Experimental Methods for Making BOFT Measuremen- 123docz.net

1. Oil Film Thickness in Engine

1.2 Does the Automotive Industry Need a Standard Engine Test

1.2.4 Experimental Methods for Making BOFT Measurements

ment is to electrically insulate the test bearing from the rest o f the engine as shown in Fig. 1.2.1. Two methods have been used to accomplish this task. In one instance, a sufficient amount o f metal is removed from the inside diameter of the bearing housing and replaced w ith a thin (nominally 0.1 mm thick) plastic insulating sheet ( 2 , 3 , 4 , 5 , 6 ). In the second instance, the metal removed from the diameter o f the bearing housing is replaced w ith a comparable th ic k ness of an " alumina ceramic" coating (7, 8 ). In either case the bearing can be sufficiently insulated.

B earing

Figure 1.2.1: Engine Test Bearing fo r Making Resistance or Capacitance Film Thickness Measurement

In addition to insulating the bearing, a second requirement o f these methods is to electrically ground the crankshaft. Most references have accomplished this task by simply draping a braided metal wire, one end o f which is connected to ground, over the portion of the shaft extending outside o f the engine. By con

necting electrical leads to the side o f the test bearing and to "ground", an electri

cal circu it can be completed.

Depending on whether measurements of electrical resistance or capacitance are to be used in calculating film thickness values, two d ifferen t methods fo r apply

ing a voltage across the bearing are employed. Spearot and Murphy (9) reviewed both methods and described the techniques fo r applying voltages in detail.

In the case of resistance measurements, a low voltage, DC signal, is used as shown in Fig. 1.2.2, and the resistance o f the oil film , R0 , in the bearing is calcu

lated directly from circuit analysis. The value of the self-generated voltage, f , produced during bearing rotation can be determined from measurements of bearing voltage w ith o u t the excitation signal applied. In the case o f capacitance measurements, a high frequency, AC signal is used as shown in Fig. 1.2,3, and the capacitance, C0 , o f the oil film is calculated from calibration curves gene

rated by using the circuit w ith known capacitors in place o f the test bearing.

The output voltage produced by the bearing in either case is measured by means of a data acquisition system capable o f collecting data at speeds comparable to the time required to rotate a few crankangle degrees.

R ESISTANCE MODEL O F T E S T BEARING

MEASURED BY W AVEFORM ANALYZER

RESISTANCE TECHMOUE ELECTRICAL EXCITATION

Figure 1.2.2: Total Resistance Measurement Method, Reprinted w ith permis

CA P A C ITA N C E MODEL OF T E 8 T BEARING

CAP AC ITAN C E TECHNIQUE ELEC TR IC AL EX C ITATIO N

" V W --- 100.000 Hz 3.0 k O 3.0 V RMS A C FUNCTION

g e n e r a t o r

IN270

I— K H

(N270 V 0.05 N/F

—vW —1

100 kO MEASURED I V V Q WAVEFORM ANALYZER

Figure 1.2.3:

Total Capacitance Measure

gineers, Inc.

Having measured the resistance or capacitance o f the total oil film at any instant, the minimum film thickness, hm , in the bearing can be calculated directly from either Equation 1 or 2, respectively.

hm = ô

*m = ô

y i ^ Z n W R j f y 1 J

/ f 2ffWRBD0 k I 2 '

^ J .

(1)

(2) In these equations, W is the w id th , Rg is the radius, and 5 is the radial clear

ance of the test bearing. Dq is the dielectric constant, a is the electrical con

ductivity o f the oil, and k is tne p e rm ittivity of free space.

By coupling the film thickness values provided by Equations 1 or 2 w ith a measure of angular rotation of the crankshaft, plots of minimum film thickness in the bearing versus crankangle during a single combustion cycle can be con

structed using either method as shown in Fig. 1.2.4 fo r a 2.51 L-4 engine. It has been demonstrated (9), that the resistance and the capacitance technique can provide comparable film thickness values when used in the same engine at a fixed set o f conditions w ith the same o il. However, statistical analyses have also demonstrated that it is easier to measure film thickness w ith greater precision using the capacitance technique than it is using the resistance technique. The explanation fo r this finding is that the dielectric constant o f the o il, D0 , is much less a function of temperature and oil chemistry, than is oil conductivity, a. The precision w ith which dielectric constants can be measured is reflected in greater precision in calculated values o f hm .

