Lubricants and Lubrication / D. Dowson et a]. (Editors) (0 1995 Elsevier Science B.V. All rights reserved.
355
Analysis of Friction in a Modern Automotive Piston Ring Pack ('
D. Radcliffe and D. Dowson
Ilepartment of Mechanical Engineering, The University of Leeds, Ilceds, LS2 9JT, UK. 'The lubrication regimes experienced by piston rings vary from full film hydrodynamic in the midstroke positions to mixed or boundary at the ends of the stroke. Full analyses of the mixed lubrication of piston rings have been rnade, but this work uses a simplified approach to analyse lubrication and friction in a modern automotive ring pack. Predictions for friction show good correlation with test measurements, showing that the simplified analysis is useful and that the simplifications are justified. 1. Introduction This paper describes piston and piston ring friction investigations made during a recent P h D st,udy [l]. The piston and ring pack is generally f t i e greatest contributor to IC engine frictional losses and can account for 2030% of the total ['L], with the split between piston and ring pack )wing approximately 30/70%. Engine manufact.itrers are still looking for improvements in power output and efficiency and a reduction in piston iitld ring pack losses will assist this quest. A current trend in engine oils is the use of lower viscosity grades such as multigrade 10W30, for reduced viscous frictional losses, at the expense of thinner lubricant films. Increasingly strict emissions legislation dictates that engine oil consumption should be kept to a minimum, necessitating higher piston ring contact pressures and thinner oil films. Both these factors mean that modern piston rings operate with a high degree of mixed lit brication. Mixed lubrication of piston rings is often analysed by considering the Average Reynolds' Equa(,ion of Patir and Cheng [3] with Greenwood and 'l'ripp's [4]asperity contact model, [571. This work, however, makes a simplified approach to the determination of mixed friction based on a rc,presentation of the Stribeck curve. The ring pack analysed in the work was from a current automotive engine with dimensions 490mrn x 71mm, in the configuration shown in
figure 1. The top ring is a barrel faced chrome plated compression ring, the second ring a tapered Napier scraper ring and the oil control ring is in 3 pieces, 2 chrome plated steel rails with a spring steel expander.
Piston 3 piece Oil Control Ring
I
Nopier Scroper Ring
Eorrel Top
Ring
Cylinder liner
Figure 1. The configuration of the ring pack
2. Lubrication Model Lubrication of the rings was analysed using the traditional approach based upon numerical solutions to the Reynolds' equation after [8]. It is briefly summarised here for completeness. The top ring was represented by a parabola as shown in figure 2, while the scraper ring was represented by a plane inclined slider bearing, as
356 Table 1 Nomenclature the variation in thickness of the oil film. gas pressure at oil film inlet, 21 gas pressure a t oil film outlet, 2 4 the hydrodynamic pressure in the oil film oil flow rates a t 22, upstream and inlet ring leakage factor, (normal = 1.0) ring axial coordinate start of oil film inlet oil film cavitation position oil film reformation position oil film outlet position physical lower limit of oil film physical upper limit of oil film crank angle increment for iteration lubricant viscosity film thickness parameter, = hmln/o coefficient of boundary friction angular velocity of crankshaft rads/sec composite surface roughness, ring & liner
= 6qU = 1270 C, D, E = integration constants for Reynolds’ equation
A
B
I1
12 13 I4
J$
J 3 .dx
s$ J$
J+*dx ring profile’s parabolic radius of curvatnre sliding speed of ring, +ve for downstrokes 11 b axial height of ring 0% sqneeze velocity hydrodynamic pressure gradient offset of parabolic ring profile, +ve up e friction force per unit length S the minimum oil film thickness (MFT) h,,, film thickness a t 2 2 where =0 h,, film thickness available on the liner h’ h:craper film thickness left by scraper ring downstrokes crankshaft position from TDCflrlng(rads) e Is
R
2
2
cording to Reynolds’ cavitation conditions:
shown i n figure 3. The contacting faces of the oil control ring rails are assumed to be sections of cylinders, so that each rail was considered as a parabolic ring. The following assumptions were made in the model: The ring face and liner are smooth surfaces separated by a thin film of oil, surface roughness effects being neglected.
