DOCTORAL THES I S

Lule University of Technology

Department of Applied Physics and Mechanical Engineering
Division of Machine Elements
:ooo:;o|issx: +o:-+|isrx: i+u-i+ -- oo ;o -- sr
:ooo:;o
Properties of Oil and Refrigerant Mixtures
- Lubrication of ball bearings in refrigeration compressors
Roger Tuomas
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration
compressors
Roger Tuomas
Machine Elements
Dept. of Applied Physics and Mechanical Engineering
Lule University of Technology
Preface
This thesis is based upon five papers concerning lubrication with oil and refrigerant
mixtures used as lubricants in rolling element bearings in refrigeration compressors.
The presented research was carried out at Lule University of Technology at the
Division of Machine Elements.
I would like to thank my supervisor, Dr. Ove Isaksson, for his help, discussions and
encouragement. I would also like to thank Dr. Ulf Jonsson for introducing me to the
Division and the field of lubrication of refrigeration compressors and for the many
interesting discussions and valuable support both within and outside our field of
research.
Special thanks go to the participating companies SKF, Trane Company, CPI
Engineering, York Refrigeration and Carrier, for their engagement and financial
support.
Thanks also to the Swedish Energy Agency for financial support.
Finally, I would like to thank all my other colleagues at the Division of Machine
Elements, my family, Anna, Cajsa and Johannes for their encouragement and
support.
Abstract
A refrigeration compressor contains highly loaded mechanical contacts that have to
be lubricated. The lubricant used in refrigeration compressors consists of an oil and
refrigerant mixture, and the refrigerant concentration can be up to 40 wt%. However,
the oil in the refrigeration system is a contaminant that lowers the efficiency of the
refrigeration system. Therefore, it is important that the oil and the refrigerant are
miscible with each other so that oil following the refrigerant out in the system is
returned to the compressor and not accumulated in the system. HFC refrigerants are
not miscible with the mineral oils used in CFC refrigeration systems. Hence, new
synthetic ester lubricants were developed for this application. It was soon evident,
though, that the lack of chlorine and that the new ester oils did not have the same
good lubrication properties as mineral oil and CFC refrigerants.
The objective of this thesis is to increase the understanding of lubricating rolling
element bearings in refrigeration compressors. The thesis will also give
recommendations for lubrication of bearing in refrigeration compressors based on
the results.
The thesis is based on an experimental study of parameters that are important for life
of rolling element bearings. The thesis consists of five papers that describe a test
apparatus designed to measure and monitor the lubrication status during operation
with capacitance. The effect of two phosphate additives in refrigeration oil is also
examined and properties such as viscosity, pressure-viscosity coefficient,
compressibility and shear strength coefficient of oil and refrigerant mixtures are
measured.
The general conclusion of the thesis is that the refrigerant affects all of the measured
properties negatively on the subject of lubrication and life of the rolling element
bearing. The usage of oil-free compressors solves many of the problems that
lubricants consisting of a mixture of oil and refrigerant cause. If the oil has to be in
the system it is of great importance that the lubrication issues are taken care of early
in the design process. The design parameters for temperature, viscosity, pressure-
viscosity coefficient, contamination etc. must be followed when the compressor is in
use.
Table of content
1 Refrigeration, compressors and lubricants.................................................. 1
1.1 About this thesis ........................................................................................................4
2 Bearing design and lubricant properties...................................................... 5
2.1 Life of rolling element bearings ................................................................................5
2.2 The refrigerant effect on lubricant properties influencing bearing life ......................7
2.3 Bearing test apparatus with on-line monitoring of lubricating film status.................8
2.4 Refrigeration oils with additives................................................................................9
2.5 Viscosity and pressure-viscosity coefficient............................................................10
2.6 Compressibility and density ....................................................................................10
2.7 Shear strength of the lubricant at EHL pressure ......................................................11
3 Research facility ........................................................................................... 13
3.1 Bearing test apparatus..............................................................................................13
3.2 Surface roughness measurements and analysis........................................................17
3.3 Additives in refrigeration oils..................................................................................18
3.4 Viscosity and pressure-viscosity coefficient measurement......................................19
3.5 High-pressure chamber............................................................................................21
3.6 Traction measurements............................................................................................22
4 Results........................................................................................................... 25
4.1 Bearing test apparatus..............................................................................................25
4.2 Dielectric constant measurement .............................................................................29
4.3 Measurement of vibrations ......................................................................................30
4.4 Polyolester oils with additives.................................................................................32
4.5 Refrigerant effect on viscosity and pressure-viscosity coefficient...........................36
4.6 The effect of refrigerant on lubricant density ..........................................................37
4.7 The refrigerant effect on lubricant shear strength. ...................................................39
5 Concluding remarks ..................................................................................... 41
6 Recommendations for further work ............................................................ 45
7 References..................................................................................................... 47
Paper A.....................................................................................................................51
Paper B.....................................................................................................................71
Paper C.....................................................................................................................89
Paper D.....................................................................................................................99
Paper E...................................................................................................................111
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
1
1 Refrigeration, compressors and lubricants
Jacob Perkins invented mechanical refrigeration in 1834 [1], using ether as
refrigerant in a vapour compression cycle. The basic principle of the technique is
still used in modern refrigeration systems, Figure 1. Compressors at this time used
the same principle as the steam piston engine, i.e. steam was produced in large
volumes and the loss of steam was negligible. In the refrigeration machine, the
engine was used to compress and pump the refrigeration gas in the circuit.
Figure 1 Refrigeration cycle.
However, the leakage now became essential when using the steam engine as a
refrigeration compressor, since the filling of refrigeration gas is not done during
operation. To reduce leakage it was necessary to have small clearances. With small
clearances and complicated parts, the production cost of the compressor became
high. The compressors were mainly used in industries like brewing, yeast production
and places with large cooled storage. Journal bearings were used in the first
compressors based on the steam engine. The bearings were lubricated manually by a
lubrication engineer who pressed lubricant into the bearings. The sealing problems
and cost of production forced scientists to develop cheaper and more reliable
refrigeration compressors.
In 1907, the self aligning ball bearing was invented, making it possible to design
machines with narrow clearances. Lars Lysholm at the KTH (Royal Institute of
Technology) in Sweden used the self aligning ball bearing invention to get control
of the clearances in his 1930 invention of the screw compressor. The screw
compressor was patented by SRM (Svenska Rotormaskiner), which refined the
compressor with new patents for new rotor profiles, capacity regulation and the
injection of oil to the bearings and screws.
Mechanical refrigeration had a major drawback when absorption refrigeration was
introduced in household applications. Electrolux started mass production of
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
2
refrigerators using the absorption refrigeration technique, for which the patent
expired in 1950. Since then, many manufacturers have developed absorption
refrigerators. It was not until 1960 when the piston compressor became price
competitive in house hold refrigeration.
Today, a wide range of different types of refrigeration compressors is used, such as
reciprocating piston-, scroll-, screw- and vane-compressor.
In the early days of mechanical refrigeration different types of refrigerants were
used, viz. ammonia, sulphur dioxide, methyl ether, methyl chloride, dichloroethene,
carbon dioxide and a mixture of petrol ether and naphtha called chemogene. Most of
these refrigerants are flammable, toxic or both, and are therefore harmful for humans
in case of leakage. Carbon dioxide is not directly toxic for humans, but the working
pressure is high and demands refrigeration systems to be designed at much higher
pressures.
The invention and introduction of CFC (Chlorofluorocarbon) refrigerants in the
1930s solved the problems with pressure and toxicity and made CFC refrigeration
systems cheap and easy to produce. Research by Molina and Rowland [2] in 1974
was the beginning of the end for CFC refrigerants. Their report showed that chlorine
was a key factor in the destruction of the ozone layer in the upper atmosphere.
Ozone in the atmosphere filters the ultraviolet carcinogenic radiation from the sun.
To meet environmental and customer driven demands changes in refrigeration
technology were needed.
The replacement of CFC refrigerants to those more environmentally friendly was
not easily achieved. The oil and refrigerant should be miscible to assure that the oil
is transported out of the refrigeration circuit and not accumulated in the system. By
the end of the 1980s the first non-chlorinated refrigerant, R-134a, was on the market.
R-134a is an HFC (hydrofluorcarbons) used to replace the chlorinated refrigerant R-
12. R-134a was found to not be miscible with mineral oil and new oils had to be
developed. Nowadays, the phase out of CFCs is almost done all over the world.
The lubricants primary objective in the refrigeration compressor is to lubricate
mechanical contacts such as bearings, mechanical seals, gears and other heavily
loaded contacts. Such contacts often refer to EHL (elastohydrodynamic lubrication),
since surfaces are separated by the lubricant and the pressure in the contact is so
high that the surfaces deform elastically. The lubricant used in a refrigeration
compressor is a mixture of oil that is diluted by the refrigerant in the compressor
cavity or compressor sump. Kruse and Schroeder [3] pointed out several other
functions of the lubricant in the compressor. The lubricant is used to transport heat
and contamination particles out of the compressor. In some applications the
lubricant acts as a seal between the high and low pressure sides in the compressor.
Short [4] showed that the lubricant must have good oxidation resistance, a wide
operating temperature range, good non-foaming properties, be hydrolytically stable
and be compatible with the materials used in the compressor.
However, in the refrigeration circuit the lubricant acts as a contamination that
reduces the system thermodynamic efficiency. The oils vapour pressure determines
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
3
how much oil will go out in the refrigeration cycle. To assure that the oil gets out of
the system, the oil and refrigerant must be miscible with each other and have the
right pour point.
PAG (Polyalkylene glycols) and POE (polyolesters) were the strongest candidates to
replace the traditional mineral oil. Both have good miscibility with the new
alternative refrigerants and possess good electrical insulation that is important in
hermetic compressors where the electrical motor is exposed to the lubricant. The
main disadvantages with the new oils were their decompositions at relatively low
temperatures, Sanvordenker [5]. When converting a refrigeration system with a new
refrigerant and oil, the new oil must be miscible with the old oil in the system. POE
has good miscibility with mineral oils, whereas PAG oils have poor miscibility. POE
oils is also more hydrolytic stable than PAG oils.
Depending on the type of compressor different kinds of bearings and bearing
arrangements are chosen. Rolling element bearings are commonly used in
compressors with small clearances and good linearity. Journal bearings are used
when the cost of the compressor has to be reduced and the demands on clearances
and stability are lower.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
4
1.1 About this thesis
The general conclusion from reviewing the relevant literature is the numerous
experiments and research done related to general lubrication and life of rolling
element bearings. There is still lack of experimental data and research of bearings
operating in refrigeration compressors to improve their design criteria.
The objective of this thesis is to increase the fundamental understanding of rolling
element bearing lubrication with oil and refrigerant mixture as lubricant, where R-
134a mixtures are of special interest. Knowledge of the lubricant properties of the
mixture is vital importance, i.e. viscosity, pressure-viscosity coefficient,
compressibility, additives and shear strength. To obtain experimental data a test
apparatus is developed where lubrication of rolling element bearings can be tested in
similar conditions to a refrigeration compressor.
The bearing test apparatus is designed and evaluated in Paper A. A capacitive
measuring technique is used to indicate changes in film thickness and detect metal-
to-metal contact between the bearing elements and inner- and outer- races during
operation.
Paper B investigates phosphate based additives to improve the lubricity of chlorine
free replacement refrigerants, HFC. In this paper, the ability to achieve a proper
lubricating film thickness and the wear of the bearing surfaces were investigated in
the bearing test apparatus. The base lubricant was POE oil mixed with either a
phosphate ester additive or an acid phosphate additive.
In Paper C a falling ball viscometer was used to measure the viscosity and the
pressure-viscosity coefficient as a function of the amount of dissolved refrigerant.
The experiments were done with a polyolester oil and refrigerants R-22, R-134a, R-
410a and R-32 at refrigerant concentrations from 0 wt% to 30 wt%, at temperatures
of 40 and 80 C and pressures up to 34 MPa.
Lubricant compressibility and density for lubricants containing refrigerant are
studied in Paper D. Experiments are performed in a high pressure chamber. The
lubricants tested in this work contain pure POE oil and the POE diluted with the
non-chlorinated HFC refrigerant R-134a, a naphthenic mineral oil and the mineral
oil diluted with the chlorinated HCFC (hydrochloroflourocarbon) refrigerant R-22.
Paper E investigates the refrigerants influence of the lubricant traction coefficient.
In the investigation a ball-on-rod test apparatus was used at a contact pressure of 2.5
GPa. Three different POE oils were tested with and without refrigerant R-134a. A
naphthenic mineral oil and refrigerant R-22 was used as reference.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
5
2 Bearing design and lubricant properties
Bearing and lubrication system design is often the second step considered during the
development of a new compressor. The compressor layout, screw geometry or
impeller design are typically selected as a first step towards a new machine, and
initial load calculations and other data are used to drive the bearing selection. It is
critical that the bearing and lubrication system design be done with enough fidelity
as early as possible in the design process. During the design phase the different parts
and subsystems selected for the compressor need to provide the performance
required to meet the machine specification and have the necessary life to meet the
expected duty cycle under the given environmental conditions.
2.1 Life of rolling element bearings
In 1947, using the Weibull probability distribution of metal fatigue and Hertzian
contact parameters, Lundberg and Palmgren [6] developed a method to calculate
bearing fatigue life, eq.1. The equation was standardized by ISO 1962. The theory
assumes that the probability of a given volume element surviving N stress cycles
and then failing is proportional to its size and is a function of its location and the
number of cycles.
p
n
P
C
L
|
.
|

\
|
=
(1)
In the formula, L
n
is the statistic nominal life in million of revolutions, which n % of
the bearings will fail. C is the dynamic load capability of the bearing and P is the
equivalent dynamic bearing load. The exponent p varies depending on the type of
bearing; p = 3 for ball bearings and 10/3 for roller bearings.
Lundbergs and Palmgrens method to calculate bearing life did not consider
lubrication, i.e. no lubricant separating the interacting surfaces. Research has shown
that the premature failure of a bearing is often the result of inadequate contamination
control, which should be incorporated into the design process. Long service life
requires satisfactory lubrication and control of the contamination level. Furthermore,
if adequately lubricated and not disturbed by foreign debris, fatigue will not occur in
the bearing as long as the load is below the fatigue load limit. Controlling the
bearing manufacturing process better and improving the quality of the bearing steel
have extended the bearing life.
It was not until 1960 when Tallian [7] correlated the life theory with the separation
of the surfaces. The method was improved during the 20
th
century, and Wuttkowski
and Ioannides introduced the new life model in 1985 [8], shown below in eq.2.
The model includes the adjustment factor a
SKF
that considers the lubricant viscosity
(), level of contamination (
c
) and the fatigue load (P
u
) of the bearing material.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
6
p
SKF nna
P
C
a a L |
.
|

\
|
=
1
(2)
The relation of the above mentioned parameters and a
SKF
are given in SKFs general
bearing catalogue [9].
a
SKf
is a function of ( ) P P
u c
and and depends also on the bearing family, p. is
the ratio between the actual viscosity of the lubricant, , and the minimum required
viscosity,
1,
eq.3.
1

=
(3)
Wardle et al [10] discovered that when refrigerant is mixed in the oil and used as
lubricant, such mixtures containing less than 75% oil by weight will not sustain an
oil film in rolling element bearings and are therefore unsuitable for lubrication
purposes. Their experiments measured the lift off speed of rolling element bearings
by measuring the capacitance between the races and balls in the bearings. The lift off
speed is defined as the minimum speed when the surfaces are completely separate
by the lubricating film thickness. The experiments were repeated at different
refrigerant concentrations and the lift off speed was measured as a function of outer
ring temperature and cavity pressure. In refrigeration compressors bearing failures
associated with high refrigerant concentrations forced the researcher to refine the
bearing life theory for bearings operating in a refrigeration environment [11], eq.4.
1
72 , 0
min
1
3

|
.
|

\
|

= =
eral
adj
adj
(4)
In this equation, the actual lubricant viscosity is adjusted by the ratio of the pressure-
viscosity coefficients of the actual lubricant, , and a reference pressure-viscosity
coefficient value set to the coefficient for pure napthenic mineral oil,
mineral
. The
pressure-viscosity coefficient describes the lubricants viscosity dependence of
pressure.
1
is adjusted by a factor 3 for HFC refrigerants and a factor 2 for HCFC.
The new value is then used to determine the correction factor a
SKF
in the bearing
life model.
Using the newly adjusted -value resulted in oversized bearings, particularly at low
refrigerant concentrations. Maintaining reasonably dimensioned bearings in
refrigeration compressors is essential to avoid trouble with large clearances. The size
itself also causes problems to fit the bearings into the design and increasing the cost
of the machine.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
7
2.2 The refrigerant effect on lubricant properties
influencing bearing life
According to the life theory, the fatigue life depends on the number of stress cycles
and the magnitude of the stress. However, the development of a superior quality
bearing material has changed the source of bearing failure. Ioannides et al. [12]
found that fatigue failures are initiated more from the surface/surface conditions than
from cracks starting from dislocations and inhomogeneity down in the bearing
surface. Several authors [13,14] have shown correlations between surface roughness
and reduced life of rolling element bearings. Tripp et al. [15] demonstrated that the
von Mises stress is much greater when the surface is rough than in the smooth case.
This led to more attention about the surface finish and contamination levels in the
lubricant. The RMS (Root-Mean Square) slope and RMS wavelength have shown to
be important surface roughness characterization parameters. To reach an optimum
load-carrying capacity, the surface should initially run-in. The ability to build up a
lubricating film is more favourable if the asperity slopes are low and have a long
wavelength. The surface slope should be 10 to 150 times less for a bearing that has
run-in than for a new bearing surface, Jacobson [16].
2.2.1 Film thickness
To obtain long life, the moving surfaces should be completely separated by a
lubricating film. When the film thickness is too thin, surface asperities penetrate the
oil film and metal-to-metal contact occurs. Contaminating particles can also be
trapped in the contact and cause contact between the moving bodies. Contacts result
in denting and plastic deformation of the surfaces with high local stresses, resulting
in surface wear and increase risk of fatique. Rolling element bearings operate in the
EHL regime, where the surfaces in the contact are completely separated by a
lubricant film thickness and the high pressure deforms the surface elastically. In the
thesis the derived expression for the minimum film thickness at elastohydrodynamic
lubrication for an elliptical contact by Hamrock and Dowson [17] is used to
calculate film thickness, eq.5.
( ) ( )
k
x
z
x
x
e
R E
w
E
R E
U
R h

|
|
.
|

\
|

|
|
.
|

\
|

=
68 , 0
073 , 0
'
49 , 0
'
68 , 0
'
min
1 63 , 3

(5)
In the equation, h
min
is the minimum film thickness in the elliptical contact, R
x
the
effective radius of the contact spot in x-direction, U the relative speed of the two
interacting surfaces, the viscosity of the lubricant, E the effective module of
elasticity of the materials in contact, the pressure-viscosity coefficient of the
lubricant, w
z
the applied load and k the ellipticity parameter.
The film parameter is used to determine the lubricating regime the bearing is
operating in, Hamrock [18]. The film parameter is the ratio between the minimum
film thickness and the RMS roughness value of the bearing race and the ball, see
eq.6.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
8
qball qring
S S
h
+
=
min (6)
According to Hamrock [18], should be higher than 3 to assure that no contact
occurs and the regime is elastohydrodynamic. When the film thickness decreases,
the surfaces are rough or both the regime is changed to mixed lubrication. In mixed
lubrication, sporadic metal-to-metal contact occurs in the contact. Cann et al. [19]
re-examined the lambda value as a tool to secure safe operation and showed it to be
far too crude as a tool at low -values. This must be under consideration when using
the lambda as tool to evaluate the lubrication, especially when the value is under 1.
2.3 Bearing test apparatus with on-line monitoring of
lubricating film status
To study lubricants in rolling element bearings used in refrigeration compressors
under realistic environments, an experimental apparatus designed by Hansson and
Jonsson [20]that simulates the running condition is redesigned. The apparatus is
equipped with an on-line capacitance monitoring system that measures film build
up, from mixed to full film lubrication, in a complete rolling bearing system without
any modification.
The most commonly used techniques to monitor the lubrication status of rolling
element bearings on-line are based on resistance, capacitance or both, with the
benefit of being used on a complete machine part with a complete bearing
arrangement. This means that a machine containing several bearings can be
monitored on-line during operation. The capacitive method utilizes the bearing inner
ring, the balls and the outer ring as a multiple of variable capacitances in series and
parallel, see eq.7.
|
|
|
|
|
|
|
.
|

