Energy Conversion & Management 42 (2001) 173±181

www.elsevier.com/locate/enconman

On the Curzon±Ahlborn eciency and its connection with

the eciencies of real heat engines
Jincan Chen a,*, Zijun Yan a, Guoxing Lin a, Bjarne Andresen b
a
Department of Physics, Xiamen University, Xiamen 361005, People's Republic of China
b
érsted Laboratory, University of Copenhagen, Universitesparken 5, DK-2100, Copenhagen é, Denmark
Received 21 September 1999; accepted 13 April 2000

Abstract

It is acknowledged that the Curzon±Ahlborn eciency ZCA determines the eciency at maximum
power production of heat engines only a€ected by the irreversibility of ®nite rate heat transfer
(endoreversible engines), but ZCA is not the upper bound of the eciencies of heat engines. This is
conceptually di€erent from the role of the Carnot eciency ZC which is indeed the upper bound of the
eciencies of all heat engines. Some authors have erroneously criticized ZCA as if it were the upper
bound of the eciencies of endoreversible heat engines. Although the eciencies of real heat engines
cannot attain the Carnot eciency, it is possible, and often desirable, for their eciencies to be larger
than their respective maximum power eciencies. In fact, the maximum power eciency is the allowable
lower bound of the eciency for a given class of heat engines. These important conclusions may be
expounded clearly by the theory of ®nite time thermodynamics. 7 2000 Published by Elsevier Science
Ltd.

Keywords: Finite time thermodynamics; Irreversible cycle; Maximum power eciency; Maximum eciency;
Fundamental optimum relation

1. Introduction

In 1975, Curzon and Ahlborn [1] considered the in¯uence of ®nite rate heat transfer between
the external heat reservoirs and the working ¯uid on the performance of a Carnot heat engine

* Corresponding author. Fax: +86-592-218-1673.

E-mail address: [email protected] (J. Chen).

0196-8904/01/$ - see front matter 7 2000 Published by Elsevier Science Ltd.

PII: S 0 1 9 6 - 8 9 0 4 ( 0 0 ) 0 0 0 5 5 - 8
174 J. Chen et al. / Energy Conversion & Management 42 (2001) 173±181

and obtained the eciency of a Carnot engine at maximum power output as

p
Zm ˆ 1 ÿ TL =TH  ZCA , …1†
where TH and TL are the temperatures of the hot and cold heat reservoirs, respectively.
Although this eciency was derived independently by Curzon and Ahlborn in 1975, similar
systems were considered as early as 1957 by Chambadal and Novikov [2,3], so the eciency
might be called, chronologically, the Chambadal±Novikov±Curzon±Ahlborn eciency [4]. For
the sake of brevity and since it is a well-established name, the eciency will be simply referred
to as the CA eciency, i.e. ZCA :
Curzon and Ahlborn [1] emphasized that ``Eq. (1) has the interesting property that it serves
as quite an accurate guide to the best observed performance of real heat engines.'' Because of
this statement, some authors have mistaken the CA eciency as the upper bound on the
eciencies of heat engines. So far, the consequences of the CA eciency have not been
understood very well, and some diverging points of view on the CA eciency have appeared in
journals and textbooks. Thus, it is essential to discuss the meaning of the CA eciency further.
The correct understanding of this issue will be helpful in proper application of ®nite time
thermodynamics to real systems.