0
The oil film cavitates at a location
x2
ac
= pin at
p
= 0
at
x=x2
(1) (2)
dP = 0 dx
at
X = X ~
(3)
The ring is axisymmetric with the cylinder bore, the oil film is so thin in the z direction, compared to its x length that the cross film flow is neglected, reducing the problem to 2 dimensions.
The lubricant is incompressible and Newtonian with viscosity 11, constant throughout the film a t any crank position. The ring axes set has its origin a t the point of minimum film thickness, as shown on figure 2. The ring is considered to be fixed while the liner moves relative to it with velocity U , +ve for downstrokes.
x = Xl
p
0
Furthermore, circumferential flow (y) and ring end effects are neglected so the ring is considered to have unit length in the y direction and the 1 dimensional form of Reynolds’ equation can be used:
( h 3 2 ) = 6qlJ
dh
ax

dh
12~RdB
(4)
357
ring coordinala
mtem
Figure 3. The model of the scraper ring
Figure 2. The model of the top ring
For a parabolic ring face profile the oil film thickness h has a shape function of,
h
=z
hmin
X' +2R
(5)
and for a scraper ring h = hmjn
+ x x tan (angle of scraper)
dp A _ _(6)
Integration of equation 4 gives an expression for the hydrodynamic pressure gradient.
Bxdh d p  A+ +d x  h2 h3 dB
C h3
(7)
Integration again yields an expression for the hydrodynamic pressure generated in the oil film. p = AIi
+ B dh z Z z + CI3 + D
(8)
For the solution of the film outlet region, between x 3 and 2 4 , it is assumed that at some point x 3 , downstream of the cavitation region, the flow reforms and the pressure rises t o match that of the gas at the ring's exit position 2 4 . The oil film thickness a t the point x 2 , where = 0, is h,. So substituting into equation 7
2
d for 5 = 0 , x = 2 2 and h = h, and assuming that the squeeze effects are negligible, the integration constant C for the outlet region can be determined. The pressure gradient in the outlet region is then:
dx
h2
Ah,,, h3
(9)
Integrating again and using the following boundary conditions gives the pressure. p = pout
at
x =24
(10)
x 3 is found by solving for p = 0 in the outlet region. By integrating the hydrodynamic pressure distribution along the ring from x1 to x4 the load carrying capacity of the oil film can be found. Assuming the load (spring tension and gas pressure) on the ring is balanced by the load carrying capacity, equating the two gives a value for $, the squeeze term, the last remaining unknown. The minimum film thickness hmin and squeeze term $ are then solved cyclically using a forward stepping iteration from an initial estimate of hmjn.
358 2.1. Scraper Ring Many modern ring packs incorporate a tapered scraper ring and a new analysis based upon plane inclined slider bearing theory has therefore been developed to take this trend into account. The procedure for solving h,,i, for the scraper ring is easier because there is no cavitation and no reformation region, the boundary conditions on pressure being:
= pin at x = x1 p = pout at x = 23 The analysis for the scraper ring predicts zero oil film thickness on the downstrokes, but this is clearly not realistic. Accordingly, the scraper ring model allows a fixed nominal amount of oil to leak past the scraper on its downstrokes. Investigation of this parameter suggested that a value equal to the surface roughness (R,) of the liner was suitable. This was confirmed by analysing the scraper as a highly offset parabolic ring, shown in figure 4. Comparison of the two solutions for oil film thickness is shown in figure 5. p

0.3
Analysis of Scraper Ring at 2000 rpm, 100
Parabolic analysis Scraper analysis ...
