Engineering Data Book III

Chapter 15
Thermodynamics of Refrigerant Mixtures
and Refrigerant-Oil Mixtures
SUMMARY: The principles of phase equilibrium of zeotropic mixtures are presented in an abbreviated
form, including the phase diagram and preparation of enthalpy profiles for describing their evaporation
(or condensation) over a temperature glide. Next the thermodynamic description of refrigerant-oil
mixtures is presented, including methods to predict the local oil mass fraction, the local bubble point
temperature rise caused by the oil, enthalpy changes for their evaporation (or condensation) and the effect
of oil on log mean temperature differences. This chapter focuses on the thermodynamics of mixtures
during their evaporation but these methods are equally applicable to condensation.

15.1 Introduction
A basic working knowledge of phase equilibria is required to understand the evaporation (and
condensation) of zeotropic mixtures and also is an important prerequisite for the design of evaporators
that properly take into account the effect of oil on thermal performance. The objective of this chapter is
to provide this necessary background. Topics covered include the phase equilibrium diagram, the dew
point temperature, the bubble point temperature, the boiling range or temperature glide, the phase
equilibria of refrigerant-oil mixtures, enthalpy curves for mixtures and a few other concepts important to
the subject. For a more detailed introduction to phase equilibria, the reader is referred to Chapter 12 in
Collier and Thome (1994, 1996). Instead, for a comprehensive treatment of prediction of phase equilibria,
numerous chemical engineering textbooks are available, such as that by Prausnitz (1969), or can be found
in the chapter by Smith, Block and Hickman (1963) appearing in the Chemical Engineer's Handbook
(available in more recent editions).
A zeotropic mixture refers to a mixture whose components have different mass fractions in the liquid
phase than in the vapor phase at equilibrium conditions. An example of such a mixture is R-407C, which
has three components. In contrast, an azeotropic mixture refers to a mixture whose components have the
same compositions in both phases. A well-known example is refrigerant R-502, which is a binary
refrigerant mixture at the azeotropic composition of its two components. An azeotropic fluid functions
just like a pure fluid and hence pure fluid design methods can be applied. Instead, a zeotropic mixture
evaporates (or condenses) over a temperature range and hence this must be taken into account in thermal
design methods.
A refrigerant-oil mixture functions as an azeotropic mixture in which the refrigerant (either a pure
refrigerant such as R-134a or a zeotropic mixture such as R-407C) is combined with a miscible,
lubricating oil. The oil for our purposes here can be considered to be one component in such a mixture,
even though normally lubricating oil is a multi-component mixture including various additives, longchain molecules, etc. In this chapter, only refrigerant-oil mixtures that are completely miscible will be
considered. These mixtures act as zeotropic mixtures and the refrigerant and oil are miscible in the liquid
phase. Some refrigerant-oil mixtures are not miscible, which means they do not mix together in
equilibrium conditions, such as is also true for water and oil. Immiscible refrigerant-oil mixtures are not
addressed here. One must be careful, however, as some refrigerant-oil mixtures are miscible only within a
certain range of temperature or up to a certain oil mass fraction, so that part of the refrigerant cycle or test
facility fluid loop may pass outside the miscible range.

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-1

Engineering Data Book III

15.2 Simple Principles of Phase Equilibrium

This section presents a very brief description of the thermodynamics of zeotropic mixtures. For a more
extensive treatment, refer to a chemical engineering handbook or textbook. The following discussion is
for a binary mixture, i.e. a mixture with two different fluids for its components. These ideas can be
extended to ternary mixtures and multi-component mixtures in general.

15.2.1 Phase Equilibrium Diagram

Phase equilibriums of binary mixtures are represented on phase diagrams. Figure 15.1 depicts a phase
diagram for a binary mixture system at constant pressure. Temperature is plotted along the vertical axis
while vapor and liquid compositions are plotted along the horizontal axis, in this case shown in mole
fractions. The dew point line represents the dew point temperatures of all the possible mixture
compositions, where Tdew is defined as the temperature at which a superheated vapor mixture will first
begin to condense upon cooling. The bubble point line represents the bubble point temperatures of all the
possible mixture compositions, where Tbub is defined as the temperature at which a subcooled mixture
will first begin to evaporate upon heating. The composition of one of the components, in this case the
component with the lowest boiling point, is depicted on the lower axis, where the equilibrium vapor
composition Y is that which corresponds to the
liquid composition X at the same temperature.
In Figure 15.1 the saturation temperature of one
pure component is on the left vertical axis while
that of the other component is on the right
vertical axis. The boiling point temperatures for
evaporation of the mixtures correspond to the
bubble point line. The condensing temperatures
for condensation of the mixtures correspond to
the dew point line. The temperature difference
between the dew point temperature and the
bubble point temperature for a specific liquid
composition is defined as the boiling range (BR
= Tdew-Tbub). For evaporation of a mixture, the
parameter BR is used in modeling the effect of
mass transfer. The boiling range is also referred Figure 15.1. Phase equilibrium diagram at
to as the temperature glide in literature referring constant pressure for a binary mixture.
to zeotropic refrigerant mixtures.

15.2.2 Definition of Mixture Compositions and Components

On a phase diagram, the compositions can be depicted in units of mass fraction, mole fraction or volume
fraction. Refrigerant-oil mixtures are usually referred to in ASHRAE publications in terms of weight
fraction or weight percent, where the first of these is a colloquial term for mass fraction. The mass
fraction of a component is defined as its mass in kg (or lb.) divided by the total mass of the mixture in kg
(or lb.), respectively, in a particular phase (liquid or vapor). For instance, if a liquid refrigerant-oil
mixture weighs 10.0 kg (11.0 lb.) and the oil in this mixture is measured to weigh 0.5 kg (1.1 lb.), then
the mass fraction of the oil in the mixture is 0.05 and its mass percent is 5.0%. The mass fraction of the
refrigerant is thus 0.95 (i.e. 1.0-0.05) and its mass percent is 95.0%.