Figure 1.2.4: Minimum Bearing Oil Film Thickness fo r an Oil During a Single Combustion Cycle as Determined by Both the Total Capacitance and Total Resistance Methods. Reprinted w ith permission, Copyright 1988, Society o f Autom otive Engineers, Inc.

Although the precision o f minimum oil film thickness values calculated has been shown to be very good when determined using the capacitance method (9), there are still sources o f error in the development o f Equations 1 and 2 which could affect the accuracy of hm values. It is assumed in the derivation o f these euqations that the geometry o f both the bearing and the journal are circular. It is known that this is not true fo r the bearing; the two bearing shells form a slight oval. However, the deviation from circular geometry is not expected to greatly influence the calculation o f hm as long as the minimum film thickness does not occur near the bearing "sp lit line".

A more critical concern involves the possible presence o f cavitation in the bearing. It has been shown (9) that the presence o f cavitation can influence the values o f hm calculated either by the capacitance or the resistance methods.

Because the dielectric constant o f air is closer to that o f D0 than the electrical conductivity o f air is to that o f a, cavitation affects the values determined with the capacitance technique less than it does the resistance method. However, the presence o f cavitation in the bearing w ill increase the calculated values o f hm over the actual film thickness in the bearing no matter which method is used.

The greater the extent o f cavitation in the bearing, the greater w ill be the effect on calculated values o f hm . A t the same time, the smaller the value o f hm (greater eccentricity), the less effect any fixed amount o f cavitation w ill have on the calculated film thickness.

12 .5 Effects of Oil Rheology on Bearing Performance Single-Grade Oils

An example o f how oil rheological properties affect film thickness in an engine journal bearing can be readily provided by a series o f single-grade oils as shown in Fig. 1.2.5. For the fro n t main bearing o f a 3.8I V -6 engine at the con

ditions indicated, the effect of viscosity grade in going from an SAE 20 to an SAE 40 oil is to merely shift the minimum film thickness curve to higher values.

In interpreting the curves in Fig. 1.2.5, it is im portant to remember that every point along each curve is a minimum film thickness at a particular time during the combustion process ( in this instance the data are referenced to the com

bustion process in cylinder # 1, the closest cylinder to the fro n t main bearing).

A t some point during the cycle, this locus o f minima, forms an absolute minimum. It is this absolute minimum which is o f interest because it defines the point at which the bearing and the journal come closest to touching during the combustion process (in this instance the data are referenced to the com

bustion process in cylinder # 1, the closest cylinder to the fro n t main bearing).

A t some point during the cycle, this locus o f minima forms an absolute m ini

mum. It is this absolute minimum which is o f interest because it defines the point at which the bearing and the journal come closest to touching during and load, and the type o f bearing being analyzed. Presumably because of greater inertial loads, con-rod or “ big-end" bearings have an absolute minimum film thickness which falls w ith in the induction or exhaust strokes o f the combustion cycle more often than during the power stroke (8 ).

Many published studies have demonstrated that BOFT values correlate well w ith a variety o f different measures o f single-grade oil viscosity. In some studies (3 , 5), linear correlations were developed between film thickness and kinematic viscosity. In some (3, 5), linear correlations have been developed between film thickness and high-temperature, high-shear viscosity (defined in this w ork as viscosity measured at 150°C and 106 s- 1 ). In other studies (7, 8 ), conducted over wider temperature ranges, excellent linear correlations have been developed between film thickness and the square root o f viscosity measured either in kinematic viscometers or at HTHS conditions. For a range of engine speeds and loads, Deysarkar (10) has demonstrated a good correlation fo r single-grade oils between film thickness in a main bearing and the square root o f a Sommerfeld Number, S, defined as:

S = (3)

where N is the engine speed in RPM, n is the HTHS viscosity in cP, and P is the o utput torque o f the engine in N • m. As shown in Table 1.2.1, fo r all o f these correlations the Indices o f Determination (R 2 values) range from a low o f 0.71 to a high o f 0.99 demonstrating a clear relationship between the viscometric properties o f these single-grade oils and journal bearing film thickness in operat

ing engines.