$4
.I: 0.25
0
0
90
180 270 crank angle ldegl
i8
04
02
0
02
04
06
08
1
12
14
rlng x cmmllnala (mm)
Figure 4. Talysurf trace of the unworn scraper ring, compared with a parabolic profile of radius SOmm.
2.2. Lubricant starvation The analysis assumes that there is sufficient oil on the cylinder wall, below the oil control ring,
I
360
Figure 5. Film thickness results for the unworn scraper ring.
to fully flood the lower rail on the downstrokes. The remainder of the pack has its oil supply reduced due to the interaction of the rings. A ring is said to be starved if its inlet is not completely filled with lubricant, as depicted in figures 2 & 3. The continuity of flow condition determines the thickness of oil left on the liner after a ring has is the oil which passed. This thickness, h* = will be available to lubricate the following ring in the pack. For a ring which has its inlet starved of lubricant, the inlet position x1 is initially unknown. Position 21 is also found from considering flow continuity. The flow rate at the inlet of the conjunction, due to Couette and Poiseuille flow, will equal the flow rate upstream on the liner. The degree to which each ring starves the others of lubricant is initially unknown, so the analysis starts by assuming the rings to be fully flooded. The assumption of fully flooded lubrication is relaxed as the cyclic film thickness is solved for each ring in turn. The procedure starts with the oil control ring lower rail and once this has been analysed for a complete cycle, the oil available for the ring above will be known. This procedure is repeated, working up to the top ring and then back down through the pack again, to yield the starved cyclic film thicknesses for each ring. The calculation of oil flow within the pack is complicated by the fact that the rings occupy different positions on the liner a t any one crank angle, for this reason oil flow w a s monitored in
y,
I
C
359 Ii tier coordinates. The model assumes no bore distortion and perfect ring conformability. In practice, bore ovality and distortion will increase the oil flow past the piston rings [9].For this reason the oil flow past the oil control ring rails was increased by a factor of 2.0 for all of the analyses. 3. Data for ring pack modelling
[email protected] he1 JM= 2 0 rnm
Figure 7. Top ring profile.
splr u h r c u t 50
04mm
40
30 20 10
;
(6) 20 hours running
10
41
(
0 8
0 6
0 2 0 02 x mordinate(mm)
04
04
06
08
1
06
08
1
06
08
1
(C) 200 hours wnnlng
i
10
0 1
0 8
0 6
0 4
0 2 0 02 x coordmte (mm)
04
(D) 600 hours runnlng
O.1
0 8
0 6
0 4
0 02 x modlnale (mm) 0 2
04
Figure 6. Change in the scraper ring profile due t.o wear.
The lubrication model was evaluated by comparing its predictions of friction with measure
ments on a motored engine test rig. The rig was specially designed and built with a floating liner assembly to measure piston and ring pack friction. The rig will be described in detail elsewhere, but its basic parameters are given in table 2. Details of the ring pack are given in table 3. Although the special analysis was developed for the scraper ring, its sharp edge wore rapidly during the tests as shown in figures 6(a),(b) and so it was modelled as a parabolic ring of very large offset. The marked change in the Talysurf profiles of the scraper ring can be seen in figure 6. The traces (a) and (b) come from the new/motored test rig while the 200 and 600 hour rings had been run in other firing engine tests. When modelling real ring profiles it is useful to include limits on the extent of the contacting face to account for features such as edge radii and the scraper ring undercut. Figure 7 shows the Talysurf trace of the top compression ring as well as the parabola used to model it and the limits on the contacting face. 3.1. Oil Film Thickness Results An example of the oil film thicknesses predicted by the starved analysis of the test rig ring pack is shown in figure 8. The results are for an analysis with the cylinder subjected to compression pressure to simulate firing conditions. The inter ring pressures were calculated using an orifice and vol
