1

10. 2 Refrigerators

(Hiroshi Matsuoka)

A refrigerator is a heat device that converts work into heat through a cyclic process so that it
can transfer energy from a colder heat reservoir to a hotter heat reservoir, which is typically the
air outside the refrigerator. In this text, the term refrigerator therefore refers to any heat device
that extracts heat from a colder heat reservoir and transfers it as heat to a hotter heat reservoir
and its examples include regular refrigerators in kitchens and air conditioners for buildings.
Clearly, it would be ideal if we could construct a refrigerator that could absorb some heat from
the colder heat reservoir and transfer it completely as heat into the hotter heat reservoir. It turns
out that we cannot construct such a refrigerator. As we will discuss in the next chapter, this
statement of the impossibility of a perfect refrigerator is nothing but the second law of
thermodynamics.
Refrigerator as a heat engine running in reverse
Any real refrigerator must therefore accept some energy as work from the outside so that this
extra work W is added to the input heat Qin flowing from the colder reservoir and gets wasted as
a part of the heat !Qex discarded to the hotter heat reservoir. Note that the input heat Qin is
positive or Qin > 0 as it flows into the refrigerator, whereas the exhaust heat Qex is negative or
Qex < 0 as it flows out of the refrigerator. As the figure below shows, the refrigerator is basically
a heat engine running in reverse.

As after each cycle there is no net change in the internal energy of the refrigerator, by
applying the first law of thermodynamics we find

Qin + Qex + W = !U = 0
so that

W = !Qex ! Qin .

2
As W > 0 , we find

!Qex > Qin ,

so that the amount of heat discarded to the hotter reservoir is larger than the heat extracted from
the colder reservoir because of the extra energy due to the input work.
Coefficient of performance of a refrigerator
To improve the refrigerator, we must then aim at increasing the coefficient of performance
COP defined by
Q
COP ! in ,
W
which is basically the ratio of the benefit, namely how much heat Qin we can take away from the
colder heat reservoir, to its cost or how much energy must be wasted as the work. To improve
the refrigerator, we must therefore understand what determines the efficiency of the refrigerator
so that we can find a way to increase its coefficient of performance.
The coefficient of performance COP can be also expressed in terms of Qin and Qex as
COP =

Qin
Qin
1
=
=
W
!Qex ! Qin !Qex Qin ! 1

so that we can calculate the coefficient of performance directly from Qin and Qex . This equation
indicates that to increase the coefficient of performance, we must make the discarded heat !Qex
as close to the input heat Qin as possible.
Coefficient of performance is not everything
Clearly, the coefficient of performance of a refrigerator is not the only factor that dictates the
design of the particular refrigerator for a particular purpose, which requires, for example, a
particular size for input heat Qin Once we have incorporated all the specifications required for
the particular refrigerator, we then try to maximize its coefficient of performance.
Ideal refrigerators based on quasi-static cyclic processes
Applying the second law of thermodynamics, we will show later that the coefficient of
performance of a refrigerator reaches a maximum value when the cyclic process for the
refrigerator becomes completely quasi-static so that the quasi-static cyclic process provides an

3
upper limit for the coefficient of performance of the refrigerator. We will therefore examine
quasi-static cyclic processes as idealized models for refrigerators.
10.2.1 The Carnot refrigeration cycle: an idealized model for a refrigerator
As an idealized model for a refrigerator, a Carnot refrigeration cycle is a quasi-static cyclic
process in which a substance inside the refrigerator exchanges heat with two heat reservoirs at
two different temperatures, Tc and Th , where Tc < Th . In the Carnot refrigeration cycle for a
refrigerator, the substance absorbs heat from the colder reservoir through a quasi-static
isothermal process at temperature Tc and discards heat to the hotter reservoir through another
quasi-static isothermal process at temperature Th . Following the isothermal process at
temperature Tc , the substance increases its temperature to Th through a quasi-static adiabatic
process before it goes through the quasi-static isothermal process at temperature Th followed by
another quasi-static adiabatic process that decreases the temperature back to Tc . As we have
seen in Sec.9.4, along a quasi-static adiabatic process, we can change the temperature of the
substance by changing its volume so that it is always possible to construct the Carnot
refrigeration cycle for the refrigerator.

