Modeling of Liner Finish Effects on Oil Control Ring Lubrication in
Internal Combustion Engines Based on Deterministic Method
by
Haijie Chen
B.Sc., Automotive Engineering
Tsinghua University, 2006
Submitted to the Department of Mechanical Engineering in Partial Fulfillment of the
Requirements of the Degree of
Master of Science in Mechanical Engineering
at the
Massachusetts Institute of Technology
June 2008
© 2008 Massachusetts Institute of Technology. All rights reserved.
Signature of Author:
I
,
Department of Mechanical Engineering
May 15, 2008
Certified by:
Dr. Tian Tian
Principle Research Scientist, Department of Mechanical Engineering
A
Thesis Supervisor
Accepted by:
MASsAcHtrss wsTrmurE
OF TEOHNOLOGY
JUL 2 9 2008
LIBRARIES
J
:rofessor Lallit Anand
Chairman, Department Committee on Graduate Studies
Department of Mechanical Engineering
Modeling of Liner Finish Effects on Oil Control Ring Lubrication in
Internal Combustion Engines Based on Deterministic Method
by
Haijie Chen
Submitted to the Department of Mechanical Engineering on May 20, 2008 in Partial Fulfillment
of the Requirements for the Degree of Master of Science in Mechanical Engineering
Abstract
Twin-land oil control ring is widely used in the automotive diesel engines, and is gaining more
and more applications in the modern designs of gasoline engines. Its interaction with the
cylinder liner surface accounts for around 10% of the total frictional losses within an internal
combustion engine, and is the most important factor that affects the lubricant oil consumption.
A twin-land oil control ring model is developed based on the deterministic hydrodynamic
method by Li et al. [31] and the Greenwood Tripp asperity contact model [39]. Unlike the
traditional methods of piston ring pack modeling determined by the ring face macro profile
which hardly exists on the twin-land oil control ring, the model considers the liner finish micro
geometry, and uses a correlation method to predict the behavior of the ring liner interaction in
both frictions losses and oil control. The model is used to study the effect of some key design
parameters of the twin-land oil control ring, including the unit load pressure, the ring tension and
the ring axial land width. The results show a large potential of the model in the optimization of
the twin-land oil control ring design.
Although many non-conventional cylinder liner finishes are now being developed to reduce
friction and oil consumption, the effects of surface finish on ring-pack performance is not well
understood. To solver this mystery, the twin-land oil control ring model and Li's deterministic
hydrodynamic model are used to study the effect of the liner finish micro geometry. Several key
parameters of liner finish texture are examined, and the analytical results show important effects
of some of these key parameters on the twin-land oil control ring friction losses as well as the
lubricant oil control. These key parameters include the surface wavelength of the plateau part,
the frequency of deep valleys and the honing cross-hatch angle. This thesis work has opened a
window on the deterministic study of the functionality of cylinder liner surface texture.
Thesis Supervisor:
Dr. Tian Tian (Principle Research Scientist, Department of Mechanical Engineering)
Acknowledgements
There are many people who I would like to thank for their contributions to this research, and to
my past two years study at MIT. These contributions have given me many opportunities for
developments on both a personal and professional level.
First and Foremost, I would like to thank my supervisor, Dr. Tian Tian for his support and
guidance throughout the course of my work. I have learned a great deal through my exposure to
his depth of knowledge, insight, and logical approach to problem solving. I would also like to
thank my peer worker Yong Li, who initiated this project. I absorbed numerous knowledge and
ideas through the intense and inspiring discussions with him. I couldn't have finished the work
without his continuous help in these entire two years.
I would also like to take this opportunity to thank our research sponsors, Mahle GmBh, AB
Volvo, Peugeot PSA, Renault SA, and Dana Corporation for their financial support, and more
specifically, their representatives, Rolf-Gerhard Fiedler, Eduardo Tomanik, Bengt Olson,
Fredrick Stromstedt, Sebastien messe, Gabriel Cavallaro, James Labiak, Max Maschewske,
Randy Lunsford, Remi Rabute, Bernard laeer, Paulo Urzua Torres and Juergen Willand for their
continued encouragement over the years, and for sharing their extensive experience with me.
Our regular meetings provided not only a motivation for completing work, but also an invaluable
opportunity to share knowledge and obtain constructive feedback. Without their help and
guidance this research would not be possible.
I would also like to thank the members of the Sloan Automotive Laboratory for their support and
friendship. In particular I would like to thank students of the Lubrication Consortium and my
office mates, Steve Preszmitzke, Eric Senzer, Fiona McClure, Ke Jia, Dongfang Bai, Jeff Wong,
and Tairin Hahn for their help to make the stressful time more fun.
Finally, I would like to thank my friends and family for support throughout my time here.
Table of Contents
A bstract ........................................................................................................
...........
3
A cknow ledgem ents......................................................................................................
5
Table of Contents ........................................................................................................
7
L ist of Figures .............................................................................................................
9
L ist of Tables .................................................................................................................... 13
15
1 Introduction ...............................................................................................................
1.1
Project M otivations....................................... .............................................. 15
1.1.1
Oil Control Ring Friction in an Internal Combustion Engine................ 15
1.1.2
Control of Oil Consumption [3].................................... ............................. 16
1.2
Twin-land Oil Control Ring...................................................................
16
Surface Finish on Modem Cylinder Liners .......................................... 17
1.3
1.4
Modeling the Twin-land Oil Control Ring and Liner Interaction.................. 19
1.5
Deterministic Hydrodynamic Modeling .....................................
....... 20
1.6
Scope of Thesis Work.............................................................................
20
2 A Deterministic Hydrodynamic Model .........................................
.......... 23
2.1
The Basic Assumptions...........................................................................
23
2.1.1
Assumptions for Lubrication Approximations ..................................... 23
2.1.2
A Cavitation Theorem...................................................................
24
2.2
The Governing Equations .....................................................................
25
2.3
The Boundary Conditions ...................................................
26
2.4
Hydrodynamic Pressure Recovery..........................................
27
3. The Twin-land Oil Control Ring Model .........................................
.......... 31
3.1
Model Simplifications.............................................. 31
3.1.1
Ring Face Profile ......................................................................................
31
3.1.2
Axial and Circumferential Conformability ......................................
33
3.2
H ydrodynam ic Part ............................................................ ......................... 37
3.2.1
Oil Viscosity and Ring Sliding Speed ........................................
..... 37
3.2.2
Squeeze Oil Effect ...................................................... 40
3.3
Asperity Contact Part.............................................................................
40
3.4
Load B alance Part .............................................................. ......................... 41
M odel Results .................................................................. ........................... 42
3.5
3.6
Assumptions Validation and Sensitivity Analysis ..................................... 48
Oil Viscosity and Ring Sliding Speed .....................................
3.6.1
... 48
3.6.2
Squeeze Oil Effect ...................................................... 50
3.6.3
Ring Tilt Effect ................................................. .................................. 53
3.6.4
Influence of Liner Surface Measurement Resolution ............................ 56
3.7
Validation of Model Results ...................................................................
70
4. Effects of Twin-land Oil Control Ring Design.................................
........ 79
4.1
Effect of Ring Land Width ..................................... ......
............... 79
4.2
Effect of Ring Tension..................................................
83
4.3
Competing Effects of Different Parameters..............................
........88
4.4
Effect of Liner Finish Evolution .....................................
........... 92
5. Effects of Liner Surface Micro Geometry ........................................
........ 93
5.1
Importance of Liner Finish ...............................................................................
93
Effects of Detailed Liner Surface Micro Geometry ..................................... 97
5.2
.......... 98
Effect of Plateau Wavelength ........................................
5.2.1
101
5.2.2
Effect of Valley Distance ..............................
102
5.2.3
Effect of Valley Depth................................
107
Effect of Cross-hatch Angle................................
5.2.4
6. Conclusions and Future Research ...............................
113
113
Twin-land Oil Control Ring Model Development..........................