360 Table 3 Data for the test rig piston rings. Ring R e dim,, top scraper rail A rail B
(mm)
(mm)
(mm)
15.0 35.0 11.0 3.5
0.27 0.0 0.1 0.03
0.87 0.985 0.23 0.15
Table 2 Parameters of the motored Bore Stroke Rod Length Piston Height Piston Radial Clearance Speed Temperature Peak Cylinder Pressure Lubricating Oil Boundary Friction Coeff. Liner Roughness
Position from
dim,,
(mm)
b (mm)
Tension
(Nlmm2)
TDC ( m m )
0.43 0.493 0.07 0.20
1.575 1.969 0.66 0.66
0.254 0.169 1.037 1.037
6.32 12.65 17.33 20.67
test rig 90.0 mm 71.0 mm 131.0 mm 65.0 mm 0.010 mm nominal 600  3000 rpm 25°C  100°C 17bar TOSAE 1OW30 pb = 0.12 R, = 0.8pm
ume type model [lo], driven by measured cylinder pressure. 4. Friction Models
4.1. Piston Ring Friction The operating conditions for a piston ring vary greatly in load, entraining velocity and lubricant viscosity, consequently the hydrodynamic oil film and the friction force generated varies widely from midstroke to endstroke positions. It is useful to consider a Stribeck type curve, with friction force plotted against the film thickness parameter A; A =  hmin U
Such a curve for a single piston ring under fully flooded conditions is shown in figure 9 and was used as the basis for the simple friction model presented in this work. With high values of X the ring is subject to full hydrodynamic lubrication and the resulting viscous friction is calculated by considering the shearing of the oil film. With the assumption
0.5 I 0.4
I top ring scraper ring .OCR r a i l A . . ' ' OCR r a i l B  
.
0
90
180 270 crank angle (deg)
360
Figure 8. Predicted ring pack oil film thicknesses at lOOOrpm and 100°C. that the fluid is Newtonian, the shear stress; dU
r=qdz The velocity profile in the fluid is given by
so the viscous stress on the liner is given by; Tzo
q U hdp = 7)du =  dz h 2dx
(16)
Integrating along the oil film from inlet to exit gives the friction force per unit length in the y direction.
f=
1;'
rzo. dx
(17)
2
In the cavitation region the term is zero. The region is thought to contain a 2 phase flow of oil and gas, so the oil film being sheared is not complete. By considering continuity of flow, the
36 1 The variation of p with
Fully Flooded Top Ring
0.12
X given by this scheme
2 can be seen in figure 9. Analysis of a single fully
0
1
2
3 4 lambda
5
6
Figure 9. Stribeck curve for piston ring with = 0.08.
0.7pLm and pb
I 7
(T
flooded piston ring shows that a generous oil film is generated, however, when all the ring pack is present it is found that oil starvation limits the maximum values of X to about 3.0 or 4.0 and so the ring pack operates mainly in the mixed lubrication regime. Initially a value of boundary coefficient of friction pb = 0.08 was assumed. Friction measurements on the test rig suggested a more suitable value to be pb = 0.12, this value was adopted in all the friction analyses.
=
*.
viscous shear calculated for the region 2 2 , 23 is reduced by a factor The calculation of hydrodynamic friction is Insed on the lubricant behaving as a Newtonian fliiid. However, a t high shear rates lubricating oils ( 1 0 not exhibit Newtonian behaviour. In practice i t is found that oils reach a limiting traction coefficient of about 0.07 at shear rates above 106sl, t.he friction model applies this value as an upper limit for viscous friction. At lower values of X the film thickness is insufficient for fluid film lubrication and the ring is subjected to mixed lubrication. Work on mixed friction often assumes that the heights of the asperities on the rough surfaces have a Gaussian distribution. The statistics of such a distribution dictate that the asperities will start to contact when X M 3.0 and are completely meshed between X = 1.0 and X = 0. From this, the friction model makes the following assumptions:
5 1.0, boundary friction
with p = pb
X > 3.0 the friction is calculated by hydrodynamic theory. 1.0 < X < 3.0, the friction due to asperity contact changes linearly from pb to 0. This value is used for the friction if it is greater than that calculated by the hydrodynamic theory. Otherwise the friction is as predicted by the hydrodynamic theory.