\
|

+
|
|
|
|
|
|
|
.
|

\
|

+
|
|
|
|
|
|
|
.
|

\
|

=
on
on
in
in
o
o
i
i
o
o
i
i
h
A
h
A
h
A
h
A
h
A
h
A
C

1 1
1
..........
1 1
1
1 1
1
2
2 2
1
1
1
1
(7)
In Equation 7, C is the total capacitance over all bearing contacts. the dielectric
constant of the lubricant, A
i
the nominal contact area of the inner ring and ball and A
o
the contact area of the outer ring and ball. h
i
is the film thicknesses between the ball
and the inner ring and h
o
is the thickness between the ball and the outer ring. The
total capacitance C, over the bearing is the sum of the capacitances at every contact
in the bearing, i.e. there are eight contacts parallel to each other in a bearing with
eight balls and every parallel contact contains two in series.
Since the measurement includes several contacts in the bearing, it is not possible to
point out where the lubrication problem occurs, except for possibly somewhere in
the bearing, bearing arrangement or machine part. Jacobson [21] demonstrated this
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
9
problem, finding wear on the bearing surfaces even when the capacitance measuring
indicated full separation.
Dyson et al. [22] measured the capacitance between two discs and estimated the oil
film thickness between them. The capacitance was measured with a frequency
bridge. The frequent bridging of the film directly between the asperities of the
opposing surfaces or through conducting particles trapped between them, disturbed
the bridge balance. Choosing higher carrier frequency and filtering the bridge output
reduced the disturbance.
ten Napel and Bosma [23] studied the influence of surface roughness on the
capacitance measurements of film thickness and found that the deviation between
measured and theoretically predicted values is mainly due to surfaces roughness.
The closer the surfaces were to a perfectly smooth surface, the better the correlation
between theoretical and experimental estimated film thickness. Using the
capacitance to monitor film thickness in an elastohydrodynamic contact provides
reliable results when the surfaces are completely separated, but unreliable
capacitance readings when metal-to-metal contact occurs.
To study mixed lubrication, the measurable range including in-contact and out-of-
contact conditions needs to be considered. Several workers have designed devices
combining the resistance and capacitance measurements, i.e. Cheng and Zhang [24],
Lord [25] and Heemskerk et al. [26]. Lord developed the capacitance technique
further to measure film thickness even when contact occurred. The capacitance
measures the film thickness when the surfaces are completely separated and the
conductivity gives information of the film thickness when incidental metal-to-metal
contact starts. However, the measured resistance will be a combination of resistance
and capacitance in the contact resulting in a filtering effect.
Several researchers have used the apparatus developed by Heemskerk et al. to study
elastohydrodynamic contacts. Jacobson [33] and Masen et al. [27] measured the lift-
off speed and Wikstrm and Jacobson [28] studied the time to lubricant
breakthrough when contact begins in an oil lubricated spherical bearing.
2.4 Refrigeration oils with additives
If the film parameter is lower than 3, metal-to-metal contact is assumed to occur
and it is recommended that additives be used to support the lubrication [9]. In
refrigeration applications, CFC and HCFC refrigerants include chlorine in the
molecule that act as EP additives [29, 30, 31, 32], making it possible to run-in the
bearing surfaces without early failures. The chlorine reacts with the steel and form
iron chlorine on the bearing surfaces. In particular, refrigerants with two or more
chlorine atoms in the molecule showed improved lubrication properties at mixed
lubrication, Murray et al. [32]. The lack of chlorine in HFC refrigerants has shown
insufficient lubrication with increased wear and a resulting shortened bearing life,
Meyers [11]. Jacobson [33] reported that the wear rate of the bearing surfaces
typically requires 50% higher viscosity when lubricated with a polyolester oil/R-
134a mixture than for a bearing lubricated with mineral oil/R-22.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
10
Experiments to find EP-additives that improve lubrication in the anaerobe
environment of refrigeration systems have been conducted with different additive
packages containing phosphor. The experiments were performed in the bearing test
apparatus described in Figure 2. Nixon et al. [34] and Wan et al. [35] both showed
that EP-additives containing sulphur-phosphor do not always extend bearing life.
Bearing life was reduced by a factor of 4.5 when EP additives were present in the
lubricant. The EP additive induces micro structural alternations on the bearing
surface due to the reactivity of the additives. The additives can promote
corrosion/diffusion mechanisms in the contact under stress and accelerate bearing
failures.
2.5 Viscosity and pressure-viscosity coefficient
A key factor for proper lubrication of rolling element bearings is a sufficient
lubricant film thickness. The film thickness in an EHL lubricated bearing is strongly
dependent on the viscosity, , and the pressure-viscosity coefficient, . Several
authors have reported that an increasing amount of refrigerant decreases both the
viscosity and pressure-viscosity coefficients, and causes a drop in film thickness,
[36,37,38]. The concentration of refrigerant used to dilute the oil depends on
temperature and pressure. Henderson [39] shows that the solubility of refrigerant in
the oil depends on the molecular branching of the ester molecule. The molecular
weight ratio between the oil and refrigerant is essential to predict the behaviour of
viscosity and pressure-viscosity coefficients when the refrigerant dilutes the oil. In a
compressor, a lighter refrigerant such as ammonia dilutes the oil by only 3-5%.
Heavier refrigerants like R-22 and R-134a are usually found at concentrations of 5
to 40 wt%, depending on the running conditions.
The effect of refrigerant on the viscosity and pressure-viscosity coefficients is
essential data to achieve the right lubricating conditions for the compressor.
2.6 Compressibility and density
An important parameter to reduce stress and improve the life of the bearing is
compressibility. The lubricant compressibility affects the magnitude of the pressure
spikes in the pressure distribution, Hamrock et. al [40]. Because the real surfaces are
rough, the pressure distribution also becomes rough. Pressure spikes appear at all
contact spots throughout the contact, Tripp et al. [15]. Investigations based on two
different approaches have been conducted to describe the density-pressure
relationship for lubricating oils, but no oils mixed with refrigerant are investigated.
The first attempts used a test apparatus to change volume as a function of pressure
under static conditions, i.e. no influence of possible transient loading. Dowson and
Higginson [41], Hamrock et al. [42], Jacobson and Vinet [43] and Sthl and
Jacobson [44] all performed investigations under static conditions. The second
approach has been to understand if the effect of structural relaxation is significant
for elastohydrodynamic lubricants. Lindqvist et al. [45] used a modified Split-
Hopkinson pressure bar set-up to determine the density-pressure relation of
lubricants, concluding that long, straight molecules found in esters, polyglycol and
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
11
polyalpholefin are rather compressible, whereas naphthenic and paraffinic mineral
oils have more complex molecules and show a stiffer behaviour. Several authors
have contributed to pressure-density relations, and some models have been used to
calculate film thickness. Dowson and Higginson model, the most commonly used
for the pressure-density relation, is empirical and based on measured pressure-
density data up to moderate 0.4 GPa. The interatomic forces in compressed solids
and liquids are similar; hence, Jacobson and Vinet [43] approximated the equation
of state for solids, which was derived by Vinet et al. [46], for the equation of state
for liquid lubricants. To obtain the parameters included in the compressibility
relation, Jacobson and Vinet fitted data published by Hamrock et al. [42] for
pressures as high as 2.2 GPa.
2.7 Shear strength of the lubricant at EHL pressure
The theoretical life of a rolling element bearing is dependent on a number of factors,
one of them being the number of load cycles. However, if the equivalent stress is
lower than the corresponding fatigue limit the fatigue life is infinite. The equivalent
stress is a combination of stress produced by the normal load, shear load, induced
stress by contact between asperities, contamination particles etc. An important
determining factor in fatigue life is the depth below the surface where the maximum
stress appears. If the stress maximum is located close to the surface, the material
fails easier than if the maximum is deep in the material. When sliding occurs in the
contact a shear stress,
L
, component will influence the equivalent stress. Jacobson
and Hamrock [47] used
L
as the limiting shear stress that the oil can sustain.
L
,
combined with the pressure, P, is called the shear strength-pressure coefficient, ,
(friction coefficient of the solidified lubricant) and is given by eq.8.
P
L

=

(8)
Hglund [48] showed that higher causes the von Mises stress maximum to move
closer to the surface and slightly increase the maximum stress. Several authors have
used different techniques to measure the shear strength of lubricants, e.g. Hglund
and Jacobsson [49] used a high-pressure chamber, hrstrm [50, 51] designed and
used a ball-on-rod test apparatus to measure the shear-strength at the impact of the
ball on the rod, and Muraki and Sano [52] made experiments using a ball-on-disk
EHL test apparatus to measure the shear strength of a polyolester oil and three
different refrigerants. The shear strength is affected by the composition of the
lubricant, temperature, pressure and relative speed of the interacting bodies. Typical
values of the shear strength are dependent on the application. A lubricant used to
reduce friction has shear strength values of around 0.03 and a traction fluid 0.14 at
40 C.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
12
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
13
3 Research facility
To better understand lubrication of rolling element bearings and find the effects of
the refrigerant on lubrication, the presented research is based on experiments
performed in different kinds of test apparatuses that measure temperature, viscosity,
vibrations, rotational speed, traction, surface characteristics, dielectric constant of
the lubricants and lubricating film status using a capacitance method.
3.1 Bearing test apparatus
The Bearing Test Apparatus shown in Figure 2 is designed to test rolling element
bearings in a refrigeration environment and is used in Papers A and B of this thesis.
Hansson and Jonsson [20] originally designed the apparatus. The bearing test
apparatus is redesigned in this work by introducing a new control and regulating
system, new oil circulating system, filtering of the electric power from the frequency
converter, capacitance measuring system and viscosity measurements on-line. Paper
A describes the bearing test apparatus.
a) b)
Figure 2 a) Test apparatus b) Schematic overview of the main part of the test
apparatus.
In the investigation, angular contact ball bearings (7210BEP) are mainly used, due
to their common usage in refrigeration screw compressors. A frequency converter
regulates the rotating speed of the test apparatus from 0-12,000 rpm 100 rpm. The
radial load is applied by a hydraulic cylinder acting on the test chamber housing.
The axial load is applied by a load cell placed between the test bearing and the
hybrid support bearing. The load cell contains 12 springs; altering the number of
springs or the spring constant the load can be changed. The load ranges from the
bearings critical minimum load up to 11,500N 100 N.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
14
Two gear pumps with a capacity of 5 litres per minute provide lubricant to the test
bearing, hybrid bearing and the mechanical seal. The lubricant temperature is kept
constant at 40 1C by a water-cooled heat exchanger controlled by a PID
regulator.
The monitoring system includes a vibration sensor attached to the test chamber. The
sensor measures shock pulses in dBsv. The vibration level is used as an indicator of
bearing failure. By monitoring vibrations produced by the test bearing, the
experiment can be stopped before failure occurs.
3.1.1 Data acquisition and control system
To control and acquire data, Labview software is used with the data acquisition
system Fieldpoint and an oscilloscope board. Fieldpoint measures the
temperatures, vibration level, system pressure, speed and viscosity. The high
frequency data acquired from the capacitance measurement of the film thickness
status is captured by the oscilloscope. The control program in Labview uses
temperature, vibration, pressure, speed and viscosity data to control the experiment.
A Labview program graphically visualizes the measured data and stores it. The
program also uses the measurements to control the lubricant temperature, alarms and
power to the electrical motor during the experiment. The alarms are coupled to the
temperature sensors, vibration sensor and pressure sensor. The minimum
temperature limit is set to 30C and the high limit to 50C, vibration sensor limit is
70 dBsv and the pressure sensor limit is 0,1 MPa.
3.1.2 Measurement of refrigerant concentration
The lubricant entering the test chamber passes through a 3m filter. The refrigerant
concentration is controlled during the experiments in the test apparatus by
measuring the viscosity on-line with the viscometer. The on-line viscometer allows
the lubricant viscosity that enters the test chamber to be monitored. The viscometer
uses a simple and reliable electromagnetic concept where two coils magnetically
move a piston back and forth at a constant force. The travelling time is measured for
the piston to move back and forth and is used to calculate the absolute viscosity of
the fluid. The accuracy of the measurement is 1%.
3.1.3 Capacitance measurement
To monitor the lubrication status online, the capacitance method developed by
Heemskerk et al [26], Lubcheck, is used. Due to the combined conductivity and
capacitance technique, film thickness variation can be detected at both
elastohydrodynamic and mixed lubrication. The instrument output voltage, V
cap
, is a
measure of surface separation where a high V
cap
value indicates full separation and a
low value indicates contact. The technique uses a high frequency alternating current
of 410 kHz, which allows the device to measure high-frequency oscillations in
capacitance when incidental asperity contacts occur. To not damage the bearing by
sparking, the voltage feed to the bearing is 90 mV. In data acquisition, an
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
15
oscilloscope is used and no filters are applied. The sample rate is 100,000 samples
each second, but only 40,000 samples are captured and used in the measurement.
Figure 3 shows a set of measured data when contact occurs.
The test apparatus is electrically insulated to allow capacitive and conductive
measurements of a single bearing in the apparatus. The outer ring is in contact with
the insulated housing and the inner ring to the shaft of the test apparatus, see Figure
2b.
To identify what should be counted as metal-to-metal contact, it is necessary to
define a signal level that can be used as a threshold. Henceforth, if the output signal
drops below the threshold level in Figure 3, it is recognised as metal-to-metal
contact. The threshold in the experiments is set to 95% of the mean of the 40,000
measured V
cap
values. The level is set to 95% after experiments that showed that a
lower level is to low to identify possible contacts. On the other hand a higher level
makes it impossible to sort out the contacts from the noise in the measured data. The
figure shows 40,000 samples acquired at a sample rate of 100,000 samples each
second.
Figure 3 Output signal from capacitance measurements, V
cap
. The figure shows
40,000 samples acquired with a sample rate of 100,000 samples each second. A data
point below the threshold line is defined as a contact.
In Figure 4, V
cap
is measured during a three-step experiment with decreasing oil
concentration, i.e. increasing refrigerant concentration.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
16
Figure 4 Example of use of V
cap
to monitor lubrication status.
When the refrigerant concentration increases, the viscosity and V
cap
decrease. The
decrease in V
cap
after each filling, marked in the figure with circles, indicates poor
separation of the surfaces. The depth and time necessary for the recovery in Figure 4
is essential to reduce metal-to-metal contact and denting. Note that every point in the
figure consists of the mean of 40,000 samples. The experiment in Figure 4 is
performed with an angular contact ball bearing (7210BEP) at constant load of
11,500 N, temperature 40C and a speed of 1,500 rpm, but an increased refrigeration
concentration in three steps.
3.1.4 Dielectric constant
The dielectric constant of the lubricant is a vital parameter when using a capacitive
measuring method. Relative capacitance measurements can be done when the
medium is the same in the measurements. However, if the medium differs, i.e.
another oil and refrigerant, it is important to know the dielectric constant of the
mediums to compare measurements.
A test apparatus is designed with a fixed area and controlled gap that allow
capacitance (dielectric constant) measurements of the lubricants presented in this
thesis. The test apparatus can be seen in Figure 5.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
17
Figure 5 Apparatus to measure dielectric constant.
The relative dielectric constant is obtained by measuring the capacitance with air
between the plates C
air
in the variable capacitance (Figure 5) and with lubricant as
dielectric medium between plates C
lub
. The relative dielectric constant can then be
calculated according to eq.9.
air
lub
R
C
C
=
(9)
To estimate the capacitance in the gap between the plates, C
x
, eq.10, compensation
for the inner capacitance C
i
of the test apparatus is necessary to the measured value
before the dielectric constant can be calculated according to eq.9.
i m x
C C C + = (10)
The dielectric constant is measured at 401C with a distance between the capacitor
plates of 0.1 mm. The pressure during the experiments is, according to the gas
pressure of the refrigerant at the measured concentrations, approximately 3, 7 and 10
bar for R-134a and 5, 9 and 12 bars for R-22.
3.2 Surface roughness measurements and analysis
After the bearing test apparatus experiments, the bearings surfaces were further
investigated in a SEM (Scanning Electron Microscope) and a topometer. The
topometer used is a Wyko NT 1100, which uses interferometry to measure the
surface topography and generate a 3-D representation of the surface, Figure 6. The
representation consists of 3-D data that are used to calculate surface parameters.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
18
Figure 6 Topographic image of the inner ring surface.
In the SEM, the mode of backscattered electrons is used to generate high-magnified
images of the surfaces in the running track. The images were used to examine the
wear of the surfaces at different running conditions. The image in Figure 7 shows
the new inner ring of a 7210 BEP angular contact bearing.
Figure 7 SEM image of a new 7210 BEP angular contact ball bearing.
3.3 Additives in refrigeration oils
Paper B presents experiments in a bearing test apparatus, Figure 2. The bearings are
tested with running conditions similar to those used in refrigeration compressors.
The on-line capacitance measurement apparatus monitors the lubrication status
during the experiments.
3.3.1 Test procedure
Before an experiment starts a new bearing is mounted into the test apparatus and the
spring package loads the test bearing axially with 11,500 N. Lubricant is added and
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
19
the circulation of oil is turned on. The experiment begins after 2 hours of filtering
the lubricant and when the temperature is stabilised. The shaft is speed up to a speed
of 1,500 rpm. Refrigerant is then added to the first of three viscosity-levels
(concentration-levels) measured by the on-line viscometer. The three levels are set
to correspond to realistic refrigerant concentrations in refrigeration compressors.
When the viscosity is constant the V
cap
measuring starts and measures the film status
every second for two hours. After two hours, more refrigerant is added to reach the
second viscosity-level. The power to the electrical motor and the V
cap
measurement
are on during the filling. The bearing then runs for two more hours at this new
concentration and the procedure is repeated to attain the lowest viscosity. When the
experiment is finished the measurement is stopped and the power is turned off. The
refrigerant is removed from the system to a bottle and the oil is drained.
The lubricant temperature is regulated during the experiment and maintained at a
constant temperature of 40 1C. The system pressure depends on refrigerant
content and varies between 3-12 bars.
The bearings are then examined in the SEM and topometer. The SEM provides high
resolution images of the bearing surfaces and hence give information about the
surfaces wear, denting and other disturbances. The topometer measures surface
parameters for roughness and the composition of the roughness, i.e. the slopes of the
roughness, wavelength between the asperities and abbot curve.
Measurements of the relative dielectric constants are performed for the four
lubricants at the refrigerant concentrations used in the investigation. The
measurements were done with the variable capacitance, Figure 5, that is attached to
the test apparatus, Figure 2. The oil is filled in the test apparatus and the oil
circulation is started. Refrigerant was filled to the oil until the mixture had the right
viscosity corresponding to the desired refrigerant concentration. When the
temperature levels out at 40C the measurement is done with an instrument
measuring the capacitance. The inner capacitance of the variable capacitor was
measured with a distance of 5 mm between the plates. The inner capacitance was
measured with the oils and air between the plates.
3.4 Viscosity and pressure-viscosity coefficient
measurement
The viscosity and pressure-viscosity relationship measured in Paper C were done in
a falling ball viscometer, Figure 8. Jonsson and Hglund [53] designed the
viscometer, allowing the viscosity to be measured at a pressure up to 34 MPa. The
viscosity is also measured on-line in the bearing test apparatus during experiments
conducted in the apparatus, see 3.1.2.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
20
Figure 8 Falling ball viscometer used in Paper C.
The viscosity is determined by measuring the time it takes for the ball to travel 100
mm in a tube containing the test liquid. Two photo sensors time the ball through two
pairs of sapphire windows. To pressurize the fluid the viscometer is equipped with a
hydraulic piston that is operated by a manual hydraulic pump. The test fluid in the
viscometer is always pressurized above the vapour pressure of the refrigerant to
assure the test fluid is in liquid state during the experiment.
The acquired time for the ball to travel the 100 mm is then processed to calculate the
viscosity. A spreadsheet program is used to calculate the viscosity. The program
compensates for the effects of pressure and temperature and the dimensions of the
ball and the bore. The viscosity is measured at three pressures and the Barus
equation [54] is used to estimate the pressure-viscosity coefficient at each pressure,
temperature and concentration, see eq.11.
p
e

