Lubrication of bearings in refrigerating machines ABSTRACT The - - PDF document

Ove Isaksson, Ph.D. Department of Machine elements, Luleå University of Technology, Luleå, ove.isaksson@mt.luth.se Roger Tuomas, Ph.D. Student Department of Machine elements, Luleå University of Technology, Luleå, tuomas@mt.luth.se

Lubrication of bearings in refrigerating machines

ABSTRACT The bearings in a modern refrigeration screw compressor are lubricated with a mixture of oil and refrigerant. However, little or no published bearing life data is available for the new generation non-chlorinated refrigerants. The work presented in this report concerns the development of a measuring technique and experimental equipment for bearing life studies. The equipment is intended to provide data about bearings lubricated with mixtures of oil/refrigerant for use by compressor designers. Bearing life is affected by the working lubricant’s ability to form a film to separate the contact surfaces. To provide a sufficient film thickness in an elastohydrodynamic lubricated (EHL) contact, the lubricant’s viscosity, η, and pressure-viscosity coefficient, α, both play an important role. The film thickness in an EHL contact lubricated with an oil/refrigerant mixture with increasing amounts of refrigerant and different load ratios have been measured experimentally. The lubricant mixture tested consisted

• f a VG68 polyolester refrigeration oil, Solest 68, with R-134a refrigerant. To measure the film thickness, an on-

line capacitance method, SKF´s Lubcheck was used. The amount of refrigerant in the oil was increased until the lubricating film broke down and asperity contact occurred. The refrigerant’s influence on the rheological properties of the oil was studied in a high pressure Höppler

• viscometer. R-134a, R-32, R-410a and R-22 refrigerants mixed with a VG68 polyolester oil, Mobil 68 Arctic,

were investigated. A test procedure has been developed to determine the refrigerant concentration at which metal-to-metal contact

• ccurs. The work showed that film formation in contacts lubricated with oil/refrigerant mixtures is more

sensitive to load than other investigations have indicated. The results also showed that run-in behaviour appears in bearings used in refrigeration applications and that the viscosity and pressure-viscosity coefficient decrease with increasing dilution by the refrigerant. INTRODUCTION Bearings in a screw compressor are used to provide accurate radial and axial positioning of the rotors. This makes it possible to design compressors with small clearances, which reduces leakage and thus increases the efficiency. Commonly used bearing types are single row angular contact bearings and cylindrical roller bearings; although other types such as deep groove ball bearings, four-point contact ball bearings, needle and taper roller bearings and nowadays also the CARB bearing are used. The lubricant used in a compressor is expected to be able to provide a lubricating film thick enough to prevent contact between asperities on the interacting surfaces. The lubricant also transports wear particles out of the compressor and helps cool the machine elements. In the screw compressor, the lubricant also plays a significant role in reducing the temperature of the compressed gas and providing a seal between the rotors. Oil is injected into the bearings and between the rotors, which brings the oil into contact with the refrigerant, resulting in the oil becoming diluted by the refrigerant. The lubricated bearings in a refrigeration screw compressor are often assigned to the elastohydrodynamic lubrication (EHL) regime, i.e. the surfaces are separated by a lubricating film and the lubricated surfaces deform elastically. To maintain the thermodynamic efficiency of the refrigeration system, the refrigerant is separated from the oil before it continues through the refrigeration cycle. However, whilst it is relatively easy to remove oil from the refrigerant, it is far more difficult to remove any remaining refrigerant from the oil. Fore this reason, the oil can become diluted by the refrigerant in concentrations of up to 30-40 %. Wardle et al [1] discovered that 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. Refrigerants and refrigeration oils CFC´s (Halogenated hydrocarbons, Freons), have been used as refrigerants since the 1930´s. CFC´s

were initially used because they were considered to be harmless to humans. However, Molina and Rowland [2] showed in1974 that refrigerants that contain chlorine were a key factor in the destruction

