In real engines this is realized by bolting the engine to the ground.
In most of the publications on endoreversible systems the only carrier taken into account
is the entropy, and thus only entropy contacts and so-called `work' contacts are consid-
ered, the latter implying that there is no entropy transported through that contact and
that the carrier can be neglected.
And nally the Novicov engine can serve as an example for another simplication: There
is no `work' reservoir, where the power output of the engine is delivered to. Instead the
power just leaves the endoreversible system.

9
3.3.2 Simplifying the description by eliminating contact variables
To further simplify the characterization of the system one often tries to reduce the
number of variables by eliminating some of them explicitly using the given constraints.
In our above example we did this by solving for the power P which nally depends only
on the temperature TiH .

4. The performance of endoreversible systems

We now turn to the analysis of endoreversible systems. A large number of dierent
systems have been analyzed in the literature, using dierent schemes and levels of
sophistication. The level of mathematical sophistication needed depends crucially on
the question, whether time-dependent contact variables and subsystem extensities are
present or not. If they are not present the level of mathematical sophistication reduces
considerable. Time-dependent endoreversible systems on the other hand require often
the use of control theory or of the calculus of variation for the discussion of their per-
formance extremes [20,21].
Due to the limited space we conne this review to endoreversible systems which operate
either in a stationary mode or in a cyclic mode with time-independent ows and contact
intensities during the branches of the cycle. Eventhough this seems to be a rather
restrictive selection, the majority of the published work can be treated this way. Time-
dependent endoreversible systems will be treated separately and will the subject of an
upcoming review on `Endoreversible Dynamics'.

4.1 The performance of stationary endoreversible systems

We rst consider the steady state operation of an endoreversible system. In such an
operation all contact variables are time-independent, for instance reservoirs are charac-
terized by stationary intensities. Then all the contact variables become simple variables.
Some of them may have externally assigned xed values, for instance the temperatures
of some heat bath, while others remain undetermined, but connected by the constraints
discussed above.
In a geometric sense the whole endoreversible system is thus characterized by a hyper-
surface of all possible operating points in the multidimensional space spanned by all the
contact variables.
Sometimes one is not only interested in discussing the performance of an endoreversible
system as a function of its thermodynamic (contact) variables, but also as a function of
certain external parameters zm . These can be for instance the heat conduction areas in
heat exchangers. Due to economic constraints the total heat exchanger inventory might
be given, and the question might be how the performance is changed by dierent alloca-
tions. In such a case the contact variables are supplemented by these parameters to form
a higher dimensional space, and the added constraints together with the thermodynamic
constraints lead to a hyper-surface of possible operating and design points.

4.2 The performance of cyclic endoreversible systems

The cyclic operation of endoreversible systems is a little more complicated than the
stationary operation. Usually the cycle time ttot after which the system has regained its

10
initial state (apart from the changes in the reservoirs) is divided into several branches b
of duration tb X
ttot = tb : (30)
b
The simplest case is the one in which all the contact variables remain constant during
one branch. This case includes the important Carnot cycle of an engine coupled to two
innite heat baths.
In the cyclic case each contact is assigned b triples of contact variables (Y;b;
X;b =
tbJ;b ;
E;b = tbI;b ), each obeying
E;b = Y;b
X;b , and the balance equations for
an engine subsystem become XX
0= ;r

Xi;b (31)
b r
and XX ;r :
0=
Ei;b (32)
b ;r
Usually most of the uxes will vanish, as the interaction { for instance the heat conduc-
tion from a reservoir to the engine { operates only during one or two of the branches.

4.3 Performance measures

As the characterization of an endoreversible system by its hyper-surface is quite com-
plicated the usual next step in the analysis of an endoreversible system is the study
of certain performance measures [22], which are dened on the contact variables (and
the external parameters) and thus on each (operating) point on the hyper-surface. In a
way such a performance measure can be considered as a projection of the complicated
hyper-surface onto one dimension.
Sometimes also two-dimensional projections are considered, where one performance mea-
sure is discussed as a function of another. Examples for performance characteristics of
this sort are the power{eciency curve [23] and the COP vs. cooling load curve.
Often the extreme values which performance measures can achieve are used in the char-
acterization of endoreversible systems. In our introductory example the power as one
performance measure was maximized, and the eciency of the engine at that operating
point was determined.
In the literature a large number of dierent performance measures is used, depending on
what type of energy transformation device is studied. Here we review a few performance
measures used in conjunction with the discussion of endoreversible systems. Note that
these are in general all dierent but some of them can become equivalent under certain
constraints.

4.3.1 Power and work per cycle

For heat engines the power (in steady state operation) and the work per cycle (in cyclic
operation) are important performance measures. Due to the irreversible transport of
heat in many cases a given endoreversible system can produce only a limited amount
of power. Then the maximum attainable power is naturally an important property
of the system, and thus in many publications the (average) power is considered as an
optimization goal. Note that in some situations a careful distinction between power and
work per cycle (average power) is needed [24,25].

11
4.3.2 Eciency
For a heat engine producing power P from an input heat ow q, or producing an amount
of work W from an amount of heat Q in one cycle the eciency  is dened as
 = Pq = WQ: (33)

De Vos [26] denotes this standard eciency as `exhaustible energy conversion eciency'
to distinguish it from the `renewable eciency' [26], which is of importance in the context
of renewable energy conversion, as for instance in solar thermal power plants. The
renewable eciency r is dened as the ratio of the work output W and the maximal
amount of heat Qmax which could be extracted from the heat source if it was linked
directly to the low temperature sink, i.e.
r = W=Qmax : (34)

4.3.3 The coecient of performance of coolers and heat pumps

A common performance measure for characterizing refrigerators and heat pumps is the
coecient of performance COP. Both, refrigerators and heat pumps absorb the heat QL
at a temperature TL , and reject the heat QH at a higher temperature TH . The work
needed to accomplish this task is W. The COP is dened as
COPR = QWL and COPP = QWH (35)
for refrigerators and heat pumps respectively.
For heat driven refrigerators or heat pumps the COP is typically (see [27], for example)
dened as the ratio of the absorbed heat QL or the rejected heat QM , respectively, and
the high temperature heat QH which drives the device
COPHR = Q QL and COP = QM : (36)
HP
H QH
4.3.4 Entropy production
In many discussions the entropy production (per cycle) or entropy production rate
X
= JiS;r (37)
i;r
has been used as a performance measure. As entropy production is considered to describe
losses, often the operating point with minimal entropy production has been the goal in
the optimization of endoreversible systems. One should note however that the amount
of power foregone due the irreversibilities depends on the temperatures of the heat baths
to which the entropy is discharged [28].
The concept of entropy minimization has been used extensively. In particular, Bejan
has published numerous articles and books on the topic (see [29{32], for example).
Recently, he reviewed [33,34] the topic tracing the development and adoption of the
method of entropy generation minimization in several sectors of mainstream thermal
engineering and science, such as cryogenics, heat transfer, storage systems, solar power
plants, nuclear and fossil power plants, and refrigerators.

12
4.3.5 Ecological performance measure
An ecological performance measure for heat engines has been proposed by Angulo-Brown
[35,36]. The criterion combines the power P and the entropy generation rate of the power
plant , and maximizes the function
E = P ? TL  (38)
where TL is the heat sink temperature.
Angulo-Brown considers the criterion as a long-range goal in the sense that it is com-
patible with ecological objectives. Optimizing the Curzon{Ahlborn heat engine using
the ecological criterion gave 80% of the maximum power but only 30% of the entropy
production that one would get if using power alone as objective function. The corre-
sponding eciency of the engine is almost equal to an engine optimized for maximum
power output alone. The ecological criterion was also used for modeling the standard
air Otto cycle [37] yielding similar results.

4.3.6 Economic performance measures

Economically optimal energy conversion systems are not designed to optimize any of the
before mentioned performance measures. In economic optimizations also prices have to
be taken into account. Economic performance measures may for instance use the costs
for fuel and investments for building the plant, or the money income achieved from the
sale of the output product, e.g. electrical power. Taking this into account neither the
reversible nor the maximum power operation might be desired. The higher the power
output, the faster the investments will be recovered { and the higher the eciency, the
better the expenses for the fuel will be recovered.
In a number of publications a variety of economic performance measures have been
proposed using dierent approaches of determining the overall costs to operate thermo-
dynamic plants subject to dierent economic requirements [6,38{42]. In most cases a
linear combination of investment costs and operating costs is used for the total cost
C = aI + bF ; (39)
where a; b are the respective prices and I is the measure for the size of the plant and F
is the measure for the use of fuel.
An intriguing and general result showing the tradeo between fuel and investment costs
is the following. If the fuel costs of a power plant are the predominate part then the
working point should be near the reversible operation to obtain a high eciency. In
the reverse case of inexpensive renewable energy, like solar or wind energy, the working
point should be near the maximum power point to obtain maximum power output.
Instead by using economic performance measures, economic interests can also be intro-
duced to thermodynamic models by cost constraints. For instance the total costs for
heat exchangers or insulations etc. might be given, which then have to be optimally
allocated [43].

4.3.7 Other performance measures

There are a number of further performance measures which have been used for the
discussion of endoreversible systems. These include for instance the loss of availability,
the work deciency and eectiveness [7,22,28,44].

13
 TH
KH QH
TiH

TiL Fig. 2 : Curzon{Ahlborn model

KL QL of an endoreversible heat engine
with nite heat transfer to and
TL
from the heat reservoirs.

5. Heat engines
Heat engines producing power from a heat ow have been central to the development of
endoreversible thermodynamics, and a large fraction of the literature is devoted to them.
In this section we will bring the formal description of endoreversible thermodynamics
to life by applying it to the Curzon{Ahlborn engine [45], a famous `work horse' in this
eld. In this model the only irreversibilities are due to the heat transfer.
We then present the eects of dierent heat transfer laws, of heat leaks, and of internal
irreversibilities (though this violates in principle the endoreversibility assumption) on
the performance of endoreversible engines, and we discuss engine cycles other than the
Carnot cycle.