=


0
(11)
3.4.1 Test procedure
Before an experiment starts the viscometer is cleaned and filled with 800 grams of
oil. The ball is placed in the cylinder and the end-cap is screwed in place. The air is
pressed out and the hydraulic piston pressurizes the viscometer. The refrigerant is
filled to the first concentration. A sample is taken to measure the concentration; see
3.1.2 for details about the concentration measurement. The pressure is then
increased to the first measuring pressure around 3 MPa. The temperature is
stabilized within 1C before the measuring can begin. Three measurements are
done at each pressure, concentration and temperature. The procedure is the same for
all three pressures, i.e. 3 MPa, 17 MPa and 34 MPa.
The refrigerant concentration is increased to the next level and the measurements are
repeated for the three pressures. This procedure is done for each refrigerant type at
40 1C and 80 1C. The tested lubricants consist of a mixture of POE oil and
four different refrigerants shows the properties of the oil and the refrigerants.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
21
Table 1 Tested oil and refrigerants
Oil/Refrigerant Refrigerant type Molecular mass
[g/mol]
Viscosity grade Pressure-viscosity
coefficient [GPa
-1
]
at 40 C
POE - 633 VG68 23
R-134a HFC 102.03
R-22 HCFC 86.48
R-410a HFC mixture 72.58
R-32 HCFC 86,48
3.5 High-pressure chamber
The compressibility measurements presented in Paper D are done to investigate if
compressibility of oil is affected by refrigerant dilution in the oil. The measurements
are done in a high-pressure chamber developed and described by Jacobson [55] and
further described by Sthl and Jacobson [44]. An overview of the high pressure
chamber can be seen in Figure 9a, with the high-pressure cylinder (3) and the
plungers (7) in Figure 9b. In the compressibility test the lower plunger is fixed
whereas the upper plunger is movable. The test lubricant is placed in the cylinder
between the plungers when the lower plunger is in position in the cylinder.
Figure 9 a) Overview of the high-pressure chamber, b) High pressure cylinder and
the two plungers.
The lower plunger is modified with a hole through the plunger and provided with a
non-return valve that also functions as a seal. This allows the filling of lubricant
when both plungers are in place. A hydraulic jack (1) pressurizes the lubricant and
controls the movement and force of the compression. To withstand high-pressures,
the high-pressure chamber cylinder is press fitted inside a pre-stressed container (10)
and pre-compressed axially by a screw assembly (9), with a maximum pressure
limited to 4.5 GPa.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
22
A force transducer measures the applied load that is then used to calculate the
corresponding pressure. A linear potentiometer measures the position of the
movable plunger. The compressibility is calculated from the measured position and
pressure. For safety, an enclosure made of lexan (4) surrounds the hard metal parts.
The hydraulic jacks are pressurized and supplied with fluid through the valves and
pressure regulator (6) from a hydraulic pump (5).
3.5.1 Test procedure
All parts are carefully cleaned with an alcohol solvent and assembled. The
refrigerants are pressurized during the experiment to assure the refrigerant is in
liquid phase.
The oil and refrigerant are mixed in a cylinder placed on a scale. To reach the right
concentration, 100 g of oil followed by the correct amount of refrigerant is applied.
The oil/refrigerant mixture is placed into a hydraulic cylinder where the pressure is
increased up to 7 MPa. The lubricant sample is then pressed into the high pressure
chamber through the hole in the plunger and the one-way valve. To ensure that no
air is present in the chamber during the experiments, the seal at the top of the
chamber is set open and then slowly moved until the leakage stops. During leakage,
the lubricant fills the entire chamber while evacuating the air. The experiments start
as soon as the lubricant sample is placed inside the high-pressure chamber.
The change in volume and pressure are derived by applying the load. The hydraulic
jack (4) moves the upper plunger (2) downwards. While the pressure increases, the
control program captures and measures the force and position of the plunger. All
experiments are performed three times for each lubricant to ensure the repeatability
of the measurements.
3.6 Traction measurements
The measurements of oil/refrigerant mixtures in Paper E are done in a ball-on-plate
test apparatus, Figure 10, to investigate if the shear strength coefficient of oil is
affected by refrigerant dilution. The test apparatus is designed by hrstrm [50,51]
and is capable of measuring the shear strength of the lubricant at pressures up to 3
GPa.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
23
Strain
gauges
Transient
recorder
Bridge
amp.
Accelero-
meter
Computer
Accelero-
meter
Charge
amp.
Charge
amp.
Tungsten carbide end plate
with lubricated surface
4
0
.
0
4
0
0
0
9
4
0
Steel ball
o 16
/
Fixture on
moving sled
Figure 10 Apparatus for traction measurements.
3.6.1 Test procedure
The experiments were carried out at ambient pressure at a temperature of 21 1C.
The ball and the rod end surface were thoroughly cleaned by solvent. The lubricant
sample is placed on the tungsten carbide plate located on the end of the rod. The ball
is placed in the movable sled, and the sled is released from a certain position
corresponding to the desired contact pressure in such way that an oblique contact
occurred between the ball and the tungsten carbide plate with the lubricant sample.
The contact pressure in the experiments is 2.5 GPa. Accelerometers and strain
gauges measure the shock waves in the rod and provide information that helps to
calculate the transient normal force and transverse force acting during the impact. A
change in the shear strength of the lubricant affects the wave pattern and
consequently the forces at the end plane of the rod. From these measurements the
shear strength is calculated for the transient experiment. In this thesis only the
maximum shear strength is presented.
To assure full film separation, a potential difference of 100 mV is applied between
the ball and the tungsten carbide plate. Any potential equalization, resulting from
asperity contact, is monitored on an oscilloscope. Boundary lubrication
measurements are not relevant and are therefore excluded.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
24
Three different polyolester oils were used Esters A, B and C. All three are POE
refrigeration oils but with different viscosity grades and acid branching. The POE
oils were diluted with the HFC refrigerant R-134a. A mineral oil diluted with the
chlorinated HCFC refrigerant R-22 was used as reference. Table 2 details these oils
together with the refrigerants used.
Table 2 Details of the four lubricants used in the experiments.
Lubricant Chemical structure Viscosity grade Branched acids Refrigerant used
Ester A C
39
H
72
O
8
VG68 70% R-134a
Ester B C
31
H
56
O
8
VG68 8%+67% R-134a
Ester C C
33
H
62
O
6
VG46 100% R-134a
Mineral - Base viscosity 57cst
at 40C
- R-22
3.6.2 Refrigerant evaporation
The ball on plate apparatus operates under ambient pressure. However, at ambient
pressure the refrigerant evaporates and its concentration decreases. To know the
actual concentration at the time of the traction measurement, experiments were
carried out to determine the change in refrigerant concentration at ambient pressure
over time for the four different mixtures. Oil weighing 27 grams is first scaled and
placed in a small bottle and then diluted with 5 grams refrigerant. The pressure in
the bottle is decreased until the refrigerant starts to boil out of the oil. The weight of
the mixture is recorded versus time. The decrease in refrigerant concentration is then
calculated.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
25
4 Results
4.1 Bearing test apparatus
The experiment in the bearing test apparatus is done to find methods to monitor and
measure the lubrication status during operation of the bearing.
4.1.1 Contact detection using the capacitance method
Measurements were carried out with a gradually reduced viscosity to confirm the
occurrence of contacts. Figure 11 shows measurements where the viscosity is
reduced by increasing the refrigerant concentration step-wise up to about 25 wt%.
During the experiment the number of contacts is counted, and increased significantly
when the concentration level exceeds 20 wt%. The measurements were repeated
with a new bearing, but the experiment was stopped at 20 wt% refrigerant
concentration, see Figure 11b. The output signal from the two measurements yields
comparatively similar results up to 20 wt% refrigerant concentration.
a) b)
Figure 11 Counted number of contacts during 10 hour testing with Solest 68/R-134a
mixture at 3,000 rpm and bearing load ratio C/P = 5.7; a) Step by step increased
refrigerant concentration up to 27 wt%, b) Step by step increased refrigerant
concentration up to 20 wt%.
Typical SEM images in Figure 12 show denting on the surfaces. The grinding marks
from the manufacturing being worn during the experiment. The bearing was also
observed to run at a lower viscosity, i.e., lower film thickness correlating well with
the counted number of contacts. The calculated film thickness and film parameter
are 259 nm and 2.7 at 20 wt% concentration, and 177 nm and 1.9 at 34 wt%
concentration.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
26
a) b)
Figure 12 SEM image of the inner bearing ring surfaces after the experiments. The
experiments are carried out with POE oil and R-134a mixture at 3,000 rpm and
bearing load ratio C/P=5.7. a) Step by step increased refrigerant concentration up to
34 wt%, b) Step by step increased refrigerant concentration up to 20 wt%.
The same experiment was carried out with R-22 containing chlorine, Figure 13.
During the measurement, the number of contacts detected with R-22 was
significantly lower than with R-134a.
Figure 13 Counted contacts and refrigerant concentration versus time for R-22 and
POE. Test condition; constant speed of 3,000 rpm and load ratio C/P = 5.7 and step
by step increased refrigerant concentration.
Figure 14 shows SEM images of the bearing inner ring surfaces after tests with R-
134a and R-22, clearly showing denting on the surfaces of both R-134a and R-22.
However, R-134a has more dents than R-22, which correlates well with the
measurements of contacts detected. The calculated film thickness and film parameter
for the highest refrigerant concentration for R-134a are 177 nm and 1.9, and for R-
22 are 130 nm and 1.4.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
27
a) b)
Figure 14 SEM images of the bearing inner ring surfaces after the experiments are
shown. The experiments are carried out at 1,500 rpm and bearing load ratio C/P=5.7.
a) R-134a with POE ester oil the refrigerant concentration is increased in three steps
up to 34 wt%, b) R-22 with napthenic mineral oil the refrigerant concentration is
increased in three steps up to 27 wt%.
V
cap
versus time during filling of the refrigerant is shown in Figure 15. The
refrigerant is filled in liquid state into the reservoir and mixed with the lubricant.
The measurement does not indicate the occurrence of a complete film breakdown.
The running speed was 1,500 rpm, the load C/P=5.7 and the lubricant temperature
40 C.
Figure 15 Measured V
cap
versus time with decreasing lubricant viscosity. R-134a
and POE oil was used in the experiment. In A) viscosity decreased from 10 to 6
mPas and film parameter from 1.7 to 1.2, B) viscosity from 22 to 14 mPas and film
parameter from 3.1 to 2.2, and C) viscosity from 30 to 22 mPas and film parameter
from 3.9 to 3.1.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
28
The lift-off is studied in Figure 16. The number of contacts is counted as the shaft
speed is increased successively from stationary to 3,000 rpm at 150 rpm/s, and the
bearing load is maintained at C/P = 11.4 in Figure 16a. The measurements were
repeated with higher load, C/P = 5.7, Figure 16b.
a) b)
Figure 16 Number of contacts with increasing shaft speed for two different load
ratios; a) C/P=11,4 and refrigerant concentration of 20,4 wt%, and b) C/P=5,7 and
refrigerant concentration of 19,6 wt%. R-134a and POE oil is used as lubricant
A frequency converter used to regulate the shaft speed transforms the original sine
wave by cutting the wave and producing a new one with the desired frequency. It
was discovered that the converter produced noise coinciding with the cutting
frequency on the output current. Figure 17 shows experiments with and without the
output filter. The up-going peeks coincide with the cut frequency of 33 Hz produced
by the frequency converter at the speed of 1,000 rpm. Output filters are used in
Figure 17b, reducing the noise and achieving a constant V
cap
reading. The noise
shown in Figure 17a can be mistaken as metal-to-metal contact and draw wrong
conclusions.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
29
a) b)
Figure 17a) shows the V
cap
readings when no output filter is used and b) when two
parallel output filters are used.
4.2 Dielectric constant measurement
Relative dielectric constant measurements show that refrigerant mixed in oil
influences the relative dielectric constant and thus the V
cap
reading, see Figure 18.
The behaviour between POE esters and the mineral oil is the same, though at
different levels of relative dielectric constant. This is of concern when comparing the
V
cap
readings to each other after different oil types have been used. As well, the
dielectric constant varies with pressure. In this investigation the relative dielectric
constant is measured at moderate pressures of approximately 5-10 bars. Dyson et al.
[22] show data of the dielectric constant at ambient pressure and 345 MPa, for
mineral and synthetic oils. The trend is that the dielectric constant increases with
increased pressure, except for two of the tested lubricants that did not show any
effect of the high pressure.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
30
a)
b)
Figure 18 a) Change in relative dielectric constant due to refrigerant concentration.
b) influence on calculated V
cap
. POE oil with R-134a refrigerant, Mineral oil
with R-22 refrigerant
4.3 Measurement of vibrations
A highly loaded bearing normally fails due to fatigue. Often, vibration, temperature
or both are monitored to detect bearing failure. In Figure 19, the measured V
cap
and
vibration level are plotted versus time. The experiments were carried out with a
mixture of POE oil and refrigerant R-134a at a concentration according to a
viscosity of 5.7 mPas. The shaft speed during the experiment was 1,500 rpm and the
load ration C/P = 5.7. The vibration and V
cap
data was captured every hour with a
sample rate of 40,000 samples/sec and presented in the figure as the mean each
second.
The vibrations reached the maximum allowed level and the experiment was stopped
after 430 hours. At these running conditions the life according to the life theory, L
10
life should be 1,700 hours. The vibrations level clearly increases just before failure
in both tests. However, the V
cap
signal based on the mean of 40,000 samples does
not clearly indicate any forthcoming bearing failure, since V
cap
is constant.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
31
a) b)
Figure 19 The measurement is done during a life experiment over 430 hours. a) V
cap
measured with the capacitance apparatus as function of time. b) vibration level
detected by the vibration sensor as function of time. The lubricant used is a POE oil
with R-134a refrigerant.
Figure 20 shows another bearing that failed after 130 hours of operating under the
same running conditions as in. During this experiment the contacts are counted
instead of V
cap
, see 3.1.3. The figure shows run-in, the steady-state period and the
increased number of contacts before failure.
Figure 20 Contacts counted during a life experiment.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
32
4.4 Polyolester oils with additives
The ability to form a lubricating film and the wear behaviour was investigated for
two POE oils with and without additives and a napthenic mineral oil. The POE oils
were diluted with R-134a refrigerant and the mineral oil with R-22. The two
additives used was phosphate ester and acid phosphate.
Table 3 shows the measurements of the relative dielectric constants and parameters
during the experiments. The film parameter is obtained by using the calculated film
thickness at the running conditions and using the S
q
value of a new bearing inner
ring and ball. The film parameter in parenthesis is calculated using S
q
of the tested
bearing.
Table 3 Tested oil/refrigerant pairs and running conditions during the experiments.
Test
fluid
Oil Relative
Dielectric
constant
Viscosity
[mPas]
0.1
Concentra
-tion
[wt%]
Pressure-
viscosity
coeff.
[GPa
-1
]
h
min
[nm]
Film
parameter
Fluid
A
Pure POE +
R-134a
3.9
4.3
5.1
14.5
5.7
2.8
8
19
34
22
19
15
227
111
61
2.40
1.17
0.65(0.75)
Fluid
B
POE +
0.25% Acid
Phosphate +
R-134a
3.8
4.4
6.2
13.5
5.9
2.8
9
18
34
21
19
15
215
115
61
2.27
1.21
0.65(0.79)
Fluid
C
POE + 1.5%
Phosphate
ester +
R-134a
3.9
4.7
5.6
16
5.7
2.7
7
19
35
22
19
15
225
111
59
2.58
1.17
0.63(0.73)
Fluid
D
Napthenic
Mineral oil
+ R-22
2.9
3.1
3.4
13.3
5.4
2.5
7
14
27
22
19
15
215
109
57
2.27
1.15
0.60(0.65)
Figure 21 and Figure 22 show the results of the mean V
cap
values measurements.
The V
cap
value consists of 40,000 samples taken each second. The V
cap
reading
changes when the refrigerant concentration increases. The signal drops significantly
each time the refrigerant concentration is changed, and eventually increases to a
higher more stable level, indicating that the lubricating film thickness is decreasing
and contact occurs. New asperities have to be run-in after each increase in
refrigerant concentration before the signal stabilises at a constant level. At a higher
concentration it takes longer time to reach the constant level. During the drop in V
cap
and during the stabilising phase, wear occurs on the bearing surface due to metal-to-
metal contact that smoothes the asperities. Denting on the bearing surfaces occurs
when the wear particles passes the contact.
Figure 21a) shows measured V
cap
with Fluid A, where V
cap
decreases to the lowest
level of all tested lubricants. Figure 21b) and Figure 21c) show the experiments with
Fluids B and C, including the phosphate additives. Clearly, additives promote film
built-up. Fluid C including the phosphate ester gives the highest output reading,
even higher than Fluid D. The better film formation of the phosphate ester compared
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
33
to acid phosphate might be explained by the difference in the amount of additive,
0.25% for acid phosphate versus 1.5% concentration of phosphate ester.
The results with the chlorinated Fluid D, Figure 21d), give higher separation than
Fluid A, indicating that the R-22 refrigerant plays a role in film build-up. However,
comparing the different types of oil/refrigerant is difficult, since the dielectric
constant can differ, and hence the output reading.
a) b)
c) d)
Figure 21a Output voltage V
cap
versus time with decreasing lubricant viscosity as
parameter, a) Fluid A no additives present, b) Fluid B with acid phosphate additive,
c) Fluid C with phosphate ester additive and d) Fluid D chlorinated.
The curve in Figure 22a fits the steady-state V
cap
values measured during the
experiments presented in Figure 21. V
cap
is plotted versus the theoretical film
parameter for each tested fluid. The capacitance measurements clearly indicate a
higher separation between the balls and the ring when additives are present,
especially when lubricated with Fluid C at a high refrigerant concentration (low ).
In Figure 22b, the V
cap
curves for Fluids A, B, C and D are compensated for the
higher dielectric constant compared to the mineral oil at the lowest refrigerant
concentration by using Equations 1 and 2. In Figure 22b the compensated V
cap
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
34
values of fluid B and C shows increasing V
cap
values when the film parameter
decreases i.e. improved lubrication, this is not the expected behaviour. The
behaviour can be explained by the effect of the additives.
a) b)
Figure 22a) The steady-state level of output voltage V
cap
versus film parameter of
the four tested lubricants. 4b) Compensated V
cap
curves for fluid A, B, C and D due
to higher dielectric constant.
The run-in process changes the surface structure, and material is removed, deformed
plastically or moved in the contact until a steady-state condition can be maintained.
The aim is to get as smooth a surface as quickly as possible. An increase in the load-
carrying capacity of the surface is characteristic of the run-in process. The surface
slope of a used bearing should be 10 to 150 times less after run-in than of a new
bearing surface [16]. The film build up is more favourable if the slopes of the
asperities are low. Figure 23 shows the SEM images of the bearing inner rings
surface from the four conducted experiments. The image shows the history of all
contacts and denting during the experiments, as shown in Figure 21. Extensive
scattered debris denting is visible when running with Fluids A and C, whereas
considerably less denting can be noticed with Fluids B and D. Loose fine debris with
a grain size of less than 3 m is suspended in the lubricant due to the wear process
of the surfaces and the lack of a fine enough filter to remove contamination.
According to capacitance measurements Fluids B and C separate the surfaces best.
However, the SEM-images show excessive scattering denting on the surface of Fluid
C, whereas the surface on Fluid B is almost unaffected. It is evident from studying
the V
cap
curves in Figure 21 more carefully that the run in after each drop in
viscosity is much tougher for Fluid C than for Fluid B. Fluid C take much longer to
reach the steady-state value, i.e. more denting and wear takes place.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
35
The image of Fluid B, including acid phosphate additives, shows the lowest degree
of denting and wear of the four experiments. The POE without additives (Fluid A)
shows the worst surface where the original surface is almost worn away.
a) b)
c) d)
Figure 23 SEM images of bearing inner ring surface; Image in background 1,000X
and in foreground 4,000X, respectively. a) Fluid A no additives present, b) Fluid B
with phosphate ester additive, c) Fluid C with acid phosphate additive and d) Fluid
D chlorinated.
An examination of the surface parameter measurements shows that running makes
the bearing surfaces better to build up a lubricating film that assures trouble-free
operation. In all four tests, the RMS angle is reduced from 0.110 to about 0.040
degrees. A low RMS- angle indicates the sharpness of the roughness. However, tests
show that the RMS wavelength increases from 5 to 120 m. These changes of the
surface topography promote the build up of the film thickness and increase the load-
carrying capability, Figure 24.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
36
a) b)
Figure 24 a) RMS angle of the inner rings of the tested bearings, b) RMS wave
length of the inner ring surfaces in the running track.
4.5 Refrigerant effect on viscosity and pressure-viscosity
coefficient
In paper C, viscosity and pressure-viscosity coefficient measurements are performed
with POE oil and four different refrigerants. Figure 25 shows how refrigerant affects
the viscosity and pressure-viscosity coefficient of the lubricant when the refrigerant
concentration increases from 0 to 30 wt%. The figure shows a fast reduction of
viscosity and pressure-viscosity coefficient when refrigerant is added to the POE oil.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
37
Temp 40
o
C
Concentration refrigerant [%]
0 10 20 30
V
i
s
c
o
s
i
t
y

[
c
P
]
0
20
40
60
80
R-32
R-134a
R-410a
R-22
Temp 40
o
C
Concentration refrigerant [%]
0 10 20 30
P
r
e
s
s
u
r
e
-
V
i
s
c
o
s
i
t
y

c
o
e
f
f
.

[
G
P
a
-
1
]
10
15
20
25
R-32
R-410a
R-22
R-134a
a) b)
Figure 25 Viscosity and Pressure- viscosity coefficient reduction due to refrigerant
dilution.
The experiments show that the molecular weight of the refrigerant controls the
degree of curvature of the reduction in viscosity and pressure-viscosity coefficient.
The lighter R-32 gives a faster decrease in viscosity compared to the heavier
refrigerant R-134a.
4.6 The effect of refrigerant on lubricant density
In paper D, a high-pressure chamber is used to study lubricant compressibility when
refrigeration oil is diluted by refrigerant. The results of the compressibility
measurements are shown in Figure 26 for the tested lubricants. The change in
volume with increasing pressure for pure mineral oil corresponds to measurements
done by previous authors. The relative change in volume for mineral oil at 1 GPa is
0.17 in the current study compared to 0.18 in Lindqvist et al. [45] and Sthl and
Jacobson [44]. The stiffness of the POE/R-134a mixture is even higher than that of
5P4E given in [56]. POE diluted with 20 wt% of R-134a has a relative volume
change of 0.1, whereas 5P4E has 0.14 at 1 GPa pressure.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
38
a) b)
Figure 26a) Compressibility as function of pressure, for pure mineral oil and mineral
oil diluted with 20 wt% R-22, b) compressibility as function of pressure, for pure
POE and POE diluted with 20 wt% R-134a
By adding the non-chlorinated refrigerant R-134a to the polyolester oil, the
compressibility decreases by approximately 40% at 1 GPa. The decrease in
compressibility is larger than for the R-22 and mineral oil. The mixture of POE and
R-134a is even stiffer than 5P4E.
Figure 27 shows the density change due to pressure for measurements of POE oil
with and without refrigerant. The figure also shows the model proposed by Dowson
and Higginson for mineral oil. The Jacobson and Vinet model is fitted to the
measured data. The measurements fit the Jacobson and Vinet at pressures below the
solidification pressure.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
39
Figure 27 Comparison of models regarding relative density as a function of pressure.
The figure shows the Jacobson-Vinet model fitted to measurement data.
4.7 The refrigerants effect on lubricant shear strength.
In paper E, the friction coefficients of both diluted and undiluted refrigerants are
tested. Figure 28 shows the results of the experiments. Each box represents five
measurements, with the height of the box indicating the deviation of the
measurement. This deviation is due to refrigerant boiling out of the oil during the
experiment. Experiments to measure the refrigerant concentration during the shear
strength experiments are done and the concentration of refrigerant is 6-9 wt% for the
polyolester oil and R-134a and 5-6 wt% for napthenic mineral oil and R-22. The
results indicate that the refrigerant in all cases increases the friction coefficient.
E
s
t
e
r
A
E
s
t
e
r
A
R
-
1
3
4
a
E
s
t
e
r
B
E
s
t
e
r
B
R
-
1
3
4
a
E
s
t
e
r
C
E
s
t
e
r
C
R
-
1
3
4
a
M
i
n
e
r
a
l
M
i
n
e
r
a
l
R
-
2
2
F
r
i
c
t
i
o
n

