Energy Conversion & Management 41 (2000) 607±619

www.elsevier.com/locate/enconman

E€ects of internal irreversibility and heat leakage on the

®nite time thermoeconomic performance of refrigerators
and heat pumps
Ali Kodal a, Bahri Sahin b,*, Tamer Yilmaz b
a
Department of Aeronautical Engineering, Istanbul Technical University, Maslak, 80626, Istanbul, Turkey
b
Department of Naval Architecture, Yildiz Technical University, Besiktas, 80750, Istanbul, Turkey
Received 10 December 1998; accepted 14 June 1999

Abstract

By using ®nite time thermodynamic theory, a performance analysis based on the thermoeconomic
criterion has been performed for an irreversible refrigerator and a heat pump. The thermoeconomic
objective function is de®ned as the cooling load for the refrigerator and the heating load for the heat
pump per unit total cost. The optimal performances and design parameters which maximize the
objective functions are investigated. The design parameters under the optimal conditions have been
derived analytically, and then, the e€ects of internal irreversibility, heat leakage and the economical
parameter on the global and optimal performances have been discussed. # 1999 Elsevier Science Ltd.
All rights reserved.

Keywords: Finite time thermodynamics; Thermoeconomic optimization; Irreversible refrigerator; Irreversible heat
pump

1. Introduction

The concept of an equilibrium and reversible Carnot cycle has played an important role in
the development of classical thermodynamics. The coecient of performance of a reversible

* Corresponding author. Fax: +90-212-261-6659.

E-mail address: [email protected] (B. Sahin).

0196-8904/00/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 9 6 - 8 9 0 4 ( 9 9 ) 0 0 1 2 9 - 6
608 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619

Nomenclature

a investment cost parameter for heat exchangers

A heat transfer area
b1 investment cost parameter for compressor and its driver
b2 energy consumption cost parameter
b b ˆ b1 ‡ b2
C cost
F objective function
IDS internal irreversibility parameter
k k ˆ a=b
ncu national currency unit
Q_ rate of heat transfer
R heat leakage parameter
S entropy
T temperature
W_ power input
U overall heat transfer coecient
b coecient of performance

Subscripts
e energy consumption
H heat sink
hp heat pump
i investment
l heat leakage
L heat source
max maximum
ref refrigerator
X warm working ¯uid
Y cold working ¯uid

Superscripts
 optimum conditions

inverse Carnot cycle has been used as an upper bound for refrigerator and heat pump
performances. This upper bound of performance can only be achieved through the in®nitely
slow processes required for thermodynamic equilibrium. Therefore, it is not possible to obtain
a certain amount of cooling or heating load by using heat exchangers with ®nite heat transfer
areas. If a Carnot refrigerator and heat pump were used to generate a certain amount of load,
the necessary heat exchanger areas would be in®nite. Thus, the coecients of performance of
reversible cycles do not have great signi®cance and are poor guides to the performances of real
refrigerators and heat pumps. Chambadal [1] and Novikov [2] and, later, Curzon and Ahlborn
 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619 609

[3] extended the reversible Carnot cycle to the endoreversible Carnot cycle by taking the
irreversibility of ®nite time heat transfer into account. By introducing ®nite time processes, a
new ®eld `®nite time thermodynamics' was born, and thus, more realistic limits have been
developed for real heat engines, refrigerators and heat pumps. Like the reversible Carnot cycle
in classical thermodynamics, the endoreversible Carnot cycle has played a major role in the
development of ®nite time thermodynamics. Bejan [4] has written a recent review about
optimization studies based on ®nite time thermodynamics. Recent optimization studies
considering ®nite time and ®nite size constraints have been focused on the performance of
endoreversible and irreversible refrigerators and heat pumps [5±25]. In these studies, the
optimization criteria chosen are usually cooling load, coecient of performance, total heat
transfer area and speci®c cooling load (cooling load per unit total heat transfer area). The
above chosen criteria for the performance optimization of refrigerators and heat pumps are not
quite sucient. This is because, in these studies, one of the criteria is chosen as an objective
function while taking the rest as constraints. The coecient of performance a€ects the energy
consumption cost, and the total heat transfer area and compressor capacity a€ect the
investment cost. For a more realistic optimization, the objective function which is desired to be
optimized must include the coecient of performance, the total heat transfer area and the
compressor capacity. In this context, Sahin and Kodal [26] have very recently introduced a new
optimization criterion which consists of the total of the investment and energy consumption
costs. They applied this criterion to the endoreversible refrigerators and heat pumps to ®nd the
optimal design conditions.
In this paper, the optimization work performed by Sahin and Kodal [26] for the
endoreversible refrigerators and heat pumps model is extended to an irreversible model by
considering internal irreversibility and heat leakage e€ects. The e€ects of the irreversibilities
and economical parameters on the global and optimal performance are investigated. We
obtained some important conclusions that have more realistic signi®cance than those of
previous studies on this subject.