6.1
114
6.2
Effects of Twin-land Oil Control Ring Design...............................
114
.....................................
Surface
Micro
Geometry
6.3
Effects of Liner
6.4
Discussion of Probable Future Work.................................
115
References ...................................................
117
List of Figures
Chapter 1
Fig.1. 1 Breakdown of Total Diesel Engine Energy, Mechanical Friction and Ring Pack
Friction [1] ................................................................................................................ 15
Fig. 1. 2 Twin-land Oil Control Ring Face Profile ......................................
....... 17
Fig. 1. 3 Liner Surface M easurement....................................................
.................... 18
Fig. 1. 4 Average Film Thickness h.......................................................................
19
Chapter 2
Fig.2. 1 Lubrication Approximations....................................................................
24
Fig.2. 2 Boundary Conditions .................................................
27
Fig.2. 3 Local Steady State Hydrodynamic Solver...............................
.........29
Fig.2. 4 Comparison of the Hydrodynamic Pressure Distributions of the Different
Pressure Recovery M ethods.........................................................................
30
Chapter 3
Fig.3. 1 Numerical Predicted Worn TLOCR Face Profile...............................
..... 32
Fig.3. 2 TLOCR Tilt and Force Configuration .......................................
......... 33
Fig.3. 3 TLOCR Tilt Face Profile .....................................................
35
Fig.3. 4 TLOCR Conformability to the Bore Distortion ......................................
36
Fig.3. 5 Circumferential Distribution of Normal Force .....................................
. 36
Fig.3. 6 Inter-asperity Hydrodynamic Pressure Generation ................................... . 37
Fig.3. 7 Correlation of average hydrodynamic pressure and film thickness ratio at a
constant viscosity [to and sliding speed Vo ................. ....... .. .... . . ... . ..
. . . . . ... 39
Fig.3. 8 Liner surface measurement......................................................
..................... 39
Fig.3. 9 N ormal Load Balance ........................................................... ........................ 42
Fig.3. 10 Cycle Variation of Oil Film Thickness Ratio .......................................
43
Fig.3. 11 Cycle Variations of TLOCR Frictions...............................
............ 43
Fig.3. 12 Influence of Engine Speeds on Oil Control Ring FMEP............................... 44
Fig.3. 13 Influence of Engine Speeds on Oil Passage Rate ....................................
. 45
Fig.3. 14 Comparison of Two Liner Surfaces............................
....
............. 46
Fig.3. 15 Comparison of the Hydrodynamic and Asperity Contact Pressure Generation on
Two Different Liner Surfaces ....................................................
47
Fig.3. 16 Comparison of the Oil Control Ring FMEP's on Two Different Liner Surfaces
................................................................................................................................... 47
Fig.3. 17 Comparison of the Oil Passage Rates on Two Different Liner Surfaces .......... 48
Fig.3. 18 Linearity of Hydrodynamic Pressure to piV at Film Thickness Ratio A=2 ...... 49
Fig.3. 19 Linearity of Hydrodynamic Pressure to gV at Film Thickness Ratio A•=6 ...... 49
Fig.3. 20 Linearity of Hydrodynamic Friction to piV at Film Thickness Ratio A•=2....... 50
Fig.3. 21 Linearity of hydrodynamic friction to gV at film thickness ratio A=6 ............ 50
Fig.3. 22 Squeeze Oil Effect Testing Results (Low Sliding Speeds) ............................ 51
Fig.3. 23 Squeeze Oil Effect Testing Results (Medium Sliding Speeds) ...................... 52
Fig.3. 24 Squeeze Oil Effect Testing Results (High Sliding Speeds)........................... 52
54
Fig.3. 25 Tilt Angle Configurations ..................................................................................
Fig.3. 26 Tilt Effect (X=3).......................................... ................................................ 55
Fig.3. 27 Tilt Effect ( =4) ................................................................................................. 56
57
Fig.3. 28 Liner Surface Measurement..................................................................
............. 58
Fig.3. 29 Division of Four Sub-surfaces .......................
58
Fig.3. 30 Local Clearances under Different Mesh Sizes .......................................
Fig.3. 31 Local Hydro Pressures under Different Mesh Sizes ................................... 59
59
Fig.3. 32 Local Oil Densities under Different Mesh Sizes ....................................
Fig.3. 33 Hydrodynamic Pressures under Different Mesh Sizes ................................... 61
Fig.3. 34 Comparison of Mean Sub-surface Pressures and the Pressure of the Entire
Surface ...................................................................................................................... 63
64
Fig.3. 35 Test Surface 1-1 - 2-2 ...............................................................................
. 65
Fig.3. 36 Fourier Analysis of The Two Original Surfaces ....................................
Fig.3. 37 Comparison of Clearance and Hydrodynamic Pressure for Surfaces 1-1 and 1-2
................................................................................
. . . 66
66..........
67
1-1
and
1-2........................
for
Surface
Fig.3. 38 Hydrodynamic Pressure Generation
Fig.3. 39 Comparison of Clearance and Hydrodynamic Pressure for Surfaces 2-1 and 2-2
....
...
68
68.........................
Fig.3. 40 Hydrodynamic Pressure Generation for Surface 2-1 and 2-2........................ 69
69
Fig.3. 41 Distribution of Aliasing Errors for Surface 2-2....................................
Fig.3. 42 Effects of Liner Finish Roughness on Piston Ring Pack Friction .................. 71
Fig.3. 43 Surface Profiles of a Smooth and Rough Liner Surfaces .............................. 72
Fig.3. 44 Numerical Results on TLOCR Frictions for the Smooth Surface at 1500 rpm. 73
Fig.3. 45 Numerical Results on TLOCR Frictions for the Rough Surface at 1500 rpm.. 74
Fig.3. 46 Comparison of Average Clearance for the Two Surfaces at 1500 rpm.......... 74
Fig.3. 47 Numerical Results on TLOCR Frictions for the Smooth Surface at 3000 rpm. 75
Fig.3. 48 Numerical Results on TLOCR Frictions for the Rough Surface at 3000 rpm .. 76
Fig.3. 49 Comparison of Average Clearance for the Two Surfaces at 3000 rpm.......... 76
Chapter 4
...... 79
Fig.4. 1 Liner Finish Measurement.................................................................
Fig.4. 2 Hydrodynamic Pressure Distribution under Different Ring Widths ................ 80
Fig.4. 3 Ring Width Effect on Hydrodynamic Pressure Distribution........................ 81
Fig.4. 4 Ring Width Effect on Average Hydrodynamic Pressure Generation.............. 81
Fig.4. 5 Influence of ring land width on oil control ring FMEPs ................................. 82
............. 84
...
Fig.4. 6 Stribeck Curves of Friction Coefficient..........................
85
Fig.4. 7 Stribeck Curves of FMEP.......................................................................
Fig.4. 8 Stribeck Curves of Velocity Weighted Asperity Contact Pressure .................. 86
.......... 87
Fig.4. 9 Stribeck Curves of Oil Passage Rate .......................................
Fig.4. 10 The Competing Effect of Ring Land Width and Ring Tension 1.................. 89
Fig.4. 11 The Competing Effect of Ring Land Width and Ring Tension 2.................. 89
Fig.4. 12 Competing Effect of Ring Land Width and Unit Load Pressure 1................ 90
Fig.4. 13 Competing Effect of Ring Land Width and Unit Load Pressure 2................ 91
Chapter 5
Fig.5. 1 Surface Profile of the Medium Rough Liner ......................................
.....
94
Fig.5. 2 Comparison of Average Clearance for the Three Surfaces at 1500 rpm.......... 95
Fig.5. 3 Comparison of Total Friction for the Three Surfaces at 1500 rpm .................. 95
Fig.5. 4 Comparison of Average Clearance for the Three Surfaces at 3000 rpm.......... 96
Fig.5. 5 Comparison of Total Friction for the Three Surfaces at 3000 rpm .................. 96
Fig.5. 6 A rtificial Liner Surface........................................................................................