2. Eciency at maximum power output of heat engines with additional irreversibilities or other
constraint conditions

The important signi®cance of Eq. (1) is to provide the eciency at maximum power output
of a Carnot heat engine only a€ected by the irreversibility of ®nite rate heat transfer. When
other irreversibilities or constraint conditions are also considered, in general, the eciency of a
heat engine at maximum power output will be smaller than ZCA : Several representative
examples are listed below.
When the in¯uence of ®nite rate heat transfer as well as ®nite compression ratio are taken
into account, the eciency of a Carnot heat engine at maximum power output is reduced to [5]
r ÿ p
2
TL 1 ÿ TL =TH 1=…g ÿ 1†
Zm ˆ 1 ÿ ÿ < ZCA , …2†
TH 2 ln rv
where g is the speci®c heat ratio, and rv is the compression ratio.
When the in¯uence of ®nite rate heat transfer and irreversibilities inside the working ¯uid are
taken into account, the eciency of a Carnot heat engine at maximum power output is given
by [6±8]
r
ITL
Zm ˆ 1 ÿ < ZCA , …3†
TH
where I > 1 describes the internal irreversibility occurring in the working ¯uid.
When the in¯uence of ®nite rate heat transfer and heat leak loss are taken into account, the
eciency of a Carnot heat engine at maximum power output is [9,10]
 J. Chen et al. / Energy Conversion & Management 42 (2001) 173±181 175
ÿ p
2
1 ÿ TL =TH
Zm ˆ p < ZCA , …4†
1 ‡ …Ci =K†ZC ÿ TL =TH
where Ci is the coecient of the heat leak loss, K is the equivalent thermal conductance of the
heat engine and ZC is the eciency of a reversible Carnot heat engine.
When the in¯uence of ®nite rate heat transfer and a ®nite heat source are taken into account,
the eciency of a heat engine at maximum power output is [11]
v
  
u Q1
u
u C1 ln 1 ÿ
p t …C1 TH †

Zm ˆ 1 ÿ TL =T H ˆ 1 ÿ TL < ZCA , …5†
Q1

where C1 is the heat capacity of the ®nite heat source, and Q1 is the heat absorbed from the
®nite heat source by the working ¯uid per cycle.
When the in¯uence of heat transfer and ¯uid ¯ow irreversibilities are taken into account, the
eciency of a heat engine at maximum power output is [12]
 p 
n
Zm ˆ 1 ÿ TL =TH < ZCA , …6†
n‡1
where n covers the range from laminar ¯ow …n ˆ 1† to turbulent ¯ow in the fully rough regime
…n ˆ 2).
The results quoted above show that when the in¯uence of other irreversibilities or constraint
conditions in addition to ®nite rate heat transfer is considered, the eciency of a heat engine at
maximum power output is smaller than the CA eciency. However, this does not imply that
the CA eciency is the upper bound on the eciencies of heat engines. For example, when the
Curzon±Ahlborn heat engine is operated under ``ecological'' optimization conditions, the
eciency is [13,14]
p
TL 1 ‡ TH =TL
ZE ˆ 1 ÿ > ZCA , …7†
TH 2
where TL is equal to the environment temperature. This behavior may be clearly seen from the
power output versus eciency curves of heat engines.

3. Power output versus eciency of heat engines: which is the optimal operation?

In order to understand the meaning of the CA eciency, it is necessary to derive the

fundamental optimum relation between the power output and the eciency of a Carnot heat
engine. When we only consider the in¯uence of ®nite rate heat transfer on the performance of
a Carnot heat engine, the fundamental optimum relation is given by [15±20]
 
1 ÿ ZC
pˆZ 1ÿ , …8†
1ÿZ
176 J. Chen et al. / Energy Conversion & Management 42 (2001) 173±181

where p is the dimensionless power output. Using Eq. (8), we can generate the power output
versus eciency curve [6,8,9,21,22] shown in Fig. 1. It is clearly seen from Fig. 1 that when the
power output of a Carnot heat engine attains the maximum, the eciency is equal to the CA
eciency. When p < pmax , there are two di€erent eciencies for a given power output, where
one is smaller than the CA eciency and the other is larger than the CA eciency. When the
eciency lies in the region of Z < ZCA , the eciency decreases as the power output decreases.
Obviously, the region of Z < ZCA is not desirable. When the eciency lies in the region of
Z > ZCA , the eciency increases as the power output decreases, and vice versa. Thus, the
optimal operation of an endoreversible Carnot heat engine should be the region of ZC >
ZrZCA [15,19,22]. The ``ecological'' optimization condition mentioned above is just in this
region. Fig. 1 also shows clearly the fact that when the eciency of a heat engine is larger than

Fig. 1. The dimensionless power output versus eciency curve of an endoreversile Carnot heat engine a€ected by
®nite rate heat transfer.
 J. Chen et al. / Energy Conversion & Management 42 (2001) 173±181 177