c
o
e
f
f
i
c
i
e
n
t
0,026
0,028
0,030
0,032
0,034
0,036
0,038
Figure 28 Friction coefficients of the undiluted oils and oils diluted with refrigerant.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
40
Once the lubricant sample is placed in ambient pressure and temperature on the
tungsten carbide plate, the refrigerant will begin to evaporate. Because the time to
perform a test in the test rig was known, the refrigerant concentration in the oil at the
time of measuring could be assumed for each oil mixture.
Experiments concerning the evaporation time for refrigerant to leave the oil are done
for the different oils, resulting in different lubricants containing different amounts of
refrigerant under the same conditions, i.e. ambient pressure and temperature. The
data is used to predict the refrigerant concentration in the friction experiments.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
41
5 Concluding remarks
The research presented in this thesis gives information about parameters that
influence the life of rolling element bearings. Test apparatuses and test methods are
developed and used to evaluate the lubricating conditions in refrigeration
compressors, where oil mixed with refrigerant is used as lubricant. The general
conclusion is that R-134a refrigerant negatively affects the lubrication. The
refrigerant decreases the viscosity and pressure- viscosity coefficient, and increases
the shear-strength and compressibility of the lubricant. The effects of the refrigerant
decrease the film thickness, and increase the von Mises stress due to increased
shear-strength and compressibility.
Test apparatus
The objective of this investigation was to develop a test apparatus to simulate the
operation of bearings in a refrigeration compressor lubricated with oil/refrigerant
mixtures and determine the lubricant properties of the mixture. A capacitive
measuring technique is used to study lubricant film build-up in the bearing. The test
apparatus is able to reproduce and monitor the running conditions of a bearing
operating in a refrigeration environment. V
cap
measurements give uncertain
information about forthcoming bearing failure. There is good correlation between
the observed denting/wear on the bearing surface and the detected number of
contacts. The detected number of contacts gives a good indication of change in film
thickness in the contact. The run-in behaviour can be studied by using the
capacitance method to detect metal-to-metal contacts in the bearing. The threshold
level affects the resolution of the experiments. The use of a threshold level that is
95% of the mean of 40,000 samples is tested to obtain good resolution in the
experiments.
The dielectric constant is a key parameter when measuring capacitance. As long as
the medium is the same, experiments can be compared to each other. If the medium
changes, i.e. higher refrigerant concentration, different oil or refrigerant, the
measurements must be compensated for the change in dielectric constant.
When using a frequency converter to power the electrical motor an output filter is
required between the converter and motor. The noise produced by the converter can
be interpreted as occasional contacts when V
cap
is measured and as a contact when
using the threshold level to count contacts.
POE lubricants with additives
Experiments with rolling element bearings used in a refrigeration environment are
performed with two polyolester oils with two different phosphate additives. The
lubrication status is monitored during the experiments using the capacitance method.
The tested bearing rings were examined in the SEM and the 3D surface topometer.
The capacitance measurement shows that polyolester oil, including phosphate
additives in an R-134a environment, increase the lubricating film thickness.
Phosphate ester additive in the POE/R-134a reduces the wear of the bearing surface
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
42
and shows the same amount of wear as mineral oil/R-22. The amount of denting on
the bearing surfaces is related to the amount of contact that occurs during operation.
The relative dielectric constant increases with increasing refrigerant concentration
and the constant is higher with POE oil than with mineral oil.
The surface topography of the bearings changes to a more favorable profile with a
lower RMS angle and a longer RMS wavelength that promotes load-carrying
capacity and film build-up. The skewness, RMS roughness, S
q
, average maximum
height and S
z
are improved during running with all POE/R-134a lubricants.
Viscosity
The viscosity and the pressure-viscosity coefficients of various refrigerants and
polyolester pairs were investigated using a high-pressure falling ball viscometer.
Increasing the refrigerant concentration decreases the viscosity and pressure-
viscosity coefficients. Already at a refrigerant concentration of 10 wt% the viscosity
is reduced to 25% of that of the base oil. A strong correlation was found between the
molecular mass of the refrigerant and the reduction of viscosity and pressure-
viscosity coefficients with increasing concentration. A refrigerant with a light
molecule shows a more rapid behaviour than a heavy molecule.
Compressibility
A high pressure chamber is used to derive the compressibility of oil/refrigerant
mixtures in pressures up to 3.1 GPa. A polyolester oil mixed with the HFC
refigerant R-134a and a napthenic mineral oil mixed with the HCFC refrigerant R-
22 were used in the investigation. By adding the non-chlorinated refrigerant R-134a
to polyolester oil, the compressibility decreases by approximately 40% at 1 GPa.
The decrease in compressibility is larger than for the R-22 and mineral oil. The
mixture of POE and R-134a is even stiffer than 5P4E. A stiffer lubricant affects the
increases in magnitude of the pressure spikes in the pressure distribution. The model
suggested by Jacobson and Vinet fits well with the experimental data.
Traction
Measurements of the shear strength coefficient, , at extreme pressures is conducted
on four refrigeration oils; undiluted and diluted with refrigerant. The objective of the
work is to examine the effect refrigerants have on . Since is coupled to the
limiting shear stress, any increase in will also increase the maximum stress in the
material and cause the maximum stress to act closer to the surface.
The results show that at a pressure of 2.5 GPa increases when a refrigerant is
present at a concentration of approximately 9%. The investigation also agrees with
the results of earlier investigations indicating that the molecular structure of the
lubricant affects the shear strength pressure coefficient [] for a given application,
but the results differ from Muraki and Sano [37].
The work also showed that the concentration of R-22 in mineral oil evaporates much
faster than R-134a in the polyolester mixtures.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
43
General conclusion
Lubrication of rolling element bearings in refrigeration compressor causes much
trouble for the compressor manufactures. In this thesis parameters affecting the life
of rolling element bearings have been measured. All of the measured parameters are
negatively affected by the refrigerant content in the oil. One solution is to design oil-
free compressors which use just a small amount of oil or only the refrigerant as
lubricant.
If it is not possible to run oil free, it is of great importance that lubrication has
priority during the design process, so the right running conditions for the bearings
are found and so that the right bearings are chosen in the design. When the
compressor is in use, the running conditions must be similar to those for the design,
which means right temperature of the oil, control of the refrigerant concentration and
control of contamination particles. Thinner film thickness due to high temperature or
high refrigerant concentration increases the damages from contamination particles in
the lubricant.
Another way to get problem free lubrication of refrigeration compressor is to use
hybrid rolling element bearings. In a hybrid bearing the balls are made of a ceramic
material and the rings are made of steel. The balls in the bearing have a healing
function on the steel rings, SKF [57], since the balls are much harder than the steel
rings and of a different material and therefore the balls smooth the rings surface.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
44
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
45
Recommendations for further work
Investigation of the stress levels in the bearing surface that has been running and
lubricated with an oil and refrigerant mixture, would give valuable information of
the expected fatigue life of the bearing and a tool to set the right operating
parameters for the bearing.
Endurance experiments need to be done for rolling element bearings lubricated with
an oil and refrigerant mixture. The experiments would give an answer to the
refrigerants influence of the reduction of the bearing life.
The increasing global warming places new demands on HFC refrigerants that act as
greenhouse gases. The consequences will be that taxes, tougher with regulations or
both will eliminate HFC refrigerants from the market. Three alternatives are
proposed as replacements, viz. carbon dioxide (CO
2
), ammonia and propane. Both
carbon dioxide and ammonia were used in the early days of mechanical
refrigeration. Today, the ability of advanced engineering and manufacturing makes
it possible to deal with the problems that engineers struggled with at the beginning
of the 20
th
century. The new refrigerants will put new demands on the refrigeration
oils. The lack of understanding regarding lubrication with new refrigerants and oils
needs further research, combined with variable-speed drive of the compressor that
allows variable cooling capacity of the refrigeration system. This will place new
demands on engineers when designing a compressor with good lubrication at all
speeds.
New materials of parts in the compressors and the bearings that allows tougher
lubricating conditions is a demand that the new generation of oil-free compressors
will drive.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
46
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
47
6 References

1 C. Englund. Den Svenska Kylteknikens Historia. Svenska kyltekniska
freningen, Sweden, ISBN: 91-86590-07-3, 1992.
2 M. J. Molina and F. S. Rowland. Stratospheric sink for chlorofluoromethanes:
chlorine atomic catalysed destruction of ozone. Nature, vol. 249, pages 810-
814, 1974.
3 H. H. Kruse and M. Schroeder. Fundamentals of Lubrication in Refriegration
Systems and Heat Pumps. Ashrae Journal, vol. 26, 5, pages 5-9, 1984.
4 G. D. Short. Development of Synthetic Lubricants for Extended Life in
Rotary-Screw Compressors. Lubrication Engineering, vol. 40, 8, pages 463-
470, 1984.
5 K. S. Sanvordenker. Durability of HFC 134a Compressors- The role of the
lubricant. Ashrae Journal, vol. 33, 2, pages 42-56, 1991.
6 G. Lundberg and A. Palmgren. Dynamic capacity of rolling bearings. Acta
Polytechnica, Mechanical Engineering, vol.1, 3, 1947.
7 T. E Tallian. On compeating failures modes in rolling contacts. ASLE
Transactions, vol. 10, pages 418-439, 1967.
8 J. G. Wuttkowski and E. Ioannides. The new life theory and its practical
consequences. Ball Bearing Journal, ISSN 0308-1664, pages 6-10, 1989.
9 SKF. General Catalogue. Edition 5000E, 2003.
10 F. P. Wardle, B. Jacobson, H. Dolfsma and A. Butterworth. Lubrication of
refrigerant compressor bearings. US Patent, 5469713, 1995.
11 K. Meyers. Creating the right environment for the compressor bearings.
Evolution, Number 4, 1997.
12 E. Ioannides and G. Bergling and A. Gabelli. The SKF formula for bearing
life. Evolution, Number 1, 2001.
13 D. Epstein, T. Yo, J. Wang, L. M. Keer, H. S. Cheng and S. Lui. An Efficient
Method of Analysing the Effect of Roughness on Fatigue Life in Mixed-EHL
Contact. Tribology Transactions, vol. 46, 2, pages 273-281, 2003.
14 D. Epstein, L. M. Keer, Q. J. Wang and H. S. Cheng. Effect of Surface
Thopography on Contact Fatigue in Mixed Lubrication. Tribology
Transactions, vol. 46, 4, pages 506-513, 2003.
15 J. H. Tripp, J. van Kuilenburg, G. E. Morales-Espejel and P. M. Lugt.
Frequency response functions and rough surface stress analysis. Tribology
Transactions, ISSN: 1040-2004, vol. 46, 3, pages 376-382, 2003.
16 B. Jacobson. Nano-Meter Film Rheology and Asperity Lubrication. Journal
of Tribology, vol. 124, pages 595-599, 2002.
17 B. J. Hamrock and D. Dowson. Isothermal Elastohydrodynamic Lubrication
of Point Contacts Part IIIFully Flooded Results. Journal of Lubrication
Technology, Transactions ASME, vol. 99, 2, pages 264-276, 1977.
18 B. J. Hamrock. Fundamentals of fluid film Lubrication. NASA Reference
Publication 1255, 1991.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
48

19 P. Cann, E. Ioannides, B. Jacobson and A. A. Lubrecht. The lambda ratio a
critical re-examination. Wear, vol.175, pages 177-188, 1994.
20 U. J. Jonsson and N. Hansson. Lubrication Limits of Rolling Element
Bearings in Refrigeration Compressors. Proceedings of 1998 International
Compressor Engineering Conference at Purdue, vol. 1, pages 129-134, 1998.
21 B. Jacobson. Lubrication of screw compressor bearings in presence of
refrigerants. Proceedings of 1994 International Compressor Engineering
Conference at Purdue, vol. 1, pages 115-120, 1994.
22 A. Dyson, H. Naylor and A. R. Wilson. The Measurements of Oil-Film
Thickness in Elastohydrodynamic Contacts. Proceedings of Institution of
Mechanical Engineers, vol. 180, Pt 3B, pages 119-134, 1965-66.
23 W. E. ten Napel. and R. Bosma. The influence of surface roughness on
capacitive measurement of film thickness in elastohydrodynamic contacts.
Proceedings of Institution of Mechanical Engineers, vol. 185, Paper 37/71,
pages 635-639, 1971.
24 Z. Cheng and P. S. Zhang. Simplified Tribological Designing Calculation and
an Effective EHL Measurement Techinque for High-Speed Bearings.
Proceedings of the 8th International Colloquim on Tribology, Teknische
Akademie Esslingen, 1992.
25 J. Lord. Mixed and Full Film EHL Contact Condition Analysis by
Simultaneous Acquisition of its Resistance and Capacitance. Presented at the
ASME/STLE Joint Tribology Conference, 26-27 October 2003, Ponte Vedra
Beach, Florida, USA. 2003.
26 R. S. Heemskerk, K. N. Vermeiren and H. Dolfsma. Measurment of
Lubrication Conditioning in Rolling element Bearings. ASLE Transactions,
vol. 24, pages 519-527, 1982.
27 M. A. Masen, C. H. Venner, P. M. Lugt and J. H. Tripp. Effects of Surface
Micro-Geometry On the Lift-Off Speed of an EHL Contact. Tribology
Transactions, vol. 45, pages 21-30, 2002.
28 V. Wikstrm and B. Jacobson. Loss of lubricant from oil-lubricated near
starved spherical roller bearings. Institution of Mechanical Engineers, vol.
211, part J, pages 51-66. 1996.
29 S. Komatsuzaki and Y. Homma. Antiseizure and Antiwear Properties of
Lubricating Oils Under Refrigerant Gas Environments. Lubrication
Engineering, vol. 47, 3, pages 193-198, 1989.
30 K. Mizuhara, M. Akei and T. Matsuzaki. The friction and Wear Behavior in
Controlled Alternative Refrigerant Atmosphere. Tribology Transactions, vol.
37, 1, pages 120-128, 1994.
31 K. Mizuhara. Lubrication in Alternative HFC Refrigerants. Japanese Journal
of Tribology, vol. 38, 5, page 414, 1993.
32 S. F.Murray, R. L. Johnson and M. A. Swikert. Difluorodichloromethane as a
Boundary Lubricant for Steel and Other Metals. Mechanical Engineers, vol.
78, pages 233-236, 1956.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
49

33 B. Jacobson. Ball Bearing lubrication in refrigeration compressors.
Proceedings of 1996 International Compressor Engineering Conference at
Purdue, vol. 1, pages 103-108, 1996.
34 H. P. Nixon and H. Zantopulos. Lubricant Additives Friend or Foe, What the
Equipment Designer Engineer Needs to Know. Lubrication Engineering,
ISSN: 0024-7154, vol. 51, 10, pages 815-822, 1995.
35 G. T. Y. Wan, H. Lankamp, A. de Vries and E. Ioannides. The effect of
extreme pressure (EP) lubricants on the life of rolling element bearings.
Journal of Engineering Tribology, vol. 208, 4, pages 247-252, 1994.
36 M. Akei and K. Mizuhara. The Elastohydrodynamic Properties of Lubricants
in Refrigerant Environments. Tribology Transactions, vol. 40, 1, pages 1-10,
1997.
37 M. Muraki, T. Sano and D. Dong. Rheological Properties of Refrigerant-
/Lubricant Mixture Under a Refrigerant Atmosphere. Japanese Journal of
Tribology, vol. 44, 2, pages 121-127, 1999.
38 U. Jonsson. Elastohydrodynamic Lubrication and Rheology in Refrigeration
Compressors, Doctoral Thesis, Lule University of Technology, Sweden,
ISSN:1402-1544; 1998:17, 1998.
39 D. Henderson. Solubility, viscosity and density of refrigerant/lubricant
mixtures. The Air-Conditioning and Refrigeration Technology Institute. Rep.
DOE/CE/23810-34, 1994.
40 B. J. Hamrock, P. Pan and R. T. Lee. Pressure Spikes in Elastohydro-
dynamiclly Lubricated Conjunctions. Journal of Tribology. vol. 110, pages
279-284, 1988.
41 G. R. Higginson and D. Dowson. Elastohydrodynamic Lubrication, The
Fundamentals of Roller or Gear Lubrication. Pergamon Press, Oxford, 1966.
42 B. J. Hamrock, B. O. Jacobson and S.-I. Bergstrm. Measurement of the
Density of Base Fluids at Pressures to 2.2 GPa. NASA TM-87114, 1987.
43 B. O. Jacobson and P. Vinet. A Model for the Influence of Pressure on Bulk
Modulus and the Influence of Temperature on the Solidification Pressure for
Liquid Lubricants. ASME Journal of Tribology, vol. 109, pages 709-714,
1987.
44 J. Sthl and B. O. Jacobson. Compressibility of Lubricants at High Pressures,
Tribology Transactions, vol. 46, 4, pages 592-599, 2003.
45 S. Lindqvist. Compressibility and Density of Lubricants in Transient
Loading. Licenciate Thesis, Lule University of Technology, Sweden,
ISSN:0280-8242; 1995:35L, 1995.
46 P. Vinet, J. Ferrante, J. R. Smith and J. H. Rose. A Universal Equation of
State for Solids. Solid state Physics, vol. 19, pages 467-473, 1986.
47 B. O. Jacobson and B. J. Hamrock. Non-Newtonian Fluid Model Incorporated
into Elastohydrodynamic Lubrication of Rectangular Contacts. NASA
Technical Memorandum 83318, 1983.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
50