98
Fig.5. 7 Artificial Surfaces of Different Plateau Wavelengths ..................................... 99
Fig.5. 8 Hydro Pressure Generation for Artificial Surfaces of Different Plateau
W avelengths............................................................................................................ 100
Fig.5. 9 Artificial Surfaces of Different Groove Densities...........................
101
Fig.5. 10 Influences of groove densities on the hydrodynamic pressure generation...... 102
Fig.5. 11 Hist Graph of the Surface Height .....................................
103
Fig.5. 12 Definition of Valley Depth Factor A .....................................
104
Fig.5. 13 Clearance Profile of Different Valley Depth Factor A ................................ 105
Fig.5. 14 Oil Density Profile of Different Valley Depth Factor A ............................ 105
Fig.5. 15 Hydrodynamic Pressure Profile of Different Valley Depth Factor A ......... 106
Fig.5. 16 Average Hydrodynamic Pressure Trace of Different Valley Depth Factor A 106
Fig.5. 17 Change of Cross-Hatch Angle...................................................................... 108
Fig.5. 18 Shrink the Curve by Changing the Grid Size of the Measured Sample .......... 108
Fig.5. 19 Grid sizes related to different honing angles .....................................
109
Fig.5. 20 Influences of the wavelengths in X and Y directions on the hydrodynamic
pressure generation ................................................................................................. 110
Fig.5. 21 Influence of different honing angles on the hydrodynamic pressure generation
................................................................
111
List of Tables
Chapter 3
Table 3. 1 Comparison of rpq of the two liner surfaces............................. ..... . 46
Table 3. 2 Errors of Hydrodynamic Pressure Induced by Squeeze Oil Effect .......
. 53
Table 3. 3 Standard Deviations of the Errors of Hydrodynamic Pressures on Different
M esh Sizes ................................................................................................................ 6 1
Table 3. 4 The Comparison of the Roughness Statistical Parameters [21] ................... 73
Chapter 5
Table 5. 1 The Comparison of the Roughness Statistical Parameters [21]................... 94
1
Introduction
1.1
Project Motivations
In the modem world, internal combustion (IC) engines are widely used in the area of
transportation. The use of IC engines has been a major fossil fuel consumer, as well as
an important air pollution contributor. As a result, two of the most important topics of
the IC engine research are improving the efficiency of the energy use and the emission
control.
1.1.1 Oil Control Ring Friction in an Internal Combustion Engine
Mechanical friction loss accounts for around 10% of the total energy in the fuel for a
typical diesel engine, as illustrated in Fig. 1.1 [1]. Among the mechanical friction loss,
piston ring pack is responsible for about 20%. Meanwhile oil control ring is the major
contributor to the piston ring pack friction loss for IC engines due to the high ring tension
used. An approximately 1-1.5% of the total diesel fuel energy is lost in the friction
between the oil control ring and the engine cylinder bore surface.
Total Engergy Breakdown
Mechanical Friction Breakdown
Mechanical Friction
.......
Ring Friction Breakdown
Rinas
Ot
(41
)utput
fau~41%)
dRods
ng
5%6)
%)
44~81(
Fig.1. 1 Breakdown of Total Diesel Engine Energy, Mechanical Friction and Ring
Pack Friction [1]
Therefore there is a large potential in improving the diesel engine efficiency by reducing
the friction between the oil control ring and the engine cylinder bore surface. Reducing
oil control ring friction also reduces the thermal load on the cooling system of the engine
by reducing the amount of heat generated in the power cylinder. Furthermore, the
improved oil control ring efficiency would also help engines to meet the more and more
stringent C02 emission regulations. The challenge in finding the strategy for lowering
oil control ring friction is not to bring adverse effects in oil consumption, blow-by,
excessive wear, and failure. And this requires a deep understanding of the interaction
between the solid surfaces of the oil control ring face and cylinder bore surface with the
existence of lubricant oil in between.
1.1.2 Control of Oil Consumption [3]
Oil consumption from the piston-ring-liner system contributes significantly to total
engine oil consumption [2] [3]. Engine oil consumption is recognized to be a significant
source of automotive engine emissions in modem engines. Unburned or partially burned
oil in the exhaust gases contributes directly to hydrocarbon and particulate emissions [3]
[4] [5]. Moreover, chemical compounds in oil additives can poison exhaust gas treatment
devices and can severely reduce their conversion efficiency [3] [6] [7]. As a result,
engine oil consumption is a very important index of modem engine performance and
needs to be controlled properly.
Numerous studies have been carried out to analyze the impact of different parameters of
the piston-ring-liner system on oil consumption. It was recognized that oil consumption is
affected by the geometric details of the piston and rings [8] [9] [10] [11] [12], liner
surface finish [13] [14] [15], cylinder bore distortion [16] [17], component temperatures
[18], oil properties [19] [20], and engine operation conditions such as speed, load, and
whether the engine operates in a steady state.
1.2
Twin-land Oil Control Ring
Twin-land oil control ring (TLOCR) is most widely used in automotive diesel engines,
and is gaining more and more applications in gasoline engines. The face profile of a
typical TLOCR is shown in Fig. 1. 2. In order to seal the oil in the crank case from the
combustion chamber, a high normal force is exerted by the oil control ring spring to
conform the ring onto the cylinder bore. Consequently the TLOCR suffers very severe
work condition and it is the major friction loss source for the piston ring pack.
Land Width = 0.2nun
Fig.1. 2 Twin-land Oil Control Ring Face Profile
TLOCR is critical in controlling the oil film thickness left on the liner, which is a
parameter important for both the top two ring lubrication and engine oil consumption.
The trade-off between the good lubrication condition and the less oil consumption makes
the oil control ring design rather complicated. In order to optimize the TLOCR
performance, a thorough understanding of the interaction mechanism of the TLOCR and
cylinder bore liner finish is necessary.
1.3
Surface Finish on Modern Cylinder Liners
The liner surfaces of modem engines, manufactured with the typical three honing
processes, are usually consisted of two different parts, the plateau part with a smaller root
mean square (RMS) roughness and the valley part with a larger RMS roughness. A
typical liner finish geometric profile is shown in Fig. 1. 3. The plateau part of the surface
is formed by the final fine honing process, while the valley part comes from the early
honing processes, and appears as the sparsely distributed grooves of several microns
depth.
Fig.1. 3 Liner Surface Measurement
In general, when another nominally flat surface, in this thesis the twin-land oil control
ring land surface, slides over the liner surface with a normal load, it is in the plateau part
that all the asperity contact occurs. With the existence of oil, high oil pressure also tends
to be generated in the plateau area to drive the oil around the asperities. Due to this
special topology, the composite RMS roughness generally used to represent the
roughness of a nominally flat surface is not sufficient and should be replaced by rpq [21],
the RMS roughness of the plateau part and other statistical parameters to represent the
valley area. Thus, the definitions of some other terminologies need to be clarified before
the detailed technical discussion.
Average film thickness h: The average film thickness is defined as the mean height of
the oil control ring surface minus the mean height of plateau part of the liner surface.
(See Fig.1. 4)
Film thickness ratio (Xratio): The widely accepted definition of k ratio is modified as
ratio of the average film thickness (defined above), to the RMS roughness of the plateau
part of the liner surface, X=h/rpq.
I
I
I
ckness h
Fig.1. 4 Average Film Thickness h
1.4
Modeling the Twin-land Oil Control Ring and Liner
Interaction
A number of works were devoted to understand how liner surface feature influences the
ring pack friction as well as wear, and somehow improve its behavior through modifying
the surface texture [22] [23] [24] [25] [26] [27] [28] [29] [30]. These works either
intended to correlate the function / performance of the ring or ring pack with the
statistical parameters derived from height distribution or neglected the unsteady nature of
the oil redistribution within asperities. Furthermore, there have been no studies dedicated
to the interaction between the oil control ring and the rough liner, which arguably is the
most critical step toward understanding the effects of the liner finish on the outcome of
the piston ring pack.