5.1 The Curzon{Ahlborn engine

5.1.1 Power-optimal cyclic operation
The Curzon{Ahlborn heat engine is depicted in Figure 2. It consists of two heat baths at
constant temperatures TH and TL and a reversible Carnot engine operating between the
temperatures TiH and TiL . The thermal couplings between the reversible Carnot engine
and the heat reservoirs are irreversible and obey the Newtonian heat transfer law.
We here consider a cyclicly operating engine. Delivering the work W per cycle, the
engine absorbs the heat QH from the hot temperature reservoir during the time tH and
rejects the heat QL to the low temperature reservoir during the time tL :
QH = KH tH (TH ? TiH ) (40)
QL = KL tL (TiL ? TL ) (41)
where KH and KL are the respective thermal conductances.
The time spent in the isentropic branches of the Carnot cycle is considered to be negli-
gible compared to the isotherms such that the total cycle time is the sum of the times
spent in the isothermal branches:
ttot = tH + tL : (42)
For the Carnot engine subsystem the energies exchanged during one cycle balance ac-
cording to equation (32)
0 = QH ? QL ? W : (43)

14
 Pmax

0.20
Power
0.10
ηmax
Fig. 3: Power vs. eciency plot
0.00
for an endoreversible Curzon{
0 0.2 0.4 0.6 0.8 Ahlborn engine with nite heat
Efficiency η transfer.

The balance equation of the carrier (31), i.e. the entropy, can be written as
0 =
SH ?
SL = TQH ? TQL (44)
iH iL
if we take into account that no entropy is transported through a work contact.
The goal is now to determine the maximum work W per cycle (which also maximizes
the maximum average power output as the total cycle time is xed) and the eciency at
that operating point. Instead of following the original work of Curzon and Ahlborn [45]
we present here another derivation due to Gordon et al. [46], in which the optimal time
allocation for the branches separates nicely from determining the optimal temperatures
for the Carnot engine.
Let us rst dene  = TiL =TiH . Using (43) and (44) we obtain for the work
W = QH(1 ? ) = KH tH (TH ? TiH )(1 ? ) : (45)
From (40), (41), and (44) follows
=Q QL = KL tL (TiL ? TL ) = KL tL(TiH ? TL ) : (46)
H KH tH (TH ? TiH ) KH tH (TH ? TiH )
Solving (46) for TiH and inserting into (45) yields
W = (1 ? )(TH ?  ?1 TL) KKHt K+L tKHtLt = CTH  1C?? (47)
HH LL
where we introduced the eciencies
C = 1 ? TL=TH (48)
 = 1 ? TiL =TiH = 1 ?  (49)
and  ?1  ?1
C = K 1t + K 1t = K 1t + K (t 1 ? t ) : (50)
HH LL HH L tot H

A plot of the average power P = W=(ttot versus eciency characteristics (47) is depicted
in Figure 3.
One can obtain the maximum work point by setting both (@W=@)tH and (@W=@tH )
equal to zero. Since C depends on tH and not on  the problem factorizes as mentioned
above.

15
The condition (@W=@tH ) = 0 gives a relation between the branch times and the con-
ductances: p
tH=tL = KL =KH : (51)
The condition (@W=@)tH = 0 leads to the Curzon{Ahlborn eciency
p
(Pmax ) = CA = 1 ? TL =TH (52)
at maximum work, which is obtained by inserting (52) and (51) into (47)
Wmax = ttot ?p KH KpL
2 TH ? TL :
p p 2
(53)
KH + KL
Note that Wmax still depends on the conductances KH and KL .
The Curzon{Ahlborn eciency is much closer to observed eciencies than the corre-
sponding Carnot eciencies [30,45]. Nonetheless a careful analysis of the dissipative
processes as well as of the optimization goals in real engines remain important [47].
5.1.2 Power-optimal design
We now turn to the problem of the optimal allocation of heat exchangers KH and KL
to the hot and cold thermal coupling for a given heat exchanger inventory [43,48{50]:
Ktot = KH + KL (54)
This constraint is used to substitute KH in (53), which leads to the optimality condition
for maximum work
@Wmax =@KL = 0 : (55)
A simple calculation results in the optimal allocation of the conductances
KH;opt = KL;opt = Ktot =2 (56)
and the optimized work
Wmax = ttot K8tot TH ? TL
p p 2
(57)
Note that dierences to results in the literature are due to dierences between steady
state and cyclic operation.
Often, the conductance K = A is dened as the product of the surface area A of
the contact and the conductivity . Then for given conductivities H and L the heat
exchanger areas AH andpAL can beppower-optimally allocated for the constraint Atot =
AH + AL . The result is HAH = L AL [51].

5.2 Heat transfer laws

The Novicov and the Curzon{Ahlborn engine both showed that irreversibilities due to
nite heat conduction drastically in uence the behavior of endoreversible systems. In
both cases the Newtonian heat conduction was chosen to model the entropy interactions
between engine and heat baths.
In general these interactions can be much more complicated, and the form of the heat
transfer law will signicantly in uence the behavior of an endoreversible system. In the
following, we give an overview over the dierent heat transfer laws that have been used
in the literature in general and for the Curzon{Ahlborn engine in particular. All the
transfer laws have in common that the energy ow (heat ow) is given as a function of
the temperatures of the two contacts connected.

16
5.2.1 Newton
The so-called Newton heat transfer law has already been used in this paper. It assumes
that the ow of heat q is proportional to the temperature dierence between two contacts
1 and 2
q = K(T1 ? T2 ) : (58)
Heat ux in solids can be well approximated by a linear heat transfer law. But the
Newtonian heat transfer may also be applied to other heat transfer mechanisms like
convection if the temperature dierences are small, as a dominating linear term fre-
quently occurs in the Taylor expansion of more general, non-linear heat transfer laws.
Due to its simplicity the Newton heat transfer law is widely used and many authors have
studied the performance of endoreversible engines, coolers and heaters based on such a
linear heat transfer law [1{3,5,6, 32,45,52{60].

5.2.2 Fourier
For the Fourier heat transfer law the heat ux q is proportional to the dierence of the
inverse temperatures,
q = K(1=T2 ? 1=T1) (59)
where K is an Onsager coecient. This form of heat transfer is often found in con-
junction with linear irreversible thermodynamics, as there the dierence of the inverse
temperatures is the force corresponding to the heat ux.
The Fourier heat conduction law has been used in endoreversible systems [61], like the
Curzon{Ahlborn model for example. If the Fourier heat transfer law is used in both heat
exchangers of the Curzon{Ahlborn model, the analysis [62,63] shows that the eciency
at maximum power output
p
(Pmax ) = [1 + KH =KL](1p? TL =TH ) (60)
2 + (1 + TL =TH) KH =KL
depends on the external temperatures and the Onsager coecients. But still the depen-
dence is only on their ratio KH =KL not on both KH and KL separately.
In the limit KL =KH ! 1 the Curzon{Ahlborn model reduces to the Novikov model
with a Fourier heat transfer law. In this case the eciency
 
(Pmax ) = 21 1 ? TTL (61)
H
depends on external temperatures only [26].
For the special case KH = KL the eciency becomes [64]
(Pmax ) = 1 ? 13++3T L =TH
TL =TH : (62)

5.2.3 Radiation
Electromagnetic radiation from a hot body like the sun can serve as a source of heat
for heat engines and heat-driven refrigerators. For some systems, in particular those
operating at high temperatures, radiation is the major transfer mechanism for heat.

17
Radiative heat transfer is typically described by the Stefan-Boltzmann law for black-
body radiation, and the heat ux between two radiating bodies at temperature T1 and
T2 is given by
q = K1 T14 ? K2 T24 : (63)
The coecients K are proportional to the Stefan-Boltzmann constant, the emittances of
the two radiating bodies, and geometry factors. Solar collectors are typical applications
where radiative heat transfer is involved as an interaction, for further applications see
Sections 8.3, 8.4, and 8.5.

5.2.4 Dulong-Petite
In some physical situations the heat transfer between two subsystems has conductive
as well as radiative components, where the latter part can not be ignored [65]. One
attempt to describe a combined conductive-convective and radiative heat transfer in a
simplied fashion is the so-called Dulong-Petit law [36,52]
q = K(T1 ? T2 )n (64)
where K is a proportionality constant. The value of the exponent n is usually in the
range between 1:1 and 1:6 [36].
Angulo-Brown and Paez-Hernandez [36] investigated the Curzon{Ahlborn model using
the Dulong-Petit heat conduction law with n = 5=4 and presented examples of power
plants where the predicted theoretical eciencies are in very close agreement (99%) to
the observed eciencies. Chen and Yan et al. [66{68] examined the in uence of the
Dulong-Petite law on the performance of a forward and reverse Carnot cycle using the
Curzon{Ahlborn model. The analysis was done for arbitrary n and is quite general.
Expressions and optima for various interesting performance parameters like maximum
power, coecient of performance, heating rate and the internal temperatures have been
derived.

5.2.5 Generalized heat transfer law

Some authors generalized the above discussed heat transfer laws and analyzed systems
obeying a non-linear heat transfer law of the form
q = K1 T1n ? K2 T2m ; (65)
which includes the Newton (n = m = 1), Fourier (n = m = ?1), and radiative (n =
m = 4) heat transfer laws as special cases.
An example of such an analysis has been provided by De Vos [23,26] who used a heat
transfer law of the form q = K(THn ? TiHn ) for the Novikov heat engine. The power output
is then given by
P = q = (1 ? TL =TiH )[K(THn ? TiHn )] : (66)
The maximum power condition @P=@TiH = 0 leads to an equation for the unknown
temperature TiH
nTiHn+1 ? (n ? 1)TLTiHn ? THn TL = 0 : (67)
For an arbitary n one cannot solve (67) analytically, but has to compute TiH numerically.
Substituting TiH into (66) gives the maximum power Pmax .
Figure 4 shows the results of the analysis for the Novikov power plant at its maximum
power operating point. The eciency is plotted versus the temperature ratio TL =TH for
dierent n.

18
 1.0
Carnot n=-1
n=1
0.8

Efficiency η(Pmax)
n=2
n=3
0.6
: Energy conversion e-
n=4
Fig. 4
0.4 ciency  of a Novicov engine at
maximum power generation as a
0.2
function of the temperature ratio
0.0 TL =TH for dierent exponents in
0 0.2 0.4 0.6 0.8 1 the heat transfer law q = K (THn ?
Temperature Ratio TL/TH TiHn ).

TH
QH
TiH Pmax

ηmax
W

Power P
Q leak

TiL
ηCarnot
QL

0.0
TL 0 ηmax ηCarnot

Efficiency η

Fig. 5: Schematic diagram of an endoreversible engine with a bypass heat leak (left) and plots
of average power (work per cycle time) P versus eciency , for an engine with a heat leak
(right). The dashed line corresponds to a model without heat leak. The solid line shows the
loop-type behavior of a model with heat leak. Both, power and eciency vanish in the thermal
short-circuit limit of very fast operation and in the limit of very slow engine operation. The
maximum power and the maximum eciency point are relatively close together (see [47]).