• f ozone in the upper atmosphere. In time, this lead

to industry developing more environmentally friendly refrigerants, such as hydroflourocarbons, which did not cause damage to the ozone layer. The first of these new refrigerants, R-134a, was introduced as early as the beginning of 1980. However this refrigerant did not become a commercial product until 1990 because of problems with miscibility with lubricating oil’s in the

• compressor. To help overcome this problem, new
• il families were developed such as

polyalphaglycol (PAG), polyolesters (POE) and polyalphaolefin (PAO). There is still a considerable amount of development to be undertaken since, these new oils have shown high wear rates on the machine elements in the compressors when diluted with the new generation

• f refrigerants. Jacobson [3] demonstrated

experimentally that roller bearings lubricated with chlorinated R-22 refrigerant have lower wear rates than the new R-134a under the same running

• conditions. One explanation is that chlorine in the
• ld chlorinated refrigerants, acts as an anti-wear

additive in the bearing. BEARING LIFE THEORY Lundberg and Palmgren [4] developed a method for calculating bearing life, which is shown below.

p

P C L       =

10

(Eq 1) The formula was standardized by ISO in 1962. In the formula, L10 is the statistic nominal life, in million of revolutions, which 90 % of the bearings will survive. C is the dynamic load capability and P is the equivalent dynamic bearing load. The exponent p varies depending on the type of bearing; p = 3 for ball bearings and p = 10/3 for roller

• bearings. 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 as well as the number of cycles. This formula is useful in many applications, but in

• rder to include other parameters that affect bearing

life, a modified life theory was developed and standardized by ISO in 1977:

10 3 2 1

L a a a Lna ⋅ ⋅ ⋅ =

(Eq 2) where Lna is the modified life in millions of revolutions, a1 is used if another reliability than 90 % is required for the bearing, a2 describes the bearing material. All bearing materials used today are better than those used in 1977 so a2 ≥ 1, and a3 is an operating parameter and is mostly dependant

• n the lubrication of the bearing. If the bearing is

running at normal contamination levels the value of a3 is equal to κ (kappa); the ratio between the actual viscosity of the lubricant, ν, and the minimum viscosity, ν1, needed to separate the surfaces.

1

ν ν κ = (Eq 3) The rated life L10 will decrease linearly with increasing load in a logarithmic plot. The adjustments factors, a1, a2 and a3 do not change the slope of this line. A complement to the bearing life models was developed by Wuttkowski and Ioannides [5] and is shown below.

p SKF nna

P C a a L       ⋅ =

1

(Eq 4) This new adjustment factor, aSKF considers the lubricating film (κ), level of contamination (ηc) and the fatigue load (Pu). For this adjustment factor, a complex relationship of the above mentioned parameters are given in SKF’s general bearing catalogue [6] as a function of

( )

P P η

u c

for different values of κ and depends on the bearing family. Problems associated with high concentrations of refrigerant in the oil have lead SKF to refine their bearing life theory for bearings operating in a refrigeration environment [7]:

1 72 , min 1

3 ν α α ν ν ν κ ⋅       ⋅ = =

eral adj adj

(Eq 5) In this equation, the actual viscosity is adjusted by the ratio of the pressure-viscosity coefficients of the actual lubricant, α, and a reference value αmineral, for pure mineral oil. The pressure-viscosity coefficient describes the viscosity change with pressure. ν1 is adjusted by a factor 3 for HFC refrigerants and a factor of 2 for HCFC (hydrochloroflourocarbons). This new value of κ is then used to determine the correction factor aSKF in SKF´s bearing life model. RHEOLOGY OF OIL/REFRIGERANT MIXTURES The chemical composition and viscosity data for a given oil are seldom available from the oil

• manufactures. It is even harder to find viscosity and

pressure-viscosity coefficient data for different types of oil/refrigerant mixtures. Jonsson and Lilje [8] developed an empirical model that can be used to predict the pressure-viscosity coefficients of mixtures of polyolesters and R-134a based on the amount of branched acids in the oil.