Unlike the top two rings, which both have a macro shape contributing to the
hydrodynamic pressure generation between the ring face and the liner surface, the twinland oil control ring usually exhibits the flat running faces after running in. Due to the
constraint between the two lands and a high normal load, the two flat faces are practically
parallel to the liner surface. As a result, the hydrodynamic pressure generation between
the ring face and the liner, if any, is solely inter-asperity pressure, due to the interaction
of the surface micro geometries, rather than the pressure developed with the macro shape
of the ring running surfaces. The average hydrodynamic method, which is typically used
in the numerical models for the top two rings, is based on the macro geometry of the
running surfaces, thus would give zero hydrodynamic pressure for the oil control ring.
Therefore, it is not able to correctly predict the behavior of the oil control ring liner
interaction. Instead, the deterministic method based on the 3D measurement of the
surface profile should be used.
1.5
Deterministic Hydrodynamic Modeling
Deterministic method has been widely used in the numerical study of the point contact
lubrication. However, not much work using the deterministic method with proper
boundary conditions has been done for the ring lubrication yet.
In order to apply the full deterministic method in evaluating the full stroke behavior of
the ring liner interaction, two difficulties need to be addressed. First of all, 3D surface
measurements usually have a limited measurement range, which could hardly be
extended to the entire liner surface. To address this difficulty, Bolander et al. [34]
presented a model based on the numerically created surface statistically equivalent to the
real measured surface. However, the second difficulty, the trade off between the
calculation efficiency and accuracy still remained as a major challenge to the researchers.
In order to attain reasonable time efficiency, coarse meshes and large time steps have
been used in some published works.
1.6
Scope of Thesis Work
The scope of this thesis work includes a novel approach to model the lubrication and
friction between the twin-land oil control ring (TLOCR) and liner with consideration of
micro structure of the liner surface finish [32]. The approach is based on an unsteady
deterministic hydrodynamic model proposed by Li et al [31]. The model is able to
examine the full cycle performance of the TLOCR based on a 3D deterministic
measurement of the engine liner surface patch with the achievements of both high
efficiency and deterministic accuracy.
In this thesis, the model is used to show some basic analysis of the twin-land oil control
ring performance under different engine running conditions and ring design parameters
including ring land width and ring load.
Furthermore, this thesis also examines the influence of different liner finish micro
geometries on the TLOCR performance [33]. In order to do this, Li's model is used to
study some critical features of the liner finish and their effects are presented.
2
A Deterministic Hydrodynamic Model
This chapter introduces a deterministic method of hydrodynamic modeling for the
lubrication of two nominally flat surfaces with relative motion. The method is originally
proposed by Elrod [35], while Li et al. [31 ] first used it in the modeling of piston ring
dynamics.
The chapter would start with a section to introduce the basic assumptions of the method,
followed by the detailed explanation of the governing equations as well as the boundary
conditions of the method for its usage in twin-land oil control ring modeling. The last
section of this chapter would introduce a problem in pressure generation of the proposed
method, and some basic studies to address the problem.
2.1
The Basic Assumptions
2.1.1 Assumptions for Lubrication Approximations
In the area where there is enough oil to fully fill the gap between the twin-land oil control
ring (TLOCR) face and the liner surface, the oil flow is governed by the Reynolds
equation:
d (ph)
d
=(hV -
3
ph
't12p
1 v (ph)
Vp[31]
2 ox
In the equation, p refers to oil density. h refers to the local clearance. pL
refers to the
dynamic viscosity of the lubricant oil. p refers to the hydrodynamic pressure of the oil,
and V is the sliding speed of ring. Here it is further assumed that the lubricant oil is
incompressible so that the oil density p is a constant. And the assumptions of lubrication
approximations need to be satisfied.
I
V
SRing
Oil
Ah
Liner
A
Fig.2. 1 Lubrication Approximations
For the oil flow between the TLOCR face and the liner surface, the assumptions of
lubrication approximations could be interpreted as the following (Fig.2. 1):
Ah
Ah
h2
<< 1, Reh - <<1 and - <<1
yT
Ax
Ax
Here y refers to the kinematic viscosity of the lubricant oil, and T is the characteristic
time constant, which in this case could be chosen to be the time for each engine stroke.
For a typical finished liner surface, - is around 0.1. And the Reynolds number of the
Ax
Ah
situation is around 1, therefore Reh -
h2
is around 0.1. - is in the scale of 10- '. As a
'Ax
yT
result, all the conditions mentioned above are approximately satisfied, and the Reynolds
equation could be used.
2.1.2 A Cavitation Theorem
Lubricating oil can not exist at pressures below its cavitation pressure. Once the pressure
in the flow field drops to a critical value (the cavitation pressure), the oil will cavitate,
separating into liquid and vapor. According to the Jakobson-Floberg-Olsson (JFO)
theory, as described by Elrod [35], the oil domain can be divided into two distinct zones,
a full film region and a partial film region.
In the full film region the oil flow is governed by the Reynolds equation as mentioned in
the previous section. In the partial film region, the pressure is assumed to be constant and
the oil density is the dependent variable. Because of the zero pressure gradients in the
partial film region, the pressure driven flow term disappears in the cavitation zone. The
oil flow is governed by a pure hyperbolic oil transport equation:
d(ph)
dt
2.2
V a(ph) [31]
2 ax
The Governing Equations
By introducing an index variable to distinguish cavtation zone and full film zone, Elrod
presented a universal numerical scheme to solve whole field [35]. This method avoids
tracking the cavitation boundary and the result will automatically satisfy mass
conservation. Other researchers reported this method shows numerical instability around
the cavitation boundary [35] [36]. Payvar and Salant presented a way to avoid the
numerical instability by controlling the index variable [37]. Instead of switch the index
variable between zero and one, they used a small relaxation variable to control the
stability. The method needs much more iterations to converge in cases with cavitation
than cases without cavitation. In Li et al.'s model [31], the advantages of existing models
are integrated together. Improvements were made to iteration scheme to gain better
robustness and efficiency.
Instead of using compressibility to relate density and pressure, Li et al. only introduce the
index variable to switch between Reynolds equation and oil transport equation. Without
the huge lubricant compressibility coefficient, the density error of point switch from
cavitation zone to full film zone will cause less numerical instability. [31]
The index variable makes the status of local point. To get a uniform governing equation,
we need to write the pressure and density as functions of universal dependent variable
[37]. Define the universal dimensionless dependent variable 0 and index variable F as
p = Fpef +Pc
p =Pc +(1-F)lp,
Then the Reynolds equation becomes
d(1- F)h = V
VF
dt
V (1-F)h# V ah dh
2
8x
2 x dt
in which
Prefh 3
12p
1
This is a convection-diffusion equation of 0. at the full film zone. It degenerates to a
convection equation in cavitation zone. Variable 0 has different physical meanings in
different zones. In full film zone, it is a dimensionless pressure. In cavitation zone, its
absolute value is the ratio of volume occupied by vapor/gas. The variable r serves as a
diffusion coefficient. It is proportional to the cubic of film thickness and decreases
dramatically around contact point. [3 I]
Furthermore, instead of updating both F and 0 through small relaxation number, Li
proposed to only update 0 . In iteration F serves as a switch function that and takes zero
or one as value.
When index variable F is fixed, the Reynolds universal equation loses nonlinearity, and
iteration can converge quickly. [31]
2.3
The Boundary Conditions
The boundary conditions of the problem include the boundary condition for pressure and
density. (Fig.2. 2) The hydrodynamic pressures on the four edges of the two lands of the
twin-land oil control ring are determined by the third land and crank case pressures.
Normally with the well behaved top two rings, crank case and third land pressures are not
much off by the atmospheric pressure. In this thesis work, all the four boundary
pressures are set to be one bar:
PA = P2 = P3 = P 4 =1 atm
Compared to the pressure boundary conditions, the oil density boundary conditions are
more complicated. When the piston is moving downward in the axial direction, it can be
assumed that the leading edge of the bottom land is fully flooded with oil, so that p, = 1.