These correlations all appear valid because of three reasons. First, single-grade oils are essentially Newtonian, and it is o f little consequence to the correlations what shear rate exists in the bearing or in the laboratory viscometer used to measure their viscosity. Second, since the heat transfer characteristics o f these mineral oil-based fluids are all approximately the same, and since viscous heating is a uniform function o f oil viscosity, any temperature rise in the bearing is hidden in the film thickness/viscosity correlations. Finally, over the range of pressures generated in a hydrodynamic journal bearing, the pressure-viscosity coefficients o f the oils are constant and roughly equivalent. Thus any pressure effects are constant across the sampling o f test oils investigated. If single-grade, Newtonian oils were all that were commercially available, these data demon

strate that current SAE J300 kinematic viscosity specifications would adequately predict bearing performance.

Table 1.2.1: Correlations between Bearing Oil Film Thickness and Single-Grade Oil Viscosity AuthorsReferenceBearingCorrelating VariableIndex of Determination RJ

Comments Spearot, Murphy, and Rosenberg3MaintJKin<100°C)0.71 Spearot, Murphy and Rosenberg3Main’IHTHS0.73 Girshick and Craig5Maini}Kin (100°C)-- -- -- -(a) Girshick and Craig5Main^HTHS-- -- -- -(a) Bates and Benwell7Con-RodtoKin)0'50.98(b) Bates and Vickars8Con-Rod (>JKin>a50.98(c) Bates and Vickars8Con-RodtoHTHS*0,50.98(d) Deysarkar10MainS0.50.90(e) Comments (a) RJ value for BOFT correlation with single-grade oil viscosity not specified, but qualitatively the correlation appears excellent. (b) Viscosity calculated from kinematic viscosity times density measured at bearing temperatures. (c) Viscosity calculated at bearing temperatures. (d) Viscosity calculated at 10* s—1 and bearing temperatures. (e) Sommerfeld Numbers calculated using tjhthS-

Multigrade Oils

As apparently simple and straightforward as the interpretation o f BOFT data fo r single-grade o il is, the analysis o f similar data fo r multigrade oils is complex and ambiguous. As shown in Table 1.2.2, fo r some o f the same studies which demonstrated consistent correlations between bearing film thickness and visco

sity in the case o f single-grade oils, a wide disparity o f results are documented fo r multigrade formulations. Indices o f Determination range from 0.0 t o 0.90 for differenct correlations. In addition to providing lower values o f R2 in every case, the regression equations between BOFT and multigrade oil viscosity are statistic

ally different from regressions between BOFT and single-grade oil viscosity.

Since the engine is not aware o f the form ulation differences between single- and multigrade oils, the lack of agreement between correlations involving these tw o classes o f lubricants must be due to either 1) measuring their viscosities at temperatures and shear rates which are not representative o f bearing operation or 2 ) the effect of some other engine or oil variable (possibly oil elasticity) which has ye t to be taken into consideration.

Recently, Spearot, Murphy, and Deysarkar (11) developed techniques fo r obtaining better estimates o f the temperatures and shear rates to which oils are subjected in the fro n t main bearing o f a 3.8I engine. By assuming that viscous heating in the bearing is minimized when extremely low-viscous, New

tonian oils are evaluated at low sump oil temperature, a "tru e " BOFT versus Sommerfeld Number curve in the presumed absence o f viscous heating is generated fo r the test bearing. Using this curve and the measured BOFT values for higher viscosity oils at other engine conditions, the "tru e " Sommerfeld Number for the bearing can be estimated. Since the engine speed and load are known, the viscosity in the bearing can be calculated and using the well known Walther Equation, ASTM D446, the temperature corresponding to this viscosity can be determined.

Using the difference in temperature between the oil in the bearing and the oil in the sump as a dependent variable, a multi-variable, linear regression technique was used to develop an equation to describe this temperature difference in terms o f sump temperature, sump oil viscosity, engine speed, engine load, and minimum oil film thickness in the bearing. As shown in Fig. 1.2.6, the temperature difference predicted from the regression equation agrees reasonably well w ith that calculated from the previously described Sommerfeld approximations. More im portantly, both analyses predict that under certain engine operating conditions, the temperature o f the o il in the bearing can be more than tw enty degrees greater than that in the sump. Thus, assuming the oil in the bearing to have the same temperature as that in the sump can be a gross miscalculation.