ZCA , its power output is not the maximum; and the larger the eciency is, the smaller the
power output is. This is one important conclusion of ®nite time thermodynamics.
In a recent article [23], Gyftopoulos pointed out that ``the good correlation between the
thermal eciencies of the old power plants and Eq. (1) must be a numerical accident for at
least two reasons. First, the plants that are considered in the comparisons were not designed to
achieve the optimum thermal eciency of ®nite time thermodynamics and, therefore, it would
be a miracle if each self-optimized itself to the point of best performance. Second, and perhaps
more importantly, last year General Electric announced the great achievement of a gas-turbine,
combined-cycle power plant with a thermal eciency of 60%.'' Gyftopoulos' statement, which
has been proliferated in another article [24], just exempli®es that the eciencies of real heat
engines designed optimally may be de®nitely larger than the CA eciency. This is obviously
correct because it is an observed experimental fact. However, it is worthwhile to note that this
experimental ``discovery'' is in full agreement with the important conclusion of ®nite time
thermodynamics mentioned above because the CA eciency is not the upper bound on the
eciency.
Although Gyftopoulos has noted this experimental observation, it is a pity that he has not
interpreted it correctly. He stated that ``if this eciency (i.e., 60%) correlated with the results
of ®nite time thermodynamics, it would imply a ratio TL =TH ˆ 0:16 and, therefore, an
available energy or exergy of the products of combustion of the gas of about 84% of the
exergy of natural gas! As a result, we would be compelled to conclude that the loss of exergy
of the fuel upon combustion is only 16% though every calculation and every measurement
yield a loss of about 30%.'' In this statement [23,24], Gyftopoulos ascribes the lack of
agreement between the measurements and the CA eciency to basic shortcomings of ®nite
time thermodynamics. This sweeping conclusion is not well founded, because an irrational
assumption was hidden in the statement.
A closer look at Gyftopoulos' claim reveals that he has assumed implicitly that the power
plant with a thermal eciency of 60% is operated at the state of maximum power output.
Following Gyftopoulos' point of view, if other irreversibilities besides ®nite rate heat transfer
were considered in the model, the di€erence between the theoretical result and the
measurement would become larger because the maximum power eciency of a heat engine
a€ected by ®nite rate heat transfer and other additional irreversibilities is, in general, smaller
than ZCA (Section 2).
When interpreting such discrepancies between experimental ®ndings and model calculations,
one must always (i) check that the real operating conditions and the modeling objectives are
consistent and (ii) consider the aim of the model (e.g. high accuracy, transparency, or a
bound). Neither was done here. The CA objective was maximum power production, whereas
real power plants have other objectives, usually economic ones [25]. Further, ®nite time
thermodynamic models are not explicitly designed for maximum agreement with experimental
detail Ð that is the realm of engineering design Ð but rather for physical insight and easy
interpretation. Thus, it is evident that Gyftopoulos' claim is incorrect. In general, real heat
engines, particularly heat engines with high eciencies, are not operated at the state of
maximum power output. One must not take the CA eciency as the upper bound on the
eciency, much less denigrate ®nite time thermodynamics, in general, based on this
misinterpretation.
178 J. Chen et al. / Energy Conversion & Management 42 (2001) 173±181