Table 1 used refrigerants.

Refrigerant Chemical name Type Molecular weight [g/mol] R-134a Tetrafluoro-ethane HFC 102.03 R-22 Chlorodifluoro-methane HCFC 86.48 R-410a 50/50% R-32/R-125 HFC Blend 72.58 R-32 Difluoro-methane HFC 52.02

The first experiments the relationship between viscosity, pressure-viscosity coefficient and refrigerant concentration has been investigated. Also a method to estimate the pressure-viscosity coefficient of oil/refrigerant mixtures based on the difference in molecular weight of the refrigerants are shown. Viscosity and pressure-viscosity coefficients for mixtures of polyolester oil and commonly used refrigerants were measured using a modified high- pressure Höppler viscometer, designed by Jonsson and Höglund [9]. The instrument’s accuracy is a few cP, which allows measurements of

• il/refrigerant mixtures at high levels of refrigerant

to be made. The viscometer can be pressurised up to 34 MPa, and the pressure-viscosity coefficient of the oil/refrigerant mixtures were calculated according to Barus [10]:

p

e ⋅ ⋅ =

α

η η (Eq 6) where, η is the measured viscosity, η0 the absolute viscosity at p=0 (atmospheric pressure) and α the pressure-viscosity coefficient. The oil tested was a refrigeration grade base stock Mobil Arctic 68, ISO VG68, containing less than 100 ppm water. The refrigerants used were R-134a used as a replacement for R-12, and R-410a as a

• blend. R-22 was included as a reference. More data

about these refrigerants is given in Table 1. The experimental results presented in Figure 1 show the viscosity and pressure-viscosity coefficients plotted against refrigerant

• concentration. The results show a decrease in

viscosity and pressure-viscosity coefficient when refrigerant is added to the oil. The results also show that a refrigerant with a heavier molecule, have a more rapidly decrease of viscosity and pressure- viscosity coefficient then a lighter refrigerant. Eyrings equation (Eq 7), described by Akei and Mizuhara [11], was used to estimate the pressure- viscosity coefficient for three different refrigerants mixed with the Mobil polyolester oil.

( ) ( )

lubr refr lubr refr refr mix

α 1 1

• m

s α α s m α + + − ⋅ = (Eq 7) where, αmix is the pressure-viscosity coefficient of the mixture, αrefr and αlubr are the pressure viscosity coefficient for pure refrigerant and oil respectively, srefr is the mass fraction of refrigerant and m is the apparent molecular mass ratio between the refrigerant and oil, m=M*

lubr/Mrefr, where M* lubr is

the apparent molecular mass of the oil and Mrefr is the molecular mass of the refrigerant. In Figure 2, the pressure-viscosity coefficient estimated using Eyrings equation is presented for the three mixtures of polyolesters and refrigerants. For the estimations, the pressure-viscosity measurements of the polyolester oil mixed with of R-134a refrigerant and the pressure-viscosity coefficient of pure R-134a were used to find the apparent mass ratio, M*

lubr, of the polyolester.

Temp 40 oC Concentration refrigerant [%] 10 20 30 Viscosity [cP] 20 40 60 80 Temp 40 oC Concentration refrigerant [%] 10 20 30 Pressure-Viscosity coeff. [GPa-1] 10 15 20 25 Temp 40 oC Concentration refrigerant [%] 10 20 30 Viscosity [cP] 20 40 60 80 Temp 40 oC Concentration refrigerant [%] 10 20 30 Pressure-Viscosity coeff. [GPa-1] 10 15 20 25

Figure 1 a) Shows the relationship between viscosity and refrigerant concentration and, b) The pressure-viscosity coefficient vs. refrigerant concentration relation. For ▲-R-32, ♦-R-134a, ■-R-410a and ♦-R-22 mixed with Mobil 68 Arctic POE VG68 oil.