Table 1.2.2: Correlations between Bearing Oil Film Thickness and Multigrade Oil Viscosity AuthorsReferenceBearingCorrelating VariableIndex of Determination R2

Comments Spearot, Murphy, and Rosenberg3Main*?Kin HOO-C)0.00' (a) Spearot, Murphy, and Rosenberg3Main^HTHS0.01(a) Girshick and Craig5MainriKin (100°C)0.22(b) Girshick and Craig5Main^HTHS0.76(c) Bates and Vickars8Con-RodtoKin>°-50.82 to 0.83(d) Bates and Vickars8Con-RodtoHTHS)0-50.88 to 0.90(d) Deysarkar10MainS0.50.64(e) Comments (a) Ra valuefor BOFTcorrelationwithmultigradeoilviscositycalculatedincluding datafor four oils which were subsequently identified as statistical ''outliers". (b) R2 value calculated for data set including both single- and multigrade oils. (c) R2 valuecalculatedfor data setincludingbothsingle- and multigrade oils. Viscosity evaluated at 106s—1 and 100°C. (d) Viscosity evaluated at bearing temperatures. Different RJ values correspond to different engines. (e) Sommerfeld Numbers calculated using r^THS-

Some o f the data in Fig. 1.2.6 suggest that there can be a negative temperature rise between the oil sump and the bearing. This finding is believed to be an artifact of the way these engine tests were conducted. The sump temperature was controlled by means o f a heater and cooling coils independent o f what speed and load conditions were applied to the engine. A t some sets o f conditions the sump temperature was controlled at a higher value than the engine would produce in the absence o f any sump controls. When this occurred, the result was a decrease in oil temperature as the oil passed from the " h o t" sump to the relatively cool engine block. This negative temperature rise could be detected by means o f the assumptions described in the preceding paragraphs.

>•

TEM PERATURE „

RISE '*

ES TIM A TED FROM SOMMERFELD

A NALYSIS. .

*C 1

TEM PERATURE RISE PREDICTED FROM REGRESSION ANALYSIS. *C

Figure 1.2.6: A Comparison o f the Temperature Rise in the Bearing Predicted by M ultilinear Regression Analysis w ith that Calculated from Sommerfeld Approxim ations. Reprinted w ith permission. Copy

right 1989, Society o f Autom otive Engineers, Inc.

Using the regression equation to calculate bearing oil temperatures and deter

mining the corresponding viscosities provides the BOFT versus Sommerfeld Number curve shown in Fig. 1.2.7. The assumptions made regarding the Som

merfeld Number and the amount o f viscous heating produce a high degree o f correlation fo r this series o f single-grade oils. Although the temperature regression equation developed is truly valid only fo r the engine bearing and operating conditions used in this work, it can be used to estimate the tempe

rature rise in the bearing fo r a series of multigrade oils over the same range o f operating conditions.

Figure 1.2.7: Absolute Minimum Oil Film Thickness as a Function of Sommer- feld Number fo r Single-Grade Oils at the Estimated Bearing Oil Temperature. Reprinted w ith permission, Copyright 1989, Society o f Autom otive Engineers, Inc.

Spearot, Murphy, and Deysarkar (11) also calculated values fo r different shear rates associated w ith the test bearing. The tw o shear rate definitions which were thought might be related to film thickness values in a journal bearing were 1) the maximum shear rate, and 2 ) the average shear rate in the loaded portion of the bearing. The maximum shear rate occurs at the minimum film thickness point in the bearing and thus the viscosity calculated at such conditions could influence BOFT. The average shear rate in the loaded portion o f the bearing is a variable which reflects the shearing conditions in the entire portion of the oil film which carries the applied load. Thus, it could also influence BOFT.

By using the temperature regression equation to calculate the temperature of the oil in the bearing and using these different shear rate definitions, the viscosity of a series o f both single- and multigrade oils were calculated at different bearing operating conditions. Using these viscosities to calculate values o f S, regressions between film thickness and Sommerfeld Number were constructed as shown in Figs. 1.2.8 and 1.2.9. In each o f these figures, (1) provides the raw data, and (b) provides the linear regressions through the data.