48 E. Hglund. Subsurface Stresses in a Lubricated Rolling/Sliding
Elastohydrodynamic Line Contact Considering Limited Strength of the
Lubricant. Doctoral Thesis, Lule University of Technology, Sweden, ISSN
ISSN:0348-8373; 32D, 1984.
49 E. Hglund and B. Jacobson. Experimental Investigation of the Shear
Strength of Lubricants Subjected to High Pressure and Temperature. Journal
of Tribology, vol. 108, pages 571-578, 1986.
50 K.-G. Sundin and B.-O. hrstrm. Method for investigation of frictional
properties at impact loading. Journal of Sound and Vibration, vol. 222, 4,
pages 669-677, 1999.
51 B.-O. hrstrm. Investigation of frictional properties of lubricants at transient
EHD-conditions. ASTM Special Technical Publication, 1404, pages 221-234,
2001.
52 M. Muraki and T. Sano. Determination of Film Thickness and Traction of
Polyolester under EHD Contact in Some Refrigerant Environments.
Tribology Transactions, vol. 43, 1, pages 15-20, 2000.
53 U. Jonsson and E. Hglund. The influence of refrigerants on the high pressure
properties of lubricating oil. Proceedings of Nordtrib 1992, vol. 3, pages 89-
96, 1992.
54 C. Barus. Isothermals isopiestics and isometrics relative to viscosity. Am. J.
Sci., vol. 45, pages 87-96, 1893.
55 B. O. Jacobson. A New High Pressure Chamber for Lubricant Investigations.
Proceedings of the International Tribology Conference, Nagasaki, Japan,
ISBN 4-9900139-5-6, pages 1199-1202, 2000.
56 E. Hglund. Influence of Lubricant Properties on Elastohydrodynamic
Lubrication. Wear, vol. 232, 2, pages 176-184, 1999.
57 H. H. Wallin and G. Morales. Hybridlager fr oljefria luftkonditionerings-
och kylkompressorer. Evolotion, Number 2, 2002.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
51
Paper A
Measurement of Lubricating Conditions in Rolling
Element Bearings in a Refrigerant Environment
Authors:
Roger Tuomas and Ove Isaksson
Reformatted version of paper originally published in:
R. Tumas and O. Isaksson. Measurement of Lubricating Conditions in Rolling
Element Bearings in a Refrigerant Environment. Submitted, 2006.
52
53
Measurement of Lubricating Conditions in Rolling
Element Bearings in a Refrigerant Environment
R. Tuomas and O. Isaksson
Division of Machine Elements
Lule University of Technology
SE-971 87 Lule
Sweden
Phone: +46 920 492414 Fax: +46 920 491047
E-mail: [email protected]
Keywords
Lubrication, condition monitoring, refrigeration, compressor, capacitance, ball
bearing.
Abstract
The ability to build up a lubricating film thickness separating the contacts in a
bearing is essential to assure long service life of rolling element bearings. Rolling
element bearings used in refrigeration suffer from poor film build up due to
decreased viscosity of the lubricant by the dilution of refrigerant in the oil. In this
paper, a bearing test apparatus is redesigned and equipped with a capacitance
measuring device able to monitor the lubrication status in the bearing on-line. The
paper includes experiments to verify the capacitance measuring technique.
The objective is to design a complete system to study film build up in rolling
element bearings in a refrigerant environment and to use and evaluate the
capacitance/resistance measuring technique.
The investigation shows good correlation between denting on tested bearings and
the identified contacts by the capacitance measuring apparatus. The method is also
useful when studying lift-off or run-in of a bearing.
Introduction
The service life of a rolling element bearing in a particular application depends on a
variety of factors, including lubrication, contamination and environmental
conditions. The lubricating film separating the rolling element contact surfaces is
critical for life of the bearing. To attain long life, the moving surfaces should be
completely separated by a lubricant film. When the film thickness is too thin surface
asperities penetrate the oil film or particles trapped between the surfaces make direct
metal-to-metal contact. These contacts result in high local stresses, and wear or
plastic deformation at the asperity level takes place. An experimental apparatus that
monitors the film build up, from mixed to full film lubrication, in a complete rolling
54
bearing system without requiring any modification is a useful tool to study new
lubricants or bearings in a tough environment.
The most commonly used techniques to monitor the status of the lubricating film on-
line are based on resistance, capacitance, or both. The benefits of the methods are
that they can be applied to a bearing in real applications, meaning that the
lubrication status is measured during operation.
Dyson et al. [1] estimated the oil film thickness between two discs from
measurements of the electric capacitance between the discs. The capacitance was
measured with a frequency bridge. The bridge balance was disturbed by the frequent
bridging of the film directly between asperities of the opposing surfaces or through
conduction particles trapped between them. Choosing a higher carrier frequency and
filtering the bridge output reduced the disturbance.
ten Napel and Bosma [2] studied the influence of surface roughness on the
capacitance measurements of film thickness and found that the deviation between
measured and theoretically predicted values was mainly due to the surface
roughness of the surfaces. The closer the surfaces are to being perfectly smooth, the
better the correlation will be between theoretical and experimental estimated film
thickness. Using capacitance to monitor film thickness in an elastohydrodynamic
contact gives reasonably good results when the surfaces are completely separated.
Studying film thickness when reaching mixed lubrication calls for other measuring
techniques than capacitance. Several devices combining the resistance and
capacitance measuring technique have been designed; Cheng and Zhang [3] and
Lord [4]. Lord concluded that his method is even capable of measuring contact
conditions in the mixed lubrication regime. When the surfaces are completely
separated the capacitance gives a measure of the film thickness; when incidental
metal-to-metal contact begins the conductivity provides information of the film
thickness. However, the measured impedance will be a combination of resistance
and capacitance in the contact. The acquired data only give a mean value of the
capacitance or resistance in the contact. Any transient changes of the contact
conditions, such as an asperity entering the contact, cause changes in resistance and
capacitance.
Heemskerk et al. [5] designed a device which main objective was to measure contact
capacitance. The electronic design does, however, view the contact as a combined
resistance and capacitance. The apparatus has been used by several researchers to
study elasto hydrodynamic contacts. Jacobson [6] and Masen et al. [7] used the
apparatus to measure the lift-off speed and Wikstrm and Jacobson [8] used it to
study lubricant starvation in a grease lubricated spherical bearing. The potential over
the bearing must be low to avoid electrical discharge and sparking in the contact.
When the apparatus indicates contact, the outer ring temperature immediately
increases, indicating asperity contact. In this paper the capacitive method utilizes the
bearing inner ring, the balls and the outer ring as multiple series variable
capacitances. Since there are two contact points at every ball, detecting exactly
where a decrease in oil film or contact occurs is not possible, only that it occurs
somewhere in the bearing. Jacobson [9] also found that even at an indicated full
55
separation, wear occurred on the bearing surface. Full separation in one contact was
shown to cause the capacitance method to indicate full film.
The objective of this work is to design a test rig to study film build up in rolling
element bearings in a refrigerant environment and to use and evaluate the measuring
technique. The capacitive method developed by Heemskerk et al. will be used to
monitor the film thickness in the bearings.
Test apparatus
The experiments are done in a modified bearing test apparatus designed by Hansson
and Jonsson [10]. The test bearings used in the apparatus are angular contact
bearings, radial roller bearings or CARB
TM
. The test apparatus is designed to sustain
pressures up to 3 MPa and is able to simulate the running conditions for bearings
operating in a refrigeration compressor. The main parts of the test apparatus are
shown in Figure 1.
a) b)
Figure 1a). Test apparatus b) schematic overview of the main part of the test
apparatus.
In the investigation, angular contact ball bearings (7210BEP) are mainly used, since
they are commonly used in refrigeration screw compressors. To get a variable shaft
speed from stationary up to 12,000 rpm, an electrical motor is supported by a
frequency converter, giving maximum speed at 400 Hz. The output current from the
frequency converter is filtered through an electrical filter to smooth the sinusoidal
wave and reduce noise that disturbs the measurements. The high-speed spindle with
the test bearings is driven by an integrated electrical motor. A spring package with
twelve springs loads the angular contact bearing axially. Altering the number of
springs or spring stiffness will change the load. Figure 2 shows the load
characteristics for 12, 6 and 3 springs.
56
Figure 2 Load characteristics of the spring package. 12 springs C/P 5.7, 6 springs
....
C/P 11.4 and 3 springs C/P 22.8
The oil circulation system consists of a 4-litre reservoir, heat exchanger, two gear
pumps, filter, viscometer, test chamber and pipes. The reservoir is equipped with
heaters to preheat the lubricant. The heat exchanger uses water to decrease and
control the lubricants temperature. The two-gear pumps circulate the lubricant in
the system. The pump capacity of each pump is 5 litres/minute and secures fully
flooded lubrication of the test bearing. The lubricant is filtered through a 3 m filter
before it reaches the test chamber and the test bearing.
To monitor the lubrication status and separation in the bearing during the test, a
capacitive measuring apparatus originally developed by Heemskerk et al. [5] is used.
The technique uses a high frequency alternating current of 410 kHz. Due to the
combined resistance and capacitance technique, a varying film thickness can be
detected at both full EHL lubrication and in mixed lubrication. The instrument
output voltage, V
cap
, is a measure of surface separation. The apparatus is capable of
measuring high-frequency oscillations in capacitance when incidental asperity
contacts occur between balls and the inner- and outer- race. To not damage the
bearing by sparking, the voltage feed to the bearing is limited to 90 mV. To acquire
data at a sample rate of 100,000 samples/sec, an oscilloscope is used and 40,000
samples are captured every second.
The capacitance measuring apparatus is attached to the inner ring through the shaft
by a mercury connector at the rear end of the shaft and to the insulated housing in
connection with the outer ring. The inner and outer rings of the test bearing are
electrically insulated from each other to allow the capacitance to be measured,
defined in eq.1, Storken [11]. The outer ring is in contact with the insulated housing
and the inner ring with the shaft of the test apparatus.
m ref
ref
cap
C C
V C
V
+

=
max
(1)
57
Since the 7210BEP bearing consists of 16 balls, 16 contacts occur between the inner
ring and ball and 16 between the outer ring and ball. These two sets of parallel
arranged capacitances in series give the measured capacitance.
Each capacitance is defined by the relation in eq.2.
h
A
C
R
m

=

0
(2)
In the equation,
R
is the relative permittivity or dielectric constant of the lubricant
captured in the contact.
The dielectric constant is estimated by using the variable capacitor shown in Figure
3. During the experiments, viscosity, temperature and pressure are controlled and
monitored by attaching the variable capacitor to the main apparatus. The capacitance
is measured between the tip of a micrometer and the corresponding plate that is
insulated from the housing. The gap between the plates can be adjusted by turning
the micrometer. The measurements are done with a gap of 0.1 mm.
Figure 3 Variable capacitor
58
The lubricants used in the experiments to evaluate the influence of the relative
dielectric constant on V
cap
are shown in Table 1.
Table 1 Properties of the oils used to evaluate the effect of relative dielectric
constant on V
cap
.
Oil Chemical composition of oil Refrigerant Density kg/m
3
15 C
Polyolester Esterified Mono-pentaerythritol R-134a 954
Polyolester +
0.25% Acid
Phosphate
Esterified Mono-pentaerythritol with
0.25% Acid phosphate additive
R-134a 954
Polyolester +
1.5%
Phosphate
ester
Esterified Mono-pentaerythritol with
1.5% Phosphate ester additive
R-134a 954
Mineral oil Napthenic mineral oil R-22 910
To monitor the viscosity during the experiments an on-line viscometer is used.
Based on a simple and reliable electromagnetic concept, its accuracy is 1% in the
viscosity range. Two coils magnetically move a piston back and forth at a constant
force. The travelling time is measured for the back and forth movement and the
cycle time are used to calculate the absolute viscosity of the fluid. The viscometer is
equipped with a temperature sensor that measures the temperature of its fluid. The
viscometer is mounted in a housing with a cross section area large enough to
decrease the fluid flow speed below the recommended maximum flow speed of 0.3
m/s.
A bearing vibration monitoring system measures the vibrations produced by the
bearing. The system includes a vibration sensor that is attached to the test chamber
and picks up vibrations. The sensor measures shock pulses in dB and converts them
into a mA output signal.
Before each experiment starts, the pumps are turned on and the lubricant circulates
through the filter for two hours to reduce the contamination level. The viscosity and
temperature are also stabilised to the correct level for the experiment. The lubricant
temperature is constant during the experiments, 401C.
The control program monitors the measurement from the thermocouples, vibration
sensor, pressure sensor, speed sensor and viscometer. The control program uses the
measurements to control the lubricant temperature, alarms and power to the
electrical motor during the experiment. Each sensor has an alarm that turns off the
test apparatus if any limit is reached. The lubricant temperature is measured by a
thermocouple before the lubricant enters the test chamber, the measurement of the
temperature is used to control the water flow to the heat exchanger. Cooling of the
lubricant is needed during operation, since the bearing and mechanical seal produce
much heat.
59
The bearing surface is analyzed in a topometer and a Scanning Electron Microscope
(SEM) provides highly magnified pictures of the surfaces. The topometer uses
optical interferometry to measure the surface roughness.
Film thickness
According to the Hamrocks & Dowsons theory [12] of EHD, the minimum film
thickness obtained by solving the coupled Reynolds, rheology and elasticity
equations for a lubricated elliptical contact is given by eq.3.
) 1 ( 63 , 3
~
68 , 0 073 , 0 49 , 0 68 , 0 min
min ,
k
x
e
e W G U
R
h
H

= =
(3)
U is the dimensionless parameter for speed, G for materials and W for load, and R
x
is the effective radius of the bearing.
The dimensionless film parameter, , is often used to define which lubrication
regime prevails. The relationship between the film parameter and the minimum film
thickness in the lubricated contact is given by eq.4.
( )
2
1
2
2
2
1
min
q q
S S
h
+
=
(4)
h
min
is the minimum film thickness, S
q1
and S
q2
are the area RMS deviations of the
surfaces ball and track. According to Hamrock [13] a -value between 1-5 indicates
partial lubrication, or at least 3 to attain elastohydrodynamic lubrication.
Results and discussion
Experiments are conducted to verify the functioning of the test apparatus and the
measuring system. The experiments also show different ways to use the measuring
technique as a tool to monitor and analyze the lubrication status.
Contact detection using the capacitance method
To identify and count the number of contacts, a signal level that can be used as a
threshold to be counted as a contact must be defined. Henceforth, if the output signal
drops below the threshold level in Figure 5 it is recognised as a metal-to-metal
contact. The threshold in the experiments is set to 95% of the mean of 40,000
measured V
cap
values each second.
60
Figure 5 Output signal from capacitance measurements, V
cap
. An output reading
below the straight line is defined as contact.
To confirm the occurrence of contacts, measurements were carried out with a
gradually reduced viscosity so that contacts could be detected. Figure 6 shows
measurements where the viscosity is reduced by increasing the refrigerant
concentration in steps up to about 25 wt%. During the experiment the number of
contacts was counted, and increased significantly when the concentration level
exceeded 20 wt%. The measurements were repeated with a new bearing, though the
experiment was stopped at 20 wt% refrigerant concentration, see Figure 6b. The
output signal from the two measurements yields comparatively similar results up to
20 wt% refrigerant concentration.
61
a) b)
Figure 6 Counted number of contacts during 10 hour testing with Solest 68/R-134a
mixture at 3,000 rpm and bearing load ratio C/P = 5.7; a) Step by step increased
refrigerant concentration up to 27 wt%, b) Step by step increased refrigerant
concentration up to 20 wt%.
Typical SEM images Figure 7 show denting of the surfaces and the grinding marks
of the surfaces from being worn during the experiment. The bearing has also been
observed to run at a lower viscosity or lower film thickness, Figure 7a shows more
denting and wear, correlating well with the counted number of contacts. The
calculated film thickness and film parameter are 259 nm and 2.7 at 20 wt%
concentration, and 177 nm and 1.9 at 34 wt% concentration.
a) b)
Figure 7 SEM image of the inner bearing ring surfaces after the experiments. The
experiments are carried out with POE oil and R-134a mixture at 3,000 rpm and
bearing load ratio C/P=5.7. a) Step by step increased refrigerant concentration up to
34 wt%, b) Step by step increased refrigerant concentration up to 20 wt%.
62
The same experiment was carried out with R-22 containing chlorine, Figure 8.
During the measurement, the number of contacts detected with R-22 was
significantly lower than with R-134a. Figure 9 shows SEM images of the bearing
inner ring surfaces after test with R-134a and R-22, clearly showing denting on the
surfaces of both R-134a and R-22. However, R-134a has more dents than R-22,
which is in good correlation to the measurements of contacts detected. The
calculated film thickness and film parameter for the highest refrigerant concentration
for R-134a are 177 nm and 1.9, and for R-22 130 nm and 1.4.
Figure 8 Counted contacts and refrigerant concentration versus time for R-22 and
POE. Test condition; constant speed of 3,000 rpm and load ratio C/P = 5.7 and step
by step increased refrigerant concentration.
a) b)
Figure 9 SEM images of the bearing inner ring surfaces after the experiments are
shown. The experiments are carried out at 1,500 rpm and bearing load ratio C/P=5.7.
a) R-134a with POE ester oil the refrigerant concentration is increased in three steps
up to 34 wt%, b) R-22 with napthenic mineral oil the refrigerant concentration is
increased in three steps up to 27 wt%.
V
cap
versus time during filling of the refrigerant is shown in Figure 10. The
refrigerant is filled in liquid state into the reservoir and mixed with the lubricant.
63
The measurement does not indicate the occurrence of a complete film breakdown.
The running speed was 1,500 rpm, the load C/P=5.7 and the lubricant temperature
40 C.
Figure 10 Measured V
cap
versus time with decreasing lubricant viscosity. R-134a
and POE oil was used in the experiment. In A) viscosity decreased from 10 to 6
mPas and film parameter from 1.7 to 1.2, B) viscosity from 22 to 14 mPas and film
parameter from 3.1 to 2.2, and C) viscosity from 30 to 22 mPas and film parameter
from 3.9 to 3.1.
The lift-off is studied in Figure 11. The number of contacts is counted as the shaft
speed is increased successively from stationary to 3,000 rpm by 150 rpm/s and the
bearing load is kept constant at C/P = 11.4 in Figure 11a. The measurements were
repeated with higher load, C/P = 5.7, Figure 11b.
64
a) b)
Figure 11 Number of contacts with increasing shaft speed for two different load
ratios; a) C/P=11.4 and refrigerant concentration of 20.4 wt%, and b) C/P=5.7 and
refrigerant concentration of 19.6 wt%. R-134a and POE oil is used as lubricant.
Since a frequency converter is used to regulate the shaft speed, the converter
transforms the original sine wave by cutting the wave and producing a new one with
the desired frequency. It was discovered that the converter produced noise
coinciding with the cutting frequency of the output current. Figure 12 shows
experiments with and without the output filter. The up-going peeks coincide with the
cut frequency of 33 Hz produced by the frequency converter at the speed of 1,000
rpm. In Figure 12b output filters are used, reducing the noise and achieving a
constant V
cap
reading. The noise shown in Figure 12a can be mistaken as metal-to-
metal contact and wrong conclusions can be drawn.
65
a) b)
Figure 12a) shows the V
cap
readings when no output filters are used and b) when two
parallel output filters are used.
Dielectric constant
Measurements of the relative dielectric constant show that refrigerant mixed in the
oil influences the relative dielectric constant and thus the V
cap
reading, see Figure
13. The behaviour for POE esters and the mineral oil is the same but at different
levels of relative dielectric constant. This is of concern when the V
cap
readings are
compared to each other after different oil types have been used. As well, the
dielectric constant varies with pressure. In this investigation the relative dielectric
constant is measured at moderate pressures of approximately 5-10 bars. Dyson et al.
[1] show data of dielectric constant at ambient pressure and at 345 MPa, for mineral
and synthetic oils. The trend is for the dielectric constant to increase with increased
pressure, except for two of the tested lubricants.
66
a)
b)
Figure 13a) Change in relative dielectric constant due to refrigerant concentration. b)
influence on calculated V
cap
. POE oil with R-134a refrigerant, Mineral oil with
R-22 refrigerant
Measurement of vibrations
A highly loaded bearing normally fails due to surface fatigue. Often, vibration,
temperature or both are monitored to detect bearing failure. In Figure 14, the
measured V
cap
and vibration level are plotted versus time. The experiments were
carried out with a mixture of POE oil and refrigerant R-134a at a concentration
according to a viscosity of 5.7 mPas. The shaft speed during the experiment was
1,500 rpm and the load ration C/P = 5.7. The vibration and V
cap
data were captured
every hour with a sample rate of 40,000 samples/sec and presented in the figure as
the mean each second.
The vibrations reached the maximum allowed level and the experiment was stopped
after 430 hours. Under these running conditions life according to the life theory, L
10
,
should be 1,700 hours. The vibration levels clearly increase just before failure in
both tests. However, the V
cap
signal based on the mean of 40,000 samples does not
clearly indicate any forthcoming bearing failure, since V
cap
more or less constant.
67
a) b)
Figure 14 The measurement is done during a life experiment over 430 hours. a) V
cap
measured with the capacitance apparatus as function of time. b) vibration level
detected by the vibration sensor as function of time. The lubricant used is a POE oil
with R-134a refrigerant.
Figure 15 shows another bearing that failed already after 130 hours at the same
running conditions as in Figure 14. During this experiment the contacts are counted
instead of V
cap
, as the number of data points under the specified threshold level, see
Figure 5. The figure shows run-in, the steady-state period and the increased number
of contacts before failure.
Figure 15 Contacts counted during a life experiment.
68
Conclusions
The objective of this investigation was to develop a test apparatus to study film build
up in a complete rolling bearing system in a refrigerant environment, and to evaluate
the capacitive measuring technique. The following conclusions can be made:
The capacitance measurement is sensitive to electrical noise, necessitating
an output filter between the frequency converter and a motor in the test
apparatus.
The relative dielectric constant increases with increasing refrigerant
concentration and is different for the tested lubricants.
There is good correlation between denting and wear on the bearing surface
and the detected number of contacts.
The method gives a good indication of change in film thickness in bearing
contact.
The method is useful when studying lift-off or run-in of a bearing.
The V
cap
measurement gives unreliable information about forthcoming
drastic bearing failure.
Acknowledgements
The author would like to thank SKF, CPI Engineering, Trane Company, York
Refrigeration and Carrier for support and help during this work. Thanks also go to
the Swedish Energy Agency for financial support.
69
References

1 A. Dyson, H. Naylor and A. R. Wilson. The Measurements of Oil-Film
Thickness in Elastohydrodynamic Contacts. Proceedings of Institution
Mechanical Engineers, vol. 180, Pt 3B, pages 119-134, 1965-66.
2 W. E. ten Napel. and R. Bosma. The influence of surface roughness on
capacitive measurement of film thickness in elastohydrodynamic contacts.
Proceedings of Institution Mechanical Engineers, vol. 185, Paper 37/71,
pages 635-639,1971.
3 Z. Cheng and P. S. Zhang. Simplified Tribological Designing Calculation and
an Effective EHL Measurement Techinque for High-Speed Bearings.
Proceedings of the 8th International Colloquim on Tribology, Teknische
Akademie Esslingen, 1992.
4 J. Lord. Mixed and Full Film EHL Contact Condition Analysis by
Simultaneous Acquisition of its Resistance and Capacitance. Presented at the
ASME/STLE Joint Tribology Conference, 26-27 October 2003, Ponte Vedra
Beach, Florida, USA. 2003.
5 R. S. Heemskerk, K. N. Vermeiren and H. Dolfsma. Measurment of
Lubrication Conditioning in Rolling element Bearings. ASLE Transactions,
vol. 24, pages 519-527, 1982.
6 B. Jacobson. Ball Bearing lubrication in refrigeration compressors.
Proceedings of 1996 International Compressor Engineering Conference at
Purdue, vol. 1, pages 103-108, 1996.
7 M. A. Masen, C. H. Venner, P. M. Lugt and J. H. Tripp. Effects of Surface
Micro-Geometry On the Lift-Off Speed of an EHL Contact. Tribology
Transactions, vol. 45, pages 21-30, 2002.
8 V. Wikstrm and B. Jacobson. Loss of lubricant from oil-lubricated near
starved spherical roller bearings. Institution of Mechanical Engineers, vol.
211, part J, pages 51-66. 1996.
9 B. Jacobson. Lubrication of screw compressor bearings in presence of
refrigerants. Proceedings of 1994 International Compressor Engineering
Conference at Purdue, vol. 1, pages 115-120, 1994.
10 U. J. Jonsson and N. Hansson. Lubrication Limits of Rolling Element
Bearings in Refrigeration Compressors. Proceedings of 1998 International
Compressor Engineering Conference at Purdue, vol. 1, pages 129-134, 1998.
11 J. Storken. Lubcheck Mk3 User Manual, SKF Engineering & Research
Centre B.V., 1997.
12 B. J. Hamrock and D. Dowson. Isothermal Elastohydrodynamic Lubrication
of Point Contacts Part IIIFully Flooded Results. Journal of Lubrication
Technology, Transactions ASME, vol. 99, 2, pages 264-276, 1977.
13 B. J. Hamrock. Fundamentals of fluid film Lubrication. NASA Reference
Publication 1255, 1991.
70
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
71
Paper B
The effect of phosphate additives on the lubrication
of rolling element bearings in a refrigerant
environment
Authors:
Roger Tuomas and Ove Isaksson
Reformatted version of paper originally published in:
R. Tumas and O. Isaksson. The effect of phosphate additives on the lubrication of
rolling element bearings in a refrigerant environment. International Journal of
Refrigeration, Vol. 30, Issue 1, January 2007, Pages 28-36.
72
73
The effect of phosphate additives on the lubrication of
rolling element bearings in a refrigerant environment
Roger Tuomas and Ove Isaksson
Lule University of Technology
Division of Machine Elements
SE-971 87 Lule,
Sweden
Phone: +46 920 492414
E-mail: [email protected]
Abstract
Chlorine free replacement refrigerants, HFC (hydrofluorocarbons) and HC
(hydrocarbons), mixed with polyolester oil have shown poor lubricating properties.
Phosphate based additives were used to improve the lubricity with refrigerant R-
134a, demonstrating positive effects. In the present paper, the ability to form
lubricating film and reduce wear of two additives, phosphate ester and acid
phosphate, were investigated in a bearing test apparatus.
The results show that phosphate additives in polyol-ester oil, in an R-134a
environment, increase the lubricating film and reduce wear. Surface topography
during the initial run-in changes to a more favorable profile with lower RMS angle
and longer wavelengths that promote load-carrying capacity and film build-up.
Keywords
Phosphate additive, elastohydrodynamic, lubrication, refrigerant, R-134a, wear
Introduction
Important lubricating properties change when oil is diluted with refrigerant. The
relationships of viscosity coefficient- and pressure-viscosity coefficient to
refrigerant concentration are described by several authors [1, 2, 3, 4]. Increasing the
amount of refrigerant decreases both the viscosity and pressure-viscosity
coefficients, and causes a drop in film thickness [3, 4, 5]. It has also been reported
that the lubricating film is too thin without extreme pressure (EP) additives [6].
Chlorofluorocarbons (CFC) and hydrochlorofluorocarbon (HCFC) refrigerants
include chlorine components that act as EP additives [7, 8, 9, 10], making it possible
to run-in the bearing surfaces without reducing the bearing life.
Most research has been done in test rigs under sliding conditions. Further
investigations using non-chloride refrigerant are still of interest, particularly under
rolling conditions conducted in a refrigerant environment. In this investigation a
phosphate ester and acid phosphate are used as additives in POE, polyolester, oil.
Chlorinated refrigerant R-22 mixed with a napthenic mineral oil used as a reference.
The objective of the paper is to investigate if the additives can improve lubrication
74
of rolling element bearings in refrigerant environments, the first being POE with
1.5% phosphate ester. The additive in this sample migrates to the metal surface and
is adhered by van der Waals forces when actual metal to metal contact occurs. The
second additive, a POE with 0.25% acid phosphate, is considered more aggressive
than phosphate ester and is adsorbed to the surface.
The experiments are performed in a bearing test apparatus, with the bearings being
tested under realistic running conditions for bearings used in refrigeration
compressors. The lubrication status during the experiments is monitored by an on-
line capacitance measurement apparatus (Heemskerk et al. [11]), which estimates
the lubricating film thickness. Other authors have used this method to indicate lift-
off and run-in of rolling element bearings [12,13].
The surface structure is changed by the run-in process, where material is removed,
deformed plastically or moved in the contact until a steady-state condition can be
maintained. The film parameter, , criterion is widely used to evaluate the
lubrication condition, though it is far too simple a tool and is therefore invalid at
conditions when < 1, Cann et al. [14].
The RMS (Root-Mean Square) slope and wavelength have shown to be important
surface roughness characterization parameters. To reach an optimum load-carrying
capacity, the surface slope should be 10 to 150 times less for bearings that are run-in
than that of a new bearing surface, Jacobson [15]. Also, the film build up is more
favorable if slopes of the asperities are low.
Different surface characterization parameters are used to determine the run-in of the
bearing surface.
Bearing test apparatus and surface analysis
The experiments were carried out in a test apparatus designed by Jonsson and
Hansson [16], and redesigned by Tuomas [17], Figure 1. The apparatus simulates
the running conditions of rolling element bearings in a refrigeration compressor. In
the investigation single row angular contact ball bearings were used. The bearing,
called 7210BEP, is axially loaded with a spring package. The applied axial load of
11,500 N corresponds to a load ratio of 5.7. The support bearing is a hybrid bearing
whose steel balls are replaced by ceramic balls.
75
a) b)
Figure 1 a) Test apparatus b) Schematic overview of the main part of the test
apparatus.
The oil is filtered through a 3 m filter and the viscosity, , is measured by an on-
line viscometer before it reaches the test bearing.
All the surface roughness measurements in this investigation are done on the inner
ring of the bearing surface. A 3-D surface topometer and Scanning Electron
Microscope (SEM) are used to analyse the bearings topography. The topometer
uses optical interferometry to do absolute measurements of the surface in 3-D.
Highly magnified images of the surfaces at 1,000X and 4,000X are captured by the
SEM.
Capacitance measurement
To monitor the lubrication status and separation in the bearing during the test, a
capacitive measuring apparatus originally developed by Heemskerk et al. [11] is
used. The instrument measures the capacitance between the inner and outer rings of
the bearing and screens it as a voltage, V
cap
defined in eq.1.
m ref
ref
cap
C C
V C
V
+

