Elastohydrodynamic Lubrication Properties of Polyol Ester Lubricants-R134a Mixtures

Transcription

1 Prde University Prde e-pbs International Compressor Engineering Conference School of Mechanical Engineering 1998 Elastohydrodynamic Lbrication Properties of Polyol Ester Lbricants-R134a Mixtres U. J. Jonsson Llea University of Technology K. C. Lilje CPI Engineering Services Follow this and additional works at: Jonsson, U. J. and Lilje, K. C., "Elastohydrodynamic Lbrication Properties of Polyol Ester Lbricants-R134a Mixtres" (1998). International Compressor Engineering Conference. Paper This docment has been made available throgh Prde e-pbs, a service of the Prde University Libraries. Please contact epbs@prde.ed for additional information. Complete proceedings may be acqired in print and on CD-ROM directly from the Ray W. Herrick Laboratories at Herrick/Events/orderlit.html

2 ELASTOHYDRODYNAMIC LUBRICATION PROPERTIES OF POLYOL ESTER LUBRICANTS-R134a MIXTURES Ulf J. Jonsson Llea University of Technology Division of Machine Elements SE Llea, Sweden Kenneth C. Lilje CPI Engineering Services, Inc. Midland, MI, 48642, USA ABSTRACT A modern screw compressor contains several highly loaded metal-to-metal contacts lbricated with oil, inclding, rolling element bearings, gears, and the rotors. These types of contacts are normally referred as EHD or EHL contacts (Elasto-Hydro-Dynarnic-Lbrication), de to the elastic deformation of the srfaces. The expected life of sch bearings is to a large degree related to the lbricant film thiclmess in the bearings. The properties of the lbricant that determine the film thickness in an EHD-contact are normally the dynamic viscosity, T], and the pressre-viscosity coefficient, a. In a refrigeration system, both those parameters are affected by the presence of refrigerant. A falling ball viscometer has been designed that can be pressrized p to 34MPa in order to measre the viscosity and pressre-viscosity of oil-refrigerant mixtres. Eight different polyolesters mixed with R-134a were evalated and the data presented. The reslts show that the degree of branching has a large impact on the pressre viscosity coefficient. An empirical model has been developed that can be sed to predict the pressre-viscosity coefficient based on viscosity and amont of branched acids in the lbricant. INTRODUCTION A modern screw compressor contains several highly loaded metal-to-metal contacts inclding rolling element bearings, gears and the rotors. These types of contacts are normally referred to as Elasto-Hydro-Dynamic- (EHD) contacts, de to the elastic deformation of the srfaces. The maximm pressre in sch contacts can exceed 4 GPa. The most important task for a lbricant in these compressors is the formation of a film that separates the interacting srfaces. The;! expected fatige life of a ball bearing is to a large degree related to the film thickness in the bearing de to the risk of asperity contact. This is very relevant in a refrigeration compressor since the working viscosity is often very low de to diltion of refrigerant. Contact between the srfaces will case locally high stresses reslting in fatige of the bearing material. Asperity contact may also reslt in wear of the bearing srfaces. The rate of this wear is fond very dependent on the type of oil and refrigerant sed. Jacobson [1] has tested anglar contact bearings lbricated with oil-refrigerant mixtres in a bearing cavity simlator. He fond that prevention of wear of the bearing srfaces typically reqires 50% higher viscosity when lbricated with HFC-134alpolyol ester mixtre than for HCFC-22/mineral oil lbricated bearings. The properties of the oil that determine the film thickness in EHD-contact are normally the dynamic viscosity, T], and the pressre-viscosity coefficient, a. In a refrigeration system, both these parameters are affected by the presence of refrigerant. Speaker and Spaschs (2] present reviews of varios methods of measring viscosity of oilrefrigerant mixtres. Jonsson and Hoglnd [3] sed a concentric cylinder viscometer to measre viscosity and pressre-viscosity coefficient of oil refrigerant mixtres. 123