(b ) R E G R E S S IO N S

ô-

S IN Q L E -G R A D E OILS

R l-£ S B 2 _ _ _

M U L TIG R A D E OILS R ' - 0.9 81 __________

A B S O L U T E M IN IM U M

O IL F IL M 3H T H I C K N E S S .

f i m

B O T H O IL S R* - 0 .9 8 6

OIL FILM TH IC K N E S S C O R R E C TE D FOR ENGINE D R IF T

SO M M E R FE LD N U M B E R C O R R E C TE D FOR TE M P E R A TU R E A N D SH EAR RATE

10 —r -

13 14 —1—

13 S Q U A R E R O O T O F S O M M E R F E L D N U M B E R

A B S O L U T E M IN IM U M

O IL FILM T H I C K N E S S .

Atm

(b ) R E G R E S S IO N S

S IN G L E -G R A D E OILS (V _ -_ 2 ]9 8 S _ _ _

/ / M U L TIG R A D E O ILS

R1 - 0.070 /

/ / / B O T H O IL S

R* - 0 .9 7 9 yy

yy yy

OIL FILM TH IC K N E S S C O R R E C TE D FOR ENG IN E D R IF T

S O M M ER FELD N U M B E R C O R R E C TE D FOR TE M P E R A TU R E A N D S H E AR RATE

2 4 1 0 12

S Q U A R E R O O T O F S O M M E R F E L D N U M B E R

Figure 1.2.9: Relationship between Bearing Oil Film Thickness and Sommer

feld Number at the Maximum Shear Rate. Reprinted w ith Per

In the case o f viscosities based on the average shear rate in the loaded quadrant of the bearing, as shown in Fig. 1.2.8, the film thickness values provided by multigrade oils are statistically indistinguishable from those provided by single

grade oils at a confidence level o f 95 percent. Using this combination o f tempe

rature and shear rate conditions, it would be concluded that multigrade oils perform the same as single-grade oils in a journal bearing and that there is no elastic benefit due to the presence o f the high molecular weight polymer in the multigrade form ulation.

In the case of viscosities based on the maximum shear rate in the bearing, as shown in Fig. 1.2.9, the film thickness values provided by multigrade oils are greater than those provided by single-grade formulations. The regression curves based on these data are statistically different at the 95 percent confidence level.

Although the differences do not appear large, certain multigrade oils, parti

cularly at low values o f S, provide as much as a 25 percent greater film thickness than single-grade oils at the same value o f S. Using this combination o f tempe

rature and shear rate conditions, one might conclude that the polymer in m u lti

grade oil formulations does provide an additional benefit to journal bearing performance over the provided by its viscometry properties.

One possible explanation for some additional benefit associated w ith multigrade oils is that of fluid elasticity. The question o f whether or not high molecular weight polymer blended into a multigrade oil could produce sufficient elastic forces to influence bearing operation has been debated fo r many years. As of yet there has been no definitive proof o f such elastic benefits. Bates, Williamson, Spearot, and Murphy (12) attempted to relate bearing film thickness measure

ments to both viscous and elastic parameters. A multivariable linear regression o f the form :

h m = C0 +C ,t? + C20 (4)

was used where r? and 6 are the viscosity and the relaxation time o f the oil, respectively, caculated at the bearing temperature and maximum shear rate. As shown in Fig. 1.2.10, Equation 4 which gives a predicted film thickness based on fluid properties was f it to a collection o f measured film thickness data for both single- and multigrade oils w ith a reasonable degree o f success (R1 = 0.73).

Although this is not absolute proof o f the importance of oil elastic properties, it does lend credibility to theories which include the influence of such effects.

The question o f which o f the tw o analyses described in Figs. 1.2.8 and 1.2.9 is correct w ill have to be determined from further research into the operation of journal bearings as well as research on the rheological properties o f engine oils, particularly at high temperatures and shear rates. The numerical solution to bearing design equations fo r both non-Newtonian and elastic fluids should provide an understanding of what characteristic shear rate is required fo r de

Experimental Methods for Making BOFT Measurements

Description of the Engine Oil Additives