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-2

Engineering Data Book III

Referring again to the phase diagram in Figure 15.1, the more volatile component (also referred to as the
lighter component) is defined as the fluid that has the lower boiling point temperature at the pressure of
the diagram. This is the fluid with its saturation temperature on the right vertical axis in Figure 15.1. This
fluid is more volatile in the sense that this component is always above its normal boiling point. For a
refrigerant-oil mixture, the refrigerant will always be the more volatile component since the boiling point
temperatures of lubricating oils are in the range from 300 C to 510 C (570-950F), and are thus much
larger than those of refrigerants. Hence, the oil will be the less volatile component (also referred to as the
heavier component).
For the present situation, a refrigerant-oil mixture can be considered as a zeotropic binary mixture, which
is a mixture with two components that does not form an azeotrope. One component is the pure
refrigerant, be it R-134a, R-22, or any other single-component refrigerant (R-123, R-125, ammonia, etc.)
and the second component is the oil. Even though lubricating oils are multi-component blends of oil plus
additives, for the present purposes it is more than adequate to consider them here as a single component
fluid.

Figure 15.2. Phase equilibrium diagram for a binary mixture of R-134a and oil at a
pressure of 3.43 bar (49.7 psia).
Figure 15.2 depicts an approximate phase diagram for R-134a mixed with Mobil Arctic EAL 68
lubricating oil at 3.43 bar (49.7 psia). In this case, the horizontal axis shows the oil composition in mass
%. At the left axis of the diagram, the saturation temperature of pure R-134a is 4.44C (40F). Near the
right axis, the dew and bubble point temperatures rise rapidly towards the saturation temperature of the
pure oil (not shown), which is in the neighborhood of 350C (662F). The bubble point curve is nearly
horizontal with a slightly upward slope from left to right at low oil compositions. It then begins to rise
more rapidly at intermediate compositions and then rises rapidly when above 70 mass % oil. The actual
location of the dew point curve is not known for refrigerant-oil mixtures; it was drawn assuming
negligible amounts of oil enter the vapor phase at liquid oil compositions below 70 mass %, i.e. the part
of the dew point line on the left y-axis at 0.0 mass % oil. Above this value its composition is represented

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-3

Engineering Data Book III

here by a simple extrapolation towards the pure oil value. The method for predicting the bubble point
curve shown here is given later in this Chapter.

15.2.3 Piston-and-Cylinder Analogy for a Binary Mixture

Figure 15.3. Isobaric process path and enthalpy curve of a binary mixture.
To better illustrate these concepts, consider now a situation excerpted from Collier and Thome (1994,
1996). Figure 15.3 shows a piston-and-cylinder setup that contains a binary liquid mixture whose mole
fraction of the more volatile component is Xo and which is held at a fixed pressure of p and temperature
of T1. What happens to the liquid mixture as it evaporates is conveniently depicted on the temperature-

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-4

Engineering Data Book III

composition phase diagram at the right, where the mole fraction of the more volatile component is plotted
along the horizontal axis and the temperature of the mixture on the vertical axis. Denoting the initial
conditions in the vessel by point Q, the piston-and-cylinder is now heated under constant pressure
maintaining an isothermal temperature throughout the vessel. When the temperature T2 is reached, liquid
will begin to evaporate at point R. The vapor formed at temperature T2 has a composition Yo as shown by
point S. This vapor is richer in the more volatile component than the liquid. If this procedure is repeated
for the complete range of initial liquid compositions, then a series of points R, T and X will be found. The
locus of the curve formed by points AXTRB is known as the bubble point curve and the curve formed by
points AWVSB is the dew point curve.
If the vessel is heated above point R, the composition of the liquid will change because of the loss of
some of its more volatile component to the vapor. The boiling point temperature will rise to some new
temperature T3 and the compositions in both the liquid and vapor phases will change, both becoming
richer in the less volatile component represented by points T and V, respectively. Since no material is lost,
the relative volumes occupied by the two phases also change during this process; the volume occupied by
the liquid decreases while that of the vapor increases. On still further heating to temperature T4, all the
liquid is evaporated and the vapor in the vessel at point W has the same composition as the original liquid
Xo. The last drop of liquid in the vessel disappears at the temperature T4 with a composition of X1,
indicated by point X on the phase diagram, and is very rich in the less volatile component.
During this experiment the amount of heat added to the system could be noted. In addition, there is a clear
analogy between this isobaric evaporation process in the piston-and cylinder chamber and evaporation of
a refrigerant-oil mixture inside an evaporator tube. Similar to evaporation of the liquid mixture in the
vessel, a refrigerant-oil mixture will evaporate along its bubble point curve. However, since essentially
none of the oil enters the vapor phase because of the extremely low vapor pressure of the oil relative to
the refrigerant (on the order of one millionth that of the refrigerant), the situation is simplified. Thus, as
one moves along an evaporator tube from the inlet towards the outlet, the local oil mass percent in the
remaining liquid increases, the vapor phase composition remains pure refrigerant, and the local boiling
point temperature Tbub increases. The amount of heat necessary to evaporate the refrigerant-oil mixture is
therefore equal to the sensible heating of the liquid and vapor phases caused by the increase in the bubble
point temperature along the tube plus the latent heat for vaporizing the refrigerant from the liquid to the
vapor phase. These points will be discussed in more detail in the following sections.

15.2.4 Oil "Contamination" Approach

Until recently, a small fraction of lubricating oil in a refrigerant charge has been treated by refrigeration
and air-conditioning system manufacturers as a contaminant in an otherwise pure refrigerant. Thus, pure
refrigerant saturation temperatures and properties have been utilized for determining heat transfer
coefficients while the oil composition has been considered an independent "contamination" parameter. In
practice, this means that during flow boiling experiments with oil to determine local heat transfer
coefficients, the local wall temperature difference (Tw-Tsat) has been determined using Tsat from the vapor
pressure curve of the pure refrigerant and the measured pressure, where Tw is the inner tube wall
temperature.
This oil contamination approach has hindered a better understanding of the problem and created a
roadblock to development of accurate generalized design methods. The reason for this is that the
"contamination" approach is not thermodynamically correct. That is, it ignores the influence of the oil on
the boiling point temperature, specific heat, latent heat, viscosity, density, etc. This in turn alters energy
balances, local boiling temperature superheats, local vapor qualities and so on in the data reduction
process to calculate heat transfer coefficients and two-phase pressure drops. To put this approach in

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-5

Engineering Data Book III

perspective, it is analogous to assuming that a zeotropic refrigerant mixture, such as R-407C, may be
modeled using the vapor pressure curve of only one of its three components, say R-134a. Clearly that is
not correct thermodynamically. Thus, the "contamination" approach will not lead to a resolution of the
long-standing problem of how to model the influence of oil on flow boiling of refrigerants.