However the density on the rest of the three edges would depend on the amount of oil
passing the leading edge of the bottom land and the density on the two edges of the top
land also depends on the free surface oil film development between the two lands.
During the upward strokes, the oil density boundary conditions on the four edges depend
on the oil amount left by the oil control ring during the downward strokes, the
redistribution of the oil film by the top two rings, and the oil film free surface
redistribution.
In this thesis, all the four boundaries edges of the TLOCR are set to be fully flooded with
oil for simplification. By doing this, the hydrodynamic pressure of the lubricant oil could
be overestimated. It should be left for future study to decide the influence of this
simplification.
p4 ,P4
P3, P3
Pa2P 2
PlXpA
Fig.2. 2 Boundary Conditions
2.4
Hydrodynamic Pressure Recovery
The deterministic method proposed in Li et al's paper [31] has a limitation in handling
the local points with very small clearances. The result pressure could be rather high with
the deterministic method. These pressure spikes are found to be unreal, and it is
diagnosed to be mainly caused by the local lack of spatial and time resolutions. In order
to address this problem, two after treatment methods have been tried, namely the cut-off
strategy and the post steady state solver.
The cut-off strategy picks up the points of clearances below a certain threshold value (in
this thesis 0.1 micron) and substitute their pressure with the average hydrodynamic
pressure of their neighboring points where the clearances are larger than the threshold
value. This method is based on the conclusion that it is insufficient to solve the
hydrodynamic pressure of these low clearance points with the current spatial and time
resolution. Since the real hydrodynamic pressure of these low clearance points are
usually higher than their neighbors, this method tend to underestimate the value of the
hydrodynamic pressure generation on these local positions.
The post steady state solver method goes beyond that. The used spatial resolution is not
enough to solve the pressure of these low clearance points accurately, while it's neither
affordable in time nor necessary to calculate the pressure on a finer mesh basis. However,
through observation of the unsteady behavior of the hydrodynamic pressure generation,
the pressure spikes are found to be quite steady during the entire ring sliding process
except in the entrance and exit regions. A compromised method to solve the local
pressure distribution of these points of very small clearances on a finer spatial resolution
basis is tried. The solver is run after the original hydrodynamic pressure solver, and the
pressure and density information of the neighboring points are used as the boundary
conditions. Most importantly, the local pressure generation of these points is assumed to
be steady state. As a criterion to distinguish these points that need a pressure recovery,
the solver is launched at those points where the local pressure difference is too large (as
opposed to the assumption of continuous pressure generation in the flow field). An
illustration of the solver is demonstrated in Fig. 2.3.
0.2t
0
YVi
1
t
X(aJgl
)
4
1
Vc
40 bar!
I
-A
X(1
,M)
1000
.0
0.8
'10.7
~500
0
0.6
-0.5
4I
0.4
S4
0.3
0.2
Y
n
0.1
-
.4
'
-
x (rn)
S
VU
1
49
x (wn)
Fig.2. 3 Local Steady State Hydrodynamic Solver
The comparison of the hydrodynamic pressure distributions of the two proposed pressure
recovery methods is presented in Fig. 2.4. The Y axis of the plot shows the contribution
to the average hydrodynamic pressure from the points whose pressure is specified by the
X axis. It can be seen that both the two proposed methods can successfully remove the
abnormal oscillations of the pressure distribution at high pressure regimes in the result
from the original method. And the remained lower pressure regime is left smooth and
consistent.
(h
AA
L-
0
a,
0o
0.
o
II-
"r"
o
<a0
0
0p,
0
0
a,
n
0
L-
co
u_
10
10
10
10,
100
10"
10'I
10"
Hydrodynamic Pressure (pa)
Fig.2. 4 Comparison of the Hydrodynamic Pressure Distributions of the Different
Pressure Recovery Methods
3.
The Twin-land Oil Control Ring Model
The twin-land oil control ring (TLOCR) model conducts the full cycle evaluation of the
TLOCR performance based on a 3D liner surface measurement with a high calculation
efficiency and deterministic accuracy [32].
The model is consisted of three basic parts:
The hydrodynamic part evaluates the hydrodynamic pressure generation between the ring
face and the liner surface with the deterministic method introduced in the previous
chapter [31].
The asperity contact part evaluates the nominal contact pressure between the ring face
and the liner surface using the Greenwood Tripp asperity contact model [39].
The load balance part conducts the load balance and solves for the transitional film
thickness ratio A of the oil control ring in the whole engine cycle with the results from
the hydrodynamic and asperity contact parts as its input.
The chapter would start with a section of some basic simplifications of the model,
followed by the detailed introduction of each key part of the model. Section 3.5 would
cover some validation works of the basic assumptions made previous the calculation.
3.1
Model Simplifications
3.1.1 Ring Face Profile
In a well-behaved modern internal combustion engine ring pack, two sorts of
conformability need to be satisfied with the TLOCR in order to get a good oil control,
namely circumferential conformability and the axial conformability. Due to the bore
distortion, the cylinder bore surface is not a perfect cylinder. Meanwhile the ring
dynamics tend to tilt the ring axially. In order to maintain the oil control, the TLOCR is
designed so that the ring could well conform to the cylinder bore circumferentially, and
the ring dynamics should not cause either land of the ring to lose the contact with the
liner surface. It is assumed in this thesis work that both of the two conformabilities are
satisfied. The detailed discussions and validations of these two assumptions are covered
in Section 3.1.2 and 3.6.3.
With the axial and circumferential conformabilities of the TLOCR, the TLOCR land
usually exhibits a flat face profile, especially after worn. As a result, different from the
top two rings, the TLOCR does not have a macro face profile either advance or diminish
the hydrodynamic pressure generation between the ring face and the liner surface.
Fig. 3.1 shows a numerical prediction of the worn TLOCR face profile. The calculation
is based on an average twin-land oil control ring model [38]. It can be seen that due to
the distribution of piston tilt and inertia force, the worn profile of both lands exhibits a
twist angle. However, the face profile regardless of the twist is practically flat with a
drop of 3 nm on the edge.
The clearance between the TLOCR land face and the liner surface mean is typically in
the scale of 0.5 micron. Therefore, the 3 nm drop on the ring face profile could just be
neglected, and the ring face could be treated as flat for modeling.
Determine the Face Profiles of Oil Control Ring
> 100
a.
2
Averaged every 30* along the circumference
-
•1
-05
0
04
1
I5
2
Face axial axis [xO.lmmj
Lower land
3nw
I 800rpmNull load
Curvature radius: 5m
Fig.3. 1 Numerical Predicted Worn TLOCR Face Profile
3.1.2 Axial and Circumferential Conformability
Both the piston dynamic tilt and the friction force on both ring lands could cause the
TLOCR to tilt in the axial direction. Fig. 3.2 shows the force configuration on the
TLOCR. The ring load force and the inertia are exerted through the center of mass and
thus do not contribute to the torque to induce the ring tilt. When the ring tilts axially, the
normal contact and hydrodynamic force would be unequal for the two ring lands:
FL * FR
And consequently, the inequality would contribute to a torque on the ring to balance the
torque exerted by the piston tilt and land frictions.
Piston
V
4-------
Finertial
fL
fR
Fload
Fig.3. 2 TLOCR Tilt and Force Configuration
A well performed TLOCR requires both of its land to contact with the liner surface for a
good sealing. Meanwhile, both the normal contact force and the normal hydrodynamic
force are quite sensitive to the clearance change. A simple force analysis could help to
evaluate the rough scale of the TLOCR tilt.