=
max
(1)
Since the 7210BEP bearing consists of 16 balls, 16 contacts occur between the inner
ring and ball and 16 between the outer ring and ball. These two sets of parallel
arranged capacitances in series give the measured capacitance.
Each capacitance is defined by the relation in eq.2.
76
h
A
C
R
m

=

0
(2)
In the equation,
R
is the relative permittivity or dielectric constant of the lubricant
captured in the contact. The dielectric constant is estimated by using variable
capacitor shown in Figure 2. During the experiments, viscosity, temperature and
pressure are controlled and monitored by attaching the variable capacitor to the main
apparatus. The capacitance is measured between the tip of a micrometer and the
corresponding plate that is insulated from the housing. By turning the micrometer
the gap between the plates can be adjusted.
Figure 2 Variable capacitor
Film thickness and lubricant film parameter
The experiments are conducted when the lubricant viscosity is decreased in steps by
adding refrigerant to the oil. The load ratio is 5.7 and the shaft is continuously
rotating at a speed of 1,500 rpm. The bearing turns 360,000 revolutions between
each filling and is always operated under fully flooded conditions. The lubricant
temperature at the test bearing is controlled and held at 40 1C.
The pressure-viscosity coefficient of the refrigerant-oil mixture is given by eq.3,
Akei and Mizuhara [4],
1 1 2
1 ) 1 (
) ( +
+

=
m c
c m
mix
(3)
where
mix
is the pressure-viscosity coefficient of the mixture, c is the weight
fraction of the refrigerant and m the molecular weight ratio of the oil and refrigerant.

1
used in this investigation is from Tuomas and Jonsson [1]. Also,
2
is estimated
by using the measured data in [1]. The oil molecular weight values were obtained by
77
the oil manufacturer and the refrigerant molecular weight was taken from Akei and
Mizuhara [4].
In the investigation according to the oil manufacturer, the small amount of additive
doesnt influence the viscosity, pressure-viscosity coefficient, density and m-value
and those parameters are assumed to be the same for the pure POE and POE +
additives.
The minimum film thickness h
min
is given for elliptical contacts by eq.4, Hamrock
and Dowson [18],
) 1 ( 63 , 3
68 , 0 073 , 0 49 , 0 68 , 0
min
k
e W G U h

=
(4)
To get an indication of the lubrication status the film parameter, , is estimated,
eq.5. In calculations of the S
q
values for the ball and the race are used.
( )
2
1
2
2
2
1
min
q q
S S
h
+
=
(5)
Tested lubricant
The lubricants used in this investigation are POE oil and a napthenic mineral oil.
The POE oil is tested with and without two different phosphate additives. Table 1
shows properties of the oils used in this investigation. The refrigerants and their
properties can be found in Table 2
Table 1 Properties of the oils used in the experiments.
Oil Chemical
composition of oil
Oil Viscosity
[mPas] at
40 C
Viscosity
Index
VI
Density
kg/m
3
15 C
Solubility
Polyolester Esterified Mono-
pentaerythritol
69 108 954 Good solubility
with HFC and
HCFC
Polyolester +
0.25% Acid
Phosphate
Esterified Mono-
pentaerythritol
additized with
0.25% Acid
phosphate
69 108 954 Good solubility
with HFC and
HCFC
Polyolester +
1.5%
Phosphate ester
Esterified Mono-
pentaerythritol
additized with
1.5% Phosphate
ester
69 108 954 Good solubility
with HFC and
HCFC
Mineral oil Napthenic mineral
oil
68 _ 910 No solubility with
HFC soluble with
HCFC
78
Table 2 Properties of refrigerant used in the experiments.
Refrigerant Chemical composition of
oil
Viscosity [mPa s] at
25 C
Density
kg/m
3
25 C
R-134a Tetrafluoroethane 0.20
Liquid
1206
R-22 Chlorodifluoro-methane 0.16
Liquid
1194
Test procedure
Before an experiment starts, a new bearing is mounted into the test apparatus and
loaded by a spring package to 11,500 N. The lubricant is added and the circulation
of oil is turned on. The experiments start after 2 hours of filtrations and stabilising
the temperature. The first step is to speed up the shaft to 1,500 rpm by starting the
electric motor. The viscosity is measured on-line and refrigerant is added to the first
viscosity-level (concentration-level). V
cap
, pressure, viscosity, temperature and speed
are measured when the viscosity reaches a constant readout. V
cap
is measured for 2
hours at each viscosity level at a sample rate of 100,000 samples/sec. Every second,
40,000 samples are captured and the mean is screened and saved. After two hours of
testing, more refrigerant is added to reach the second viscosity-level. The procedure
is repeated at one more viscosity level. The power to the electrical motor and the
data acquisition are on during the filling of refrigerant. The temperature of the
lubricant is regulated during the experiment by a water cooled heat exchanger and
kept constant at 40 1C. The pressure in the system depends on refrigerant type
and concentration, and varies between 3-12 bars.
The relative dielectric constant is obtained by measuring the capacitance with air
between the plates C
air
in the variable capacitance (Figure 2) and with lubricant as
dielectric medium between plates C
lub
. Hence:
air
lub
R
C
C
=
(6)
To estimate the capacitance in the gap between the plates, C
x
, eq. 7, a compensation
for the inner capacitance C
i
of the test apparatus is necessary to the measured value
before the dielectric constant can be calculated according to eq.6.
i m x
C C C + = (7)
The measurements are done at 401C with a distance between the capacitor plates
of 0.1 mm. The inner capacitance, C
i
, is measured with a distance of 5 mm between
the plates.
79
Results and discussion
Table 3 summarises the running conditions used during the experiments for the inner
ring of the test bearing (7210BEP). The viscosity is decreased and maintained at
three different levels by changing the bulk viscosity of the lubricant on-line. The
minimum film thickness, h
min
, and the film parameter, , are calculated at each
concentration level. The film parameter reaches almost the same value 0.6 0.65 at
the end of each test. However, during the test, the surface of the bearing rings and
balls becomes smoother compared to a new bearing. For a new bearing the
roughness were measured to S
qb
= 30 nm and S
qr
= 90 nm. The film parameter
calculated in Table 3 is based on the measurements of the new bearing, within
brackets are the film parameter shown, based on the measurements of the tested
bearing. The measurements show that the dielectric constant increases with
increasing refrigerant concentration.
In this investigation the relative dielectric constant is measured at moderate
pressures. Dyson et al. [19] investigated the influence of pressure on the dielectric
constant for two mineral oils, and found that the dielectric constant increased when
the pressure was increased for both of the oils.
Table 3 Tested oil/refrigerant pairs and running conditions during the experiments.
Test
fluid
Oil Relative
Dielectric
constant
Viscosity
[mPas]
0.1
Concentra-
tion
[wt%]
Pressure-
viscosity
coeff.
[GPa
-1
]
h
min
[nm]
Film
parameter
Fluid
A
Pure POE +
R-134a
3.9
4.3
5.1
14.5
5.7
2.8
8
19
34
22
19
15
227
111
61
2.40
1.17
0.65(0.75)
Fluid
B
POE +
0.25% Acid
Phosphate +
R-134a
3.8
4.4
6.2
13.5
5.9
2.8
9
18
34
21
19
15
215
115
61
2.27
1.21
0.65(0.79)
Fluid
C
POE + 1.5%
Phosphate
ester +
R-134a
3.9
4.7
5.6
16
5.7
2.7
7
19
35
22
19
15
225
111
59
2.58
1.17
0.63(0.73)
Fluid
D
Napthenic
Mineral oil
+ R-22
2.9
3.1
3.4
13.3
5.4
2.5
7
14
27
22
19
15
215
109
57
2.27
1.15
0.60(0.65)
Capacitance measurement
Figures 2 and 3 show the measurement results of the mean V
cap
values. It is
important to note that each V
cap
value consists of a 40,000 samples taken each
second. The V
cap
reading decreases when the refrigerant concentration increases.
The signal drops significantly each time the refrigerant concentration changes, and
eventually increases to a higher more stable level, indicating that the lubricating film
thickness drops and contact occurs. New asperities have to be run-in after each drop
in viscosity and pressure-viscosity coefficient before the signal stabilises at a
constant level. It takes longer at a higher concentration to reach the constant level.
80
Figure 3a) shows measured V
cap
with Fluid A, where V
cap
decreases to the lowest
level of all tested lubricants. Figure 3b) and c) show the experiments with Fluids B
and C, including the phosphate additives. Clearly, additives promote film built-up.
Fluid C with phosphate ester gives the highest output reading, even higher than
Fluid D. The better film formation of phosphate ester compared to acid phosphate
might be due to the difference in the amount of additive, 0.25% for acid phosphate
versus 1.5% concentration of phosphate ester.
a) b)
c) d)
Figure 3 Output voltage V
cap
versus time with decreasing lubricant viscosity as
parameter, a) Fluid A no additives present, b) Fluid B with acid phosphate additive,
c) Fluid C with phosphate ester additive and d) Fluid D chlorinated.
Figure 4a shows curve fits to the steady-state V
cap
values measured during the
experiments presented in Figure 3. V
cap
is plotted versus the theoretical film
parameter for each tested fluid. The capacitance measurements clearly indicate a
higher separation between the balls and the ring when additives are present,
especially when lubricated with Fluid C at a high refrigerant concentration (low ).
In Figure 4b V
cap
, the curves for Fluids A, B, C and D are compensated for the
higher dielectric constant compared to the mineral oil at the lowest refrigerant
concentration by using eq.1 and 2. In Figure 4b the compensated V
cap
values of fluid
81
B and C shows increasing values when the film parameter decreases, this is not the
expected behaviour. It can be explained by the effect of the additives and error in the
measurement of the dielectric constants.
a) b)
Figure 4a) The steady-state level of output voltage V
cap
versus film parameter of the
four tested lubricants. 4b) Compensated V
cap
curves for fluid A, B, C and D due to
higher dielectric constant.
SEM images
Figure 5 shows the bearing surface of the inner rings from the four conducted
experiments. The image shows the history of all contacts and denting during the
experiments, as shown in Figure 3. The tested lubricants fail to completely eliminate
metal-to-metal contact when the refrigerant concentration is increased, leading to a
degradation of surfaces with little or no sliding or denting on the surface. Extensive
scattered debris denting is visible when running with Fluids A and C, whereas
considerably less denting can be noticed with Fluids B and D. Loose fine debris with
a grain size of less than 3 m is suspended in the lubricant due to the wear process
of the surfaces, contaminations in the system and the lack of a fine enough filter to
remove the debris. According to capacitance measurements Fluids B and C separate
the surfaces best. However, the SEM-images show excessive scattering denting on
the surface of Fluid C, whereas the surface on Fluid B is almost unaffected. It is
evident from studying the V
cap
curves in Figure 3 more carefully that run in after
each drop in viscosity is much tougher for Fluid C than for Fluid B. It takes much
more time for fluid C to reach the steady-state value, i.e. more denting and wear
takes place.
The image of Fluid B including acid phosphate additives shows the lowest degree of
denting and wear of the four experiments. The POE without additives (Fluid A)
shows the worst surface where the original surface is almost worn away.
82
a) b)
c) d)
Figure 5 SEM images of bearing inner ring surface in to magnification; Image in
background 1,000X and in foreground 4,000X, respectively. a) Fluid A no additives
present, b) Fluid B with phosphate ester additive, c) Fluid C with acid phosphate
additive and d) Fluid D chlorinated.
Surface parameters
Measurements of surface roughness parameters show that the running smoothens the
bearing surfaces. In all four tests, the RMS angle is reduced from 0.110 to about
0.040 degrees, see Figure 6a. A low RMS- angle indicates the sharpness of the
roughness. However, tests show that the RMS wavelength increases and makes the
surface smoother with long roughness waves. These surface changes promote the
bearing to build up a thick film thickness and increases the load-carrying capability
[15], see Figure 6b. The results are presented in the figures as box plots, with each
box containing five measurements and the box showing a 95% confidence interval
and the mean of the five measurements.
83
a) b)
Figure 6a) RMS angle of the inner rings of the tested bearings, b) RMS wave length
of the inner ring surfaces in the running track.
Figure 7a show the S
q
values for the test bearings, Figure 7b the skewness of the
height distribution. For all lubricants the skewness becomes more negative and will
essentially assure deep valleys and rounded peaks. Figure 7c shows the S
z
-values,
and both S
q
and S
z
decrease for the fluids with POE. To achieve fluid film
lubrication, the surfaces need to be completely separated by a lubricant film, with
low values of S
q
and S
z
and a negative skewness being desirable. The surface
roughness slopes should be as flat as possible, i.e. low RMS angle. The RMS
wavelength should be long to get a smooth surface with good load bearing capacity.
84
a) b)
c)
Figure 7a) RMS roughness of the inner rings of the tested bearings in the running
track, b) Skewness of the inner ring surfaces in the running track, c) S
z
-value of the
inner ring surfaces in the running track.
85
Conclusions
Tests of angular contact ball bearings used in a refrigeration environment have been
carried out. The lubrication status was monitored during the experiments using a
capacitance method. SEM and a 3D surface topometer were used to analyse surface
wear and damage after an initial run-in process. The following conclusions may be
made:
The capacitance measurement shows that Polyolester oil, including
phosphate additives in an R-134a environment, increases the lubricating
film thickness compared with the experiments without phosphate additives.
Phosphate ester additive in the POE/R-134a reduces wear of the bearing
surface compared with the experiments without phosphate additives and
shows the same amount of wear as mineral oil/R-22.
The amount of denting on the bearing surfaces is related to the amount of
particles and asperity contact that are present during operation.
The relative dielectric constant increases with increasing refrigerant
concentration.
The relative dielectric constants are higher with POE oil than with mineral
oil.
The surface topography of the bearings changes to a more favorable profile
with a lower RMS angle and longer RMS wavelength that promotes load-
carrying capacity and film build-up. The skewness, RMS roughness, S
q
,
average maximum height, S
z
is improved during running with all POE/R-
134a lubricants.
86
Nomenclature
S
k
Skewness
RMS-slope Root-Mean Square (RMS) slope, mDeg
RMS-wavelength Root-Mean Square (RMS) wavelength, m
V Signal from the capacitance measurement, V
C Capacitance, F
A Area, m
2
h Separation in the capacitor, m
R
q Root-Mean Square (RMS) roughness, m
R
z Roughness height, m
S
q Area Root-Mean Square (RMS) roughness, m
S
z Area Roughness height, m
c Weight fraction of component 2 i.e. refrigerant, %
U Speed parameter
G Material parameter
W Load parameter
k Ellipticity parameter
m Molecular weight ratio between oil and refrigerant
h
min
Minimum film thickness, m
Greek letters
Film parameter,

R
Relative permittivity or relative dielectric constant

0
Permittivity of vacuum, F/m
Dynamic viscosity, mPas
Pressure-viscosity coefficient, GPa-1
Subscripts
cap Output signal from the capacitance measurement
ref Reference
m Measured
i Inner
Mix Mixture of oil and refrigerant
1 Oil
2 Refrigerant
b Bearing ball
r Bearing inner ring
Acknowledgement
The authors would like to thank Kenneth Lilje, CPI Engineering for providing oils
and guidance and SKF, The Trane Company Inc., York Refrigeration and Carrier for
support during this work. Thanks also go to the Swedish Energy Agency for
financial support.
87
References

1 R. Tuomas and U. Jonsson. Influence of Refrigerant on Viscosity and
Pressure-Viscosity Coefficient of Refrigerant Compressor Lubricants.
Proceedings of the 25th Leeds-Lyon Symposium, Elsevier Science B.V.,
1999.
2 U. Jonsson and E. Hglund. The influence of refrigerants on the high pressure
properties of lubricating oil. Proceedings of Nordtrib 1992, vol. 3, pages 89-
96, 1992.
3 U. Jonsson and E. Hglund. Determination of Viscosities of Oil-Refrigerant
Mixtures at Equilibrium by Means of Film Thickness Measurements.
ASHRAE Transactions, vol. 99, pt. 2, pages 1129-1136, 1993.
4 M. Akei and K. Mizuhara. The Elastohydrodynamic Properties of Lubricants
in Refrigerant Environments. Tribology Transactions, vol. 40, 1, pages 1-10,
1997.
5 M. Muraki, T. Sano and D. Dong. Rheological Properties of
Refrigerant/Lubricant Mixture Under a Refrigerant Atmosphere. Japanese
Journal of Tribology, vol. 44, 2, pages 121-127, 1999.
6 SKF. General Catalogue. Edition 5000E, 2003.
7 S. Komatsuzaki and Y. Homma. Antiseizure and Antiwear Properties of
Lubricating Oils Under Refrigerant Gas Environments. Lubrication
Engineering, vol. 47, 3, pages 193-198, 1989.
8 K. Mizuhara, M. Akei and T. Matsuzaki. The friction and Wear Behavior in
Controlled Alternative Refrigerant Atmosphere. Tribology Transactions, vol.
37, 1, pages 120-128, 1994.
9 Mizuhara K. Mizuhara. Lubrication in Alternative HFC Refrigerants.
Japanese Journal of Tribology, vol. 38, 5, page 414, 1993.
10 S. F.Murray, R. L. Johnson and M. A. Swikert. Difluorodichloromethane as a
Boundary Lubricant for Steel and Other Metals. Mechanical Engineers, vol.
78, pages 233-236, 1956.
11 R. S. Heemskerk, K. N. Vermeiren and H. Dolfsma. Measurment of
Lubrication Conditioning in Rolling element Bearings. ASLE Transactions,
vol. 24, pages 519-527, 1982.
12 V. Wikstrm and B. Jacobson. Loss of lubricant from oil-lubricated near
starved spherical roller bearings. Institution of Mechanical Engineers, vol.
211, part J, pages 51-66, 1996.
13 B. Jacobson. Lubrication of screw compressor bearings in presence of
refrigerants. Proceedings of 1994 International Compressor Engineering
Conference at Purdue, vol. 1, pages 115-120, 1994.
14 P. Cann, E. Ioannides, B. Jacobson and A. A. Lubrecht. The lambda ratio a
critical re-examination. Wear, vol. 175, pages 177-188, 1994.
15 B. Jacobson. Nano-Meter Film Rheology and Asperity Lubrication. Journal
of Tribology, vol. 124, pages 595-599, 2002.
88