15.2.5 Thermodynamic Approach to Modeling Refrigerant-Oil Mixtures

The thermodynamically correct approach is to treat the refrigerant-oil mixture as the real mixture that it in
fact is. This will be referred to as the Thermodynamic Approach. It requires methods to determine the
boiling point temperature and enthalpy changes during evaporation (or condensation). Methods for doing
this are now available from Thome (1995) and are presented in the next section. The additional
complexity is not that great since the oil remains very predominately in the liquid phase (the vapor
pressures of lubricating oils are on the order of one-millionth that of refrigerants) and hence its
composition in the vapor phase is negligible and happily does not have to be predicted. Consequently,
dealing with a refrigerant-oil mixture is in this context easier than modeling zeotropic refrigerant
mixtures, whose vapor-phase compositions change as the zeotrope evaporates along an evaporator tube.

15.3 Thermodynamics of Refrigerant-Oil Mixtures

15.3.1. Definition of Boiling Heat Transfer Coefficient
To include oil effects correctly in measuring and modeling flow boiling coefficients, the local bubble
point temperature and the temperature-enthalpy-vapor quality (T-h-x) relationship have to be calculated.
They are then used in determining the local temperature differences and energy balances, and hence the
local vapor quality, analogous to how these concepts are applied to zeotropic refrigerant mixtures. Adding
miscible oil to a refrigerant increases the bubble point temperature at which the fluid evaporates.
The boiling heat transfer coefficient is defined as

q
Tw Tbub

[15.3.1]

where q is the heat flux through the wall to the evaporating fluid, Tw is the wall temperature and Tbub is
the bubble point temperature of the fluid. Since the value of Tbub for a refrigerant-oil mixture is higher
than Tsat for the pure refrigerant, reducing raw test data in this correct manner (rather than reducing it
using Tsat as has been incorrectly but widely done in the past) results in an increase in the heat transfer
coefficient. On the other hand, use of Tbub rather than Tsat for the pure refrigerant in an incremental log
mean temperature difference (LMTD) calculation for an evaporator will decrease the value of the LMTD,
which partially compensates for the larger boiling coefficient when using Tbub in the definition of . In
any case, it should be remembered that the difference between using Tbub or Tsat typically only becomes
significant when reducing test data at vapor qualities greater than about 80%. Thome (1995) presented a
detailed discussion of the thermodynamic effects on boiling points and enthalpy caused by oil and these
are discussed below.

15.3.2. Bubble Point Temperatures

Takaishi and Oguchi (1987) proposed the following empirical method to predict bubble point
temperatures for a specific lubricating oil mixed with R-22 based on their measurements:

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-6

Engineering Data Book III

ln(psat ) = [A / Tbub] + B

[15.3.2]

where psat is the saturation pressure in MPa and Tbub is the bubble point temperature in K. Rearranging
[15.3.2] in order to calculate the bubble point temperature for a given saturation pressure and oil mass
fraction, it becomes:

Tbub = A /[ln(psat ) + B]

[15.3.3]

The amount of oil in the liquid-phase is given by w, i.e. in mass fraction, not mass percent. A and B are
given by the following empirical expressions:

A = a0 + a1w + a 2w3 + a3w5 + a 4w7

[15.3.4]

B = b0 + b1w + b2w3 + b3w5 + b4w7

[15.3.5]

where the values of the empirical constants are:

ao = -2394.5
a1 = 182.52
a2 = -724.21
a3 = 3868.0
a4 = -5268.9

bo = 8.0736
b1 = -0.72212
b2 = 2.3914
b3 = -13.779
b4 = 17.066

Evaluating [15.3.3] at a saturation pressure of 0.55 MPa (79.8 psia), Table 15.1 depicts the effect of oil
mass fraction on the bubble point temperature for R-22 mixed with various mass fractions of oil. The
right column shows the difference between the bubble point temperatures of the mixtures and the
saturation temperature of pure R-22 (TR-22) at this pressure.
Table 15.1. Bubble point temperatures of R22/oil mixtures at 0.55 MPa (79.8 psia).
w (mass fraction)
Tbub (C)
0.00
2.99
0.01
3.01
0.02
3.03
0.03
3.04
0.04
3.06
0.05
3.09
0.06
3.11
0.07
3.13
0.08
3.15
[] - Extrapolated value.

Tbub-TR-22 (C)
0.00
0.02
0.04
0.05
0.07
0.10
0.12
0.14
0.16

w (mass fraction)
0.09
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80

Tbub (C)
3.17
3.19
3.44
3.79
4.31
5.25
7.22
11.53
[19.95]

Tbub-TR-22 (C)
0.18
0.20
0.45
0.80
1.32
2.26
4.23
8.54
[16.96]

Thome (1995) generalized the above method for application to refrigerants other than R-22 and also
temperatures outside the original experimental range. This was accomplished by replacing the original
values of ao and bo that are specifically for R-22 with new ones for the desired refrigerant, such as R-134a.
Since the vapor pressure of oil is very small compared to that of the refrigerant, it was demonstrated that
the effect of the specific type of oil on the empirical constants a1 to a4 and b1 to b4 is negligible for oil

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-7

Engineering Data Book III

fractions below 0.50 of oil (50% mass). Rather than using fixed values of ao and bo for a particular pure
refrigerant, an accurate equation of state for the pure refrigerant vapor pressure curve is used to determine
the values of ao and bo at the specified saturation pressure instead of the simple Antoine expression, i.e.
[15.3.2]. Using these new values of ao and bo, [15.3.2] and [15.3.3] are now accurate for predicting
saturation temperatures and bubble point temperatures over narrow temperature ranges for essentially any
miscible mixture of commonly used refrigerants and oil. Thome (1997b) [also see Zrcher, Thome and
Favrat (1998b)] has also applied this method to zeotropic refrigerant blends mixed with miscible oils (in
particular R-407C/oil mixtures) and showed that it accurately matched experimental measurements over a
wide range of oil mass fractions.
The procedure to use the expression [15.3.3] for any refrigerant is thus:

Determine the pure refrigerant saturation temperature and pressure just above and just below the
design pressure with an accurate equation-of-state of the user's choice;
Use these two sets of values for Tbub and psat to solve for ao and bo in [15.3.3] with w set to zero. The
solution is straightforward with two equations and two unknowns.
These new values of ao and bo replace those of ao = -2394.5 and bo = 8.0736 for R-22, respectively.
All the other values of a1 to a4 and b1 to b4 remain the same since they only refer to the effect of the
oil on Tbub.
Equations [15.3.3], [15.3.4] and [15.3.5] are then evaluated for the desired oil mass fraction, w, to
obtain Tbub for the desired value of psat and w; instead, psat can be obtained for specified values of Tbub
and w using [15.3.2], [15.3.4] and [15.3.5].

15.3.3 Local Oil Mass Fractions

Preparation of a T-h-x (temperature-enthalpy-vapor quality) curve and determination of local
thermodynamic and transport properties of refrigerant-oil mixtures require knowledge of the local oil
mass fraction in the liquid at locations along the evaporator tube. From a thermodynamic viewpoint, the
ideal evaporation process is infinitesimally slow such that thermodynamic equilibrium exists and
consequently the liquid phase has a uniform oil mass fraction and temperature at any cross-sectional
location along the tube. Enthalpy curves are prepared assuming that thermodynamic equilibrium exists
throughout the process, exactly the same as assumed for evaporating a pure fluid.
The liquid-phase mass fraction of the oil circulating in a refrigeration system is a function of its location
in the system. Thus, to unequivocally define this mass fraction, a point is chosen where all the circulating
fluid is in the liquid-phase, which occurs in the refrigerant line between the exit of the condenser and the
entrance into the expansion valve or expansion device. The oil mass fraction at this location is defined as
winlet. When the local vapor quality x is 0.0, the local oil mass fraction w is equal to winlet. If all the
refrigerant could be evaporated out of the liquid-phase, then the local vapor quality x would be equal to
(1-winlet) at that point since all of the refrigerant would be in the vapor-phase and all the oil in the liquidphase; w would be equal to 1.0 at this point since only oil remains in the liquid-phase. Hence, the
maximum exit vapor quality is equal to (1-winlet). This is an important limiting point to remember in
designing a system with say 3 (% mass) oil, since the exit vapor quality from the evaporator has to be less
than 0.97. Also, for refrigerant-oil flow boiling experiments the maximum outlet vapor quality must be
less than (1-winlet).
From a design viewpoint, the local oil mass fraction as a function of vapor quality is required. The
expression that relates the local oil mass fraction, w, to the local vapor quality, x, and the inlet oil mass
fraction winlet is obtained from a conservation of mass and is a simple expression:

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-8

Engineering Data Book III

w = winlet /(1 x)

[15.3.6]

As an illustration, assume that 100 g of refrigerant-oil mixture enters the expansion valve with an inlet oil
mass fraction, winlet, equal to 0.05 (thus, there are 5 g of oil and 95 g of refrigerant):
1. Before the expansion valve the local vapor quality is 0.0 and w = winlet = 0.05 as indicated by [15.3.6];
2. For a local vapor quality equal to 0.20 immediately after the expansion valve, w is 0.0625 [i.e.
0.05/(1-0.20)];
3. For a local vapor quality of 0.90, w is 0.50 or 50 mass % oil, i.e. [0.05/(1-0.90)];
4. If the entire refrigerant was evaporated from the liquid mixture, the local vapor quality would be (1winlet) or 0.95 and [15.3.6] appropriately predicts that w for this situation is 1.0 [i.e. 0.05/(1-0.95)].

15.3.4 Enthalpy Curves

The change in enthalpy dh of a mixture during evaporation is referred to as the enthalpy curve. The
change in dh with respect to hL for completely saturated liquid, i.e. with respect to h at x = 0, as the
bubble point temperature rises is comprised of three contributions:

The latent heat to the fraction of liquid vaporized, x;

The sensible heat to the fraction of fluid in the liquid phase (1-x);
The sensible heat to the fraction of fluid in the vapor phase x.

In mathematical terms for a refrigerant-oil mixture this is:

dh = h LGx + (1 x)dTbubcpL + xdTbubcpG

[15.3.7]

where x is the local vapor quality, hLG is the latent heat of vaporization of the pure refrigerant since no oil
enters the vapor-phase, cpL is the specific heat of the liquid-phase refrigerant-oil mixture and cpG is the
specific heat of the pure refrigerant vapor. The values of hLG and cpG are obtained from equations for the
pure refrigerant at the local saturation temperature while cpL is a function of the local oil composition and
bubble point temperature. Equation [15.3.7] reduces to only the latent heat for a pure refrigerant without
oil.
A heat release curve is not actually implemented as a curve. Instead, it is prepared as a table of values at
set intervals of temperature or vapor quality that gives the amount of heat absorbed by the fluid per unit
mass (i.e. dh is in J/kg or Btu/lb) relative to its inlet state together with the bubble point temperature and
vapor qualities that correspond to these points.
At a standard evaporation temperature of 4.44C (40F) in a refrigeration system, for pure R-134a this
corresponds to a saturation pressure of 3.43 bar (49.7 psia) absolute. For a mixture with 3 (% mass) oil in
R-134a at this saturation pressure, Table 15.2 lists the enthalpy curve in tabular form at this pressure
covering the vapor quality range from 0.15 to 0.95 utilizing the method of Thome (1995). The oil is
Mobil Arctic EAL 68 whose density is 971 kg/m3 (60.7 lb/ft3) at 15.56C (60F) {Note: the oil's density is
utilized to predict its liquid specific heat, cf. Thome (1995) and thus that of the refrigerant-oil mixture}.
The first column in Table 15.2 shows the local vapor quality intervals along the evaporator tube. The next
two columns list Tbub and w. The total heat absorbed relative to the inlet condition is shown in the 4th
column (in kJ/kg). The contributions of the heat absorbed as latent heat dhlatent and as sensible heating of
the liquid and vapor dhsensible are given in the last two columns (in kJ/kg), respectively. As can be seen, the
rise in the bubble point temperature and oil mass fraction in the liquid is most rapid at high vapor