As mentioned above, the result of ®nite time thermodynamics shows that the optimal
eciency of endoreversible Carnot heat engines should be ZrZCA : It is, thus, clear that it is
allowable for an optimally designed power plant not to be operated in the state of maximum
power output. If the power plant is operated at a certain state of Z > ZCA , say ZE , it is
irrational to use the maximum power criterion to analyze the performance of the power plant
and to derive other conclusions. If one wants to understand the performance of a heat engine,
it is necessary to start from the fundamental optimum relation of the heat engine.
Consequently, ®nding the fundamental optimum relation of a heat engine is the kernel of ®nite
time thermodynamic cycle theory and is more important than ®nding e.g. the maximum power
eciency of the heat engine, since the latter can be derived directly from the former. In
addition, starting from the fundamental optimum relation, one can easily determine the
optimum operating region and optimize other performance parameters of a heat engine besides
the maximum power eciency.
In another journal [26], Sekulic stated that ``a ®nite time thermodynamics ®gure of merit
(i.e., the maximum power eciency ZFTT † should always be smaller than the Carnot eciency.
Consequently, the margin for improvement of a real heat engine, calculated on the basis of Eq.
(1), should always be smaller than the corresponding margin implied by the Carnot eciency.
In other words, the following sequence of thermodynamic eciency margin de®nitions and
inequalities should hold: DZCarnot ˆ ZCarnot ÿ ZActual > DZFTT ˆ ZFTT ÿ ZActual > 0: Thus, the margin for
improvement of a real system based on a ®nite time thermodynamics ®gure of merit, DZFTT ,
intends to provide a more realistic assessment of a maximum achievable improvement in
comparison to the margin DZCarnot calculated using Carnot eciency. Also note that the
following inequality must be satis®ed in all situations DZFTT 6ˆ0:'' In this last statement, Sekulic,
like Gyftopoulos above, has mistaken the CA eciency to be the maximum eciency of
irreversible heat engines. The inequality DZFTT ˆ ZFTT ÿ ZActual > 0 violates the theory of ®nite
time thermodynamics and is not true.
Early in 1988, Bejan [27] pointed out that the CA eciency is not the ``maximum'' eciency
of a heat engine. In other words, it should not be confused with the Carnot eciency. In some
earlier literature [15], Yan pointed out that the CA eciency is not the maximum eciency of
a heat engine but determines the lower bound on the optimal eciency of a heat engine
a€ected by ®nite rate heat transfer. This correct point of view is also advanced in de Vos' book
[19]. It is, thus, obvious that although the eciencies of real heat engines cannot attain the
Carnot eciency, it is possible, and is often desirable, for the eciencies to exceed the
respective maximum power eciencies.

4. Optimum operating criteria

When di€erent irreversibilities are taken into account, the maximum power eciency Zm of a
heat engine will be di€erent from the CA eciency. However, both Zm and ZCA have similar
signi®cance. They both determine the lower bound for the optimal eciencies of their
respective class of heat engines. Thus,
ZrZm …9†
 J. Chen et al. / Energy Conversion & Management 42 (2001) 173±181 179

is an optimality criterion for heat engines with the speci®ed irreversibilities. If the eciency of
a heat engine, adequately described by the choice of irreversibilities, does not satisfy Eq. (9),
the heat engine is not designed or operated optimally.
When just ®nite rate heat transfer and heat leak loss are considered, the fundamental
optimum relation of a heat engine may be used to generate the power output versus eciency
curve [9,10,21,22] of the heat engine shown in Fig. 2. In this case, the power output versus
eciency curve is of a loop shape, which is very di€erent from that of a Carnot heat engine
a€ected only by ®nite rate heat transfer. There exist a maximum power output and a maximum
eciency with non-zero power output. The optimal operating region of this heat engine is
situated on that part of the power output versus eciency curve which has a negative slope

Fig. 2. The dimensionless power output versus eciency curve of a Carnot heat engine a€ected by ®nite rate heat
transfer and heat leak loss. pmax is the dimensionless maximum power output and pm is the dimensionless power
output at maximum eciency Zmax :
180 J. Chen et al. / Energy Conversion & Management 42 (2001) 173±181

[9,10,22]. Thus, besides Eq. (9), the dimensionless power output of a heat engine must be
constrained by
prpm , …10†
where pm is the dimensionless power output at maximum eciency.
It is important to note all the information emerging from Fig. 2. When the cycle time is
equal to two particular ®nite values, the power output and the eciency will attain their
maximum values, respectively. When the cycle time tends to zero or in®nity, both the power
output and the eciency of the cycle approach zero. This shows that when certain
irreversibilities are considered, an in®nite time process may be irreversible. Gyftopoulos et al.
provided an experimental example of this [23,28]. Having ``two identical electricity storage
batteries with an internal discharge time constant of 100 days, we can discharge one battery
very slowly, say over 104 days, and the other very fast, say over 1 day, and ask `Which
discharge process is closer to reversible?' As is very well known from billions of experiments,
the fast process is very close to reversible, whereas the slow process is totally irreversible,
because in the fast process practically all the stored adiabatic availability is transferred out,
whereas in the slow process practically no adiabatic availability is transferred; it is all
dissipated in the battery.'' Such an experiment is not contrary to the theory of ®nite time
thermodynamics. According to ®nite time thermodynamic analysis [9], when the thermal
resistance (equivalent to the internal resistance of a battery) is dominant for a power plant, a
slow process is closer to reversible; but when the heat leak loss (corresponding to the internal
discharge of the battery) is dominant, a fast process is closer to reversible. Thus, with internal
irreversibilities, a fast discharge process may be closer to reversible than a slow one. Quite a
similar problem, the maximum work from an electric battery model with an internal discharge,
has been analyzed theoretically [29]. Thermal systems with internal structure were also
analyzed a long time ago [30].