The in uence of the non-linear heat transfer law q = K(T1n ?T2n ) on the Curzon{Ahlborn
model was investigated by Chen [62], Gordon [46], and Nulton et al. [69]. Orlov [64] also
investigated the Curzon{Ahlborn model but used an extraordinary heat transfer law
   9
q = K1 T1 ? T1 + K2 T1 ? T1 : (68)
1 2 1 2
Optimizing for maximum eciency at a given amount of heat he obtained a general
cycle consisting of three isothermal and three adiabatic branches.

5.3 Heat engines with heat leaks

In real heat engines heat leaks are unavoidable, and there are many features of an actual
power plant which fall under that kind of irreversibility, such as the heat lost through
the walls of a boiler, a combustion chamber, or a heat exchanger, and the heat ow
through the cylinder walls of an internal combustion engine, for example. Such losses
have a decisive in uence on the performance of the engines.
To study this a heat leak is easily introduced into an endoreversible heat engine. Figure 5
shows an extended Curzon{Ahlborn model with heat leak. Then if the endoreversible
engine operates fast, the internal temperature dierence TiH ? TiL becomes small and the
eciency of the endoreversible engine degrades. If on the other hand the engine operates

19
slow, heat is lost through the heat leak causing a decrease in eciency. Between this
extremes, there are operation modes with internal temperatures TiH and TiL for which
either the eciency or the power output of the engine is maximized.
While the heat leak does not aect the work output of the engine cycle, it enters the
expression for the eciency which is now dened as
() = Q W() + Q = W : (69)
leak H Kleak ttot (TH ? TL ) + W1?()

As W can be obtained from the Curzon{Ahlborn calculation as a function of the tem-

perature ratio  (47), equation (69) gives the dependence of  on . Thus W() (47)
and () (69) dene the (average) power{eciency curve for an engine with heat leak
as shown in Figure 5. Here the parameter  varies between TL =TH (innitely slow
operation) and 1 (innitely fast operation).
The most remarkable feature is the loop-type behavior of the P{ characteristics, which
is markedly dierent from the one for the case without heat leak. As expected, both
power and eciency vanish at fast or slow engine operation, and the dierence between
maximum power and maximum eciency points is small.
Including heat leaks brings the endoreversible engine models closer to reality. A convex
P{ characteristics (dashed line in Figure 5) is rarely observed in the real world. Loop-
shaped P{ curves on the other hand appear to be universal [30,47] and can be observed
in real heat engines [70], in which also the power as well as the eciency vanish if the
engine operates at a too slow speed.
There are several examples in the literature where a loop-type power versus eciency
characteristics has been found. A reversible heat engine with bypass heat leak was
studied by Bejan [71]. The model in Figure 5 was analyzed by several authors including
Bejan [43,50,51], Chen [72], Gordon and Huleihil [47], and Pathria et al. [73]. Moukalled
et al. [74] diskussed a slightly dierent model for nonlinear heat conduction laws in all
three heat links.

5.4 Heat engines with internal irreversibilities

For many devices such as gas turbines, automotive engines, steam turbines, and thermo-
electric generators, the nite heat transfer is not the sole contribution to the irreversibil-
ities. Other loss mechanisms, like friction or regenerator losses, etc. play an important
role, but are hard to model in detail. Therefore, one tried to lump the losses into one
(or a few) additional elements and parameters.
Some authors simply use one parameter to describe the internal losses. Such a parameter
is typically associated with the entropy i produced inside the engine during a cycle.
For a Carnot cycle, for example, one nds
I
i = Q = QL ? QH  0:
T TiL TiH (70)

Eventhough such an approach in principle violates the concept of endoreversibility it

can { from the point of mathematics { be very easily incorporated into the framework
of endoreversible thermodynamics. This is outlined below.
To proceed further with the analysis one needs to make some modeling assumptions
for i . To stay within the realm of our description, all what is needed is to have a
dependence of i on the contact variables or a combination of them. For instance, one

20
 Cycle Adiabates Isotherms Isochors Isobars
Carnot 2 2
Stirling 2 2
Ericsson 2 2
Brayton1 2 2
Otto 2 2
Diesel 2 1 1
Atkinson 2 1 1
Dual combustion 2 2 1
Rallis 2 2 2
Tab. 1 : Branches employed in common cycles (adopted from [44]).

could assume that the produced entropy is proportional to the entropy in ux on the hot
isothermal
i = ( ? 1) TQH ; (71)
iH
which can be rewritten as
 TQH = TQL : (72)
iH iL
In general   1 holds, where  = 1 corresponds to an endoreversible heat engine.
The irreversible engine can now be analyzed within the Curzon{Ahlborn model (Fig-
ure 2) using a linear heat conduction law for both heat links and assuming continuous
operation. The only dierence to the endoreversible calculation is, that the balance
equation for the entropy inside the engine is replaced by (72). The optimization of the
power output with respect to the internal temperatures gives
Pmax = (K K+H KKL ) TH ? TL 
p p 2
(73)
L H
at an eciency of r
(Pmax ) = 1 ? TTL  (74)
H
for the maximum power point [38,40,72,75,76]. The endoreversible engine result is
retrieved for  = 1.
As in the case of an endoreversible engine model, the internal irreversible ( > 1)
Curzon{Ahlborn model shows a convex power eciency characteristics with a single
maximumlike the one depicted in Figure 3. Thus internal irreversibilities have a dierent
eect on the performance than an external heat leak.
Models which employ the concept of internally irreversible engines have been described
for linear [48,77{80] and nonlinear [81] heat transfer laws, and for staged engines [82].
Chen [72] extended the model by including an external heat leak.

5.5 Engine cycles

Thermomechanical converters like heat engines, refrigerators and heat pumps are conve-
niently classied by the type of cycle the working uid undergoes. In the previous sec-
tions only the Carnot cycle has been discussed. However, practical applications are often
based on cycles dierent from the Carnot cycle. For instance non-isothermal branches
1 also known as Joule{Brayton cycle

21
 TH

1e+06

1113.0
2 3 3

Temperature T
8e+05

913.0
Pressure p
2

6e+05

713.0
4e+05

513.0
4

2e+05
1

313.0
4 TL
1

-2e+04

113.0
0 0.01 0.02 0.03 0.04 -0.2 0.3 0.8 1
Volume V Entropy S

Fig. 6 : pV- and TS-diagram of an ideal Brayton-cycle

can be constructed such that they match non-isothermal heat sources and sinks. Table
1 gives an overview on the type of branches which occur in common standard cycles.
For cycles using isochoric, isobaric, or polytropic branches heat will be transfered at
a time-dependent temperature, and thus the discussion of such engines is in principle
beyond the scope of this review. There are however a number of publications on such
cycles without considering explicit time-dependencies.
One example is a system [83,84], where the temperature of the reservoirs are time-
dependent and changed in such a manner that the dierence in temperature of the
reservoir and working uid remains constant during the heat exchanging branches. Other
publications belong to one of the following three dierent categories:

5.5.1 Optimizing ideal engine cycles

In this category the performance of ideal cycles, i.e. reversible cycles with no irreversibil-
ities on the heat in and out ux taken into account, is analyzed. Though no irreversibil-
ities on the heat transfer to and from the engine are considered, the mathematical treat-
ment is very similar to the calculations performed in endoreversible thermodynamics.
To highlight the dierence we present such an analysis of the Brayton cycle.
Figure 6 shows the ideal Brayton cycle with its two isobaric (2 ! 3 and 4 ! 1) and
two adiabatic (1 ! 2 and 3 ! 4) branches. T3 and T1 are the maximum and minimum
temperature of the cycle. For a working uid with temperature-independent specic
heat capacities C(23) and C(41) the amounts of heat exchanged on the isobaric heating
and cooling branches are
QH = C(23)(T3 ? T2 ) (75)
QL = C(41)(T4 ? T1 ) : (76)
The work output W of an ideal Brayton cycle is given by
W = QH ? QL = C(23)(T3 ? T2) ? C(41)(T4 ? T1 ) : (77)
The mathematical equivalence to the Curzon{Ahlborn engine with Newtonian heat
transfer becomes apparent by setting T3 = TH and T1 = TL , while T2 and T4 play
the roles of TiH and TiL , respectively. The heat capacities correspond to the heat con-
ductances.
Then maximizing the work output of one cycle leads necessarily to the eciency
p
(Wmax ) = 1 ? T1 =T3 (78)
at maximum work [85]. This is the well known Curzon{Ahlborn eciency again, al-
though the cycle was assumed to be ideal!

22
There is a number of heat engine investigations of this kind [43,47,83,84,86{91]. Lands-
berg and Le [92] for instance showed that the maximum-work eciencies of the ideal
Stirling, Otto, Diesel, and Atkinson cycle is well approximated by the Curzon{Ahlborn
eciency. All these cycles can be regarded as special cases of a `generalized' cycle con-
sisting of two adiabates and two heat transfer branches with constant heat capacities of
the working uid.
5.5.2 Engine cycles with regeneration
In a number of publications dierent engine cycles have been investigated, which make
use of regenerators. These cycles consist of two isotherms and two (or more) other
branches on which heat transfer occurs. If the heat which enters or leaves the engine
during the non-isothermal branches is deposited in a regenerator, and the regeneration
is lossless, i.e. reversible, then from the outside these engine cycles look like Carnot
cycles, with heat only exchanged with external heat baths on isotherms. Then { not
surprising { the results for the eciency at maximum power are exactly the same as for
the Curzon{Ahlborn engine. However, the maximum(average) power might be dierent.
For instance it is reduced in cyclic operation, if the regenerative branches take a nite
amount of the total available cycle time. The endoreversible Stirling [93], Ericsson [94],
and Rallis [95] cycle with perfect regeneration belong to this category. For non-perfect
regeneration internal irreversibilities occur [96].
5.5.3 The log mean temperature approach
The Brayton, Otto, Diesel, Atkinson, and Dual combustion cycle have no isothermal
branches. Thus an analysis of such a cycle would usually require a time-dependent
treatment, since the temperatures of the working uid change during the irreversible
heat transfer. Some authors [48,87,94,97{99] eliminated the explicit time dependence
of the heat transfer by approximating the variable temperature dierence by an eective
temperature dierence, the log mean temperature dierence (LMTD). The LMTD can
be motivated from the description of a heat transfer to a working uid with constant
heat capacity.
Many of the generalizations and alterations to the basic Curzon-Ahlborn engine pre-
sented above apply as well to other heat{work transformation devices, especially to the
refrigerators and heat pumps discussed below.