Fig. 1. Irreversible refrigerator model and its T±S diagram.

610 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619

2. Optimization for an irreversible refrigerator

The model of the irreversible refrigerator and its T±S diagram are shown in Fig. 1. The cycle
operates between a heat source of temperature TL (cooling space temperature) and a heat sink
of temperature TH (ambient temperature). The temperatures of the ¯uid exchanging heat with
the reservoirs at TH and TL are TX and TY, respectively. The rate of heat leakage Q_ l from the
hot reservoir at TH to the cold reservoir at TL with thermal conductance UlAl is given by

Q_ l ˆ Ul Al …TH ÿ TL †: …1†

When heat transfer obeys a linear law, the rate of heat ¯ow from the heat source to the
refrigerator is

Q_ LC ˆ UL AL …TL ÿ TY † …2†

where UL and AL are the overall heat transfer coecient and heat transfer area of the heat
exchanger between the heat source and the refrigerator, respectively. The rate of heat ¯ow
from the refrigerator to the heat sink is

Q_ HC ˆ UH AH …TX ÿ TH †, …3†

where UH and AH are the overall heat transfer coecient and heat transfer area of the heat
exchanger between the heat sink and the refrigerator, respectively. The net heat rates Q_ H and
Q_ L transferred to the hot reservoir and from the cold reservoir are

Q_ H ˆ Q_ HC ÿ Q_ l , …4†

Q_ L ˆ Q_ LC ÿ Q_ l : …5†

The ®rst law of thermodynamics requires the power input to the refrigerator to be
_ ˆ Q_ HC ÿ Q_ LC ˆ Q_ H ÿ Q_ L :
W …6†

The second law for an irreversible cycle requires that

‡
dQ_ Q_ Q_
ˆ LC ÿ HC < 0: …7†
T TY TX

In order to describe quantitatively the e€ect of the internal dissipations of the working ¯uid on
the performance, we introduce an irreversibility parameter,
S2 ÿ S3
IDS ˆ …8†
S1 ÿ S4

which characterizes the degree of internal irreversibility. The inequality in Eq. (7) can now be
written as
 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619 611

Q_ LC Q_ HC
IDS ÿ ˆ 0: …9†
TY TX
Eq. (9) shows clearly that when IDS ˆ 1, the cycle is endoreversible, and when IDS > 1 the cycle
is internally irreversible.
The coecient of performance of the refrigerator is de®ned as

Q_ L Q_ LC ÿ Q_ l
bref ˆ ˆ : …10†
W _ Q_ HC ÿ Q_ LC

By using Eqs. (7) and (10), we obtain

1 ÿ Rref
bref ˆ   …11†
TX
IDS ÿ1
TY

where the parameter Rref ˆ Q_ l =Q_ LC represents the heat leakage percentage and is assumed to
be a known constant.
In order to account for both investment and energy consumption costs, optimization of the
cooling load per unit total cost is considered. The function to be optimized is de®ned as

Q_ L
Fref ˆ …12†
…Ci ‡ Ce †
where Ci and Ce refer to annual investment and energy consumption costs, respectively. For
the investment cost, one can consider the investment costs of the main system components, the
heat exchangers and the compressor together with its prime mover. The investment cost of the
heat exchangers is assumed to be proportional to the total heat transfer area. On the other
hand, the investment cost due to compressor and its driver is assumed to be proportional to its
compression capacity or the required power input. Thus, the annual investment cost of the
system can be given as
ÿ