5. Conclusions

Finite time thermodynamics is a science. It is an extension of classical thermodynamics in

which the unavoidable minimum irreversible losses are considered for the purpose of gaining
insight. Such a new branch of learning is rich in content, not just the maximum power
eciency of a simple Curzon±Ahlborn heat engine or endoreversible cycles. Over the last two
decades, ®nite time thermodynamics has been successfully used in many regions in physics and
other subjects. It is necessary to further develop and generalize the theory of ®nite time
thermodynamics.
In order to understand the performance of an endoreversible or irreversible heat engine, one
must ®nd the fundamental optimum relation of the heat engine. Only in this way can one
realize the genuine meaning of the maximum power eciency. An irreversible process may take
part in a ®nite or in®nite interval in time. The performance of such an irreversible process may
be expounded very well by using the theory of ®nite time thermodynamics.
 J. Chen et al. / Energy Conversion & Management 42 (2001) 173±181 181

References

[1] Curzon FL, Ahlborn B. Am J Phys 1975;43:22.

[2] Chambadal P. Les Centrales Nucleaires. Paris: Armand Colin, 1957.
[3] Novikov II. Atomnaya Energiya 1957;3:409.
[4] Bejan A. J Appl Phys 1996;79:1191.
[5] Gutkowwicz-Krusin D, Procaccia I, Ross J. J Chem Phys 1978;69:3898.
[6] Ibrahim OM, Klein SA, Mitchell JW. J Engng Gas Turbines Power 1991;113:514.
[7] Wu C, Kiang RL. Energy 1992;17:1173.
[8] Ibrahim OM, Klein SA, Mitchell JW. J Sol Energy Engng 1992;114:267.
[9] Yan Z, Chen L. J Phys A: Math Gen 1995;28:6167.
[10] Chen J. J Phys A: Math Gen 1998;31:3383.
[11] Yan Z, Chen L. J Phys A: Math Gen 1997;30:8119.
[12] Ikegamil Y, Bejan A. J Sol Energy Engng 1998;120:139.
[13] Angulo-Brown F. J Appl Phys 1991;69:7465.
[14] Yan Z. J Appl Phys 1993;73:3583.
[15] Yan Z. J Engng Therm 1985;6:1.
[16] Chen J, Yan Z. J Xiamen Univ 1988;27:289.
[17] Chen J, Yan Z. Phys Rev A 1989;39:4140.
[18] Chen L, Yan Z. J Chem Phys 1989;90:3740.
[19] de Vos A. Endoreversible thermodynamics of solar energy conversion. Oxford: Oxford University Press, 1992.
[20] Bejan A. IntJ Heat Mass Transfer 1995;38:433.
[21] Gordon JM, Huleihil M. J Appl Phys 1992;72:829.
[22] Chen J. J Phys D: Appl Phys 1994;27:1144.
[23] Gyftopoulos EP. Energy Convers Mgmt 1997;38:1525.
[24] Moran MJ. In: (ECOS'98) Eciency, Cost, Optimization, Simulation and Environmental Aspects of Energy
Systems and Processes, Nancy, France, 8±10 July, vol. II. Berlin: Springer, 1998. p. 1147.
[25] Salamon P, Nitzan A. J Chem Phys 1981;74:3546.
[26] Sekulic DP. J Appl Phys 1998;83:4561.
[27] Bejan A. Advanced engineering thermodynamics. New York: Wiley, 1988.
[28] Gyftopoulos EP, Stoukides M, Mendez M. In: Granger RA, editor. Experiments in heat transfer and
thermodynamics. New York: Cambridge University Press, 1994.
[29] Bejan A, Dan N. Energy 1997;22:93.
[30] Andresen B, Rubin MH, Berry RS. J Phys Chem 1983;87:2704.