6. Refrigerators and heat pumps

Refrigerators and heat pumps are another major group of energy transformation de-
vices, which have been subjected to an endoreversible analysis. Figure 7 shows a model
which is used for endoreversible refrigerators and heat pumps. It consists { like the
Curzon{Ahlborn model { of an internally reversible Carnot engine and two nite heat
conductances to the temperature baths. The Carnot engine operates in a reverse fashion
such that it uses (mechanical) power to drain the heat ux qL (the cooling load) from
the low temperature bath TL and delivers the heat ux qH to the high temperature bath
TH .
The endoreversible analysis of refrigerators and heat pumps is markedly dierent from
that of a heat engine due to two reasons:
First of all the internal temperatures of heat engines have to lie between the high tem-
perature TH and the low temperature TL and thus the amount of energy transfered from

23
 TiH TiH
KH qH KH qH

TH TH
K L q leak

P P

TL TL
KL qL KL q L
TiL TiL

Fig. 7: Schematic of an en- Fig. 8 : Schematic diagram

doreversible refrigerator or of an endoreversible refriger-
heat pump with nite heat ator with heat leak [100].
transfer

the hot bath to the engine is limited. The internal temperatures of refrigerators or
heat pumps on the other hand lie outside this interval and might be very high or small
respectively. Thus the heat ow to the hot reservoir is not limited.
The second dierence is due to the appropriate measure of performance. In most cases
the COP as dened in section 4.3.3 for refrigerators and heat pumps is considered.
Several authors used the model of Figure 7 with Newtonian heat transfer to analyze
refrigerators [49,95,101{103] and heat pumps [104]. A typical analysis of the endore-
versible refrigerator model of Figure 7 in steady state operation gives the coecient of
performance as a function of the cooling load
COPR = T T?L T? q+L =C (79)
H L qL=C
with C = (1=KH + 1=KL)?1 . One nds that the COP is maximum for a cooling load
equal to zero and decreases monotonically with the cooling load, a behavior indeed
observed in real refrigerators. Thus for refrigerators (and for heat pumps as well) there
is no analogue to the maximum power calculation in heat engines. Curve (a) in Figure 9
shows the COP as a function of the specic cooling load.
For a given cooling load the COP-optimal allocation of heat exchanger inventory and
of heat exchanger areas can be determined, alternatively the analysis can focus on the
optimal allocation of heat exchanger inventory or area for maximum heating or cooling
rates at a given power [38,49,105,106].
p Then
p for given conductivities H;L, and total
area Atot = Ap H +AL thepresult is H AH = L AL . From this the optimal C  = Atot 
with  = (1= H +1= L )?2 is obtained. The model was also investigated for Dulong-
Petit heat transfer [66,68], and the generalized heat transfer law [67,68,107].
A more advanced model for refrigerators which includes also a heat leak between the
temperature reservoirs is shown in Figure 8 [100]. This model was investigated for
a linear heat transfer law in all three heat links and under steady state conditions
[51,73,105].
A thorough analysis of the eect of internal irreversibilities in the above described models
is provided in [106,108{116]. Chen and Wu [106] for instance found the coecient of
performance as a function of the specic cooling load r = qL=Atot

COPR;irr = T =TL??Tr=+irrr= (80)
H L irr

24
 10

8 Fig. 9 : The COP for an refrig-

a
erator versus the specic cooling
6 load r for TH = 300 K and TL =

COP
4 b
270 K. The curve (a) corresponds
to an endoreversible refrigerator,
2 c while the curves (b) and (c) cor-
respond a refrigerator with inter-
nal irreversibilities ?1 = 1, 1.1,
0
0 20 40 60 80
r/K
*
and 1.2 respectively [106].

TH TH2 TH
KH qH KH
TH qH
TH1 P1
K
P1 P2

TL1 TH1 qM K
KM qM KM
TH2 qM TL2

P2 P2
P1 K
TL2
qL TL
KL qL KL
TL TL1 TL

Fig. 10 : Two-staged power plant (left), refrigerator (middle), and complicated arrangement
(right) with three heat exchangers.
p
with irr = (1= H  + 1=pL )?2, and  being the irreversibility introduced above.
Figure 9 shows its dependence on .
The results of the theoretical models have been compared to the experimental data ob-
tained from measurements on actual refrigerators and heat pumps. It has been demon-
strated for various types of devices that the theoretical modeling is capable of describing
the fundamental aspects of refrigerator and heat pump behavior [30,105,111,117].

7. Staged and combined endoreversible systems

In many applications several energy transformation devices are coupled together. A
typical example are power plants where the waste heat from a high temperature cycle
is used as the heat source of a low temperature cycle. For the endoreversible analysis of
such a staged power plant the model shown left in Figure 10 is used. It consists of two
endoreversible subsystems with a nite thermal coupling to the temperature reservoirs
and between the endoreversible subsystems.
A typical problem for such a model is to determine the internal temperatures in steady
state operation which maximize the overall power output P = P1 + P2 for given heat
bath temperatures TH and TL and conductances KH , KL , and KM . The expression for
the maximum power [50,118]

Pmax = K K +KK
H KM KL
hp p i2
TH ? TL (81)
M L HKL + KH KM

25
has the same temperature dependence as the one for a single system (53). The staging
does not in uence the eciency at maximum total power which turns out to be the
Curzon{Ahlborn eciency
r
(Pmax ) = CA = 1 ? TTL : (82)
H
Rubin and Andresen [54] investigated the model in great detail for cyclic operation and
found the same result.
For a constrained total heat exchanger inventory Ktot = KH + KM + KL the power{
optimal allocation of the thermal conductances yields [50] KH;opt = KM;opt = KL;opt .
Chen and Wu [118] constraint the total area Atot = AH + AM + AL of the heat ex-
changers for given thermal conductivities H , L, and M. Then the power-optimal area
distribution leads to pH AH;opt = pM AM;opt = pL AL;opt : Subsequently, Chen and
Wu generalized the results to n-staged endoreversible power cycles [119,120].
De Mey and De Vos [121] investigated a staged system with three heat exchangers and
two endoreversible Carnot heat engines as shown at the right-hand side of Figure 10.
Using the total power output as the objective function gave an eciency of
  1=3
 = 1 ? TTL (83)
H
for each of the heat engines, a remarkable result as the square root occuring in the
Curzon{Ahlborn eciency is replaced by a cubic root.
Two-stage and multi-stage combined refrigeration cycles are often employed when ma-
terial limitations in refrigeration plants restrict the span of the temperatures between
the heat source and heat sink. A simple model of a two-staged refrigerator is depicted
in middle of Figure 10. The model has been analyzed for a linear heat transfer law
[57,122,123]. The dependence of the optimal coecient of performance on the cooling
load was found. For constrained heat exchanger areas or thermal conductivities the
optimal allocations have been derived, which are the same as for the heat engine case.
Another important combination of energy transformation devices is between a heat en-
gine and a refrigerator, where the heat engine's power output is used to drive the refrig-
erator. Combined arrangements of this kind were suggested as models for heat-driven
refrigerators (see section 8.1) [124{127] and heat driven heat pumps [128], respectively.
Adopting the respective COP as an objective function, performance parameters such
as the (specic) cooling load have been obtained and the optimal allocation of heat
exchanger areas was derived for the case of Newtonian heat ow.

8. Selected applications of endoreversibility

In this section we present a few classes of devices where endoreversible modeling has
been used. The systems treated are still small and simple, yet they show already that
with the endoreversible approach very dierent elds of engineering can be investigated.

8.1 Absorption refrigerator

The absorption refrigeration system uses heat instead of work to achieve cooling. It
primarily consists of a generator, an absorber, condenser and an evaporator, as shown

26
 Q M2 QH

Condensor Generator

Pump

Fig. 11 : Schematic of an ab-

Exvaporator Absorber sorption refrigeration cycle. The
straight arrows indicate the ow
QL Q M1 of the working uids in the refrig-
erator, not the ow of heat.

TH
qH
TiH

qM

TiM TM

TiL
Fig. 12: Three heat source
qL
model to describe the absorption
TL
refrigeration cycle

in Figure 11. The generator receives the heat ux qH from a high temperature source
at TH and the absorber rejects the heat ux qM1 at TM1 . This heat ow drives another
heat ow from a cold space at temperature TL , where the evaporator absorbs the heat
ux qL, to the condenser, which rejects the heat ux qM2 at TM2 . Here we discuss the
case that the temperatures TM1 and TM2 are given by a common ambient temperature
TM .
The absorption refrigerator can be modeled as an endoreversible refrigerator driven
by an endoreversible heat engine [124,125,127]. In a dierent modeling approach the
absorption refrigerator is treated as a three heat source system [27,129], which is depicted
in Figure 12. The reversible subsystem of the model receives the heat uxes qH and qL
from the hot heat source and cooled space, respectively, and releases the heat qM which
corresponds to qM1 + qM2 in Figure 11 to the ambient. The three heat source model is
similar to the tricycle model introduced by Andresen, Salamon, and Berry [3].
An analysis of the three heat source model has been provided by Chen [27] for a linear
heat transfer law in all three heat links. Figure 13 shows the cooling load qL aspa function
of thepcoecient of the performance COPHR = qL=qH parameterized by B = ( H=L ?
1)=( H=M + 1), where H, M , L are the corresponding conductivities pHAH +ofpthe heat
links. Here the optimal area allocation for maximum cooling load
p A for given conductivities has already been used. The same area allocation occurs L AL=
M M
if the device operates as a heat pump [130].
Endoreversible models for absorption refrigerators were further analyzed by Wu [131,
132], and Yan and Chen [133]. Gordon et al. [112,117] compared their analysis of the ab-
sorption refrigerator including internal irreversibilities with performance data from jour-

27
 1.0
a
: Cooling load of an ab-

Cooling Load QL/QL,max

b Fig. 13
c sorption refrigerator aected by
0.5 thermal resistances. Plots are
presented for TH = 120o C, TL =
15o C, TM = 40o C. Curves a, b,
and c correspond to the cases
0.0 where B=-0.5, 0, and 0.5 respec-
0 1 2 tively [27]. For an explanation
Coefficient of Performance COPHR for B see text.

nal articles, manufacturer catalogue data, and experimental measurements, and found
a good agreement with their theoretical predictions.