The friction coefficient of the ring land liner interaction is typically 0.1, so that:
fL ~ O.1FL ,and fR- 0.1FR
Consider the axial force balance:
F - f + fR + Finerial
For a typical twin-land oil control ring design, the inertial force is about one tenth of the
ring load force, namely:
Fienta
~ 0.1(F, +FR)
Assume the ring axial and radial dimensions are close. With the torque balance, one can
find:
FL -FR
F+fL +fR
2(fL +fR) + Fe,,t - 0.3(FL +FR)
Therefore
FL -1.86FR
The typical average unit load pressure of the TLOCR is around 20 bars. So we can find:
FL - 26bar, and FR - 14bar
Since the normal forces on the ring lands are quite sensitive to the clearance, such a force
difference on the two lands (12 bars) could only induce a small tilt angle on the ring
(-0.1'). As shown in Fig. 3.3, on each ring land face, the height difference between two
edges caused by the ring tilt is only in the scale of 10 nm. Therefore, in this thesis work,
the tilt angles on each face are just neglected. The TLOCR land faces are assumed to
well conform to the liner and slide parallel over the cylinder bore surface. The influence
of such a simplification on the accuracy of the model results will be analyzed in the later
sections of this chapter.
Lower land
I Innrr r lant•
3nm
~10nm
Fig.3. 3 TLOCR Tilt Face Profile
Except for the axial conformability, the TLOCR also has the issue of circumferential
conformability. The shape of the cylinder bore is not a perfect cylinder due to the
manufacturing tolerance and thermal distortion during running. Fig. 3.4 shows a typical
shape (exaggerated in radial deformation) of the cylinder bore / TLOCR interface cross
section. In order to perform a good sealing function, the TLOCR has to well conform to
the liner free surface circumferentially. A high ring load is exerted by the ring spring to
force the conformation.
In reality, due to the bore distortion, the normal reaction force on the ring land would be
unequally distributed along the circumference of the cylinder cross section, even though
the ring load force are evenly enforced by the ring spring. (Fig. 3.5) In the work
presented in this thesis, this inequality of the normal force in the circumferential direction
is not considered. It would be left for the future improvement of the model.
Fig.3. 4 TLOCR Conformability to the Bore Distortion
i~i~i~
mom=
Pbwd* + P2M
Fig.3. 5 Circumferential Distribution of Normal Force
3.2
Hydrodynamic Part
The purpose of the hydrodynamic part of the model is to determine the correlation of the
hydrodynamic pressure and the parameters including the film thickness ratio A, the oil
viscosity pland the ring sliding speed V, based on a specific liner measurement and finite
ring land width L.
3.2.1 Oil Viscosity and Ring Sliding Speed
As introduced in the previous chapter, lubricating oil can not exist at pressures below its
cavitation pressure. Once the pressure in the flow field drops to a critical value (the
cavitation pressure), the oil will cavitate, separating into liquid and vapor. According to
the Jakobson-Floberg-Olsson (JFO) theory, as described by Elrod [35], the oil domain
can be divided into two distinct zones, a full film region where the Reynolds equation
governs the pressure generation, and a partial film region at a constant pressure.
Fig. 3.6 demonstrates how inter-asperity hydrodynamic pressure is generated around a
local asperity. In the converging wedge in front of the asperity, hydrodynamic pressures
are built up, while behind the asperity, the pressure drops due to the diverging geometry
of the wedge until it reaches the cavitation pressure, and partial film is then formed. The
sum of these inter-asperity pressures supplies the hydrodynamic lift for the oil control
ring.
Partial Filhn Area
0
04ý1
0
c€,O
P4
Fig.3. 6 Inter-asperity Hydrodynamic Pressure Generation
In the full film region, Reynolds equation suggests that the inter-asperity pressure
generation goes linearly with the product of oil viscosity gt and sliding speed V when the
ring is sliding over the specific liner at a certain film thickness ratio A. This relation
doesn't hold in the partial film region since the pressure is constant there. However,
since the areas of small local film thickness with full film and high pressure development
are the major contributors to the hydrodynamic lift, it is legitimate to extend this linear
relation to the entire domain. (See the sensitivity study part 3.6.1 for detailed validation.)
Therefore, the aimed correlation could be attained through scaling from the correlation of
average hydrodynamic pressure and film thickness ratio at a constant viscosity gio and
sliding speed Vo,
Ph (A,V,p) = Ph (A)0,(PV / V0o)
The deterministic solver introduced in the previous chapter [31] is used to calculate the
correlation Ph (2)o,,, (see Fig. 3.7 for an example). The solver evaluates the inter-asperity
pressure generation at different film thickness ratios with a reference speed Vo and
viscosity go. In the hydrodynamic solver, a surface measurement of 600x 1000 grid with
4ptm mesh size (see Fig. 3.8) is used to represent the liner roughness. Since the
roughness of the oil control ring land is typically one order of magnitude smaller than the
liner roughness [40], the ring face roughness is neglected.
Similarly, the hydrodynamic friction, and oil flow rate could be evaluated with the
following scaling strategies:
fh (2,V5)= fP ()VO,/,
(V /PoV)
and
Q(A,V)=Q(A)V (v/vo)
40
-o
35
30
,
u
E
25
20
15
O 10
r 5
0
2
2.5
3.5
3
4
4.5
5
5.5
6
Film thickness ratio
Fig.3. 7 Correlation of average hydrodynamic pressure and film thickness ratio at a
constant viscosity po and sliding speed Vo
E
,*Wo
C
0
E
0E.,
'C
0
1:
CO
UI,
E
=IC
Ct)
0
0U.5
1
1.5
2
2.5
3
3.5
Sliding Direction (mm)
Fig.3. 8 Liner surface measurement
3.2.2 Squeeze Oil Effect
The squeeze oil effect, namely the hydrodynamic pressure generation caused by the
normal relative motion of the ring face is a very important mechanism for the piston top
ring dynamics to prevent the lubricant film from collapse due to the sudden increase of
the gas pressure in the combustion chamber.
However, for the twin-land oil control ring (TLOCR), the contribution from this
mechanism is rather trivial due to the relatively stable load from the spring and the
irrelevance of the in-cylinder gas pressure to the TLOCR load condition. Therefore, in
the work presented in this thesis, the squeeze oil mechanism is neglected for model
simplification. The detailed validation of this simplification is presented in the
assumption validation section 3.6.2 of this chapter.
3.3
Asperity Contact Part
Different asperity contact models could be implemented, since it's decoupled with the
hydrodynamic part. The one used in this thesis is the asperity contact model developed
by Greenwood and Tripp [39]. According to the Greenwood and Tripp model, nominal
asperity contact pressure between two rough surfaces could be calculated by the
following correlation,
Pc () = apK'E' J(z - A)2.5 (z)dz
in which
In the above two equations, Pc is the nominal asperity contact pressure between the two
surfaces, ap is the area ratio of plateau part to the entire surface, rl is the asperity density
is the asperity peak radius of curvature in the plateau
per unit area in the plateau part, 03
part, and 4(z) is the probability distribution of asperity heights.
The Greenwood and Tripp model assumes that contact is elastic, and the asperities are
parabolic in shape and identical on the contacting surfaces. In this paper, all the surface
geometry related parameters such as ap, rl, P and 4(z) are evaluated based on the 3D
deterministic surface details.
The correlation of the asperity contact friction and the film thickness ratio could be
expressed as
fc ()= CfAPc (A)
where Cfc refers to the asperity contact friction coefficient, and A is the oil control ring
face area. The asperity contact friction coefficient depends on the material of the
interacting surfaces and the lubricating oil. As an input value, it could be modified in the
model and should be assigned with the measured value for specific applications. For the
results presented in this thesis, it is assigned as 0.1 unless stated otherwise.
3.4
Load Balance Part
The load balance part conducts the load balance and solves for the transitional film
thickness ratio A of the oil control ring in the whole engine cycle.
See Fig. 3.9. The normal load exerted by the oil control ring spring is sustained by the
hydrodynamic force and the asperity contact force. Neglecting the radial inertia of the
ring, the normal force balance could be represented as
Ph (A, P)+Pc (A)= load
The correlations of the hydrodynamic pressure and asperity contact pressure are the
output of the modules introduced above. Ring sliding speed V could be calculated with
the engine geometries and operation parameters, and lubricating oil viscosity depends on
the oil used and the local temperature. With certain unit normal load pressure as the
input, the model could solve for the transitional film thickness ratio Xof the oil control
ring in the entire engine cycle.