3 A method for determining the viscosity of oil refrigerant mixtres by means of film thickness interferometry measrements has been sed by Jonsson and Hoglnd [4). This method was later developed frther by Akei et al [5, 6] to estimate the viscosity and the pressre viscosity coefficient of oil refrigerant mixtres, inclding a polyolester with R-134a. They sed the Eyring eqation, Eq.1, to estimate pressre-viscosity coefficient of oil refrigerant mixtres. m Srefr (arefr -albr) amix = ( 1 ) 1 Srefr m- + +albr In Eq. 1, is the pressre-viscosity coefficient of the mixtre, Cirefr and a 1 br are the pressre viscosity coefficient for pre refrigerant and lbricant respectively. Srefr is the mass fraction of refrigerant and m is the moleclar weight ratio of the lbricant and refrigerant. Akei and Mizhara sed data for the pressre-viscosity coefficient for R12 measred by Babb and Scot [7] to fit their data SKF [8] and Jacobson [9] sggest that the viscosity sed in calclations of bearing life is modified to compensate for the redction in pressre-viscosity coefficient compared to mineral oil The adjsted viscosity, Va.:j, for se in the evalation of bearing lbrication can be determined as follows: Va.:j=vmix (CI.miJraJ) 0 72 where: vmix and are the actal kinematic viscosity and pressre-viscosity coefficient of the oil-refrigerant mixtre at the location of the bearing and a...wera1 is the pressre-viscosity coefficient of mineral oil that is sed as a reference. Today, the oil manfactrers normally spply pressre-temperatre-concentration data and viscosity-temperatre concentration data for their prodcts. This type of data is also available in the literatre, however, detailed information on the composition of the lbricants is sally not disclosed. Pressre-viscosity coefficient data is seldom available de to the complexity of the measrements reqired to obtain the data. It is important that pressre viscosity coefficients of lbricantlrefrigerant pairs be available for new methods of bearing life calclation. This work aims at making sch data available in a format that can be sed in the refrigeration indstry. (1) TESTED LUBRICANTS Eight different lbricants were evalated with R-134a. The lbricants were all refrigeration grade basestocks containing less than 100 ppm water. They were selected to give a broad representation of commercial polyolesters for indstrial refrigeration applications. Polyolester properties can vary significantly, depending on the strctre of the acids sed in their preparation. In refrigeration application, branched acids impact properties dependent on interaction between refrigerant and the lbricant sch as solbility, miscibility and viscosity-concentration dependency. To better nderstand this phenomenon, esters with varying degrees of branching were examined. Table 1 lists the lbricants sed in this stdy and otlines their essential properties and strctral featres. The last two colmns lists the lowest temperatre a 10 respectively 20% oil-refrigerant mixtre will form a single phase. Table 1 List of tested lbricants Name Polyol %Branched Kinematic Viscosity 10%misc. 20%misc. acids viscosity, est index oc oc Ester-A Tech-PE ND ND Ester-B Mono-PE ND ND Ester-C Mono-PE <-60 <-60 Ester-D TMP Ester-E TMP Ester-F Tech-PE Ester-G Tech-PE Ester-H Di-PE The refrigerant sed in this work (R-134a, 1,1,1,3-tetraflorethane) was obtained from a commercial sorce and was sed as received. 124