8.2 The thermoelectric generator

A thermoelectric generator converts a heat ow directly into electricity and is composed
of two dissimilar conductors or semiconductors between two heat reservoirs. The typi-
cally good thermal conductivities of converter materials additionally cause a direct ow
of heat between the two reservoirs. A particular convenient aspect of the thermoelectric
generator is that its complete power-eciency curve can be realized (short-circuit to
open-circuit limit). Theoretical models can be veried experimentally this way.
Chen [134{136] derived the eciency of the thermoelectric generator, and Wu [79,80]
modeled the thermoelectric generator using the Curzon{Ahlborn model for endore-
versible heat engines (see Figure 2). Gordon [70] used an endoreversible model with
heat leak (see Figure 5) which includes the major sources of irreversibility of the thermo-
electric generator. The analysis resulted in a loop-type power eciency characteristics
which can also be observed in real devices.

8.3 Solar thermal heat engines

In solar-thermal heat engines collectors receive radiation from the sun and convert it
into thermal energy, which is then fed to a heat engine. The thermodynamic limits of
this conversion has been the topic of a book by De Vos [41], and endoreversible solar
thermal models have been widely discussed in the literature [15,23,51,58,137{148].
A simple endoreversible approach to solar thermal heat engines is the Muser model [137]
where an endoreversible heat engine is radiatively coupled to a heat source. The Muser
model is identical to the Novikov model in Figure 1 except that the linear heat transfer
law is replaced by a radiative one. In the Muser model, the hot end of the engine at
temperature TiH receives heat from a source at temperature TH { typically the sun { and
rejects heat to a sink at temperature TL . The heat ow q between sun and the engine
is given by ?

q = K TH4 ? TiH4 : (84)
A straightforward analysis (for instance [23]) of the model gives the power-eciency
characteristic
P = K[(1 (1? )4 TH4 ? TL4 ] ; (85)
? )4
with  = 1 ? TL =TiH . Then maximum power is generated if 4TiH5 ? 3TLTiH4 ? TH4 TL = 0,
which is the n = 4 case of (67).

28
 β=0.1
β=0.2
β=0.5
β=1
β=5
Fig. 14: The eect of the tem-

Pmax
perature ratio ex = TL =TH on
the maximum power delivered by
an endoreversible heat engine ra-
diatively coupled to a collector
0.00
and a radiator plotted for several
ratios  of the heat transfer coef-
0 0.2 0.4 0.6 0.8 1
Temperature Ratio TL/TH
cients.

A slightly more complicated model is discussed by Goktun et. al. [147] who employed the
Curzon{Ahlborn model with radiative heat transfer to the hot and the cold heat bath.
The analysis of the model gives an expression for the maximum power Pmax which has
to be solved numerically. Figure 14 shows the results of the calculation. The maximum
power Pmax in units of the incoming radiation ux from the hot bath is plotted against
the temperature ratio ex = TL=TH for several ratios  = KH =KL of the heat transfer
coecients. The power output Pmax is largest for small ex , although the variations are
minor if ex < 0:2. In order to increase the thermal eciency and power output,  must
be decreased. This means that the heat conduction on the low-temperature side rather
than that of the high-temperature side has to be increased for good performance.
Bejan [51] showed for the special case TL = 0 that, if the total area of the collector and
radiator is constrained, maximum power output is achieved when the low temperature
radiator area is about twice as large as the high temperature collector area. Blank
and Wu [148] investigated an endoreversible Stirling engine with radiative heat links.
Howell and Bannerot [149] studied the regenerative Stirling, Ericsson, and Brayton heat
engines, and Stirling and Ericsson cycles were also documented by Badescu [150].

8.4 Solar to wind energy conversion

Wind energy is one of the oldest used renewable energy sources. In determining its
potential one needs to know the eciency with which energy can be extracted from
wind, and one needs information about the fraction of the incident solar energy which
is converted to wind energy. The latter question has been addressed in a number of
dierent (endoreversible) approaches [30,151{160] by considering wind generation as
a solar driven heat engine. Particular interesting is the work of Sertorio et al. who
extended the endoreversible theory to a general eld description where the temperature
is a eld T(x; t) and power is produced uniformly in space [151,152,154,157]. In all
the models endoreversible thermodynamics is applied to calculate upper bounds for the
fraction of solar energy that can be converted into wind energy within the atmosphere
of a planet. De Vos summarizes and discusses several of the models in a book [41].
The heat transfer to and from the `wind engine' is an irreversible process due to radiative
heat transfer from the sun to the earth and from the earth into space. A commonly used
model of an endoreversible `wind engine' is the Curzon{Ahlborn engine with radiative
heat transfer. Although this models crudely simplies the processes in the atmosphere
of a planet and does not consider the laws governing movements of air, it gives a rst
estimate for the conversion eciency of solar energy to wind. Initially two dierent
applications of the model were considered. In one, the cold isotherm occurs in the upper
atmosphere where the air cools, and the hot isotherm occurs at the surface of the planet

29
 TH
QH Fig. 15: Schematics of an en-
doreversible model for the wind
TiH
generation and the subsequent
αW dissipation of the wind energy.
The fraction  of the (kinetic)
W
(1−α) W
wind energy is dissipated to heat
TiL at the hot temperature of the en-
QL gine, and the remaining energy
is dissipated to heat at the cold
TL temperature of the engine.

where the air is heated [155]. In the other version the hot isotherm is on the day side of
the earth, and the cold isotherm is on the night side [156].
De Vos and Flatter [156] enlarged the model of Gordon and Zarmi [155] by including the
dissipation of the wind energy. The fraction  of the (kinetic) wind energy is dissipated
to heat at the hot temperature of the engine, and the remaining energy is dissipated
to heat at the cold temperature of the engine. Figure 15 shows the corresponding
endoreversible model.
These processes serve as good example for irreversible interactions in endoreversible
systems. The process is modeled as an interaction with one power contact and two
entropy contacts. The momentum transfer, which occurs in the generation of the wind
and in its dissipation is not considered in this model, as it is of no importance within
this purely energetic consideration.
The calculations [156] gave an upper bound for the conversion eciency of solar energy
into wind energy of 8.3% for the planet Earth. Subsequently, De Vos and van der
Wel [158] rened their endoreversible model, considering several endoreversible `wind
engines' on the earth surface. A much lower boundary for the conversion eciency
was obtained by Nuwayhid and Moukalled [160] who included a heat leak from the
illuminated side to the dark side of the planet. In a recent work, De Vos and van der
Wel [159] assumed the presence of six Hadley cells (macroscopic atmospheric cycles) as
endoreversible subsystems in the Earth's wind system. The model gives an upper limit
of 1.17% for the eciency of the solar to wind energy conversion. This value compares
well with recently obtained empirical gures of 0.8% to 1.0% [161,162].

8.5 Solar cells and photo-voltaic engines

A solar cell converts light directly into electric power. If the electron transfer in the cell is
taken into account, the endoreversible model becomes more complicated than a photo-
thermal converter since the description needs not only temperatures to characterize
the device, but also chemical potentials. However, there are also simpler modeling
approaches.
A number of authors [41,163{166] analyzed solar cells as thermodynamic engines.
Baruch and Parrott [163] discussed the possibility of creating a reversible Carnot cycle
for photo-voltaic energy conversion. They used the electron-hole plasma as the working
uid of the cycle. De Vos addressed the topic with the question "Is a solar cell an en-
doreversible engine? " [164]. As an example, a single junction solar cell was investigated.
For this device, a current-voltage characteristics has been derived that is very similar
to the q{ characteristic (heat ux { eciency characteristic) of the radiative, endore-
versible Muser engine (see section 8.3). Eventually, De Vos showed that a solar cell can
indeed be described as an endoreversible system. Later he and others [166] created a

30
 high chemical µH
hot reservoir TH
potential reservoir

qH JH

TiH µiH
P P
q leak J leak
TiL µiL

qL JL

cold reservoir TL low chemical µL

potential reservoir

Fig. 16 : The models of an endoreversible heat engine and a chemical engine show many simi-
larities. The heat engine operates between two heat reservoirs at a high and a low temperature
TH and TL , respectively. The only irreversibilities are due to heat transfers QH , QL , and a heat
leak Qleak over nite temperature dierences. The internal temperatures of the engine are TiH
and TiL . The endoreversible chemical engine employs reservoirs at a high and a low chemical
potential H and L , respectively. The irreversibilities are associated with the particle transfer
JH , JL , and Jleak over nite chemical potential dierences (adopted from [168]).

thermodynamic model of a solar cell using energy and entropy transport equations. The
model consists of two surfaces which radiate to each other. The only irreversibilities are
located in two surfaces where excited electrons are thermalized. Within this model a
generalization of the Landsberg inequality, which gives a bound on the eciency of a
photo-voltaic engine, was obtained for the nite band gap case. In his book [41], De Vos
gives also a review on the photo-voltaic conversion of solar radiation.

8.6 Chemical engines

Chemical engines [167{169] and heat engines show many similarities in the way they op-
erate. While chemical engines generate work from dierences in chemical potential heat
engines convert dierences in temperature into work. Chemical potential and particle
transfer then play the analogous roles of temperature and heat transfer. Some engines
exploit dierences in both temperature and chemical potential for power production.
The exchanged extensities for chemical engines can for instance consist of electrons (so-
lar cells) or molecules (gas exchange in the respiratory system). Examples for chemical
engines are biological systems, and electro-chemical, photochemical (photovoltaics) or
solid state devices.
The similarities of heat engines and chemical engines suggest to apply the same set of
endoreversible reasoning to chemical engines that was used for the heat engine. Figure
16 shows the endoreversible model of a heat engine and a chemical engine [168].
Basically, the heat transfer equations are replaced by particle transfer equations in the
case of a chemical engine. In the model of the chemical engine, let JH and JL denote
the rate of particle transfer between the engine and the high and low chemical potential
reservoir, respectively. Additionally, there may be a particle leak which bypasses the
endoreversible chemical engine. The transport equation for the particle ows are given
by
JH = hH (H ? iH ) (86)
JL = hL (iL ? L): (87)
Supercially the chemical engine looks to be an analogon to the heat engine with New-

31
tonian heat transfer, however, a closer look reveals that the analogy between the two
engines is not perfect. While in the heat engine the linear transfer law applies to the
energy transported, in the chemical engine the linear transport law applies to the carrier
of the energy, namely a particle ux. This small dierence in the interaction leads to
completely dierent behavior of the engine.
The balance equations for energy and particle number lead to
P = iH JH ? iL JL ; (88)
where P is the produced power, and
0 = JH ? JL : (89)
Solving for iH and iL gives P as a function of J = JH = JL alone
 
J
P = J H ? L ? h ? h : J (90)
H L
Maximizing with respect to J then gives the optimal internal chemical potentials and
the optimal particle ux:
J = (H ? L ) 2(hhH+hLh ) : (91)
H L
The chemical engine serves as a good example for an endoreversible system, where the
interactions obey no balance equation for the energy. As the particle ux in each of the
interactions is conserved, and the chemical potentials are dierent for the two contacts
in an interaction, energy is obviously lost. For the performance measure maximized in
this example that is of no concern, and thus the energy uxes accompanying the particle
uxes can be savely ignored. If, however, one desires to take these also into account for
a more general characterization of a chemical engine, one can do so, see for instance the
work by De Vos [167] and Mironova et al. [169].
For the above chemical engine model Gordon [168] discussed also the cyclic operation.
They included the nite rate particle leak and gave expressions for the maximum power
output. They also discussed the more complicated particle transfer law
 
1 ? exp 2
h  i
J / exp kT kT (92)
which depends on the temperature T of the system as well.