Based on the film thickness ratio, the transitional friction could be represented by
f =f, (A,v,p) +fC (/)
and the oil flow rate will be Q (A, V) . Integrating the friction and oil flow rate for the
full engine cycle yields the value of friction mean effective pressure (FMEP) of the oil
control ring, and the total amount of oil penetrating through the oil control ring liner seal
(oil passage).
Pload
Vo
Ittiiit t
X
tt +Fhydreo
-r----------- --- h
P. ,flt<ta
Fig.3. 9 Normal Load Balance
3.5
Model Results
This section demonstrates some basic applications of the twin-land oil control ring model.
Fig. 3.10 shows the predicted variation of the oil control ring film thickness ratio on a
liner of 13L heavy duty diesel engine. The liner surface measurement is shown in figure
5, and the engine speed is fixed at 2000 rpm. Axial variation of the oil viscosity based on
the liner temperature distribution is input to the model. In the mid stroke, the increase of
the ring sliding speed helps to generate a large hydrodynamic lift, and the film thickness
ratio goes up. The larger film thickness relates to a reduction in asperity contact friction,
while the higher sliding speed increases the hydrodynamic shear force (see Fig.3.11).
The mid stroke hydrodynamic friction is in the comparable level to the asperity contact
friction in the top and bottom center areas due to the nature of small clearance and high
sliding speed. As a result, the common notion of "reducing the liner roughness could
help to reduce the friction" needs more cautions down to certain roughness level. Since
while a smoother liner usually gives higher hydrodynamic pressure generation and less
asperity contact friction, it also leads to a smaller average film thickness. The increase in
hydrodynamic friction due to the reduction of average film thickness might offset the
gain in reducing the asperity contact friction. This makes the reduction of the total
friction rather complicated.
4.5
0
4
2.5
2
0
60
200
80
100
120
140
160
180
Crank Angle (degree)
Fig.3. 10 Cycle Variation of Oil Film Thickness Ratio
0
20
40
60
80
100
120
140
160
Crank Angle (degree)
Fig.3. 11 Cycle Variations of TLOCR Frictions
The effects of different engine speeds are shown in Fig. 3.12 and Fig. 3.13. The oil
passage rate in figure 3.13 is defined as the total amount of oil left on the liner by the oil
control ring during the piston down strokes per unit time. The oil passage rate is
considered to be related to the oil consumption. It is to be noted that the value of oil
passage rate is two orders of magnitude larger than the typical diesel engine oil
consumption rate, which indicates only a small fraction of oil left by the oil control ring
is burned in the combustion chamber. Meanwhile, the oil passage rate also suggests the
amount of oil available for the top two ring lubrication, thus will have effects on the
friction of the top two rings.
9500
!CL 9000
LLJ
u. 8500
o 8000
0
0
a
7000
500
1000
1500
2000
2500
3000
Engine Speed (rpm)
Fig.3. 12 Influence of Engine Speeds on Oil Control Ring FMEP
· All
4UUU
3500
3000
,
a
2500
I
a
a
U)
U)
2000
S1500
1000
500
51
Engine Speed (rpm)
Fig.3. 13 Influence of Engine Speeds on Oil Passage Rate
Liner surface micro geometries have a strong influence on the oil control ring
performance. Two different diesel engine liner measurements are displayed in Fig. 3.14,
and their plateau RMS roughness' rpq are listed in Table 3.1. The comparison of the
hydrodynamic and asperity contact pressure generations on these two different liner
surfaces at the reference ring sliding speed and oil viscosity is shown is Fig. 3.15.
Clearly the smoother surface 1 develops roughly the same hydrodynamic pressure and a
much lower asperity contact pressure than surface 2 at any average film thicknesses. As
a result, surface 1 displays a better performance in oil control ring friction (see Fig. 3.16),
and a lower oil passage rate (see Fig. 3.17) under the same ring design and ring tension.
Surface 1
g0CE
2
o
2
4wD
1.5
ICD
I
2
&ra
Ie-
0
U
C)
E
C,)
0.5
::3
0O
I
0
I-
11
0.5
1.5
2
--
i
3
2.5
~3.5
Sliding Direction (mm)
Surface 2
E
2
E,
E 1.5
0C
0
I cO
0
C/)
1
2
'U
s=E
·
U
E
0.5
'U
oU)
0.5
1
1.5
2
2.5
3
3.5
Sliding Direction (mm)
Fig.3. 14 Comparison of Two Liner Surfaces
Table 3. 1 Comparison of rpq of the two liner surfaces
1
I
I
I
1
·
·
Surface 1 contact pressure
-
Surface 1 hydro pressure
-
Surface 2 contact pressure
60
Surface 2 hydro pressure
-
-
50
\
40
30
-K
`;=··;,,
I
=
0.45
-C~5
I
i
I
0.5
0.55
"z
12~
1,
1
0.6
.
~
~-1~
1
1
0.65
0.7
I
I
0.75
0.8
Average Film Thickness (nm)
Fig.3. 15 Comparison of the Hydrodynamic and Asperity Contact Pressure
Generation on Two Different Liner Surfaces
x 104
I
I
I
I
Surface 1
-
1.2
-
Surface 2
E
CL
1.1
LL
0
rU
O0
g-
\
0.9
0•
0.8
n7
.I
-
50(0
1000
1500
2000
2500
3000
Engine Speed (rpm)
Fig.3. 16 Comparison of the Oil Control Ring FMEP's on Two Different Liner
Surfaces
4500
4000
E 3500
- 3000
0) 2500
1000
5000
rnN
500
1000
1500
2000
2500
3000
Engine Speed (rpm)
Fig.3. 17 Comparison of the Oil Passage Rates on Two Different Liner Surfaces
3.6
Assumptions Validation and Sensitivity Analysis
3.6.1 Oil Viscosity and Ring Sliding Speed
A major approximation of this new approach is that the inter-asperity hydrodynamic
pressure generation, as well as the hydrodynamic shear force scale linearly with the
product of oil viscosity gi and sliding speed V when the ring is sliding over the specific
liner at a certain film thickness ratio A. This approximation is vital in reducing the time
cost of the calculation. In this part, certain studies are done to examine the error induced
by this basic approximation.
Based on the surface measurement shown in Fig. 3.8, tests of the linearity of the
hydrodynamic pressure and friction are done in two conditions of film thickness ratio,
A=2 when the inter-asperity pressures are very high, and A=6 when the inter-asperity
pressures are lower. Results scaled from the value of hydrodynamic pressure and friction
at the position [tV=0.015 N/m are compared to the results evaluated by the deterministic
solver. The value of gtV varies from 5E-4 N/m to 0.2 N/m, which covers the piston speed
range of 0.5 m/s to 20 m/s, and the oil viscosity range of 1E-3 kg/m,s to 1E-2 kg/m,s.
This range covers the value of pV under normal engine conditions. Test results are
shown in figure Fig. 3.18 to Fig. 3.21. For both hydrodynamic pressure and friction, in
the two tested conditions of film thickness ratios, the results show the strong linear
relations. Relative errors induced by the linear scaling are all bounded by the limit of 4% to 4%. Combined with the benefit of calculation efficiency brought by the
approximation, this is a good approximation for an engineering problem.
Y
__
500
SRelative Error of Pressure Due to Scaling
4501
0
* Calculated Hydro Presure
-
Scaled Hydro Pressure from
the point p.V-0.015 N/m
0a-
S250
E
e200
1:
2
10
3:
5
A.~,
.