4 EXPERIMENTAL METHOD The viscosity was measred with a falling ball viscometer at 40 and 80 C at pressres of 2, 17 and 34 MPa for each concentration. A few tests were made at 60"C to confirm the viscosity-temperatre relation as pblished in ASTM D341. The concentration of dissolved refrigerant was varied between 0 and 40% in steps of approximately 10% to cover the range fond in most applications.. The viscometer sed was a modified Hoppler viscometer. The working principle is shown in Figre 1 Locations of timing windows Figre I Schematic diagram of the falling ball viscometer The viscosity was determined by the time it took the ball to travel a defined distance in a tbe containing the test liqid. Two pairs of sapphire windows were sed to time the ball. The separating piston is sed to pressrize the test flid dring measrements. The test flid and the pressrizing flid are separated by two 0-rings on the piston. The volme between the 0-rings is ventilated throgh the piston rod to prevent contamination of the test flid in case of a leakage at the 0-rings. The viscometer was always pressrized above the vapor pressre of the test flid. A more detailed description of the viscometer can be fond in Ref. [10] Concentration of refrigerant was determined by taking a sample throgh a capillary tbe. The sample valve is an anglar pattern valve with no dead volmes ensring a well-defined sample. A spread sheet program was sed to calclate the viscosity for each test point. The program compensates for effects of pressre and temperatre on the dimensions of the ball and the bore. Eq (I) is sed to fit experimental data of pressre-viscosity coefficient vs concentration of refrigerant. RESULTS AND DISCUSSION Figre 2 shows the variation of dynamic viscosity, TJ, with refrigerant concentration at 40 and 80 C. The data show that increased branching reslts in a more rapid decrease in mixed viscosity at a given refrigerant concentration. This particlarly stands ot in lbricants that contain 100% branched acids (Esters E and H), bt can also be seen in the other combinations. Increased branching will also reslt in a higher solbility reslting in a frther decreased viscosity for a fixed temperatre and pressre. Figre 3 shows the pressre viscosity coefficient variation with refrigerant concentration.. Several generalities can be seen in this data. Within the ISO VG 32lbricants (esters A-D), the a.-vale shows a strong correlation to % branching, with higher branching reslting in higher a.-vales. The ndilted a.-vales for esters E and Hare the extreme examples of this. Ester E has a higher a.-vale than esters F and G despite significantly lower viscosity, sggesting that in some cases branching afford extremely high a.-vales. The data also sggests that the redction in a.-vale with increased refrigerant concentration is larger as branching increases in the lbricant strctre. This is particlarly apparent in esters which are 100% branched (esters E and H). This data sggest that cation shold be exercised when sing ndilted pressre-viscosity coefficients to predict EHD lbrication-properties of lbricant/refrigerant pairs. 125

5 C C 0 Ester A p.., /::,. Ester B " Ester C! 0 Ester D l... Ester E > Ester F " '"V:i>ot;;. EsterG Ester H... ;::.. 5 'QO 5 o Q 3 y 'DO " i> Concentration ofr-134a, %w/o Concentration ofr-134a, %w/o Figre 2 Viscosity at 40 and 80 C of the tested lbricants vs concentration,.c:tj p.., C 30 sooc 0 Ester A rj:i 26 0) EsterB v Ester C 0 EsterD EsterE T <> EsterF EsterG > EsterH.t, ;:; P:: Concentration ofr-134a, %w/o Concentration ofr-134a, %w/o Figre 3 Pressre-viscosity coefficient at 40 and 80 C vs concentration The data shown in Figre 3 is fitted with Eq. 1 in two steps. The first step was sed to estimate an effective pressre-viscosity coefficient of pre R-134a, C4efr. The reslting effective refrigerant pressre-viscosity coefficient was 11 GPa- 1. This was then inserted into Eq (1) to obtain a two-parameter model. The reslting coefficients are listed in Table 2 Table2 Reslts from crve fit POE-A POE-B POE-C POE-D POE-E POE-F POE-G POE-H 40 C O:!br m C CI!br m Figre 4 illstrates how the pressre-viscosity coefficient varies with temperatre for pre lbricants and for lbricant refrigerant mixtres. The diagram shows that the pressre-viscosity coefficient-temperatre relationship can be approximated with a linear model. 126