16 U. J. Jonsson and N. Hansson. Lubrication Limits of Rolling Element
Bearings in Refrigeration Compressors. Proceedings of 1998 International
Compressor Engineering Conference at Purdue, vol. 1, pages 129-134, 1998.
17 R. Tuomas. Bearing Lubrication in Refrigeration Screw Compressors.
Licenciate Thesis, Lule University of Technology, Sweden, ISSN 1407-
1757; 2001:20, 2001.
18 B. J. Hamrock and D. Dowson. Isothermal Elastohydrodynamic Lubrication
of Point Contacts Part IIIFully Flooded Results. Journal of Lubrication
Technology, Transactions ASME, vol. 99, 2, pages 264-276, 1977.
19 A. Dyson, H. Naylor and A. R. Wilson. The Measurements of Oil-Film
Thickness in Elastohydrodynamic Contacts. Proceedings of Institution of
Mechanical Engineers, vol. 180, Pt 3B, pages 119-134, 1965-66.
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
89
Paper C
Influence of refrigerant on viscosity and pressure-
viscosity coefficient of refrigeration compressor
lubricants.
Authors:
Roger Tuomas and Ulf Jonsson
Reformatted version of paper originally published in:
R. Tumas and U. Jonsson Influence of refrigerant on viscosity and pressure-
viscosity coefficient of refrigeration compressor lubricants. Proceedings of the 25th
Leeds-Lyon Symposium. Elsevier Science B.V. 1999.
90
91
Influence of refrigerant on viscosity and pressure-
viscosity coefficient of refrigeration compressor
lubricants.
R. Tuomas and U. Jonsson
Lule University of Technology
Division of Machine Elements
SE-971 87 Lule, Sweden
E-mail: [email protected]
Abstract
Following the introduction of non-chlorinated HFC (hydrofluorocarbon) refrigerants
and polyolester lubricants, designers of refrigeration compressors have had to re-
evaluate their knowledge of compressor bearing lubrication. The higher solubility
and the lack of anti wear protection in comparison to the traditionally
refrigerant/lubricant makes the lubrication of bearings using the modern systems a
much more difficult task than before.
A modern twin screw compressor contains several highly loaded metal-to-metal
contacts lubricated with oil, including, rolling element bearings, gears and the rotors.
These types of contacts are normally referred to as EHD or EHL contacts
(Elastohydrodynamic-Lubrication), due to elastic deformation of the surfaces. The
expected life of such bearing is to a large degree related to the film thickness in the
bearing. The properties of the lubricant that determine the film thickness in an EHD-
contact are normally the dynamic viscosity, , and the pressure-viscosity coefficient,
. In a refrigeration system, both parameters are affected by the presence of
refrigerant. This effect is depending on the oil and refrigerant composition.
A falling ball viscometer has been used to measure the viscosity and the pressure-
viscosity coefficient as a function of the amount of dissolved refrigerant. A
polyolester oil was tested with refrigerant R-22, R-134a, R-410a and R-32 at 40 and
80 C, at refrigerant concentrations from 0 % to 30 % dilution.
The results show that Eyring equation can be used to predict the pressure-viscosity
coefficient of a mixture based on the molecular mass ratio of the refrigerants.
92
Introduction
A modern refrigeration screw
compressor contains several highly
loaded rolling element bearings
lubricated with a mixture of oil and
refrigerant. These bearings are
operating in the Elasto-Hydro-
Dynamic (EHD) lubricating regime,
due oil film separation and elastic
deformation of the surfaces.
In these bearings the most important
task for the lubricant is to create a
film that separates the interacting
surfaces. The expected life of these
bearings is to a large degree
dependent on the reduction in film
thickness caused by the dilution of
refrigerant in the oil. Contact between
the surfaces will cause locally high
stresses resulting in fatigue of the
bearing material. If asperity contact
occurs it may also result in wear of
the bearing surfaces. Jacobson [1]
found that the rate of the wear
depends on the oil and refrigerant
used and that prevention of wear of
the bearing surfaces typically requires
50% higher viscosity when lubricated
with a polyolester/R-134a mixture
than for mineral oil/R-22 lubricated
bearings.
The properties of the lubricant that
determine the film thickness in a
bearing are normally the dynamic
viscosity,, and the pressure-
viscosity coefficient, . This is very
relevant in a refrigeration compressor
since these properties are affected by
the presence of refrigerant. In SKF
application handbook for bearings in
screw compressors [2] it is suggested
that the viscosity used in calculations
of bearing life should be modified to
compensate for the reduction in
pressure-viscosity coefficient
compared to mineral oil. The adjusted
viscosity,
adj
, for use in bearing life
calculations can be determined as
follows:
) / (
0.72
mineral mix mix adj
=
where:
mix
and
mix
are the actual
kinematic viscosity and pressure-
viscosity coefficient of the oil-
refrigerant mixture at the location of
the bearing and
mineral
is the
pressure-viscosity coefficient of
reference mineral oil.
There are a lot of different methods
used in the past for measuring the
viscosity of oil-refrigerant mixtures.
A review of different methods is
presented by Speaker and Spauschus
[3]. Jonsson and Hglund [4,5] have
measured the and for
oil/refrigerant mixtures using a
concentric cylinder viscometer and
later on by means of film thickness
interferometry measurements. Akei
and Mizuhara [6,7] developed the
interferometry method further to
estimate the viscosity and the
pressure viscosity coefficient of oil
refrigerant mixtures, including a
polyolester with R-134a. They used
the Eyring equation, Eq. (1), to
estimate pressure-viscosity
coefficient of oil refrigerant mixtures.
( )
( )
lubr
refr
lubr refr refr
mix

1 1 - m s
s m
+
+

=
(1)
In Eq. (1),
mix
is the pressure-
viscosity coefficient of the mixture,

refr
and
lubr
are the pressure
viscosity coefficient for pure
refrigerant and oil respectively. s
refr
is
the mass fraction of refrigerant and m
is the apparent molecular mass ratio
93
between the refrigerant and oil,
m=M
*
lubr
/M
refr
. M
*
lubr
is the apparent
molecular mass of the oil and M
refr
is
the molecular mass of the refrigerant,
calculated from the chemical
composition. M
*
lubr
does not give a
representation of the molecular mass
of the oil molecule but is rather a
measure of the average weight of the
segments in the oil.
The oils chemical composition and
pressure-viscosity coefficient data are
seldom available from the oil
manufactures. It is even harder to
find viscosity and pressure-viscosity
coefficient for different types of
oil/refrigerant mixtures. Jonsson and
Lilje [8] have developed a semi
empirical model that can be used to
predict the pressure-viscosity
coefficients of mixtures between
polyolesters and R-134a based on the
amount of branched acids in the oil.
In this investigation the viscosity and
pressure-viscosity coefficient was
measured for mixtures of an ISO
VG68 polyolester and refrigerant: R-
32, R-134a, R-410a and R-22. A
method has been developed that can
be used to predict the pressure-
viscosity coefficient of mixtures with
other refrigerants than R-134a based
on the difference in molecular weight
of the refrigerants.
Tested Refrigerants
Four different refrigerants were
evaluated with one ISO VG68
polyolester. The oil was a
refrigeration grade basestock
containing less than 100ppm water.
The refrigerants were selected to
represent common used refrigerants
in industrial refrigeration systems. R-
134a is used as a replacement for R-
12 and R-410a as an alternative to R-
22. R-22 was included as a reference.
Table 1 Refrigerants used in the
experiments.
Refrigerant Chemical name Type MM
[g/mol]
R-32 Difluoro-methane HFC 52.02
R-134a Tetrafluoro-ethane HFC 102.03
R-410a 50/50%
R-32/R-125
HFC
mixture
72.58
R-22 Chlorodifluoro-
methane
HCFC 86.48
All refrigerants used in this work
were obtained from a commercial
source and were used as received.
Experimental method
The viscosity of the mixtures was
measured using a falling ball
viscometer at temperatures of 40 and
80C and at pressures of 2, 17 and 34
MPa. The concentration of dissolved
refrigerant was varied between 0 and
40% in steps of approximately 10%
to cover the range found in most
applications. The results from the
experiments were then fitted using
Eq. (1).
Experimental procedure
The viscometer used was a modified
Hppler viscometer developed by
Jonsson and Hglund [9], shown in
Figure 1.
94
Figure 1 The modified Hppler
viscometer.
The viscosity was determined by
measuring the time it took for the ball
to travel 100 mm in a tube containing
the test liquid. Two photosensors
timed the ball through two pairs of
sapphire windows. The viscometer is
equipped with a separating piston
used to pressurize the test fluid
during measurements. The test fluid
and the pressurizing fluid are
separated by two O-rings on the
piston. The volume between the O-
rings is ventilated through the piston
rod to prevent contamination of the
test fluid in case of a leakage at the
O-rings. The viscometer was always
pressurized above the vapour
pressure of the test fluid.
The dilution of refrigerant was
determined by taking a sample
through a capillary tube into a test
tube. The concentration was
determined gravimetrically by
evaporating the refrigerant from the
mixture under vacuum. A spread
sheet program was used to calculate
the viscosity for each test point. The
program compensates for effects of
pressure and temperature on the
dimensions of the ball and the bore.
The pressure-viscosity coefficient
was calculated for each pressure,
temperature and concentration and
the results were fitted to the Eyring
equation Eq.(1).
To predict the pressure-viscosity
coefficient the value of pure
refrigerant
refr
respective pure oil

lubr
has to be known. Together with
the apparent mass ration m the
mix
can be calculated.
The apparent mass ratio, m, is
calculated from the apparent
molecular mass of the oil, M
*
lubr
, and
the molecular mass of the refrigerant,
M
refr
. To get the apparent molecular
mass of the oil pressure-viscosity
measurements have to be done for at
least a combination of a single
refrigerant and the oil at a few
concentrations. Eq.(1) is used to fit
the measured data and the regression
gives a value of M
*
lubr
. When M
*
lubr
is
known, the pressure-viscosity
coefficient can be calculated for
mixtures with all suitable refrigerants
providing that
refr
is known for the
actual refrigerant.

refr
is given either from pressure-
viscosity measurements of pure
refrigerant or extrapolated from the
regression when M
*
lubr
. is calculated.
Results and discussion
The results show the effect of
different refrigerants diluted in oil vs.
the viscosity and pressure-viscosity
coefficient, for two different
temperatures.
95
Effect of refrigerant
dilution in the oil
Figure 2 and 3 show the variation of
dynamic viscosity, , for the polyol
ester mixed with the different
refrigerants and for two different
temperatures. The degree of
curvature is depending of the
molecular weight of the refrigerant.
The lighter R-32 gives a faster
decrease in viscosity compared to the
heavier refrigerant R-134a.
Temp 40
o
C
Concentration refrigerant [%]
0 10 20 30
V
i
s
c
o
s
i
t
y

[
c
P
]
0
20
40
60
80
R-32
R-134a
R-410a
R-22
Figure 2 Viscosity vs. concentration
40C.
Temp 80
o
C
Concentration refrigerant [%]
0 10 20 30
V
i
s
c
o
s
i
t
y

[
c
P
]
0
5
10
15
R134a
R-410a
R-22
Figure 3 Viscosity vs. concentration
at 80C.
Figures 4 and 5 show the pressure-
viscosity coefficient, , for the
polyolester diluted with different
refrigerants. The same trend as for
the viscosity can be found, that a light
refrigerant causes a more rapid
decrease in pressure-viscosity
coefficient than the heavier
refrigerants. It can also be seen that
the temperature does not influence
the value as much as it affects the
viscosity. In this charts the Eyring
equation, Eq.(1), is used to fit a curve
to the data. The regression of the
curve fit gives the apparent mass ratio
for the mixture and the
refr
for the
refrigerants, later used to predict
mix
for all concentrations.
Temp 40
o
C
Concentration refrigerant [%]
0 10 20 30
P
r
e
s
s
u
r
e
-
V
i
s
c
o
s
i
t
y

c
o
e
f
f
.

[
G
P
a
-
1
]
10
15
20
25
R-32
R-410a
R-22
R-134a
Figur 4 Pressure-Viscosity coefficient
vs. concentration at 40C.
Temp 80
o
C
Concentration refrigerant [%]
0 10 20 30
P
r
e
s
s
u
r
e
-
V
i
s
c
o
s
i
t
y

c
o
e
f
f
.

[
G
P
a
-
1
]
14
16
18
R-410a
R-22
R-134a
Figur 5 Pressure-Viscosity coefficient
vs. concentration at 80C.
96
In figure 6 the solid lines are
calculated from the molecular mass
ratio obtained from measurements
with R-134 and the molecular mass
difference between the refrigerants.
From this it can be concluded that the
Eyring equation can be used
successfully to eliminate tedious
measurements for different oil-
refrigerant pairs.
Temp 40
o
C
Concentration refrigerant [%]
0 20 40 60 80 100
P
r
e
s
s
u
r
e
-
V
i
s
c
o
s
i
t
y

c
o
e
f
f
.

[
G
P
a
-
1
]
5
10
15
20
25
R-32
R-410a
R-22
Figure 6 Predicted pressure-viscosity
coefficient vs. concentration.
Conclusions
The pressure-viscosity coefficient
and the viscosity of various
refrigerants/polyol ester pairs were
investigated using a high-pressure
falling ball viscometer. A strong
correlation was found between the
molecular mass of the refrigerant and
the reduction of viscosity and
pressure-viscosity coefficient with
concentration. It has been shown that
the Eyring equation can be used to
predict the pressure-viscosity
coefficient of a mixture based on the
molecular mass ratio of the
refrigerants.
Nomenclature
dynamic viscosity, cP
pressure-viscosity coefficient,
GPa
-1
kinematic viscosity, cSt
m mass ratio, M
*
lubr
/M
refr
MM Molecular mass, g/mol
s mass fraction
Acknowledgements
The author would like to thank SKF,
CPI Engineering, Trane Company,
York Refrigeration and Carrier for
support and help during this work.
Thanks also go to the Swedish
Energy Agency for financial support.
97
References

1 B. Jacobson. Ball Bearing lubrication in refrigeration compressors. Proceedings
of 1996 International Compressor Engineering Conference at Purdue, vol. 1,
pages 103-108, 1996.
2 SKF. Bearings in Twin Screw Compressors - Application Handbook. SKF USA
Inc., 1998.
3 L. M. Speaker and H. O. Spauschus. A study to increase the state-of-the-art
solubility and viscosity relationships for oil/refrigerant mixtures. Ashrae
Research report, Project 444, 1986.
4 U. Jonsson and E. Hglund. The influence of refrigerants on the high pressure
properties of lubricating oil. Proceedings of Nordtrib 1992, vol. 3, pages 89-96,
1992.
5 U. Jonsson and E. Hglund. Determination of Viscosities of Oil-Refrigerant
Mixtures at Equilibrium by Means of Film Thickness Measurements. ASHRAE
Transactions, vol. 99, pt. 2, pages 1129-1136, 1993.
6 M. Akei, K. Mizuhara, T. Taki, and T. Yamamoto. Evaluation of film-forming
capability of refrigeration lubricants in pressurized refrigerant atmosphere.
Wear, vol. 196, pages 180-187, 1996.
7 M. Akei and K. Mizuhara. The Elastohydrodynamic Properties of Lubricants in
Refrigerant Environments. Tribology Transactions, vol. 40, 1, pages 1-10, 1997.
8 U. Jonsson and K. Lilje. Elastohydrodynamic lubrication properties of polyol
ester lubricants-R134a mixtures. Tribologia, vol. 18, 1, pages 31-37, 1999.
9 U. Jonsson. Lubrication of rolling element bearings with HFC-polyol ester
mixtures. Proceedings of Nordtrib 1992, vol. 1, pages 403-411, 1992.
98
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
99
Paper D
Compressibility of Oil/Refrigerant Mixtures in
Elasto-Hydrodynamic Contacts.
Authors:
Roger Tuomas and Ove Isaksson
Reformatted version of paper originally published in:
R. Tumas and O. Isaksson. Compressibility of Oil/Refrigerant Mixtures in Elasto-
Hydrodynamic Contacts. Journal of Tribology, v 128, n 1, January, 2006, p 218-220.
100
101
Compressibility of Oil/Refrigerant Mixtures in Elasto-
Hydrodynamic Contacts
Roger Tuomas and Ove Isaksson
Lule University of Technology
Division of Machine Elements
SE-971 87 Lule,
Sweden
Phone: +46 920 492414
E-mail: [email protected]
Abstract
A high pressure chamber is used to study lubricant compressibility when
refrigeration oil is diluted by refrigerant. The tested lubricants in this work is a POE
(polyolester) oil, POE diluted with non-chlorinated (HFC) refrigerant R-134a, a
naphthenic mineral oil and the mineral oil diluted with the chlorinated (HCFC)
refrigerant R-22.
The high-pressure chamber experiments show that adding 20 wt% of R-134a to the
polyolester oil, the compressibility of the lubricant increases by approximately 38 %
at 1 GPa and is much higher than for R-22 and naphthenic mineral oil.
Introduction
Rolling element bearings are widely used in refrigeration compressors. The lubricant
in such compressors consists of an oil/refrigerant mixture. Important lubricant
properties change when refrigerant dilutes the oil, e.g. viscosity, pressure-viscosity
coefficient and probably compressibility. The presence of a refrigerant leads to more
demanding operating conditions for the bearing since the pressure is high in the
contact between ball and ring, i.e. several GPa.
To calculate film thickness, pressure distribution and friction in an elasto-
hydrodynamic contact it is necessary to know the viscosity, compressibility of the
lubricant, and how it is affected by pressure and temperature. Investigations based
on two different approaches have been conducted to describe the compressibility for
lubricating oils. The first attempts used a test apparatus to change volume as
function of pressure under static conditions, i.e. no influence of possible transient
loading. Dowson-Higginson [1], Hamrock et al. [2], Jacobson and Vinet [3] and
Sthl and Jacobson [4] all performed investigations under static conditions. The
second approach has been to understand if the effect of structural relaxation is
significant for elasto-hydrodynamic lubricants. Lindqvist et al. [5] used a modified
Split-Hopkinson pressure bar set-up to determine the compressibility of lubricants,
concluding that long straight molecules found in esters, polyglycol and
polyalpholefin are rather compressible, while naphthenic and paraffinic mineral oils
have more complex molecules and show lower compressibility. Several authors
102
have contributed to data of compressibility and some models have been used to
calculate film thickness in an elasto-hydrodynamic contact. Dowson and Higginson
suggested the most commonly used model for compressibility. The model is
empirical and based on measured compressibility data up to only 0.4 GPa.
Since the interatomic forces in compressed solids and liquids are similar, the
equation of state for solids, derived by Vinet et al. [6], was approximated by
Jacobson and Vinet [3] for the equation of state for liquid lubricants. To obtain the
parameters included in the compressibility relation, Jacobson and Vinet fitted data
published by Hamrock et al. [2] for pressures as high as 2.2 GPa.
However, knowledge about the compressibility for lubricants containing an
oil/refrigerant mixture still lacks. The present paper attempts to contribute with
compressibility data for some widely used oil/refrigerant mixtures. The tested
lubricants in this work are a POE (polyolester) oil, POE diluted with the non-
chlorinated HFC (hydrofluorocarbon) refrigerant R-134a, the naphthenic mineral oil
and the mineral oil diluted with chlorinated HCFC (hydrochlorofluorocarbon)
refrigerant R-22. Compressibility data for the tested lubricants are fitted to the
Jacobson and Vinet model. The results for the mixtures are compared with pure oil
and earlier presented results from other authors.
High-Pressure Chamber
The compressibility measurements presented in this paper are done in the high-
pressure chamber developed and described by Jacobson [7] and further described by
Sthl and Jacobson [4]. An overview of the high pressure chamber can be seen in
Figure 1a, with the high-pressure cylinder (3) and the plungers (7) in Figure 1b. In
the compressibility test the lower plunger is fixed while the upper is movable. The
test lubricant is placed in the cylinder between the plungers when the lower plunger
is in position in the cylinder.
103
Figure 1a) An overview of the high-pressure chamber, b) High pressure cylinder and
the two plungers.
The lower plunger is modified with a hole through the plunger and provided with a
one-way valve also functioning as seal. This allows filling of lubricant when both
plungers are in place. A hydraulic jack (1) pressurizes the lubricant and controls the
movement and force of the compression. To withstand the high pressures the high-
pressure cylinder is press fitted inside a pre-stressed container (10) and pre-
compressed axially by a screw assembly (9). The maximum pressure is limited to
4.5 GPa. A force transducer measures the applied load that is used to calculate the
corresponding pressure. The position of the upper plunger is measured by a linear
potentiometer. The compressibility is calculated from the measured position and
pressure. For safety, an enclosure made of lexan (4) surrounds the hard metal parts.
The hydraulic jacks are pressurized and supplied with fluid through the valves and
pressure regulator (6) from a hydraulic pump (5).
Experimental Procedure High-Pressure Chamber
All parts are carefully cleaned with alcohol solvent and assembled. The refrigerants
used are in gas phase at ambient pressure and 20C. The refrigerant is therefore
pressurized to 0.7 MPa to be in liquid phase.
The oil and refrigerant are mixed in a cylinder placed on a scale. To reach the right
concentration, 100 g of oil followed by the correct amount of refrigerant is applied.
The oil/refrigerant mixture is then placed into a hydraulic cylinder where the
pressure is increased up to 7 MPa. The lubricant sample is then injected into the high
pressure chamber. To ensure that no air is present in the chamber during the
experiments the seal at the top of the chamber is set open and then slowly moved
until the leakage stops. During leakage, the lubricant fills the entire chamber while
evacuating all air. The experiments start as soon as the lubricant sample is placed
inside the high-pressure chamber.
104
The change in volume and pressure are derived by applying the load. The hydraulic
jack (1) moves the top mounted plunger (7) downwards. A force transducer
measures the load on the plunger and a linear potentiometer measures the position of
the plunger. While the pressure increases the force and position of the plunger are
measured and captured by the control program. All experiments are performed three
times for each lubricant to ensure the measurements are repeatable.
Tested Lubricants and Refrigerants
Four lubricants are tested at room temperature, 20 C. Table 1 shows the
investigated oils and oil/refrigerant mixtures. The POE oil is formulated from ester
base stocks to provide controlled miscibility as well as solubility with HFC based
refrigerants, including R- 134a.
Table 1 Investigated lubricants in high pressure chamber.
Test Lubricant Oil type Viscosity
grade
Refrigerant
used
1 Solest 68 POE VG68 R-134a
2 Solest 68 POE VG68 -
3 Mobil Arctic 300 Mineral VG68 R-22
4 Mobil Arctic 300 Mineral VG68 -
POE oil is commonly used as refrigeration compressor oil. The naphthenic mineral
oil is primarily intended to be used in refrigeration compressors. It is compatible
with all refrigerants except sulphur dioxide and is not recommended for use with
HFC refrigerants.
The HFC refrigerant R-134a used in this investigation does not contain any chlorine.
The chlorinated refrigerant is R-22. The experiments are done with pure oil (tests 2
and 4) and oil diluted with refrigerants (tests 1 and 3). The refrigerant concentration
is 20 wt% and the pressure is increased to 3.1 GPa in the experiments.
Calculation of Compressibility and Density
The applied load on the lubricant is reduced with the friction force between the seal
and the test chamber wall. The magnitude of the friction force is measured to
approximately 4,000 N. By moving the upper plunger and seal without the lower
plunger in place, the friction force from the seal is measured. The compressibility
can be calculated from the positions with eq.1. In the equation, V is the volume
change due to the compression of the lubricant, h
o
is the initial height of the
lubricant before the load is applied and h(p), the volume when the lubricant is
pressurized.
0
0
0
) (
h
p h h
V
V
=