_ ˆ a…AH ‡ AL ‡ Al † ‡ b1 Q_ HC ÿ Q_ LC
Ci ˆ a…AH ‡ AL ‡ Al † ‡ b1 W …13†

where the proportionality coecient for the investment cost of the heat exchangers, a, is equal
to the capital recovery factor times investment cost per unit heat transfer area, and its
dimension is ncu/year m2, and the proportionality coecient for the investment cost of the
compressor and its driver, b1, is equal to the capital recovery factor times investment cost per
unit power and its dimension is ncu/year kW. The unit ncu stands for the national currency
unit. The initial investment cost is converted to equivalent yearly payments using the capital
recovery factor [27,28]. The annual energy consumption cost is proportional to the power
input, i.e.,
ÿ

Ce ˆ b2 W_ ˆ b2 Q_ HC ÿ Q_ LC …14†

where the coecient b2 is equal to the annual operation hours times price per unit energy, and
its dimension is ncu/year kW. Substitution of Eqs. (5), (13) and (14) into Eq. (12) yields
612 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619
ÿ

Q_ LC ÿ Q_ l
Fref ˆ ÿ
…15†
a…AH ‡ AL ‡ Al † ‡ b Q_ HC ÿ Q_ LC

where b ˆ b1 ‡ b2 . Using Eqs. (1)±(6) and (9) in Eq. (15), one can obtain
1 ÿ Rref
Fref ˆ    : …16†
1 IDS TX Rref IDS TX
a ‡ ‡ ‡b ÿ1
UL …TL ÿ TY † UH TY …TX ÿ TH † Ul …TH ÿ TL † TY

Eq. (16) can be maximized with respect to TX and TY, and the results are
r
 aTH
T X ˆ TH ‡ , …17†
bUH

TL
T Y ˆ r !ÿ1 : …18†
r
UH bUH TH
1‡ 1‡
IDS UL a

By substituting Eqs. (17) and (18) into Eq. (16), the maximum of the objective function can be
obtained,
1 ÿ Rref
…Fref †max ˆ " r  #2 …19†
r
aIDS UH bUH TH aRref
1‡ ‡ ‡ ÿb
UH TL IDS UL a Ul …TH ÿ TL †

and the optimum coecient of performance is

…1 ÿ Rref †TL
bref ˆ r  …20†
aTH 1 1
IDS TH ‡ IDS p ‡ p ÿ TL
b UH IDS UL

and the ratio of heat exchanger areas at this optimum condition is

  r
AH IDS UL
ˆ : …21†
AL UH

The speci®c cooling load which is de®ned as the cooling load per unit total heat exchanger
area A ˆ AH ‡ AL [5,8] at the optimum conditions, can also be found
 
 Q_ L TL …1 ÿ Rref †
_ ref ˆ
…q†
A ref
ˆ r ! : …22†
1 1 bTH 1 1
IDS p ‡ p ‡ p ‡ p
IDS UL UH a IDS UL UH
 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619 613

3. Optimization for an irreversible heat pump

The model given in Fig. 1 for an irreversible refrigerator holds schematically for an
irreversible heat pump also. However, in this case TH is the temperature of the heated space
and TL is the ambient temperature. Therefore, Eqs. (1)±(9) can be used for an irreversible heat
pump. The coecient of performance (bhp ) of the irreversible heat pump is de®ned as

Q_ H Q_ HC ÿ Q_ l
bhp ˆ ˆ : …23†
W _ Q_ HC ÿ Q_ LC

By using Eqs. (7) and (23), we obtain

1 ÿ Rhp
bhp ˆ …24†
1 ÿ TY =IDS TX

where the parameter Rhp ˆ Q_ l =Q_ HC represents the heat leakage percentage, and it is assumed to
be a known constant.
The objective function to be optimized for the heat pump is de®ned as the heating load per
unit total cost, i.e.,