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-9

Engineering Data Book III

qualities. The contribution of sensible heat is important at high vapor qualities since its effect is directly
dependent on the rise in Tbub.
Table 15.2. An enthalpy curve for R-134a/3 (% mass) oil mixture at 3.43 bar.
x
0.150
0.230
0.310
0.390
0.470
0.550
0.630
0.710
0.790
0.870
0.950

Tbub (C)
4.509
4.516
4.525
4.536
4.550
4.570
4.598
4.643
4.729
4.954
8.289

w
3.53
3.90
4.35
4.92
5.66
6.67
8.11
10.34
14.29
23.08
60.00

dhtotal (kJ/kg)
0.00
15.64
31.28
46.92
62.56
78.20
93.85
109.52
125.23
141.05
159.76

dhlatent (kJ/kg)
0.00
15.63
31.26
46.29
62.51
78.13
93.75
109.37
124.99
140.59
156.07

dhsensible (kJ/kg)
0.00
0.01
0.02
0.03
0.05
0.07
0.10
0.15
0.24
0.46
3.69

Table 15.2 also illustrates the influence of oil on Tbub with respect to the pure refrigerant (R-134a)
saturation temperature Tsat of 4.44C (40F). For x < 0.50, the increase in boiling point tends to be 0.1C
(0.2F) or less, for 0.50 < x < 0.80, the rise increases up to about 0.3C (0.54F), and for x > 0.80 the
elevation of the boiling point can become very significant (3.845C or 6.921F at x = 0.95 in Table
15.2!). The temperature difference between Tbub and Tsat depends on the local vapor quality and the inlet
oil mass fraction.
For refrigerant-oil mixtures, it is preferable to use set intervals of vapor quality for determining the points
on a heat release curve rather than set intervals of temperature. If an evaporator is to be designed with an
inlet vapor quality after the expansion valve of 0.20 and an outlet vapor quality of 0.90, then the heat
absorbed and bubble point temperatures might be determined at vapor qualities of 0.20, 0.30, 0.40, 0.50,
0.60, 0.70, 0.80 and 0.90. The heat absorbed at x = 0.20 would be zero and would serve as the datum for
the other values. In practice, the number of intervals to use is up to the designer and should be sufficient
to accurately represent the heat release curve and also the variation in the two-phase heat transfer
coefficient as a function of vapor quality. Normally eight to ten intervals are recommended, but more can
be used as desired.
For refrigerant-oil mixtures, the following procedure is used for preparation of a heat release curve at a
fixed, isobaric pressure:
1. Select the desired pressure psat and inlet oil mass fraction before the expansion valve, winlet.
2. Determine two sets of pure refrigerant saturation temperature and pressure just above and just below
psat with an accurate equation-of-state;
3. Use these two sets of values for Tbub and psat to solve for ao and bo with w set to zero from the two
equations and two unknowns.
4. The new values of ao and bo are used in place of ao = -2394.5 and bo = 8.0736, respectively. All the
other values of a1 to a4 and b1 to b4 remain the same.

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-10

Engineering Data Book III

5. Choose the inlet vapor quality, the exit vapor quality (always less than the value of 1-winlet), and the
number of design intervals desired (for example, if the evaporator is divided into ten heat transfer
zones, then 11 points are needed in the heat release curve).
6. Calculate the local oil fractions corresponding to the local vapor qualities using [15.3.6] (for example,
determine w for each of the eleven vapor quality points above).
7. Calculate the local bubble point temperatures corresponding to the local vapor qualities and local oil
mass fractions at each point.
8. Calculate the values of hLG and cpG at these bubble point temperatures using methods for pure
refrigerants. Determine the liquid specific heat of the refrigerant-oil mixture cpL for each point based
on the local values of w and Tbub using [15.4.1] and [15.4.2].
9. Determine the heat absorbed by the refrigerant-oil fluid in each interval, dh, and then add these up to
obtain the total heat absorbed per unit mass of fluid from inlet to outlet.
The heat release curve preparation is now complete and the following values required for thermal design
of the evaporator are known at each local vapor quality x: w, Tbub and dh.

15.4 Liquid Specific Heats of Oils and Refrigerant-Oil

Mixtures
15.4.1 Liquid Specific Heat of Lubricating Oils
Liquid specific heats of lubricating oils are often not available from their manufacturers; hence a general
correlating equation is required for their estimation. Based on a literature search, Thome (1995)
recommended the following equation cited by Liley and Gambill (1973) for predicting liquid specific
heats of petroleum oils, specified to be accurate to within 5%:

(cpL )oil = 4.186{[0.388+ 0.00045(1.8T + 32)]/ s1/ 2}

[15.4.1]

where the liquid specific heat (cpL)oil is in units of kJ/kgC, the temperature of the oil, T, must be in C
(valid for -18 < T < 204C) and s is the liquid specific gravity at 15.56C (60F) and is valid for 0.75 < s
< 0.96. Literature furnished by manufacturers of lubricating oils usually provides the density or specific
gravity of the oil at a temperature of 60F (15.56C). Thus, application of [15.4.1] for general use for
lubricating oils presents no problem. In preparation of a heat release curve, the temperature, T, in [15.4.1]
is the bubble point temperature Tbub for the local vapor quality in question.
The above correlation compares well with tabular values for engine lubricating oil in Incropera and
DeWitt (1981), giving maximum deviations of -2% and +7%. This equation has also been doublechecked against numerous heavy hydrocarbons and viscous commercial heat transfer fluids with good
results over the temperature range of interest for lubricating oils. Thus it should be sufficiently accurate
since liquid sensible heat normally only contributes a small fraction of the total heat absorbed by an
evaporating refrigerant-oil mixture.
Some lubricating oils have a specific gravity as high as 1.05, i.e. not very far outside the range cited by
Liley and Gambill, and use of [15.4.1] is still recommended in the absence of values from the

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-11

Engineering Data Book III

manufacturer. As an example calculation, consider an oil at 4.44C (40F) with a specific gravity of 0.890
at 15.56C (60F). Equation [15.4.1] predicts the liquid specific heat to be 1.80 kJ/kgC (0.43 Btu/lbF).
In comparison, the values for R-134a and R-22 at 4.44C (40F) are 1.354 and 1.187 kJ/kgC,
respectively (0.323 and 0.283 Btu/lbF).