Th
! Qex

Qin
Tc

The universal coefficient of performance for the Carnot refrigeration cycles between
reservoirs at Tc and Th
The Carnot refrigeration cycle occupies a special place in the study of refrigerators, because
the value of its coefficient of performance depends only on Tc and Th so that all the Carnot
refrigeration cycles that operate with two heat reservoirs at Tc and Th share the same universal
coefficient of performance:

COPCarnot =

1
Tc
=
.
Th Tc ! 1 Th ! Tc

As with the universal efficiency for the Carnot engine cycles, this universal coefficient of
performance depends only on the temperature ratio between Tc and Th .
Deriving the universal coefficient of performance for the Carnot refrigeration cycles
between reservoirs at Tc and Th
In the Carnot refrigeration cycle, the quasi-static heat Qinqs that the substance in the
refrigerator absorbs from the colder reservoir at Tc is then

Qinqs = Tc !Sc ,
where !Sc is the change in entropy of the substance in this isothermal process, while the quasistatic heat !Qexqs that the substance in the refrigerator discards to the hotter reservoir at Th is

!Qexqs = !T h "Sh ,
where !Sh is the change in entropy of the substance in this isothermal process. In contrast,
along the quasi-static adiabatic processes, there is no heat flowing into the substance in the
refrigerator so that the entropy of the substance remains constant and the change in entropy of
the substance in these processes is zero.
As the Carnot refrigeration cycle is a cyclic process, the entropy of the substance must return
to its initial value after the cycle so that the net change in the entropy after the cycle is zero:

0 = !S = !Sh + !Sc ,
from which we obtain

!Sc = " !Sh .

We then find the coefficient of performance of the Carnot refrigeration cycle to be
COPCarnot =

1
1
1
.
=
=
qs
!Q Qin ! 1 ( !T h "Sh ) (Tc "Sc ) ! 1 Th Tc ! 1
qs
ex

5
The Carnot refrigeration cycle in the TS diagram and its universal coefficient of
performance
We can understand the above derivation of the coefficient of performance of the Carnot
refrigeration cycles between two heat reservoirs at Tc and Th by sketching a Carnot refrigeration
cycle in the TS diagram shown below.

Sh

Th
T

Tc

Sc
S
qs

In this diagram, the quasi-static input heat Qin = Tc !Sc corresponds to the area of the shaded
area below the line from state 4 to state 1, while the quasi-static discarded heat !Qexqs = !T h "Sh
corresponds to the sum of the two shaded areas below the line from state 2 to state 3. The
coefficient of performance of the cycle depends on the ratio between !Qexqs and Qinqs , which is
equal to the ratio between the two areas, which is clearly equal to the ratio between Tc and Th ,
A crucial point here is that the entropy changes along the two isothermal processes are equal in
magnitude: !Sh = !Sc . This point is also the reason why the coefficient of performance of the
Carnot refrigeration cycles between two heat reservoirs at Tc and Th is universal despite that the
values of entropy at states 1 through 4 may vary from one Carnot cycle to another.
The shaded area enclosed by the Carnot refrigeration cycle in this TS diagram also
corresponds to the input work because
W q s = !Qexqs ! Qinqs = (Th ! Tc )"Sc .

Input heat, exhaust heat, and input work in the TS diagram

In general, for a quasi-static cyclic process on the TS diagram shown below, the area
underneath the upper portion of the curve for the cyclic process represents the exhaust heat !Qexqs

6
of the process while the area underneath the lower portion of the curve for the cyclic process
represents the input heat Qinqs of the process. As

W q s = !Qexqs ! Qinqs ,
qs
the area enclosed by the curve for the cyclic process represents the input work W . Therefore,
for a given upper portion of the curve for the cyclic process, the larger the area underneath the
lower portion of the curve for the cyclic process becomes, the higher the coefficient of
performance gets.

10.2.2 The standard refrigeration cycle: an idealized model for a refrigerator

As an idealized model for a cyclic process in a refrigerator, we will examine a quasi-static
standard refrigeration cycle of a refrigerant such as a hydrofluorocarbon called HFC-134a that
has replaced the notorious freon. The standard refrigeration cycle consists of two quasi-static
isobaric processes at constant pressure and two adiabatic processes, one of which is quasi-static
while the other is not.
Quasi-static heat in an isobaric process: P = const
As we have found in Sec.9.2 the quasi-static heat QPq s flowing into a system in an isobaric
process at pressure P is simply given by

QPq s = !H ,
where !H is a change in the enthalpy of the system after the isobaric process:

7
!H = H (T f ,V f ,n) " H (Ti ,Vi ,n ) = H(T f , P, n ) " H(Ti , P, n) ,

so that if we can find the initial and final values of the enthalpy H through an analytical
expression for H or in a numerical table for H in some reference handbook, we can find the
quasi-static heat QPq s .
The standard refrigeration cycle
The standard refrigeration cycle consists of the following four processes:
(i) An quasi-static adiabatic compression of gas to increase its pressure proceeds from state
(V1 , P1 ,n) to state (V 2 , P2 ,n ) .
(ii) A quasi-static isobaric condensation of the gas into liquid proceeds from state (V 2 , P2 ,n ) to
state (V 3 ,P2 ,n) . In this process, an exhaust heat !Qexqs flows out of the gas and is related to
the change !H2 3 in the enthalpy of the refrigerant:
qs
!Qex
= ! "H2 3 = ! {H( V3 , P2 ,n ) ! H(V 2 , P2 ,n )} # ! ( H3 ! H2 ) = H2 ! H3 .

(iii) A Joule-Thomson non-equilibrium adiabatic process through throttling valve to cool down
the liquid into a mixture of gas and liquid proceeds from state (V 3 ,P2 ,n) to state (V 4 , P1 ,n ) .
(iv) A quasi-static isobaric evaporation of the mixture of gas and liquid into gas proceeds from
state (V 4 , P1 ,n ) back to the initial state (V1 , P1 ,n) . In this process, the temperature of the
mixture of gas and liquid also remains constant at T1 . In this process, an input heat Qinqs
flows into the mixture of gas and liquid and is related to the change !H4 1 in the enthalpy of
the refrigerant:

Qinqs = !H4 1 = H( V1 ,P1 , n) " H(V 4 , P1 ,n ) # H1 " H4 .

We can now sketch this standard refrigeration cycle on the PV diagram as shown below.

As the Joule-Thomson throttling process is not quasi-static, during the process the system is not
found in its equilibrium state so that we cannot use a curve to represent this process on the PV
diagram and we have used a broken line for this process instead.
The Joule-Thomson throttling process with no enthalpy change
Although the Joule-Thomson process is not quasi-static, the enthalpy of a system undergoing
this process takes the same value at both the initial and the final equilibrium states of the process.
In the standard refrigeration cycle, we therefore find

H3 = H4 .
We can derive this very useful result as follows. The throttling process can be modeled by a
slow passage of a fluid through a porous plug in a cylinder. Initially, the fluid is in one side of
the plug and by gradually compressing the fluid with a piston to keep the pressure at a constant
value P2 , we can force the fluid to pass through the plug. To keep the pressure of the fluid at a
constant value P1 after it passes through the plug, we place another piston on the other side of
the plug and gradually draw it outward until all of the fluid passes through the plug. Keep in
mind that the flow of the fluid inside the porous plug is not quasi-static as the pressure of the
fluid must be rapidly decreased inside the plug and is therefore not spatially uniform inside the
plug, which is one of the requirements for equilibrium states, so that in this process, the fluid
does not go through its equilibrium states while it flows through the plug.
As this process is adiabatic, by applying the first law of thermodynamics we find
U(V 4 , P1 ,n ) ! U (V3 ,P2 , n) = "U = Q + W q s = W q s = P2 V 3 ! P1 V4

so that

9
H4 = U (V4 , P1 ,n) + P1 V4 = U(V 3 , P2 ,n ) + P2 V 3 = H3 .

The coefficient of performance of the standard refrigeration cycle

The coefficient of performance of the standard refrigeration cycle is then given by
COP =

1
1
H1 ! H4
H ! H3
,
=
=
= 1
!Qex Qin ! 1 ( H2 ! H3 ) ( H1 ! H4 ) ! 1 H2 ! H3 ! H1 + H4 H2 ! H1

where we have used H3 = H4 . Using a table for saturated liquid and gas, we can look up the
H3 = 3444 kJ value of H1 and H3 , while using a table for superheated gas, we can also look up
the value of H2 .
Example: As an example, consider a typical household refrigerator with refrigerant HFC134a operating between two pressures,
and
P1 = 1.0 atm
P2 = 10 atm
5
(1 atm ! 1" 10 Pa ). The temperature for the mixture of liquid and gas refrigerant is at
T1 = !26.4C = 246.8 K . Using a table for saturated liquid and gas refrigerant, we
find the value of H1 for 1 kg of liquid refrigerant to be

H1 = 231 kJ

and the value of H3 for 1 kg of gas refrigerant to be

H3 = 105 kJ .
We can also use a table for superheated gas refrigerant to calculate the value of H2 for
1 kg of gas refrigerant and find it to be
H2 = 279 kJ .