Q_ H
Fhp ˆ  ÿ
 : …25†
a…AH ‡ AL ‡ Al † ‡ b Q_ HC ÿ Q_ LC

Using Eqs. (1)±(6) and (9) in Eq. (25), one can obtain
1 ÿ Rhp
Fhp ˆ    : …26†
TY 1 Rhp TY
a ‡ ‡ ‡b 1ÿ
IDS TX UL …TL ÿ TY † UH …TX ÿ TH † Ul …TH ÿ TL † IDS TX

Eq. (26) can be maximized with respect to TX and TY, the results are
TH
T X ˆ   ÿ1  , …27†
p p
1 ÿ IDS UL =UH bUL TL =a ÿ 1

p
T Y ˆ TL ÿ aTL =bUL : …28†

By substituting Eqs. (27) and (28) into Eq. (26), the maximum of the objective function can be
obtained
ÿ
1 ÿ Rhp
Fhp max ˆ "r …29†
r #2
a bUL TL IDS UL aRhp
bÿ ÿ1ÿ ‡
IDS UL TH a UH Ul …TH ÿ TL †

and the optimum coecient of performance is

614 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619
ÿ

1 ÿ Rhp IDS TH
bhp ˆ" r r ! # …30†
aTL IDS 1
IDS TH ‡ ‡ p ÿ TL
b UH UL

and the ratio of heat exchanger areas at this optimum condition is

  r
AH IDS UL
ˆ : …31†
AL UH
The speci®c heating load at the optimum conditions can also be found for a given total heat
exchanger area A ˆ AH ‡ AL ,
  ÿ

 Q_ H TH 1 ÿ Rhp
_ hp ˆ
…q† ˆ r r ! r !: …32†
A hp 1 bTL IDS 1 IDS 1
ÿ ÿ p ‡ p
IDS a UH UL UH UL

4. Results and discussion

The variations of the objective functions for the refrigerator and heat pump with respect to
the coecient of performance, for various internal irreversibility parameters are shown in Fig.
2(a) and (b), respectively. From these ®gures, one can observe the e€ect of the internal

Fig. 2. Variations of the objective function for the refrigerator (a) and for the heat pump (b) with respect to the
coecient of performance, for various IDS values, (TH ˆ 288 K, TL ˆ 243 K for the refrigerator and TH ˆ 350 K,
TL ˆ 273 K for the heat pump, Ul ˆ UH ˆ UL ˆ 0:5 kW/m2 K, k ˆ a=b ˆ 0:2, Rref ˆ Rhp ˆ 0:05, for the solid lines
Rref ˆ Rhp ˆ 0).
 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619 615

irreversibility parameter (IDS ) on the global and optimal performances. As the internal
irreversibility increases, the global and optimal performances speedily decrease. Eq. (11) for the
refrigerator and Eq. (24) for the heat pump show that when there is no ®nite rate heat transfer
irreversibility, i.e., TX ˆ TH and TY ˆ TL , b is equal to bmax . Thus, from Eqs. (11) and (24), we
obtain
ÿ
1 ÿ Rref ÿ
1 ÿ Rhp
bmax ref ˆ , bmax ˆ : …33†
TH hp TL
IDS ÿ1 1ÿ
TL IDS TH
On the other hand, when b ˆ bmax , the objective functions of the refrigerator and heat pump
become zero, as shown in Fig. 2. Therefore, the upper bounds of the performance coecients
given in Eq. (33) do not have very much instructive signi®cance for practical refrigerators and
heat pumps. It is also seen from Fig. 2 that very large di€erences exist between the optimal
performance coecient (b ) and the maximum performance coecient (bmax ). However, this
di€erence is reduced with increased internal irreversibility.
The e€ect of the heat leakage parameter (R ) on the global and optimal performances is
shown for the refrigerator in Fig. 3(a) and for the heat pump in Fig. 3(b). As can be seen from
these ®gures, the global and optimal performances gradually decrease with increasing heat
leakage, as expected. However, the reducing e€ect of the heat leakage on the performances at
b > b conditions is much greater than those at b < b conditions.
In Fig. 4(a) and (b), the e€ects of the internal irreversibility and heat leakage on the optimal
coecients of performance for refrigerators and heat pumps (bref and bhp ) can be seen more
clearly. By examining these ®gures, we observe that the reducing e€ect of the internal
irreversibility on bref is greater in comparison to that on bhp . However, the reducing e€ects of