15.4.2 Liquid Specific Heat of Refrigerant-Oil Mixtures

For preparation of a heat release curve, the liquid specific heat of the refrigerant-oil mixture is required at
each local vapor quality as a function of oil fraction, w, and bubble point temperature, Tbub. In the absence
of experimental data, the only practical approach for predicting specific heats of liquid mixtures is to
weigh the single-component values by their mass fractions, according to Liley and Gambill (1973). Thus,
heats of mixing are ignored, unless specific information is available, and ideal mixing is assumed.
Sophisticated methods in the literature involving corresponding states cannot be applied because the
molecular weight of the oil and its critical properties are not known.
Working in the units of kJ/kgK (Btu/lbF), a linear mixing law based on mass fractions of liquid phase is
used. For liquid specific heats the mixture expression is

( )

( )

cpL = w cpL oil + (1 w ) cpL ref

[15.4.2]

The above equation is evaluated with the refrigerant and oil properties at the local bubble point
temperature of the mixture. For example for 30 mass % oil (w = 0.30), if the specific heat at 4.44C for
the oil is 1.80 kJ/kgK while that of R-22 is 1.187 kJ/kgK, then the liquid specific heat of the mixture is
1.371 kJ/kgK [= 0.3 (1.80) + (1.0-0.3)(1.187)].

15.5 Example of Application of Thermodynamic Approach

Using the new Thermodynamic Approach, Table 15.3 shows a temperature-enthalpy curve in tabular
form calculated for R-134a mixed with 5 mass % oil. The oil is Mobil Arctic EAL 68 whose properties
are: specific gravity of 0.971 at 15C, viscosities of 62.5 and 8.1 mm2/s at 40C and 100C, respectively,
and assumed molecular weight of 590) at the isobar of 2.93 bar (where Tsat = 0.0C for pure R-134a at
this pressure). Column one gives the local vapor qualities used in the calculation while columns 2 and 3
list the calculated bubble point temperatures and local oil mass fractions, w. The total heat absorbed by
the fluid dhtotal (in kJ/kg of flowing fluid) is given in column 4 while the next two columns show the
individual contributions of latent heat dhlatent and sensible heat dhsensible.
As can be observed, the bubble point temperature at first increases slowly relative to Tsat of pure R-134a
and then rises rapidly at vapor qualities above 0.80. For x < 0.90, the contribution of sensible heat is less
than 1% of the total but its influence rises rapidly afterwards; thus for test data for x > 0.90, the reported
vapor qualities become ever too high. For example, if 181.41 kJ/kg of heat is added to the refrigerant-oil
mixture, its local vapor quality should be 0.902 according to Table 15.3; using only the latent heat and
neglecting sensible heat the vapor quality would be calculated to be 0.9114, i.e. 0.0112 too high. Thus, the
error in reported vapor qualities is not very significant except at vapor qualities approaching 1.0. It should
be pointed out however that many data reported in the literature incorrectly use a vapor quality based on
the local mass of refrigerant vapor divided by the total mass of refrigerant, i.e. ignoring the presence of
the oil, while instead the thermodynamic vapor quality x is the local mass of vapor divided the total mass
of refrigerant and oil. For the highest vapor quality shown in the table, x = 0.93, the bubble point
temperature is predicted to be 8.015C (14.4F) above the pure refrigerant value.

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-12

Engineering Data Book III

Table 15.3. R134a-oil temperature-enthalpy curve at 2.93 bar (42.5 psia) for 5 mass % oil at inlet.
dhtotal
dhlatent
(kJ/kg)
(kJ/kg)
0.000
0.076
5.00
0.00
0.00
0.093
0.084
5.51
18.53
18.52
0.186
0.094
6.14
37.06
37.04
0.279
0.106
6.93
55.60
55.56
0.372
0.123
7.96
74.14
74.08
0.465
0.146
9.35
92.68
92.59
0.558
0.180
11.31
111.23
111.10
0.651
0.236
14.33
129.80
129.63
0.744
0.346
19.53
148.42
148.11
0.837
0.663
30.67
167.22
166.59
0.846
0.729
32.47
169.14
168.44
0.856
0.819
34.72
171.08
170.29
0.865
0.922
37.04
173.03
172.14
0.874
1.056
39.68
175.00
173.98
0.883
1.240
42.74
177.02
175.82
0.893
1.541
46.73
179.15
177.66
0.902
1.974
51.02
181.41
179.50
0.911
2.726
56.18
183.93
181.33
0.921
4.435
63.29
187.43
183.15
0.930
8.015
71.43
192.70
184.95
(% errors determined assuming a heat flux of 10,000 W/m2).
x

Tbub (C)

w (%)

dhsensible
(kJ/kg)
0.00
0.01
0.02
0.04
0.06
0.06
0.13
0.16
0.31
0.63
0.70
0.79
0.89
1.02
1.20
1.49
1.91
2.63
4.28
7.75

5C
%Error
-1.5
-1.7
-1.9
-2.1
-2.5
-2.9
-3.6
-4.7
-6.9
-13.3
-14.6
-16.4
-19.4
-21.1
-24.8
-30.8
-39.5
-54.5
-

2C
%Error
-3.8
-4.2
-4.7
-5.3
-6.2
-7.3
-9.0
-11.8
-17.3
-33.2
-36.5
-41.0
-46.1
-52.8
-62.0
-

In the last two columns in Table 15.3, the effect of erroneously using Tsat in place of Tbub in converting
"raw" data to experimental heat transfer coefficients is shown for wall superheats of 5C (9F) and 2C
(3.6F). It is evident that existing published data using Tsat are always too low compared to their more
correct values. For instance, at a vapor quality of 0.837 and a wall superheat of 5C (9F), published
values would be 13.3% lower than the thermodynamically correct values.
For a plain tube the effect of the oil on flow boiling heat transfer becomes significant (> 5%) at a local oil
mass fraction greater than 15% while for a micro fin tube it becomes significant at w > 6.5%. At high
vapor qualities and oil mass fractions, the effect becomes very significant, substantially changing the
trends reported in the literature. Published test data reduced with the oil contamination method that fall
in these ranges should be used with caution or better yet, the heat transfer coefficients should be
recalculated.