6 E:s:tcrB,23% o F.<erE,O%.,. F.<lerE, 2010 F.<lerf,O% F.<lerF,23% 0 F.<erG,O% F.<erG,l6% F.<erH,O% (j F.<lerll, l8% - I.in<arfit Figre Temperarre, oc 80 Pressre-viscosity coefficient for different lbricant-refrigerant mixtres at 40, 60 and sooc LUBRICANT STRUCTUREwPRESSURE VISCOSITY COEFFICIENT MODEL Figre 3 shows the strong inflence that the viscosity and the amont of branched acid has on the pressreviscosity coefficients of the lbricants. The following approximation, Eq. 2, can be sed to estimate the pressre viscosity coefficient of polyolester refrigeration oils at operating temperatre. The eqation nder estimated the a vale of Ester A at 80 C with 17% bt for all the typical refrigeration oils inclded in this work, the estimated a vale is within ±:8% of the measred vale over the whole temperatre range. albr*=a+b*lnt]+c*x (2) In this model, ClJbr* is the predicted a-vale for the pre lbricant, T] is the ndilted dynamic viscosity and X is the amont of branched acids in the lbricant strctre in percent. A, B and Care constants determined to be 2.611, 3.391and 0.065, respectively. The redction in a with refrigerant diltion can then be calclated sing Eq (1) and the constants from Table 2. Temperatre corrections can also be made sing the graph in Figre 4. Figre 5 provides a graphical representation of the calclation. The ndilted a-vale is estimated by straight line extrapolation from the viscosity and the% branching. The dilted a-vale is then estimated by following the nearest crved line to the desired diltion and reading back to the y-axis "t) > 15 I ;::l Figre 5 46 Refrigerant concentration, %w/o ISO VG 68,75% Branched, 10% R-134a a. =19.2 GPa' Viscosity, cp Plot for estimation of dilted pressre-viscosity coefficient at 40 C Temperatre corrections can then be made sing the graph in Figre 4. The graph can be sed to directly estimate the Widilted a-vale from 40 to 80 C. The graph shall not be sed to adjst a-vales for dilltion at other temperatres than 40 C since the a-vale vs concentration crves are generated sing data for 40 C. The graph is generated sing the dynamic viscosity. The error introdced by sing it with kinematic viscosity as inpt can be neglected from a practical point of wiew

7 CONCLUSIONS The pressre viscosity coefficient and the viscosity of varios lbricant/r-134a pairs were investigated sing a high-pressre falling ball viscometer. The stdy shows that the lbricant strctre has a strong inflence on interactions with the refrigerant. Esters with mainly branched acids show the highest pressre viscosity coefficient. The data also sggests that the redction in pressre viscosity coefficient with increased refrigerant concentration is larger as branching increases in the lbricant strctre. Care shold be exercised when sing ndilted pressreviscosity coefficients to predict EHL-properties of lbricant/refrigerant pairs. A general model was developed to predict pressre viscosity coefficients of polyol ester/r-134a pairs based on initial viscosity and percentage branching. The effective pressre-viscosity coefficient for pre R-134a was estimated to 11 GPa" 1 A graphical method was developed that allows pressre-viscosity coeficients of pol yolester/r -134a mixtres to be determined. REFERENCES [1] Jacobson, B., "Ball bearing lbrication in refrigeration compressors". Proc. of the 1996 International Compressor Engineering Conference at Prde. pp (1996) (2] Speaker, L. M. and Spaschs H. 0. "A stdy to increase the state-of-the-art solbility and viscosity relationships for oil/refrigerant mixtres" ASHRAE Research report, Project 444.(1986) [3] Jonsson, U. and Hoglnd, E.. "The inflence of refrigerants on the high pressre properties of lbricating oil", Proceedings ofnordtrib '92. vol. 3, pp (1992) [ 4] Jonsson, U. and Hoglnd, E. "Determination of Viscosities of Oil-Refrigerant Mixtres at Eqilibrim by means of Film Thickness Measrements". Ashrae transactions, VoL 99, Pt 2 (1993) [5] Akei, M. and Mizhara, K., Taki, T., and Yamamoto, T., "Evalation of film-forming capability of refrigeration lbricants in pressrized refrigerant atmosphere", Wear 196, pp (1996) [ 6] Akei, M., Mizhara, K., "The Elastohydrodynamic properties of Lbricants in Refrigerant Environments". ASME/STLE Tribology conference in San Francisco Oct 13-17, (1996) [7] Babb, S. E. and Scot G. J., "Rogh viscosities to Bars," Jor. Chern. Phys., 40, 12. pp (1964) [8] "Bearings in Twin Screw Compressors- Application Handbook", SKF, King of Prssia, USA, To be pblished Jly 98 [9] Private commnication, Professor Bo Jacobson, Lnd, Sweden [10] Jonsson, U. and Hoglnd, E Lbrication ofrolling element bearings with HFC-polyolester mixtres. Accepted for presentation at Nordtrib'98, Dennmark 128