Fig. 3. Variations of the objective function for the refrigerator (a) and for the heat pump (b) with respect to the
coecient of performance, for various R values, (TH ˆ 288 K, TL ˆ 243 K for the refrigerator and TH ˆ 350 K,
TL ˆ 273 K for the heat pump, Ul ˆ UH ˆ UL ˆ 0:5 kW/m2 K, IDS ˆ 1:2, k ˆ 0:2).
616 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619

Fig. 4. E€ects of the internal irreversibility (a) and heat leakage (b) on the optimal coecients of performance
(TH ˆ 288 K, TL ˆ 243 K for the refrigerator and TH ˆ 350 K, TL ˆ 273 K for the heat pump, UH ˆ UL ˆ 0:5 kW/
m2 K, k ˆ 0:2).

the heat leakage on bref and bhp is of the same order of magnitude. Thus, it is more important
to take precautions to reduce internal irreversibility in the design of a refrigerator with respect
to the design of a heat pump.
The objective functions for the refrigerator and heat pump can be plotted with respect to

Fig. 5. E€ect of the economical parameter (k ˆ a=b) on the global performance for the refrigerator (a) and for the
heat pump (b), (TH ˆ 288 K, TL ˆ 243 K for the refrigerator and TH ˆ 350 K, TL ˆ 273 K for the heat pump,
Ul ˆ UH ˆ UL ˆ 0:5 kW/m2 K, Rref ˆ Rhp ˆ 0:05, IDS ˆ 1:2).
 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619 617

coecient of performance, for various k ˆ a=b values, the economical parameter, as shown in
Fig. 5(a) and (b), respectively. It is seen from Fig. 5(a) and (b) that the global and optimal
performances are greatly a€ected by the economical parameter (k ). One can see that the
performances increase with decreasing k. However, the e€ect of k on the performances
decreases as b approaches its minimum and maximum values. The variations of the optimum
coecient of performance for the refrigerator given in Eq. (20) and for the heat pump given in
Eq. (30) with respect to k are shown in Fig. 6. As may be seen from the ®gure, both of the
optimum performance coecients (bref and bhp ) decrease as k increases. The e€ect of k on bref
and bhp is greater at low values of IDS . It should be noted that when k ˆ 0, the optimum
performance coecient reaches its upper bound (bmax ), as given in Eq. (33). In real
applications, k is always greater than zero, and its value should be determined according to the
economical conditions of a country.
From Eqs. (21) and (31), we see that the optimum ratios of the heat exchanger areas
‰…AH =AL †ref , …AH =AL †hp Š are the same for the irreversible refrigerator and heat pump, and they
are only functions of UH, UL and IDS . It is interesting to ®nd that …AH =AL †ref and …AH =AL †hp
are independent of the economical parameter k. The optimum speci®c cooling load given in
Eq. (22) and the optimum speci®c heating load given in Eq. (32) depend on the parameters R,
IDS and k. It is seen from Eqs. (22) and (32) that the optimal speci®c cooling and heating loads
are linear decreasing functions of R. The variations of the optimal speci®c loads with respect to
IDS and k are given in Fig. 7(a) and (b). As can be seen from Fig. 7(a), for increasing IDS , the
optimal speci®c cooling load decreases, while the optimal speci®c heating load increases. From
Fig. 7(b), we observe that the optimal speci®c cooling and heating loads increase
monotonically as k increases. This result indicates that for a speci®ed cooling or heating load,
increasing k leads to the result of decreasing the optimal total heat transfer area.