15.6 Illustration of Physical Properties of Refrigerant-Oil

Mixtures
Thome (1993) put together a comprehensive set of prediction methods that may be the best currently
available (but would really benefit from further research). Table 15.4 shows the calculated physical
properties that correspond to the enthalpy curve in Table 15.3 at several vapor qualities using his
methods. Note that since the vapor pressures of oils are about one-ten millionth those of refrigerants, no
oil enters the vapor-phase and hence vapor properties are determined for the pure refrigerant at a
saturation temperature corresponding to the local bubble point temperature. For the present conditions,
the physical properties that demonstrate a significant change from inlet to outlet (> 5%) are: liquid density

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-13

Engineering Data Book III

(-17.2%), liquid viscosity (+9003%), liquid specific heat (+23.6%), liquid thermal conductivity (+11.7%),
critical pressure (-58.1%), critical temperature (+103%) and liquid Prandtl number (+10,021%). Also
shown is the ratio of the liquid Reynolds number relative to than for the inlet at x = 0.0, which decreases
drastically as the liquid viscosity rises. Thus, the variation is local physical properties can potentially have
a significant effect on the local heat transfer coefficients, both for nucleate pool boiling and for flow
boiling.
Table 15.4. R134a-oil properties at 2.93 bar for 5 mass % oil at inlet over a range of vapor qualities.
Local Vapor Quality
Oil Mass Fraction %
Liquid Molecular Weight
Bubble Point Temp. (C)
Heat Absorbed (kJ/kg)
Liquid Density (kg/m3)
Vapor Density (kg/m3)
Liquid Viscosity (cp)
Vapor Viscosity (cp)
Liquid Sp. Heat (kJ/kg K)
Vapor Sp. Heat (kJ/kg K)
Liquid Th. Cond. (W/m K)
Vapor Th. Cond. (W/m K)
Surface Tension (dyne/cm)
Latent Heat (kJ/kg)
Critical Pressure (bar)
Critical Temperature (K)
Liquid Prandtl Number
Relative Liquid Re Number

0.000
5.00
126.4
0.076
0.00
1276
14.5
0.398
0.011
136.3
0.889
0.094
0.012
11.7
199.1
38.9
405.7
5.75
1.000

0.279
6.93
135.9
0.106
55.6
1270
14.5
0.458
0.011
1.372
0.889
0.095
0.012
11.7
191.1
38.2
417.9
6.64
0.869

0.558
11.31
157.2
0.180
37.1
1256
14.5
0.628
0.011
1.391
0.889
0.095
0.012
11.7
1.991
36.7
445.5
9.20
0.634

0.744
19.53
197.3
0.346
148.4
1230
14.6
1.137
0.011
1.428
0.889
0.096
0.012
11.8
198.9
33.9
497.4
16.9
0.350

0.930
71.43
450.6
8.015
192.6
1057
16.8
35.83
0.011
1.684
0.906
0.105
0.013
17.2
195.5
16.3
824.7
576.2
0.011

15.7 Online Measurement of Refrigerant-Oil Mass Fractions

The Thermodynamic Approach allows the local bubble point temperature, local oil mass fraction, local
heat transfer coefficient, change in enthalpy and (potentially) local physical properties to be calculated for
an evaporator from knowledge of winlet for any refrigerant and lubricating oil mixture, as long as they are
miscible and the local oil mass fraction is below 70% oil. To implement the Thermodynamic Approach,
an accurate method is required for measurement of the oil composition. For this purpose, a new online
measurement system was developed and described in Bayini, Thome and Favrat (1995).

15.7.1 Previous Test Methods

Over the years, oil mass fractions circulating in test facilities and equipment qualification facilities have
been inferred from: (i) measuring the masses of refrigerant and oil charged into the system, (ii) measuring
the separate oil and refrigerant flow rates leaving an oil separator, or (iii) by withdrawing liquid samples
during tests. The 1st approach is often no able to be implemented. The 2nd method is not necessarily
accurate since an oil separator actually acts as a liquid-vapor separator and thus some refrigerant remains
in the oil leaving the separator as liquid but gets counted as if it were oil. The 3rd method uses ASHRAE
Standard 41.4-1994 to determine oil compositions for liquid samples withdrawn from the flow circuit.
Each sample is weighed together with its bottle and then the refrigerant is allowed to slowly evaporate off
through a valve. To be accurate, the sample must also be evacuated with a vacuum pump and heated up to

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-14

Engineering Data Book III

150C (302F) to drive off the refrigerant, then weighed again with a top loading balance accurate to 0.01
g or better. The oil composition can then be calculated from the initial and final weights. While this
method is appropriate as a standard, it is not very convenient for practice. Shortcuts to the standard
method, i.e. dispensing with the heating and/or evacuation procedures, result in overstating the oil
concentration by as much as 1.5 mass % oil or more. Consequently, it is felt that an online method is
imperative in order to measure the oil compositions conveniently and accurately over the range of test
conditions to be encountered.
Various methods for determining oil compositions online have been proposed and several have been
tested. Baustian, Pate and Bergles (1988a, 1988b, 1988c) have used a Coriolis density meter, an acoustic
velocity sensor and a viscosity meter, respectively, to measure oil concentrations with accuracies in the
range of 1-2 mass % oil. The Coriolis meter used in their tests had a density measurement accuracy of 10
kg/m3 (0.6 lb/ft3) and the effect of compressibility on the subcooled liquid passing through the meter was
not accounted for; however, newer more accurate Coriolis meters are now available and may suit
industrial use. Meyer and Saiz Jabardo (1994) built an ultrasonic acoustic velocity device and accurately
calibrated it for measuring concentrations for R-12 and R-134a with various oils (95% confidence level
accuracy of 0.23-0.26 mass % oil). Suzuki, Fujisawa, Nakarazawa and Matsuoka (1993) instead have
used a light absorption method for measuring oil concentrations online.