Figure 10. Talysurf profile of the cylinder liner and modified profile, yx2000, 2 x 20.
In practice cylinder liners are finished by a honing process and their surface distributions are far from Gaussian. On closer inspection it can be seen that the surface of a liner is dominated by the deep honing scores, see figure 10(A). Whilst the deep grooves are very important to retain a reservoir of lubricating oil on the cylinder wall they play no active part in the mechanics of asperity contact which causes the mixed friction. The presence of the deep grooves also has a great influence on R, measurements of roughness. In this friction analysis it was assumed that the deep honing grooves had no effect on the mixed friction and could be neglected from the surface
362 ( A ) original Profile
1
6
4
2
0
2
4
6
(B) Modified Profile
7
6
4
2
4
b
Figure 11. Profile height distributions about the mean line.
profile. The scheme adopted was to neglect any groove which fell below the mean line of the original profile. This can be seen in figure 10(B). The mean line and roughness of the modified profile were then calculated. As can be seen from figure 11(B), the modified surface profile had a distribution nearer to Gaussian and so the friction model could be used with the modified value of R,. To take account of surface waviness the effective R, was further reduced to 0.15prn for all the friction analyses. 4.2. Piston Skirt Friction A simple piston skirt friction model has been developed. Its purpose was to provide an approximation of piston skirt friction to add t o the ring friction predictions, so that they could be directly compared with the test measurements. It is generally accepted that pistons are lubricated hydrodynamically for most of their operating cycle [ll].So the skirt friction was calculated
Figure 12. Exaggerated oil film thickness between piston and bore. by considering the shearing of the oil in the gap between piston and bore. The geometry of the clearance space between the piston and cylinder is shown in figure 12. The piston is both oval in diameter and barrelled in profile. Assuming that the piston occupies an eccentric position in the clearance of the bore, an expression can be derived for the film thickness h at any point on its surface. The friction force is then; qudA f=h Where: d A = element of surface area, q = oil viscosity and u = speed. If the piston has a nominal radial clearance in the bore of c, then its eccentricity 6 = If the profile shape of the skirt is represented by radial difference from nominal piston q!~
E.
6=
C
363 Then the film thickness at any point is: h = c( 1
+ c cos(8) + psin(8) + 6)
(19)
Where for ovality o , /? = f . The friction for the whole piston was then determined from;
u
0
u
c
100 50 25
o 25
:.
50
I4
75
100
ti. Results
I
75
2 U
(A) Experimental Data I
'
0
I 60
120 180 240 300 crank angle (degrees)
i t
.I:
360
25
.I: 50 75
100
180 135 90
45
0
45
crank angle (deg)
90
135
1
0
I 45
180
Figure 13. Test rig cylinder pressure at 1000rpm.
Friction analyses were made for most of the conditions tested on the rig, so that direct comparison of the theoretical and experimental results could be made. Full details of these wide ranging comparisons will be reported elsewhere. 'I'he rig could be operated with or without compression in order to realistically load the rings. The variation in cylinder pressure being shown in figure 13. The friction results for analysis with and without compression pressure are shown in figures 14 & 15 respectively. The breakdown in the components of total pack friction at 75°C and 1200rpm is shown in figure 16. It can be seen how the friction is dominated by the oil control ring, which has the highest contact pressure, and which operates mainly in the mixed friction region. A similar comparison between the theoretical and measured results for a temperature of 25°C is shown in figure 17.
90
135
180
225
270
crank angle (degrees)
315
360
Figure 14. Ring pack friction at lOOOrpm and lOO"C, no cylinder pressure.