9. Conclusion
Endoreversible thermodynamics in our view is the successful attempt to include irre-
versibilities and dissipative processes into the description of thermodynamic processes,
while at the same time preserving the advantages of classical reversible thermodynamics.
The central idea is to think of a system as a network of subsystems { each undergoing
only reversible processes { which exchange energy. All irreversibilities occur only in the
interactions between the subsystems. Treating systems in this way one gets one step
closer to a realistic description of real dissipative processes.
In this review we presented the general framework for the endoreversible description of
thermodynamic systems undergoing irreversible processes. Depending on the desired
accuracy of the description a system can be separated into a larger or smaller number

32
of subsystems. This way the irreversibilities of the energy exchange between the parts
of the system can be taken into account. The discussion focused equally on obtaining a
proper mathematical theory and on the characterization of the systems performance.
We showed how the framework is applied to heat engines and gave an overview on
the dierent heat transfer laws. The optimization of ideal reversible engine cycles and
internal irreversibilities were discussed, two topics beyond the original denition of en-
doreversible systems, but closely related. Our aim was not so much to elaborate a large
number of dierent energy transformation devices, our aim was a presentation from a
systematic point of view. As a good starting point for further reading of this subject
and related topics we recommend references [33,41,170].
Due to the limitations of space we conned this review to the time-independent treat-
ment of endoreversible systems. Nonetheless, a large number of dierent systems was
discussed, which showed that endoreversible thermodynamics is a conceptually simple
and successful step towards the inclusion of irreversibilities in the description of ther-
modynamic processes.

References
[1] Rubin, M. H., Optimal conguration of a class of irreversible heat engines. I, Phys.
Rev. A, 19 (1979) 1272.
[2] Rubin, M. H., Optimal conguration of a class of irreversible heat engines. II,
Phys. Rev. A, 19 (1979) 1277.
[3] Andresen, B., Salamon, P., Berry, S. R., Thermodynamics in nite time: Ex-
tremals for imperfect heat engines, J. Chem. Phys., 66 (1977) 1571.
[4] Andresen, B., Berry, R. S., Nitzan, A., Salamon, P., Thermodynamics in nite
time. I. the step-Carnot cycle, Phys. Rev. A, 15 (1977) 2086{2093.
[5] Salamon, P., Andresen, B., Berry, R. S., Thermodynamics in nite time. II. po-
tentials for nite-time processes, Phys. Rev. A, 15 (1977) 2094{2102.
[6] Salamon, P., Nitzan, A., Finite time optimizations of a newton's law Carnot cycle,
J. Chem. Phys., 74 (1981) 3546{3560.
[7] Andresen, B., Rubin, M. H., Berry, R. S., Availability for nite-time processes.
general theory and a model, J. Phys. Chem., 87 (1983) 2704{2713.
[8] Andresen, B., Finite-Time Thermodynamics, Physics Laboratory II, University of
Copenhagen, (1983).
[9] Andresen, B., Salamon, P., Berry, R. S., Thermodynamics in nite time, Phys.
Today, 37 (1984) 62{70.
[10] Sieniutycz, S., Salamon, P., eds., Finite-Time Thermodynamics and Thermoeco-
nomics, Taylor and Francis, New York, (1990).
[11] Salamon, P., Band, Y. B., Kafri, O., Maximum power from a cycling working uid,
J. Appl. Phys., 53 (1982) 197{202.
[12] Mozurkewich, M., Berry, R. S., Optimal paths for thermodynamic systems: The
ideal otto cycle, J. Appl. Phys., 53 (1982) 34{42.

33
[13] Homann, K. H., Watowich, S. J., Berry, R. S., Optimal paths for thermodynamic
systems: The ideal Diesel cycle, J. Appl. Phys., 58 (1985) 2125.
[14] Watowich, S. J., Homann, K. H., Berry, R. S., Intrinsically irreversible light-
driven engine, J. Appl. Phys., 58 (1985) 2893{2901.
[15] Gordon, J., Nonequilibrium thermodynamics for solar energy application, in:
Finite-Time Thermodynamics and Thermoeconomics, Advances in Thermody-
namics 4, eds. Sieniutycz, S., Salamon, P., Taylor and Francis, New York, (1990),
p. 95.
[16] Salamon, P., Berry, R. S., Thermodynamic length and dissipated availability, Phys.
Rev. Lett., 51 (1983) 1127{1134.
[17] Novikov, I. I., The eciency of atomic power stations, Atomnaya Energiya, 3
(1957) 409, (in Russian).
[18] Novikov, I. I., The eciency of atomic power stations, Journal Nuclear Energy II,
7 (1958) 125{128, translated from Atomnaya Energiya, 3 (1957), 409.
[19] Falk, G., Ruppel, W., Energie und Entropie, Springer, Berlin, (1976).
[20] Homann, K. H., Optima and bounds for irreversible thermodynamic processes,
in: Finite-Time Thermodynamics and Thermoeconomics, Advances in Thermody-
namics 4, eds. Sieniutycz, S., Salamon, P., Taylor and Francis, New York, (1990),
p. 22.
[21] Sieniutycz, S., Salamon, P., Thermodynamics and optimization, in: Finite-Time
Thermodynamics and Thermoeconomics, Advances In Thermodynamics 4, eds.
Sieniutycz, S., Salamon, P., Taylor and Francis, New York, (1990), p. 24.
[22] Rubin, M. H., Figures of merit for energy conversion processes, Am. J. Phys., 46
(1978) 637.
[23] De Vos, A., Re ections on the power delivered by endoreversible engines, J. Phys.
D: Appl. Phys., 20 (1987) 232{236.
[24] Wu, C., Kiang, R. L., Lopardo, V. J., Karpouzian, G. N., Finite-time thermody-
namics and endoreversible heat engines, Int. J. Mech. Eng., 21 (1993) 337{346.
[25] Kiang, R. L., Wu, C., Clarication of nite-time thermodynamic-cycle analysis,
Int. J. Power Energy Syst., 14 (1994) 69.
[26] De Vos, A., Eciency of some heat engines at maximum-power conditions, Am.
J. Phys., 53 (1985) 570{573.
[27] Chen, J., Performance of absorption refrigeration cycle at maximum cooling rate,
Cryogenics, 34 (1994) 997{999.
[28] Homann, K. H., Andresen, B., Salamon, P., Measures of dissipation, Phys. Rev.
A, 39 (1989) 3618{3621.
[29] Bejan, A., Entropy Generation through Heat and Fluid Flow, Wiley, New York,
(1982).
[30] Bejan, A., Advanced Engineering Thermodynamics, volume 51, Wiley, New York,
(1988).
[31] Bejan, A., Entropy Generation Minimization, CRC Press Inc., Boca Raton, FL,
(1996).

34
[32] Bejan, A., Models of power plants that generate minimumentropy while operating
at maximum power, Am. J. Phys., 64 (1996) 1054{1059.
[33] Bejan, A., Entropy generation minimization: The new thermodynamics of nite-
size devices and nite-time processes, J. Appl. Phys., 79 (1996) 1191{1218.
[34] Bejan, A., Notes on the history of the method of entropy generation minimization
(nite time thermodynamics), J. Non-Equilib. Thermodyn., 21 (1996) 239{242.
[35] Angulo-Brown, F., An ecological optimization criterion for nite-time heat engines,
J. Appl. Phys., 69 (1991) 7465{7469.
[36] Angulo-Brown, F., Paez-Hernandez, R., Endoreversible thermal cycle with a non-
linear heat transfer law, J. Appl. Phys., 74 (1993) 2216{2219.
[37] Angulo-Brown, F., Fernandez-Betanzos, J., Diaz-Pico, C. A., Compression ratio
of an optimized air standard otto-cycle model, Eur. J. Phys., 15 (1994) 38{42.
[38] Ibrahim, O. M., Klein, S. A., Mitchell, J. W., A relation between thermodynamic
performance and economics for heat engines, ASME AES, 24 (1991) 15.
[39] Gordon, J. M., Huleihil, M., On optimizing maximum-power heat engines, J. Appl.
Phys., 69 (1991) 1{7.
[40] Ibrahim, O. M., Klein, S. A., Mitchell, J. W., Eects of irreversibility and eco-
nomics on the performance of a heat engine, J. Solar Energy Engng., 114 (1992)
267{271.
[41] De Vos, A., Endoreversible Thermodynamics of Solar Energy Conversion, Oxford
University Press, Oxford, (1992).
[42] De Vos, A., Endoreversible thermoeconomics, Energy Convers. Mgmt, 36 (1995)
1{5.
[43] Bejan, A., Theory of heat transfer-irreversible power plants, Int. J. Heat Mass
Transfer, 31 (1988) 1211{1219.
[44] Salamon, P., Nitzan, A., Andresen, B., Berry, R. S., Minimum entropy production
and the optimization of heat engines, Phys. Rev. A, 21 (1980) 2115.
[45] Curzon, F. L., Ahlborn, B., Eciency of a carnot engine at maximum power
output, Am. J. Phys., 43 (1975) 22.
[46] Gordon, J. M., Observations on eciency of heat engines operating at maximum
power, Am. J. Phys., 58 (1990) 370{375.
[47] Gordon, J. M., Huleihil, M., General performance characteristics of real heat en-
gines, J. Appl. Phys., 72 (1992) 829{837.
[48] Lee, W. Y., Kim, S. S., Power optimization of an irreversible heat engine, Energy,
16 (1991) 1051{1058.
[49] Bejan, A., Power and refrigeration plants for minimum heat exchanger inventory,
J. Energy Res. Techn., 115 (1993) 148{150.
[50] Bejan, A., Theory of heat-transfer irreversible power plants - II. the optimal allo-
cation of heat-exchange equipment, Int. J. Heat Mass Transfer, 37 (1995) 433{444.
[51] Bejan, A., Power generation and refrigeration models with heat transfer irre-
versibilities, J. Heat Transfer Soc. Japan, 33 (1994) 68{75.