Fig. 6. Variations of the optimum coecients of performance with respect to k (TH ˆ 288 K, TL ˆ 243 K for the
refrigerator and TH ˆ 350 K, TL ˆ 273 K for the heat pump, UH ˆ UL ˆ 0:5 kW/m2 K, Rref ˆ Rhp ˆ 0:05).
618 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619

Fig. 7. Variations of the optimal speci®c cooling and heating loads with respect to IDS for k ˆ 0:2 (a), and with
respect to k for IDS ˆ 1:2 (b), (TH ˆ 288 K, TL ˆ 243 K for the refrigerator and TH ˆ 350 K, TL ˆ 273 K for the
heat pump, UH ˆ UL ˆ 0:5 kW/m2 K, Rref ˆ Rhp ˆ 0:05).

5. Conclusions

We have presented a thermoeconomic optimization to determine the optimal operation and

design parameters for the irreversible refrigerator and heat pump which include the
irreversibilities of ®nite rate heat transfer, heat leakage and internal dissipations. The objective
functions, taken as the cooling load per unit total cost for the irreversible refrigerator and also
taken as the heating load per unit total cost for the irreversible heat pump, are optimized with
respect to the working ¯uid temperatures. As a result of the optimization procedure, equations
of optimum working ¯uid temperatures, optimum performance coecients, optimum ratios of
the heat exchanger areas and optimum speci®c cooling and heating loads were found in terms
of technical and economical parameters. The e€ects of internal irreversibility, heat leakage and
the economical parameter on the global and optimal performances were demonstrated with
plots and discussed. It is important to note that the optimum design parameters obtained in
this analysis minimize the total cost per unit cooling or heating load (the inverse of the
objective function). Therefore, the above optimization results also represent the economical
design conditions and provide signi®cant guidance to optimal design for real refrigerator and
heat pump systems.

References

[1] Chambadal P. In: Les Centrales Nucleaires. Paris: Arman Colin, 1957. p. 41.
[2] Novikov I. J Nucl Energy II 1958;11(7):125.
[3] Curzon FL, Ahlborn B. Am J Phys 1975;43:22.
 A. Kodal et al. / Energy Conversion & Management 41 (2000) 607±619 619

[4] Bejan A. J Appl Phys 1996;79:1191.

[5] Wu C. Energy Convers Mgmt 1995;36:7.
[6] Wu C. Energy Convers Mgmt 1996;37:1509.
[7] Chiou JS, Liu CJ, Chen CK. J Phys D: Appl Phys 1995;28:1314.
[8] Chen J, Wu C. Energy Convers Mgmt 1996;37:353.
[9] Chen J. Energy Ð The International Journal 1994;19:1031.
[10] Chen J. Energy Convers Mgmt 1994;35:1009.
[11] Cheng C-Y, Chen C-K. J Phys D: Appl Phys 1995;28:2451.
[12] Bejan A, Vargas JVC, Sokolov M. Int J Heat Mass Transfer 1995;38:2997.
[13] Wijeysundera NE. Energy Ð The International Journal 1995;20:123.
[14] Wu C, Chen L, Sun F. Energy Convers Mgmt 1996;37:1671.
[15] Yan Z, Chen J. J Phys D: Appl Phys 1990;23:136.
[16] Chen L, Wu C, Sun F. Energy Convers Mgmt 1998;39:113.
[17] Yan Z, Chen J. J Appl Phys 1989;65:1.
[18] GoÈktun S, OÈzkaynak S. Energy Ð The International Journal 1997;22:481.
[19] Chen L, Wu C, Sun F. Energy Convers Mgmt 1998;39:45.
[20] Chen J. J Phys D: Appl Phys 1997;30:582.
[21] Chen J. J Phys A: Math Gen 1994;27:6395.
[22] Lin G, Yan Z. J Phys D: Appl Phys 1997;30:2006.
[23] Chen J, Schouten JA. Energy Convers Mgmt 1998;39:999.
[24] Wu C, Chen L, Sun F, Cao S. Energy Convers Mgmt 1998;39:143.
[25] Chen L, Sun F, Ni N, Wu C. Energy Convers Mgmt 1998;39:767.
[26] Sahin B, Kodal A. Energy Convers Mgmt 1998;40:951.
[27] Arora JS. In: Introduction to optimum design. New York: McGraw-Hill, 1989. p. 521±35.
[28] White JA, Agee MH, Case KE. In: Principles of engineering economic analysis. New York: Wiley, 1989. p. 81±
168.