To be able to estimate the pressure-viscosity coefficient of the three refrigerants mixed with the

• il, the pressure-viscosity coefficient of the pure

refrigerants has to be known. The curvature of the curves depends on the molecular mass ratio, m. From these results it can be concluded that the Eyring equation can be used successfully to eliminate tedious measurements for different oil- refrigerant pairs.

Temp 40 oC Concentration refrigerant [%] 20 40 60 80 100 Pressure-Viscosity coeff. [GPa-1] 5 10 15 20 25 R-32 R-410a R-22

Figure 2 shows the estimated pressure-viscosity coefficient

• btained using Eyrings equation for three refrigerants: ▲-R-32,

♦-R-134a, ■-R-410a and ♦-R-22 mixed with Mobil 68 Arctic POE VG68 oil.

OIL FILM THICKNESS IN COMPRESSOR BEARINGS When the bearing life is calculated using SKF´s new bearing life theory and the corrections to the κ- value applied, the suggested bearing is often over sized, which results in reduced system efficiency and larger and heavier compressors. More research is therefore necessary to refine the theory and/or the correction factor. From Hamrock & Dowsons [12,13] theory of EHL film thickness, (Eq 8), it can be seen that the viscosity, η0, is an important parameters as far as film thickness is concerned. Another important parameter included in the equation is the pressure- viscosity coefficient, α.

) 1 ( ' ' ' ' 63 , 3

68 , 073 . 2 49 . 68 , min k x x x

e R E w R E E E V R h

⋅ − − −

−         ⋅         ⋅       ⋅ ⋅ = α η

(Eq 8) where Rx is the effective radius, V the speed, E’ the effective module of elasticity, w the load and k the ellipticity parameter. Measuring oil film thickness in operating bearings In this section the method is described of the measuring technique and experimental equipment for bearing life studies where the lubricant is a mixture of oil/refrigerant. The experiments also investigates bearing run-in when operating in a refrigeration environment and how the film parameter, Λ [14], changes during run-in. A modified test apparatus described by Jonsson and Hansson [15] has been used in the experiments. In the experiments angular contact bearings were used, although the apparatus can also be used to test radial roller and CARBTM bearings. A capacitive measuring technique to estimate the bearing oil film thickness was implemented in the test apparatus. This technique has been used in

• ther investigations, i.e. Wikström and Jacobson

[16] and Williamson and Miller [17]. The method permits an estimation of the film thickness under

• perating conditions and detects local asperity
• contacts. The method has been developed by SKF

and is called Lubcheck MK3. A description of the apparatus can be found in the Lubcheck user manual [18]. In the test apparatus, the bearing acts as two capacitances in series; one between the inner race and ball and the other between the ball and the

• uter race. The output signal is therefore a

combination of the separation in the two contacts. Figure 3 shows asperity contact data and κ-value presented for a range of refrigerant concentrations. The number of contacts was counted for a period of five minutes and then plotted against the concentration at two different load ratios. The third curve shows the κ-value, used when calculating the bearing life. It can be seen that the κ-value reduces with increasing refrigerant concentration.

Refrigerant concentration [%] 10 15 20 25 30 35 40 45 κ−value 0,0 0,5 1,0 Number of contacts 50 100 150 200

Figure 3. The κ-value (kappa) and number of contacts plotted against the refrigerant concentration, at a shaft speed of 1500 rpm.• κ-value; ▲ Load ration 6.2; ♦ Load ratio 12.4.