(1)
105
The density change due to pressure can be calculated by eq.2. In the equation is
the measured density and
0
is the density at ambient conditions.
) (
0
0
p h
h
=

(2)
Density Relations
Dowson and Higginson proposed a static density-pressure relation, eq.3.
p
p
7 . 1 1
6 . 0
1
0
+
+ =

(3)
where the unit of pressure is GPa, is the measured density of the oil and
o
is the
density at ambient pressure. Jacobson and Vinet suggested eq.4 to model the
compressibility of the lubricant.
(

=

) ) ( 1 ( ' exp ) ) ( 1 ( ) ( 3
3 / 1 3 / 1 3 / 2
o o o
o
B p

(4)
where B
o
is the bulk modulus at ambient pressure and ' a parameter in the bulk
modulus and pressure equation. The equation is slightly inconvenient, as it cannot
easily be inverted analytically. The parameters are based on static measurements of
the volume changes as a function of the pressure.
106
Results and Discussion
The pressure and the compressibility of the lubricant specimen are evaluated from
the force and position measurements. Compressibility as a function of pressure is
shown in Figure 2 for the four tested lubricants.
a) b)
Figure 2a) Compressibility as a function of pressure, for pure mineral oil and
mineral oil diluted with 20 wt% R-22, b) compressibility as a function of pressure,
for pure POE and POE diluted with 20 wt% R-134a
When R-134a is added to the POE, its compressibility compared to pure oil
decreases by approximately 38 % at a pressure of 1 GPa. The influence of R-22 on
the compressibility of the mineral oil is small and can be neglected.
The change in volume with increasing pressure for pure mineral oil corresponds to
measurements done by previous authors. The relative change in volume for mineral
oil at 1 GPa is 0.17 in the current study compared to 0.18 in Lindqvist et al. [5] and
Sthl and Jacobson [4]. The compressibility for POE/R-134a mixture is even lower
than of 5P4E given in [8]. POE diluted with 20 wt% of R-134a has a relative volume
change of 0.1 while 5P4E has 0.14 at 1 GPa pressure.
The measured data are fitted to the Jacobson and Vinet model. The derived data and
Dowson and Higginson and Jacobson and Vinet models are plotted in Figure 3. The
measurements fit the Jacobson Vinet model before the lubricant solidifies. The fit
seems to be OK only up to a pressure of about 2 GPa.
107
Figure 3 Comparison of models regarding relative density as a function of pressure.
The figure shows the Jacobson-Vinet model fitted to measured data.
Conclusions
A high pressure chamber is used to derive the compressibility of an oil/refrigerant
mixture, including refrigerants R-134a and R-22 up to pressures of 3.1 GPa. By
adding the non-chlorinated refrigerant R-134a to polyolester oil, the compressibility
decreases by approximately 38 % at 1 GPa. The decrease in compressibility is larger
than for the R-22 and mineral oil mixture. The mixture of POE and R-134a gives a
lower compressibility than 5P4E. The model suggested by Jacobson and Vinet fits
well with the experimental data up to 2 GPa.
Nomenclature
B
0
Bulk modulus at zero pressure, N/m
2
V
o
Initial volume of the compressed lubricant, m
3

h
0
Initial height of the compressed lubricant, m
V Volume change of the lubricant, m
3
h
p
Measured position of the upper hydraulic jack during the
compression, m
p Pressure, N/m
2
Density of the lubricant, kg/m
3

0
Density of the lubricant at ambient pressure, kg/m
3
Parameter in bulk modulus and pressure equation
108
Acknowledgements
The authors would like to thank SKF, CPI Engineering, Trane Company, York
Refrigeration and Carrier for their support and help. Thanks also go to the Swedish
Energy Agency for financial support and Professor Bo Jacobson at Lund University
for the kindness to let us use the high-pressure chamber in this investigation.
109
References

1 G. R. Higginson and D. Dowson. Elastohydrodynamic Lubrication, The
Fundamentals of Roller or Gear Lubrication. Pergamon Press, Oxford, 1966.
2 B. J. Hamrock, B. O. Jacobson and S.-I. Bergstrm. Measurement of the
Density of Base Fluids at Pressures to 2.2 GPa. NASA TM-87114, 1987.
3 B. O. Jacobson and P. Vinet. A Model for the Influence of Pressure on Bulk
Modulus and the Influence of Temperature on the Solidification Pressure for
Liquid Lubricants. ASME Journal of Tribology, vol. 109, pages 709-714,
1987.
4 J. Sthl and B. O. Jacobson. Compressibility of Lubricants at High Pressures.
Tribology Transactions, vol. 46, 4, pages 592-599, 2003.
5 S. Lindqvist. Compressibility and Density of Lubricants in Transient
Loading. Licenciate Thesis, Lule University of Technology, Sweden,
ISSN:0280-8242; 1995:35L, 1995.
6 P. Vinet, J. Ferrante, J. R. Smith and J. H. Rose. A Universal Equation of
State for Solids. Solid state Physics, vol. 19, pages 467-473, 1986.
7 B. O. Jacobson. A New High Pressure Chamber for Lubricant Investigations.
Proceedings of the International Tribology Conference, Nagasaki, Japan,
ISBN 4-9900139-5-6, pages 1199-1202, 2000.
8 E. Hglund. Influence of Lubricant Properties on Elastohydrodynamic
Lubrication. Wear, vol. 232, 2, pages 176-184, 1999.
110
Properties of Oil and Refrigerant Mixtures
Lubrication of ball bearings in refrigeration compressors
111
Paper E
Friction coefficient of oil/refrigerant mixtures at a
pressure of 2.5 GPa.
Authors:
Roger Tuomas
Reformatted version of paper originally published in:
R. Tumas and O. Isaksson. Friction coefficient of oil/refrigerant mixtures at a
pressure of 2.5 GPa. Proceedings of the 2002 Purdue international Compressor
conference, 2002.
112
113
Friction coefficient of oil/refrigerant mixtures at a
pressure of 2.5 GPa.
Roger Tuomas
Division of Machine Elements
Lule University of Technology
SE-971 87 Lule, Sweden
Phone: +46 920 492414
E-mail: [email protected]
Abstract
In refrigeration screw compressors, bearing lubrication refers to the elasto-
hydrodynamic, (EHL), regime. Under these conditions, bearing life is dependent on
the film thickness separating the surfaces and the stress produced by the cyclic load
on the surface material. Since the bearings in a refrigeration screw compressor are
lubricated with a mixture of compressor oil diluted with the refrigerant, it is
important to understand the behaviour of such mixtures. It is known that the
presence of refrigerant decreases the viscosity and pressure-viscosity coefficient of
the lubricant; those two lubricant properties which are most important in ensuring a
film thickness sufficient to separate the contact surfaces in an EHL contact.
The parameter controlling the location of the maximum stress in the material and, to
some extent, the magnitude of that stress, is the friction coefficient in the contact. In
this investigation the influence of refrigerant dilution of the lubricant on friction
coefficient in a contact at a pressure of 2.5 GPa has been studied.
To be able to measured the friction coefficient of the lubricant under pressures up to
2.5 GPa an test apparatus that considers an oblique impact of a steel ball on a rod
end. Accelerometers and strain gauges on the rod measure the shock waves and the
transient normal- and transverse forces, acting during the impact, are calculated.
The investigation showed that in all tested oil/refrigerant mixtures the friction
coefficient increases when refrigerant are present in the mixture. A more branched
ester oil give higher friction coefficient.
Introduction
In a refrigeration compressor, the bearings supporting the rotors are highly loaded
and lubrication is in the EHL (elasto-hydrodynamic-lubrication) regime. Under such
lubricating conditions, a very thin lubricant film and the materials in the bearing
surfaces are exposed to a high contact pressure that results in the material deforming
elastically in the contact. The lubricant in an EHL contact is also subjected to high
shear rates and, due to the high local pressures, lubricant compression will occur.
In a refrigeration screw compressor the lubricating medium is oil diluted with the
refrigerant used in the refrigeration system. It has been shown by Tuomas and
Jonsson [1] that the presence of refrigerant lowers both the viscosity, , and
114
pressure-viscosity coefficient, , of the lubricant and hence the film thickness in the
contact. Also in refrigeration applications, Jonsson [2] showed that the use of esters
with different molecular branching influence and differently. A more branched
molecule results in an increases in and and therefore, according to Hamrock and
Dowson [3,4], will result in a thicker lubricating film.
Figure 1 shows the inner ring of an angular contact bearing (7210). The bearing
failed due to surface fatigue after 1900 h. The lubricant used in the experiment was a
polyolester oil diluted with 26 wt% R-134a. The temperature was 40C, speed 1,500
rpm and load ratio C/P 12.4.
Figure 1 An angular contact bearing (7210) that failed after 1,900h running with a
polyolester oil diluted with R-134a at a concentration of 26 wt%. Operating
parameters, 1,500rpm, temperature 40C and load ratio 12.4.
The fatigue life of a surface depends on the number of load cycles the surface
material is exposed to, Lundberg and Palmgren [5]. If the stress intensity becomes
too high, cracks will start to develop from dislocations or other inhomogeneous parts
of the material. An important determining factor of the fatigue life is the depth at
which maximum stress occurs. If the maximum is located close to the surface, the
material will fail more easily than if the maximum is located deeper in the material.
Sliding in the contact will add another stress component to the combined stress that
act upon the material,
L
.
L
, being the limiting shear stress that the oil can sustain.
L
, devided by the pressure, P, is called the shear strength-pressure coefficient []
(friction coefficient of the lubricant) and is given by [
P
L

=

]. It was shown by
Hglund [6], that higher causes the stress maximum to move closer to the surface
of the material and also increase the maximum stress slightly.
This paper investigates how the shear strength-pressure coefficient [] is affected by
refrigerant concentration in the oil.
115
Method
To be able to measured the friction coefficient of the lubricant under pressures up to
3 GPa an apparatus described by hrstrm [7,8] was used, the apparatus show in
Figure 2, considers an oblique impact of a steel ball on a rod end. Accelerometers
and strain gauges on the rod measure the shock waves and from that information the
transient normal- and transverse forces, acting during the impact, are calculated. A
change in friction coefficient affects the forces on the end plane of the rod and
consequently also the wave pattern in the rod.
Strain
gauges
Transient
recorder
Bridge
amp.
Accelero-
meter
Computer
Accelero-
meter
Charge
amp.
Charge
amp.
Tungsten carbide end plate
with lubricated surface
4
0
.
0
4
0
0
0
9
4
0
Steel ball
o 16
/
Fixture on
moving sled
Figure 2 The test apparatus (hrstrm [7,8]) used to measure lubricant friction
coefficients at pressures of up to 3 GPa.
Test procedure
The ball and the rod end surface were thoroughly cleaned and a lubricant sample
was placed on the tungsten carbide plate located on the end of the rod. The ball was
116
then released in such a way that an oblique contact resulted, the height of release
being such that the required contact pressure of 2.5 GPa was achieved.
To be able to detect if it is a full film separation between the ball and the rod end, a
potential difference of 100 mV is applied between the ball and the end plane. Any
potential equalization, resulting from asperity contact, is monitored on an
oscilloscope. Boundary lubrication contact measurements are not relevant because
metal-to-metal contact occurs at the impact, and are therefore excluded.
Tested lubricants and refrigerants
Three different polyolester lubricating oils were used; Ester A, B and C. All three
are polyolester refrigeration oils but with different viscosity grades and acid
branching. The ester oils were diluted with the HFC refrigerant R-134a. A
naphthenic mineral oil diluted with the chlorinated refrigerant R-22 was used as a
reference. Details of all these oils are given in Table 1 together with the refrigerants
used. The molecular structures of the esters are shown in Figure 3.
Table 1 Details of the four lubricants used in the experiments.
Lubricant Chemical structure Viscosity grade Branched acids Refrigerant used
Ester A C
39
H
72
O
8
VG68 70% R-134a
Ester B C
31
H
56
O
8
VG68 8%+67% R-134a
Ester C C
33
H
62
O
6
VG46 100% R-134a
Mineral - Base viscosity 57cst
at 40C
- R-22
O O
O
O
O
O
O
O
A)
O O
O
O
O
O
O
O
B)
O
O O
O
O
O
C)
Figure 3 Molecular structure of the tested polyolesters. From left to right, Ester A,
B and C.
Refrigerant concentration estimation
At ambient pressure, the refrigerant will boil out of the oil. In order to know the
actual concentration at the time of the friction measurement, experiments were
117
carried out to determine the change in refrigerant concentration at ambient pressure
over time for the four different mixtures. 27 grams of oil were placed in a small
bottle and then diluted with 5 grams of refrigerant. The pressure in the bottle was
then decreased to atmospheric pressure and the refrigerant starts to boil out of the
oil. The change in weight was then measured over time and the concentration
calculated. The results are shown in Figure 4. As can be seen, refrigerant
concentration at zero time varies with the molecular structure of the oil; a more
branched acid, Figure 3C, in the ester results in higher solubility. Henderson [9] also
describes this. It can be seen that the amount of R-22 diluted in the mineral oil
decreased faster than the R-134a in ester. This means that the mineral oil will more
easily give up any diluted refrigerant if the pressure is lowered and therefore provide
an increased film thickness in the bearings.
R
e
f
r
i
g
e
r
a
n
t

c
o
n
c
e
n
t
r
a
t
i
o
n

[
w
t
%
]
Time [min]
0 10 20 30 40 50
5
6
7
8
9
10
Ester A R-134a
Ester B R-134a
Ester C R-134a
Mineral R-22
Figure 4 Refrigerant concentration at ambient pressure vs. time for the oil/refrigerant
mixtures tested.
Once the sample is placed on the tungsten carbide plate the refrigerant will begin to
evaporate. Since the time for a measurement in the oblique impact test rig was
known, the refrigerant concentration in the oil at the time of measurement could be
calculated for each oil mixture from Figure 4. These concentrations are given in
Table 2. The experiments were carried out at ambient pressure at a temperature of
21C with a contact pressure of 2.5 GPa. The actual refrigerant concentrations were
between 4.5 wt% and 8.5 wt%.
118
Table 2 Refrigerant concentrations used in the experiments.
Oil Refrigerant Refrigerant concentration [wt.%]
Ester A R-134a 7 %-8 %
Ester B R-134a 7.5 %-8.5 %
Ester C R-134a 7.5 %-8.5 %
Mineral R-22 4.5 %-5 %
Results and Discussion
Figure 5 shows the friction coefficient of the tested oils both undiluted and diluted
with refrigerant. Each box represents five measurements and the height of the box
indicates the measurement deviation. This deviation is due to refrigerant boiling out
of the oil during the experiment.
E
s
t
e
r
A
E
s
t
e
r
A
R
-
1
3
4
a
E
s
t
e
r
B
E
s
t
e
r
B
R
-
1
3
4
a
E
s
t
e
r
C
E
s
t
e
r
C
R
-
1
3
4
a
M
i
n
e
r
a
l
M
i
n
e
r
a
l
R
-
2
2
F
r
i
c
t
i
o
n

c
o
e
f
f
i
c
i
e
n
t
0,026
0,028
0,030
0,032
0,034
0,036
0,038
Figure 5 Friction coefficients of the investigated lubricants, with and without
refrigerant.
The results shown in Figure 5 clearly show that the coefficient of friction in the
contact increases when refrigerant is present. The molecule structure of different
esters also affects the friction in the contact. Long straight molecules deform more
easily when subjected to shear stress that results in a lower coefficient of friction in
the contact. Ester C, which contains 100% branched acids, had the highest friction
coefficient of all the tested lubricants.
The lubricant acting in the contact can be in three forms; liquid-, visco elastic- and
solid phases, hrstrm et al [10]. The measured friction coefficients are thus a
combination of the effects of the three phases, namely, the solidified friction
coefficient [], the visco elastic friction coefficient and the friction coefficient of the
fluid. The dominant component for the oil/refrigerant mixtures in these experiments
is from due to the high pressure. Adding refrigerant to the oil causes the viscosity
to decrease, which will also reduce the friction coefficient component from the
119
liquid phase. The water content in the lubricant during the experiments is low and
the influence to the results is neglected.
Conclusions
Measurements of the limiting shear strength-pressure coefficient [] at 2.5 GPa have
been made on four refrigeration oils; either in there undiluted form or diluted with
refrigerant. The objective of the work was to see whether changes when the oils
were diluted with refrigerant. Since is coupled to the limiting shear stress, any
increase in will also increase the maximum stress in the material and cause the
maximum stress to act closer to the surface.
The results show that, at a pressure of 2.5 GPa, increases with increasing
refrigerant concentration at least up to a level of about 9 wt%. Since in actual
refrigeration conditions refrigerant concentrations in the lubricating oil can be up to
30 wt%, the measurements presented in this investigation must be used as an
indication of the affect that the refrigerant plays on the limiting shear strength-
pressure coefficient [] .
The investigation also agrees with the results of earlier investigations [9], which
indicated that molecule structure was an important parameter for achieving the
correct limiting shear strength-pressure coefficient [] for a given application.
The work also showed that the refrigerant concentration decreases much faster for
naphthenic mineral oil with R-22 than for polyolester oil with R-134a. Therefore,
the lubrication conditions will be better in a bearing lubricated by the naphthenic
mineral oil and R-22 than with a polyoletser oil and R-134a.
Acknowledgements
The author would like to thank SKF, CPI Engineering, Trane Company, York
Refrigeration and Carrier for support and help during this work. Thanks also go to
the Swedish Energy Agency for financial support.
120
References

1 R. Tuomas and U. Jonsson. Influence of Refrigerant on Viscosity and
Pressure-Viscosity Coefficient of Refrigerant Compressor Lubricants.
Proceedings of the 25th Leeds-Lyon Symposium, Elsevier Science B.V.,
1999.
2 U. Jonsson. Lubrication of rolling element bearings with HFC-polyol ester
mixtures. Proceedings of Nordtrib 1992, vol. 1, pages 403-411, 1992.
3 B. J. Hamrock and D. Dowson. Isothermal Elastohydrodynamic Lubrication
of Point Contacts Part I- Theoretical formulation. Journal of Lubricating
Technology, Transactions ASME, vol. 99, 2, pages 223-229, 1976.
4 B. J. Hamrock and D. Dowson. Isothermal Elastohydrodynamic Lubrication
of Point Contacts Part IIIFully Flooded Results. Journal of Lubrication
Technology, Transactions ASME, vol. 99, 2, pages 264-276, 1977.
5 G. Lundberg and A. Palmgren. Dynamic capacity of rolling bearings. Acta
Polytechnica, Mechanical Engineering, vol.1, 3, 1947.
6 E. Hglund. Subsurface Stresses in a Lubricated Rolling/Sliding
Elastohydrodynamic Line Contact Considering Limited Strength of the
Lubricant. Doctoral Thesis, Lule University of Technology, Sweden, ISSN
ISSN:0348-8373; 32D, 1984.
7 K.-G. Sundin and B.-O. hrstrm. Method for investigation of frictional
properties at impact loading. Journal of Sound and Vibration, vol. 222, 4,
pages 669-677, 1999.
8 B.-O. hrstrm. Investigation of frictional properties of lubricants at transient
EHD-conditions. ASTM Special Technical Publication, 1404, pages 221-234,
2001.
9 D. Henderson. Solubility, viscosity and density of refrigerant/lubricant
mixtures. The Air-Conditioning and Refrigeration Technology Institute. Rep.
DOE/CE/23810-34, 1994.
10 B.-O. hrstrm, C. Penchinat and T. Norrby. An experimental study of the
influence of heat storage and transportability of different lubricants on
friction under transient elastohydrodynamic conditions. Journal of
Engineering Tribology, vol. 217, 1, pages 27-38, 2003.