35
[52] Gutkowicz-Krusin, D., Procaccia, I., Ross, J., On the eciency of rate processes.
power and eciency of heat engines, J. Chem. Phys., 69 (1978) 3898.
[53] Blanchard, C. H., Coecient of performance for a nite speed heat pump, J. Appl.
Phys., 51 (1980) 2471{2472.
[54] Rubin, M. H., Andresen, B., Optimal staging of endoreversible heat engines, J.
Appl. Phys., 53 (1982) 1{7.
[55] Ondrechen, M. J., Rubin, M. H., Band, Y. B., The generalized Carnot cycle -
a working uid operation in nite-time between nite heat sources and sinks, J.
Chem. Phys., 78 (1983) 4721{4727.
[56] Rebhan, E., Ahlborn, B., Frequency-dependent performance of a nonideal Carnot
engine, Am. J. Phys., 55 (1987) 423{428.
[57] Chen, J., Yan, Z., Optimal performance of an endoreversible-combined refrigera-
tion cycle, J. Appl. Phys., 63 (1988) 4795{4798.
[58] Gordon, J. M., On optimized solar-driven heat engines, Solar Energy, 40 (1988)
457{461.
[59] Andresen, B., Finite-time thermodynamics, in: Finite-Time Thermodynamics and
Thermoeconomics, Advances in Thermodynamics 4, eds. Sieniutycz, S., Salamon,
P., Taylor and Francis, New York, (1990), p. 66.
[60] Orlov, V. N., Berry, R. S., Analytical and numerical estimates of eciency for an
irreversible heat engine with distributed working uid, Phys. Rev. A, 45 (1992)
7202{7206.
[61] Chen, L., Sun, F., Chen, W., Optimal expansion of a heated working uid in a
cylinder with a movable on piston in the case of q /
(T ?1)s, Chinese J. Mech.
Engng, 29 (1993) 97.
[62] Chen, L., Yan, Z., The eect of heat-transfer law on performance of a two-heat-
source endoreversible cycle, J. Chem. Phys., 90 (1989) 3740{3748.
[63] Chen, J., Yan, Z., Optimal performance of endoreversible cycles for another linear
heat transfer law, J. Phys. D: Appl. Phys., 26 (1993) 1581{1586.
[64] Orlov, V. N., Optimum irreversible Carnot cycle containing three isotherms, Sov.
Phys. Dokl., 30 (1985) 506{508.
[65] O'Sullivan, C. T., Newton law of cooling { a critical assessment, Am. J. Phys., 58
(1990) 956{960.
[66] Chen, W. Z., Sun, F. R., Chen, L. G., The optimal relation of a carnot heat pump
for the case Q /
T n, in: Proceedings of the International Energy Conference on
Energy Conversion and Energy Source Engineering, Wuhan, China, (1990), pp.
18{20.
[67] Yan, Z., Chen, J., A class of irreversible Carnot refrigeration cycles with a general
heat tranfer law, J. Phys. D: Appl. Phys., 23 (1990) 136{141.
[68] Chen, W. Z., Sun, F. R., Cheng, S. M., Chen, L. G., Study on optimal performance
and working temperatures of endoreversible forward and reverse carnot cycles,
International Journal of Energy Research, 19 (1995) 751.

36
[69] Nulton, J. D., Salamon, P., Pathria, R. K., Carnot-like processes in nite time. I.
theoretical limits, Am. J. Phys., 61 (1993) 911{916.
[70] Gordon, J. M., Generalized power versus eciency characteristics of heat engines:
The thermoelectric generator as an instructive illustration, Am. J. Phys., 59 (1991)
551{555.
[71] Bejan, A., Paynter, H. M., Solved Problems in Thermodynamics, Massachusetts
Institute of Technology Press, Dept. Mech. Eng., Cambridge, MA, (1976).
[72] Chen, J., The maximum power output and maximum eciency of an irreversible
Carnot heat engine, J. Phys. D: Appl. Phys., 27 (1994) 1144{1149.
[73] Pathria, R. K., Nulton, J. D., Salamon, P., Carnot-like processes in nite time. II.
applications to model cycles, Am. J. Phys., 61 (1993) 916{924.
[74] Moukalled, F., Nuwayhid, R. Y., Noueihed, N., The eciency of endoreversible
heat engines with heat leak, Int. J. Energy Research, 19 (1995) 377{389.
[75] Ibrahim, O. M., Klein, S. A., Mitchel, J. W., Optimum heat power cycles for
specied boundary conditions, J. Engng. Gas Turbines Power, 113 (1991) 514{
521.
[76] Wu, C., Kiang, R. L., Finite-time thermodynamic analysis of a Carnot engine with
internal irreversibility, Energy, 17 (1992) 1173{1178.
[77] Grazzini, G., Work from irreversible heat engines, Energy, 16 (1991) 747{755.
[78] Ait-Ali, M. A., Maximum power and thermal eciency of an irreversible power
cycle, J. Appl. Phys., 78 (1995) 4313{4318.
[79] Wu, C., Performance of solar-pond thermoelectric power generators, Int. J. Am-
bient Energy, 16 (1995) 59.
[80] Wu, C., Analysis of waste-heat thermoelectric power generators, Appl. Thermal
Engng, 16 (1996) 63{69.
[81] O zkaynak, S., Goktun, S., Yavuz, H., Finite-time thermodynamic analysis of a
radiative heat engine with internal irreversibility, J. Phys. D: Appl. Phys., 27
(1994) 1139{1143.
[82] Sahin, B., Kodal, A., Steady-state thermodynamic analysis of a combined Carnot
cycle with internal irreversibility, Energy, 20 (1995) 1285.
[83] Johnson, D., The exergy of the ocean thermal resource and analysis of second-law
eciencies of idealized ocean thermal energy conversion power cycles, Energy, 8
(1983) 927{949.
[84] Lee, W.-Y., Kim, S.-S., The maximum power from a nite reservoir for a lorentz
cycle, Energy, 17 (1992) 275{281.
[85] Le, H. S., Thermal eciency at maximum work output: New results for old heat
engines, Am. J. Phys., 55 (1987) 602{610.
[86] Le, H. S., Available work from a nite heat source and sink - how eective is a
maxwell's demon, Am. J. Phys., 55 (1987) 701{705.
[87] Wu, C., Power optimization of a nite-time closed gas-turbine power plant, Int. J.
Energy, 2 (1992) 57.

37
 [88] Sokolov, M., Hershgal, D., Optimal coupling and feasibility of a solar-powered
year-round ejector air conditioner, Solar Energy, 50 (1993) 507.
[89] Aydin, M., Yavuz, H., Application of nite-time thermodynamics to mhd power
cycles, Energy, 18 (1993) 907{911.
[90] Blank, D. A., Wu, C., The eect of combustion on a power-optimized endore-
versible dual cycle, Int. J. Power Energy Syst., 14 (1994) 98.
[91] Angulo-Brown, F., Rocha-Martinez, J. A., Navarrete-Gonzalez, T. D., A non-
endoreversible otto cycle model: Improving power output and eciency, J. Phys.
D: Appl. Phys., 29 (1996) 80{83.
[92] Landsberg, P. T., Le, H. S., Thermodynamic cycles with nearly universal
maximum-work eciencies, J. Phys. A: Math. Gen., 22 (1989) 4019{4025.
[93] Blank, D. A., Davis, G. W., Wu, C., Power optimization of an endoreversible
stirling cycle with regeneration, Energy, 19 (1994) 125{133.
[94] Blank, D. A., Wu, C., Power limit of an endoreversible ericsson cycle with regen-
eration, Energy Convers. Mgmt, 37 (1995) 59{66.
[95] Wu, C., Analysis of an endoreversible rallis cycle, Energy Convers. Mgmt, 35
(1994) 79{85.
[96] Wu, C., Chen, L., Sun, F., Performance of a regenerative brayton heat engine,
Energy, 21 (1996) 71{76.
[97] Lee, W. Y., Kim, S. S., Finite-time optimizations of a heat engine, Energy, 15
(1990) 979{985.
[98] Wu, C., Kiang, R. L., Power performance of a nonisentropic brayton cycle, J.
Engng. Gas Turbines Power, 113 (1991) 501{504.
[99] Wu, C., Power optimization of an endoreversible brayton gas heat engine, Energy
Convers. Mgmt, 31 (1991) 561{565.
[100] Agrawal, D. C., Finite-time Carnot refrigerators with wall gain and product loads,
J. Appl. Phys., 74 (1993) 2153{2158.
[101] Goth, Y., Feidt, M., Thermodynamics - optimum coecient of performance for
endoreversible heat pump or refrigerating machine, C. R. Acad. Sc. Paris, 303
(1986) 19, (in french).
[102] Agrawal, D. C., Menon, V. J., Performance of a Carnot refrigerator at maximum
cooling power, J. Phys. A: Math. Gen., 23 (1990) 5319{5326.
[103] Wu, C., Maximum obtainable specic cooling load of a refrigerator, Energy Con-
vers. Mgmt, 36 (1995) 7{10.
[104] Wu, C., Specic heating load of an endoreversible Carnot heat pump, Int. J.
Ambient Energy, 14 (1993) 25.
[105] Bejan, A., Theory of heat transfer-irreversible refrigeration plants, Int. J. Heat
Mass Transfer, 32 (1989) 1631{1639.
[106] Chen, J., Wu, C., Design considerations of primary performance parameters for
irreversible refrigeration cycles, Int. J. Ambient Energy, 16 (1995) 17.