According to the new life theory, the value of κ should be greater than 1 to sustain a proper lubricating film thickness. From experimental data, this indicates that the refrigerant concentration should not be higher than 15% if a proper film is to be maintained. It can also be seen, that the number

• f contacts per seconds starts to increase more

rapidly at a certain refrigerant concentration; about

Time [s]

2000 4000 6000

Number of contacts

500 1000 1500 2000 2500 3000

Vcap [V]

1 2 3 4

Time [s]

2000 4000 6000

Vcap [V]

1 2 3 4 5

Number of contacts

1000 2000 3000 4000 5000 6000

a, b,

Time [s]

2000 4000 6000

Number of contacts

500 1000 1500 2000 2500 3000

Vcap [V]

1 2 3 4

Time [s]

2000 4000 6000

Vcap [V]

1 2 3 4 5

Number of contacts

1000 2000 3000 4000 5000 6000

a, b,

Figure 4. Number of contacts and Lubcheck output Vcap, respectively, for two different load ratios during a run-in period of 7200 seconds. a) Load ratio, C/P=6.2; b) Load ratio, C/P=12.4.

20 % refrigerant concentration when C/P is 6.2 and at 28 % when C/P is 12.4. To give the same number

• f contacts for the two different load ratios, the

refrigerant concentration can be increased by approximately 8 % for the lighter load. Run-in of bearing surfaces in an atmosphere containing air usually takes place before the bearing reaches normal operating conditions [19]. In a refrigerant environment the behaviour of the run-in process is described briefly by Jacobson [20], but it cannot be said that the run-in behaviour has been thoroughly investigated. In Figure 4 the number of measured contacts and the Lubcheck signal Vcap during a two hour run-in period are plotted. The number of asperity contacts decreases during the run-in period whilst the Lubcheck signal increases and reaches a steady state level of about 4.3 V for a load ratio C/P=12.4 and 3.8 V for a load ratio C/P=6.2, respectively. The steady state level indicates that the bearing maintains a thicker lubricating film with the lighter load than in the more heavily loaded bearing. By calculating capacitance out of the measured Vcap, it can be shown that the capacitance increases by 23 % when the load is doubled. According to Hamrock and Dowson [12,13], the change in film thickness should be just 5 %, when the load is doubled. This is clearly not the case in the present study and indicates that load is a more important parameter as far as achieving a proper lubricating film thickness than suggested by Hamrock and Dowson. The arithmetical roughness, Ra, and the root-mean square roughness, Rq, on the race for a new test bearing were measured to 0.06 µm and 0.08 µm

• respectively. During the run-in period, the surface

is flattened and smoothed and the number of contacts decreases. For a test bearing running for 47 million revolutions in an oil/refrigerant mixture with a refrigerant concentration of 23 % and load ratio C/P=6.2 the roughness reduced to 0.02 µm respective 0.03 µm. In Figure 5, the film parameter, Λ, is shown as function of refrigerant concentration. The calculations were made for the inner ring contact at 1500 rpm with an axial load of 720 N, representing a C/P of 6.2.

Figure 5 The film parameter, Λ, plotted against refrigerant

• concentration. ─ for a new bearing; ▬ for a bearing that has

been run-in for 263 hours.

The film parameter varies with refrigerant concentration, due to the reduction of film

• thickness. In addition, run-in of the surfaces change

their roughness, which is shown by the two curves and affects the film parameter, the lower curve gives the film parameter for a new bearing, whilst the higher curve is that of a bearing run-in for 263 hours. CONCLUSIONS This report presents an investigation into the lubrication of bearings in a refrigeration screw

• compressor. Lubrication in this kind of

environment is a challenge due to the presence of refrigerant, which invariably becomes dissolved in the oil and dilutes it. To provide a proper film thickness in an elastohydrodynamic contact (EHL) the viscosity, pressure-viscosity coefficient and speed all play an important role. When the oil is diluted by refrigerant, it’s properties change and

contact between bearing surfaces can occur, leading to excessive wear. It is therefore of great importance to the compressor designers to know at what oil/refrigerant dilution the contact starts, with the associated risk for reduced bearing life. The objective of investigation was to develop a measuring technique and experimental equipment for bearing life studies for bearings lubricated with mixtures of oil/refrigerant. ACKNOWLEDGEMENTS At last I wish to thank the participating companies in the project; SKF, Trane Company, CPI Engineering, York Refrigeration and United Technology for their involvements in the research and for their financial support. REFERENCES 1 Wardle, F., P., Jacobson, B., Dolfsma, H. and Butterworth, A., “Lubrication of refrigerant compressor bearings”, US Patent 5469713, 1995. 2 Molina, M. J. and Rowland, F. S., “Stratospheric sink for chloroflouromethanes: chlorine atomic catalysed destruction of ozone”, Nature 249, 1974, pp. 810-814.