In this case the increased oil viscosity has enabled the oil control ring to generate a thicker fluid film, resulting in lower friction. It can be seen from figures 16 & 17 that the piston model over estimates friction at the lower temperaturelhigher viscosity condition and underestimates it at the higher temperature/lower viscosity condition. The predictions for the rings match the measurements quite well under both conditions. 6. Conclusions
Piston ring friction was analysed by a simplified model using calculated hydrodynamic friction and a linear change from mixed to boundary friction. The predictions from the analysis correlate well with measurements made on a test rig.
364
:::
Anal sis Results
( A ) Experimental Data
5
100
Ex erimental Data
60
50
0 50
100 150
___
2nn
200
0
45
90
135 180 225 270 crank angle (deg)
315
360
Figure 16. Components of pack friction at 75'C.
(B) Friction Analysis Results I
piston only + oil ring .......
I
erimental Data
.4
0
150
200
I
I
I
1
'
0
45
90
135 180 225 270 crank angle (deg)
315
I
360

scraper ring 60 50
.
40
.............. ._
30 20
10
Figure 15. Ring pack friction at lOOOrpm and lOO"C, with compression pressure.
Accurate results from the friction analysis depend on modification of the cylinder liner surface, so that the heights of the surface asperities have a Gaussian distribution. The piston skirt model overpredicts friction at higher viscosities, although the analysis and experimental results matched well for the 75" and 100°C cases. 7. Acknowledgments
This work was carried out under a Science and Engineering Research Council CASE award in collaboration with Jaguar Cars Ltd. The authors thank both bodies for providing funding for the project. We are particularly grateful to Steve Richardson and Martin Roskilly who coordinated the project at Jaguar.
oil control ring

.
..

top ring ___ ... ........
scraper ring
.....................................
.
oil control ring piston
..................................... piston
0
Figure 17. Components of pack friction at 25'C.
REFERENCES 1. C. D Radcliffe. A n Experimental and Analytical Study of a Piston Ring Pack. PhD thesis, The University of Leeds, Department of Mechanical Engineering, 1993. 2. Y Tateishi. Tribological issues in reducing piston ring friction losses. Tribology Znternational, 27(1):1723, 1994. 3. N Patir, & H. S Cheng. Application of average flow model to lubrication between rough sliding surfaces. Journal of Lubrication Technology, Trans. A S M E , 101:220230, Apr 1979. 4. J Greenwood, & J Tripp. The contact of two nominally flat rough surfaces. The Znstitution of Mechanical Engineers, Proceedings., 185:625633,1971. 5. S. M Rohde. A mixed friction model for dynamically loaded contacts with application to
365
6.
7.
8.
9.
10
11.
piston ring lubrication. In Friction and Traction, pages 262278. Proceedings of the 71h LeedsLyon Symposium on Tribology, Westbury House, 1981. Q.B Zhou, T.Z Zhu,, & R.S Wang. Full lubrication model for rough surface piston rings. fiibology International, 21(4):211214, 1988. Y Hu, H. S Cheng, T Arai, Y Kobayashi,, & S Aoyama. Numerical simulation of piston ring in mixed lubrication  a nonaxisymmetrical analysis. Journal of Tribology, Trans. A S M E , 116:470478,1994. D Dowson, P Economou, B Ruddy, P Strachan,, & A Baker. Piston ring lubrication, part 2  theoretical analysis of a single ring and a complete ring pack. In S Rhode, D Wilcock,, & H Cheng, editors, Energy Conservation Through Fluid Film Technology, New York, 1979.ASME. S. P Edwards. The Contribution of Piston Ring Packs and Cylinder Bore Distortion to Engine Friction. PhD thesis, The University of Leeds, Department of Mechanical Engineering, 1992. B Ruddy, D Dowson,, & P Economou. The prediction of gas pressures within the ring packs of large bore diesel engines. Proceedings of the IMechE, Journal of Mechanical Engineering Science, 23(6):295304, 1981. G Knoll, & H Peeken. Hydrodynamic lubrication of piston skirts. Journal of Lubrication Technology, Trans. A S M E , 104:504509,Oct 1982.