38
[107] Chen, L., Sun, F., Wu, C., The in uence of heat-transfer law on the endoreversible
carnot refrigerator, J. Inst. Energy, 69 (1996) 96{100.
[108] Yan, Z., Chen, J., Optimal performance of an irreversible Carnot heat pump, Int.
J. Energy, 2 (1992) 63.
[109] Grazzini, G., Irreversible refrigerators with isothermal heat exchanges, Int. J. Re-
frig., 16 (1993) 101{106.
[110] Chen, J. C., New performance bounds of a class of irreversible refrigerators, J.
Phys. A: Math. Gen., 27 (1994) 6395{6401.
[111] Gordon, J. M., Ng, K. C., Thermodynamic modeling of reciprocating chillers, J.
Appl. Phys., 75 (1994) 2769{2774.
[112] Gordon, J. M., Ng, K. C., Predictive and diagnostic aspects of a universal ther-
modynamic model for chillers, Int. J. Heat Mass Transfer, 38 (1995) 807{818.
[113] Gordon, J. M., Ng, K. C., Chua, H. T., Centrifugal chillers: Thermodynamic
modeling and a diagnostic case study, Int. J. Refrig., 18 (1995) 253{257.
[114] Chen, L. G., Wu, C., Sun, F. R., Heat-transfer eect on the specic cooling load
of refrigerators, Appl. Thermal Engng, 16 (1996) 989{997.
[115] Chua, H. T., Ng, K. C., Gordon, J. M., Experimental-study of the fundamental
properties of reciprocating chillers and their relation to thermodynamic modeling
and chiller design, Int. J. Heat Mass Transfer, 39 (1996) 2195{2204.
[116] Ng, K. C., Chua, H. T., Ong, W., Lee, S. S., Gordon, J. M., Diagnostics and opti-
mization of reprociating chillers: Theory and experiment, Appl. Thermal Engng,
17 (1997) 263{276.
[117] Gordon, J. M., Ng, K. C., A general thermodynamic model for absorption chiller:
Theory and experiment, Heat Recovery Syst. & CHP, 15 (1995) 73{83.
[118] Chen, J., Wu, C., Maximum specic power output of a two-stage endoreversible
combined cycle, Energy, 20 (1995) 305{309.
[119] Wu, C., Power performance of a cascade endoreversible cycle, Energy Convers.
Mgmt, 30 (1990) 261{266.
[120] Chen, J. C., Wu, C., General performance-characteristics of an n-stage endore-
versible combined power system at maximum specic heat output, Energy Con-
vers. Mgmt, 37 (1996) 1401{1406.
[121] De Mey, G., De Vos, A., On the optimum eciency of endoreversible thermody-
namics processes, J. Phys. D: Appl. Phys., 27 (1994) 736{739.
[122] Chen, L., Sun, F., Chen, W., Optimization of the specic rate of refrigeration in
combined refrigeration cycles, Energy, 20 (1995) 1049{1053.
[123] Chen, J., Wu, C., Optimization of a two-stage combined refrigeration system,
Energy Convers. Mgmt, 37 (1996) 353{358.
[124] Yan, Z., Chen, J., An optimal endoreversible three-heat-source refrigerator, J.
Appl. Phys., 65 (1988) 1{4.
[125] Wu, C., Cooling capacity optimization of a waste heat absorption refrigeration
cycle, Heat Recovery Syst. & CHP, 13 (1993) 161{166.

39
[126] Wu, C., Maximum cooling load of a heat-engine-driven refrigerator, Energy Con-
vers. Mgmt, 34 (1993) 691{696.
[127] Chen, J., The equivalent cycle system of an endoreversible absorption refrigerator
and its general performance characteristics, Energy, 20 (1995) 995{1003.
[128] Chen, J., Andresen, B., Optimal analysis of primary performance parameters for
an endoreversible absorption heat pump, Heat Recovery Syst. & CHP, 15 (1995)
723{731.
[129] Chen, J. C., Wu, C., The R- characteristics of a 3-heat-source refrigeraton cycle,
Appl. Thermal Engng, 16 (1996) 901{905.
[130] Chen, J., Optimal heat-transfer areas for endoreversible heat pumps, Energy, 19
(1994) 1031{1036.
[131] Wu, C., Cooling capacity optimization of a geothermal absorption refrigeration
cycle, Int. J. Ambient Energy, 13 (1992) 133.
[132] Wu, C., Performance of a heat-driven endoreversible cooler, Energy Convers.
Mgmt, 36 (1995) 1053{1057.
[133] Yan, Z., Chen, J., The maximumoverall coecient of performance of a solar-driven
heat pump system, J. Appl. Phys., 76 (1994) 8129{8134.
[134] Chen, J. C., Thermodynamic analysis of a solar-driven thermoelectric generator,
J. Appl. Phys., 79 (1996) 2717{2721.
[135] Chen, J., Andresen, B., The maxiumum coecient of performance of thermoelec-
tric heat pumps, Int. J. Power Energy Syst., 17 (1996) 22.
[136] Chen, J., Andresen, B., New bounds on the performance parameters of a thermo-
electric generator, Int. J. Power Energy Syst., 16 (1996) 23{27.
[137] Muser, H., Behandlung von Elektronenprozessen in Halbleiterrandschichten, Z.
Phys., 148 (1957) 380{390, in German.
[138] ns, M. C., Bases sicas del aprovechamiento de la energia solar, Revist. Geofis.,
35 (1976) 227{239.
[139] Bejan, A., Kearney, D. W., Kreith, F., Second law analysis and synthesis of solar
collector systems, J. Solar Energy Engng., 103 (1981) 23.
[140] Badescu, V., On the theoretical maximum eciency of solar-radiation utilization,
Energy, 14 (1989) 571.
[141] Badescu, V., The theoretical maximum eciency of solar converters with and
without concentration, Energy, 14 (1989) 237.
[142] Lund, K., Application of nite-time thermodynamics to solar power conversion,
in: Finite-Time Thermodynamics and Thermoeconomics, Advances in Thermody-
namics 4, eds. Sieniutycz, S., Salamon, P., Taylor and Francis, New York, (1990),
p. 121.
[143] Badescu, V., Note concerning the maximal eciency and the optimal operating
temperature of solar converters with or without concentration, Renewable Energy,
1 (1991) 131.
[144] Badescu, V., Maximum conversion eciency for the utilization of diuse solar
radiation, Energy, 16 (1991) 783.

40
[145] Badescu, V., A study concerning the opportunity of solar power generation in
south-east rumania (dobrogea), Energy Convers. Mgmt, 33 (1992) 971.
[146] Badescu, V., Thermodynamic conversion of the radiation ux emitted by a spher-
ical source of black-body radiation, Int. J. Energy Research, 16 (1992) 717.
[147] Goktun, S., O zakynak, S., Yavuz, H., Design parameters of a radiative heat engine,
Energy, 18 (1993) 651{655.
[148] Blank, D. A., Wu, C., Power optimization of an extraterrestrial, solar-radiant
stirling heat engine, Energy, 20 (1995) 523{530.
[149] Howell, J. R., Bannerot, R. B., Optimum solar collector operation for maximizing
cycle work output, Solar Energy, 19 (1977) 149.
[150] Badescu, V., Optimum operation of a solar converter in combination with a stirling
or ericsson heat engine, Energy, 17 (1992) 601.
[151] Pollarolo, G., Sertorio, L., Energy and entropy balance for black piecewise homo-
geneous planet, Il Nuovo Cimento, 2C (1979) 335{351.
[152] Sertorio, L., Thermodynamic equations for a black planet with nearest-neighbor
surface Carnot interaction, Il Nuovo Cimento, 3C (1980) 37{44.
[153] Bejan, A., Heat Transfer, Wiley, New York, (1983).
[154] d'Isep, F., Sertorio, L., On the estimate of the mechanical power that can be
processed by earth, Il Nuovo Cimento, 6C (1983) 97{132.
[155] Gordon, J. M., Zarmi, Y., Wind energy a solar-driven heat engine: A thermody-
namic aproach, Am. J. Phys., 57 (1989) 995{998.
[156] De Vos, A., Flatter, G., The maximum eciency of the conversion of solar-energy
into wind energy, Am. J. Phys., 59 (1991) 751{754.
[157] Sertorio, L., Thermodynamics of complex systems, Singapore: World Scienitc,
(1991) 109{124.
[158] De Vos, A., van der Wel, P., Endoreversible models for the conversion of solar
energy into wind energy, J. Non-Equilib. Thermodyn., 17 (1992) 77{89.
[159] De Vos, A., van der Wel, P., The eciency of the conversion of solar-energy into
wind energy by means of hadley cells, Theor. Appl. Climatol., 46 (1993) 193{202.
[160] Nuwayhid, R. Y., Moukalled, F., The eect of planet thermal conductance on
conversion of solar energy into wind energy, Renewable Energy, 4 (1994) 52.
[161] Arpe, K., Brannkovic, C., Oriol, E., Speth, P., Variability in time and space of
energetics from a long series of atmospheric data produced by ecmwf, Beitrage zur
Physik der Atmosphare, 59 (1986) 321{355.
[162] Peixoto, J., Oort, A., Physics of Climate, American Institute of Physics, New
York, (1992).
[163] Baruch, P., Parrott, J. E., A thermodynamic cycle for photovoltaic energy conver-
sion, J. Phys. D: Appl. Phys., 23 (1990) 739.
[164] De Vos, A., Is a solar cell an endoreversible engine?, Solar Cells, 31 (1991) 181{196.

41
[165] Parrott, J. E., Thermodynamics of solar-cell eciency, Solar Energy Materials and
Solar Cells, 25 (1992) 73{85.
[166] De Vos, A., Landsberg, P. T., Baruch, P., Parrott, J. E., Entropy uxes, endore-
versibility, and solar energy conversion, J. Appl. Phys., 74 (1993) 3631{3637.
[167] De Vos, A., Endoreversible thermodynamics and chemical reactions, J. Phys.
Chem., 95 (1991) 4534{4540.
[168] Gordon, J. M., Orlov, V. N., Performance characteristics of endoreversible chemi-
cal engines, J. Appl. Phys., 74 (1993) 5303{5309.
[169] Mironova, V. A., Tsirlin, A. M., Kazakov, V. A., Berry, S. R., Finite-time thermo-
dynamics: Exergy and optimization time-constrained processes, J. Appl. Phys.,
76 (1994) 629{636.
[170] Sieniutycz, S., Shiner, J. S., Thermodynamics of irreversible processes and its rela-
tion to chemical engineering: Second law analyses and nite time thermodynamics,
J. Non-Equilib. Thermodyn., 19 (1994) 303{348.

42