33

Jacobson, Bo, “Lubrication of Screw Compressor Bearings in the Presence of Refrigerants”, Proc. 1, 1994 International Compressor Engineering Conference at Purdue, Vol. 1, pp. 115-120. 4 Lundberg, G. and Palmgren, A.,”Dynamic capacity of rolling bearings”, Acta Polytechnica, Mech.Eng. Vol.1, No.3, 1947. 5 Wuttkowski, J., G. and Ioannides, E., “The new life theory and its practical consequences”, Ball Bearing Journal, April 1989, pp. 6-12, ISSN 0308-1664. 6 SKF, “General Catalogue”, edition 4000//II S, 1994. 7 Meyers, K., “Creating the right environment for the compressor bearings”, Evolution,

• NO. 4, 1997.

8 Jonsson, U. and Lilje, K., C., ”Elastohydrodynamic lubrication properties

• f polyol ester lubricants-R134a Mixtures”,
• Proc. 1998, International Compressor

Engineering Conference at Purdue, 1998,

• Vol. 1, pp. 123-128.

9 Jonsson, U. and Höglund, E.,”The influence

• f refrigerants on the high pressure

properties of lubricating oil”, Proc. of Nordtrib, 1992, Vol 3, pp. 89-96. 10 Barus, C., ”Isothermals, isopiestics and isometrics relative to viscosity”, Am. J., Sci.,

• Vol. 45, pp. 87-96, 1893.

11 Akei, M. and Mizuhara, K., ”The Elastohydrodynamic properties of Lubricants in Refrigerant Environment”, ASME/STLE Tribology conference in San Francisco, Oct. 13-17, 1996. 12 Hamrock, B., J. and Dowson, D. “(1976a) Isothermal Elastohydrodynamic Lubrication

• f Point Contacts”, Part I- Theoretical
• formulation. J. Lubr. Technol., 1976, Vol.

98, NO. 2, pp. 223-229. 13 Hamrock, B. J., and Dowson, D., “(1977a) Isothermal Elastohydrodynamic Lubrication

• f Point Contacts”, Part III- Fully Flooded
• Results. J. Lubr. Technol., 1977, Vol. 99,
• NO. 2, pp. 264-276.

14 Hamrock, B., J., ”Fundamentals of fluid film lubrication”, NASA Reference Publication 1255, 1991. 15 Jonsson, U. and Hansson, N., ”Elastohydrodynamic lubrication properties

• f polyol ester lubricants-R-134a mixtures”,
• Proc. 1998 International Compressor

Engineering Conference at Purdue, 1998,

• Vol. 1, pp. 129-134.

16 Wikström, V and Jacobson, B., “Loss of lubricant from oil-lubricated near-starved spherical roller bearings”, Proc. Instn.

• Mech. Engrs., Vol. 211, Part J, 1997.

17 Williamson and Miller, ”Condition Monitoring of Grease Lubricated Rolling Element Bearing”, NLGI Spokesman, April 1999. 18 Storken, J.,”Lubcheck user manual” SKF Engineering & Research Services B.V., 1997-08-27. 19

• D. J. Whitehouse, “Handbook of Surface

Metrology”, ch 7.5.2. 20 Jacobson, Bo, ”Ball Bearing lubrication in refrigeration compressors”, Proc., 1996 International Compressor Engineering Conference at Purdue, 1996, Vol. 1, pp. 103- 108.