Using these values of the enthalpy, we find the coefficient of performance of this
refrigerator to be
COP =

H1 ! H3 (231 kJ) ! (105 kJ)

=
= 2.6 .
H2 ! H1 (279 kJ) ! (231 kJ)

10
The standard refrigeration cycle in the TS diagram and its coefficient of performance
compared with that of the Carnot refrigeration cycle
The coefficient of performance of the standard refrigeration cycle is less than that of the
Carnot refrigeration cycle between reservoirs at T1 and T2 as can be seen in the sketch of the
standard refrigeration cycle in the TS diagram shown below.

In the previous example for the standard refrigeration cycle in a household refrigerator,
T1 = !26.4C = 246.8 K and T2 = 49.2C = 322.4 K so that the coefficient of performance of a
Carnot refrigeration cycle between reservoirs at T1 and T2 is indeed higher than the coefficient
of performance for the standard refrigeration cycle:
COPCarnot =

(!26.4 + 273.15) K
T1
=
= 3.26 .
T2 ! T1 (49.2C) ! ( !26.4C)

Appendix to Sec.10.2.2: Temperature change due to a Joule-Thomson process

In the above, we have assumed that T4 < T3 , which is always the case when state 3 is on the
saturated liquid curve. In general, when the pressure of a gas or a liquid near the gaseous phase
is low enough that constituent molecules are relatively far apart, the potential energy of the fluid
increases through a volume expansion as the mean distance among the molecules increases to
weaken the attractive forces among them. In contrast, when the pressure of the gas or the liquid
is very high so that the molecules are being pushed against each other, the potential energy of the
fluid decreases through a volume expansion as the mean distance among the molecules decreases

11
to reduce the repulsive forces among them. To compensate for the increase or the decrease of
the potential energy, the temperature then decreases or increases.
To characterize the extent of the temperature change, we use the following Joule-Thomson
coefficient:
$ #T '
!"& ) .
% #P ( H ,n
On the PT diagram, the points at which this coefficient vanishes from a curve called the
inversion curve. In the lower pressure side of the inversion curve, ! > 0 so that the temperature
drops as the pressure drops so that we can cool the fluid by reducing the pressure. In the higher
pressure side of the inversion curve, ! < 0 so that the temperature jumps as the pressure drops so
that we can heat up the fluid by reducing the pressure.
We can express ! in terms of measurable quantities, T, v, c P , and ! , as follows. As we
have found in Sec.7.2, an infinitesimal change in enthalpy is given by

dH = CP dT ! V (T" ! 1)dP .
We then obtain
dT =

1
V
1
v
( T! " 1)dP =
dH +
dH + (T! " 1)dP
CP
CP
ncP
cP

so that

# "T &
v
! = % ( = ( T) * 1) .
$ "P ' H ,n cP
The inversion curve is therefore determined by the following equation:
T! = 1 .

For low-density gases, for which the ideal gas law is a good approximation for their
equations of state, this equation is satisfied so that their temperature cannot be changed by a
Joule-Thomson process, which is consistent with that for these gases, the potential energy is
negligible so that it does not change with the changes of their pressures.
For a gas whose equation of state can be reduced to the universal van der Waals equation of
state:
1 " T
a %'
,
P = $$
!
Zc # v ! b v2 '&

12
where Zc = 3 8 , a = 9 8 , and b = 1 3 and we also limit the range for v and T to v > b and
T > 1 , we can define the reduced Joule-Thomson coefficient by
$ #T '
P
! " && )) = c
% #P ( h T c

$P ' v
$ P v ' v
$ #T '
v
& ) = && c )) (T* + 1) = && c c ))
T* + 1 = Zc
T* + 1 ,
% #P ( h % Tc ( cP
cP
% RT c ( cP

where cP ! cP R and

+- 1 $ # v ' -/$ # P '

1 $ #v '
! " Tc ! = & ) = ,-* & ) 0-&& ))
v % #T ( P . v % # P ( T 1% # T ( v

$ #P '
&& ))
v2 v * b
1 % #T ( v
=* $ ' =
v # P
v3 T * 2 a v * b
&& ))
% # v ( T

so that

T! " 1 =

" bT " 2 a v2 " 4abv + 2ab 2

Tv3 " 2 a v " b

The inversion curve is then determined by T! " 1 = 0 Note that the denominator is positive:

Tv 3 ! 2a v ! b

"
$
T
= ! v 3 v ! b #!
$ v ! b
%

&
* ) P 2
2 a $
+ 3 ' = ! v 3 v ! b Zc ,, // > 0 ,
v $
+ )v . T
(

since
" !P %
$$ '' < 0 .
# !v & T

If T = 2 a b so that bT ! 2a = 0 , then

T! " 1 =

)
" 2 a (v " b )

"4ab v " b 2
Tv3

<0

so that there is no value for v for which T! " 1 = 0 . If T ! 2 a b , then the inversion curve in the
T - v plane must be determined by the following quadratic equation:

! bT ! 2a v 2 ! 4 a bv + 2 a b2 = 0 ,
whose two roots, v+ and v! , are

13

v =

b
bT
1m
2a

As v! < b , v! = v! T does not represent the inversion curve. If T > 2 a b = 6.75 , then v+ < 0 so
that only for 0 < T < 2 a b = 6.75 , v+ = v+ T can represent the inversion curve. Solving
v = v T for T , we can parametrize the temperature along the inversion curve using v as a

()

()

()

parameter:

2
2 a "$
b %'

Tinv (v+ ) = $ 1! ' .

b # v+ &

We can also parametrize the pressure along the inversion curve by

1 " T (v ) a % a "$ 2
3 %'
.
Pinv (v+ ) = $$ inv + ! 2 '' =
!
Zc # v+ ! b v+ & Zc $# bv+ v+2 '&

The inversion curve is then parametrized by Pinv (v+ ), Tinv( v+ ) on the P - T plane as shown in the

figure below. In the figure we also show a set of isenthalpes, along each of which the reduced
molar enthalpy of the gas is constant. The reduced molar enthalpy of the gas is given by
#
h
u + Pv
h !
=
= u + Zc P v = % cvT "
$
RTc
RT c

a & Zc
(+
v ' Zc

# T
#
a &(
v & 2 a
%%
= % cv + 1+
,
"
v
(T "
2
(
$
v " b '
v
$ v " b v '

where cv ! cv R = 5 2 , so that we can parametrize each isenthalpe by Ph ( v), Th (v) , where

Th (v) =

2 a
h +
v
cv + 1+

v
v ! b

and

1 "$ Th ( v) a %'
Ph ( v) =
! 2 .
Zc $# v ! b v '&

The inversion curve intersects each isenthalpe at its maximum. In the region to the left of the
inversion curve, we find ! > 0 so that the temperature of the gas decreases if the pressure
decreases after a Joule-Thomson process whereas in the region to the right of the inversion
curve, we find ! < 0 so that the temperature of the gas increases if the pressure decreases after a
Joule-Thomson process.

14

SUMMARY FOR SEC.10.2

1.

For a refrigerator that accepts some work W from the outside as well as an input heat Qin
from a colder reservoir and discard an exhaust heat !Qex to a hotter heat reservoir., we can
calculate its coefficient of performance COP by
COP =

2.

Qin
Qin
1
=
=
.
W
!Qex ! Qin !Qex Qin ! 1

In a Carnot refrigeration cycle, a substance absorbs heat from a colder reservoir through a
quasi-static isothermal process at temperature Tc and discards heat to a hotter reservoir
through another quasi-static isothermal process at temperature Th . Following the isothermal
process at Th , the substance decreases its temperature to Tc through a quasi-static adiabatic
process before it goes through the quasi-static isothermal process at Tc followed by another
quasi-static adiabatic process that increases the temperature back to Th . The coefficient of
performance COP of the Carnot refrigeration cycle depends only on Th and Tc by the
following universal equation:
COPCarnot =

1
Tc
=
Th Tc ! 1 Th ! Tc

15
3.

An standard refrigeration cycle of a refrigerant such as a hydrofluorocarbon called HFC134a is an idealized model for a cyclic process in a refrigerator, and consists of two quasistatic isobaric processes at constant pressure and two adiabatic processes, one of which is
quasi-static while the other through a Joule-Thomson process is not. Its coefficient of
performance depends on three enthalpy values: H1 for the initial equilibrium state of the
quasi-static adiabatic compression, H2 for the initial equilibrium state of the isobaric
condensation process, and H3 for the initial equilibrium state of the isobaric condensation
process.
COP =

1
H ! H3
= 1
.
!Qex Qin ! 1 H2 ! H1