15.7.2 Online Density Flowmeter

A very high accuracy, commercially available density flowmeter was used Bayini, Thome and Favrat
(1995) as the measuring device, a model available from Schlumberger (Liquid Density Transducer type
7847 in stainless steel). This method will be described in some detail here. The meter is a vibrating tube
type, where the tube is a 25.4 mm diameter, stainless steel corrugated tube about 1.0 meter long. The
meter has a calibrated accuracy of 0.1 kg/m3 (0.006 lb/ft3) over a wide range of density and temperature,
including compensating equations to account for nonlinear temperature effects on the instrument's
operation. A digital display gives the density and temperature and a digital signal for the density, which
was read into our data acquisition system. It operates with subcooled refrigerant-oil flowing through the
meter, such than measurements are continuous and no bypass circuit is required.
The density flowmeter is used to measure the density of subcooled liquid, either pure refrigerant or
refrigerant-oil mixture, after it leaves the flow loops condenser. To avoid cavitation in the vibrating
corrugated tube (apparently caused by its high vibration frequency), a minimum of 8C (14.4F) of
subcooling was required at its outlet. A further consideration is the sensitivity of the density flowmeter.
The density difference between pure R-134a and the oil is approximately 300 kg/m3 at the test conditions.
Thus, in order to determine oil concentrations accurate to 0.1 mass % oil, one must measure density to an
accuracy of 0.30 kg/m3 (0.019 lb/ft3) including temperature and liquid compressibility effects.

15.7.3 Oil Concentration Calibration Correlation

Measured liquid densities are not only a function of temperature and oil concentration but also liquid
compressibility. For example, the density of compressed R-134a increases by about 0.3-0.4 kg/m3 (0.0190.025 lb/ft3) for each increase of 1 bar (14.5 psi) with respect to the saturation pressure at the measured
temperature. The calibration should thus take the liquid compressibility into account, which for mixtures
in the range from 0-6 mass % oil can reasonably be assumed to match that of the pure refrigerant.
Compressed liquid densities for R-134a were reported for instance by Hou et al. (1992) over a wide range
of temperature and pressure. They fit the Tait equation to their data with separate empirical parameters for
each isotherm they tested. Refer to Bayini, Thome and Favrat (1995) for further details on this and the
calibration procedure.

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-15

Engineering Data Book III

Their calibration correlation had an average error of 0.09 mass % oil for all 246 data points measured for
the oil concentrations of 0.0, 0.5, 1.0, 2.0, 3.0, 4.0, 5.0 and 6.0 mass %, a maximum error of 0.69 mass %
oil and a 95% confidence limit of 0.21 mass % oil.

15.7.4 Industrial Application of a Density Flowmeter

As a shortcut "industrial" method, the saturation density of the refrigerant-oil mixture can be determined
using a linear mixing law of the specific volume. The refrigerant-oil saturation density is defined by this
ideal mixing law as:

= oil /[1 + (1 w)(oil ref 1)]

[15.6.1]

Solving this expression to obtain the oil mass fraction w (in mass fraction, not mass %) gives

w = [(oil )(ref )]/(ref oil )

[15.6.2]

The refrigerant saturation density ref in this equation is calculated from an accurate equation for the
refrigerant (such as ASHRAE methods). The lubricating oil density oil can either be measured as a
function of temperature or obtained from data supplied by the manufacturer. Otherwise, using a method
adapted from Thome (1992), it can be estimated from the following equation for lubricating oils:

oil = man[(Tcrit T) /(Tcrit (Tman + 273.15))]

0.29

[15.6.3]

where man is the oil density (kg/m3) at the temperature provided by the manufacturer Tman (normally
15C), Tcrit is the critical temperature of the oil (assume equal to 760 K for all oils as an approximation),
and T is the measured temperature (K). Subtracting the density difference due to compressibility from the
measured density and using the pure refrigerant and oil densities determined as discussed, the oil
composition is determined with [15.6.2]. This simplified method has been successfully compared to R134a/oil density test data of Bayini, Thome and Favrat (1995) with a mean error of 0.22 mass % oil.
Because of nonlinear variations in density with composition, the accuracy varies from about 0.1 mass %
mean error for the 0.5 mass % mixture up to about 0.5 mass % mean error for the 6.0 mass % mixture,
which is more than satisfactory for most applications.
To get the best results, the meter should be calibrated against the pure refrigerant first, i.e. measure the
pure refrigerant's density with the meter, correct this value for liquid compressibility and then compare it
against the calculated saturated refrigerant density.
------------------------------------------------------------------------------------------------------------------------------Example Calculation: Assume a density meter is used to measure oil mass fraction of a slightly
subcooled mixture of R-134a and oil at the conditions of 0C and 2.93 bar. R-134a has a density of 1276
kg/m3 at this temperature while the oil has a density of 900 kg/m3 at the 15C according to the
manufacturer. If the density measured is 1270 kg/m3, what is the oil mass fraction?
Solution: The density of the oil at 0C (273.15 K) is obtained with [15.6.3]:

oil = man[(Tcrit T) /(Tcrit (Tman + 273.15))]

0.29

oil = 900[(760 273.15) /(760 (15 + 273.15))]

0.29

= 908.2 kg / m3

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-16

Engineering Data Book III

The oil mass fraction w (in mass fraction) is obtained with [15.6.2]:

w = [(oil )(ref )] /(ref oil )

w = [(908.2 1270)(12761270)] /(1276 908.2) = 0.0117

Thus, the oil mass fraction is 0.0117 or 1.17 % oil.

Thermodynamics of Refrigerant Mixtures and Refrigerant-Oil Mixtures

15-17