IA
0.15
j V (N/m)
Fig.3. 18 Linearity of Hydrodynamic Pressure to ptV at Film Thickness Ratio A=2
.J
LI
a-
0
6
2
0:
aw
O 10
E
0
a-
68
A
A
0 W
Calculated Hydro Presure
S_.__
o2:
· 414
:3:
·e1
ii
QL
Scaled Hydro Pressure from
the point pV-0.015 /Wm
A
0.05
Relative Error of Pressure Due to Scaling
-
0.1
-----------0.15
SV (N/m)
Fig.3. 19 Linearity of Hydrodynamic Pressure to pV at Film Thickness Ratio A=6
F-M
)
100
Calculated Hydro Friction
Scaled Hydro Friction from
the point WiV40.015 N/m
r
2
'or of Hydro
a to Scaling
1.5e
a-
0
W.."
Ai
·
·t
0.15
0.2
SV (N/m)
Fig.3. 20 Linearity of Hydrodynamic Friction to pV at Film Thickness Ratio A=2
-ZE
flil
f2 25
-
•20
-.J
E:H
q
Calculated Hydro Friction
3
Scaled Hydro Friction from
the point pV-N0.015 N/Wm
2.5
Relative Error of Hydro
Friction Due to Scaling
v
c
2
1 .5 W
A
4)
L.
42
r;;
,In
U
0.5
L
C
DS
wF
II~~I
a
a
I
""""I
I~
-'
-~-----~
0.15
0.05
SV (N/m)
0
I2
Fig.3. 21 Linearity of hydrodynamic friction to pV at film thickness ratio A=6
3.6.2 Squeeze Oil Effect
The squeeze oil effect, namely the hydrodynamic pressure generation caused by the
normal relative motion of the ring face is a very important mechanism for the piston top
ring behavior to prevent the lubricant film from collapse due to the sudden increase of the
gas pressure in the combustion chamber.
However, for the twin-land oil control ring (TLOCR), the contribution from this
mechanism is rather trivial due to the relatively stable load from the spring and the
irrelevance of the in-cylinder gas pressure to the TLOCR load condition. Therefore, in
the work presented by this thesis, the squeeze oil mechanism is neglected for model
simplification.
In this section, some basic validations are made for this basic simplification. From the
results of the TLOCR cycle clearance variation in Fig. 3.10, the normal speed of the oil
control ring land can be evaluated. These results were calculated based on the
assumption that the ring normal speed does not contribute to the hydrodynamic pressure
generation. With Li's model [31], the influence of ring normal speed on hydrodynamic
pressure generation under different ring sliding speeds is tested. The liner surface
measurement shown in Fig. 3.8 is used for the testing. The results are presented in the
following plots (Fig. 3.22 - Fig. 3.24).
Sln 5
V=lm/s
a.
a.
0
0
0
0.
0.
.U
E
E
M
-6
0
P
7%
I
4
Sliding Distance (mm)
Sliding Distance (mm)
Fig.3. 22 Squeeze Oil Effect Testing Results (Low Sliding Speeds)
- ·
Y1nl
CL
£.
II ···
v=-mis
9D
0.
.2
E
'A
'a
0V
0
C
E
0
a.
.,.
.u_
E
'A
'C
0
-U
"o
-r-
2
I
Sliding Distance (mm)
Sliding Distance (mm)
Fig.3. 23 Squeeze Oil Effect Testing Results (Medium Sliding Speeds)
WnB
V=12m/s
U
CL
0
0.
P
CL
o
E
'A
V
I:>,
"1-
0
4
0
V
Sliding Distance (mm)
Sliding Distance (mm)
Fig.3. 24 Squeeze Oil Effect Testing Results (High Sliding Speeds)
The ring normal speed assigned for the testing is 120 gLm/s, which is over twice the
maximum ring normal speed presented in Fig. 3.10. The tests are conducted under
different ring sliding speeds from Im/s to 15m/s. The suck condition relates to the
positive ring normal speed, when the hydrodynamic pressure generation is diminished,
while the squeeze condition relates to the negative ring normal speed, when the normal
speed is help to build up the hydrodynamic pressure. Fig. 3.22 to Fig. 3.24 compares the
hydrodynamic pressure evaluated considering the ring normal speed (solid lines without
dots), and the hydrodynamic pressure approximated using the proposed scaling strategy
where the normal speed is neglected (solid lines with dots).
The results show that the ring normal speed has larger influences under low ring sliding
speeds, due to the low hydrodynamic pressure generation capability of low sliding speeds.
The relative errors of hydrodynamic pressure generation induced by neglecting the
squeeze effect are listed in Table 3.2.
Table 3. 2 Errors of Hydrodynamic Pressure Induced by Squeeze Oil Effect
Sliding Speed (m/s)
1
3
6
9
12
15
Error of Hydrodynamic
0.2551
0.2170
0.1121
0.0727
0.1089
0.1518
0.1766
0.2074
0.2350
0.2936
0.3925
0.5382
0.1621
0.0517
0.0177
0.0084
0.0093
0.0102
0.1573
0.0587
0.0268
0.0198
0.0182
0.0192
Pressure - Squeeze (bar)
Error of Hydrodynamic
Pressure - Suck (bar)
Relative Error of
Hydrodynamic Pressure Squeeze
Relative Error of
Hydrodynamic Pressure - Suck
The results in Table 3.2 show that the squeeze oil effect is only important for very low
ring sliding speeds. When the ring sliding speed is over 3 m/s, the error is less than 5%.
It should be reminded that the errors listed in Table 3.2 come from both the squeeze oil
effect and the error of the scaling method, which is shown to be able to induce up to 4%
errors under extreme conditions. Furthermore, for all the sliding speeds, the error
pressure is well under 1 bar, which is completely negligible comparing to the typical 20
bar unit load pressure. For all these reasons listed, the squeeze oil effect could be
reasonably neglected in the twin-land oil control ring modeling.
3.6.3 Ring Tilt Effect
It has been discussed in the previous section that due to the high load force exerted on the
ring, the actual tilt angle of the twin-land oil control ring is usually very small. Typically
the tilt angle is less than 1 minute. In the twin-land oil control ring model, the influence
of the tilt angle on the hydrodynamic pressure generation on each land is neglected. In
this section, some validations on this approximation is done, and the error induced by the
approximation is analyzed.
In the test, both the cases of converging angle and diverging angle are evaluated using
Li's deterministic method [31 ]. Fig. 3.25 shows the case configuration of both the
converging and diverging cases. In the converging case, the tilt angle is defined to be
positive, and it will contribute to the hydrodynamic pressure generation by the wedge
effect. And in the diverging case, the tilt angle is defined to be negative, and it will
diminish the hydrodynamic pressure generation.
The calculation is based on the deterministic liner surface measurement exhibited in Fig.
3.8. The ring sliding speed is fixed and the normal speed is set to zero. Two film
thickness ratios (k=3 and k=4) are evaluated, which represent the typical values of oil
control ring film thickness ratio at top/bottom center and mid stroke (see Fig. 3.10).
When changing the ring tilt angle, the film thickness ratio between the lower edge of the
ring land and the liner mean is fixed, in other words, the minimum clearance hmin is fixed.
Diverging
Converging
brain
S-
'
V" '
-
Positive Tilt Angle
"
Ac
lihlmi
V 'VV
-
-
Negative Tilt Angle
Fig.3. 25 Tilt Angle Configurations
The results of the testing are shown in Fig. 3.26 and Fig. 3.27. In these two figures, the
red lines with the square marker represents the hydrodynamic pressure generation
between two flat surfaces, which only relies on the wedge to generate the hydrodynamic
pressure. The boundary pressures are set to be one bar. The blue lines with the circle
marker represents the hydrodynamic pressure generation between the flat ring land face
and the rough liner surface. Here apart from the macro wedge, the roughness is also
contributing to the hydrodynamic pressure generation between the two surfaces.
The results show that when the tilt angle is not large, (between -1 minute and 1 minute),
the difference in hydrodynamic pressure caused by the ring tilt is mild. An error of
approximately 10% can be expected when neglecting the tilt angle of -1 to 1 minute large.
As estimated in section 3.1.2, the actual tilt angle of the twin-land oil control ring is in the
scale of several minutes. Therefore, the error caused by the negligence of the ring axial
tilt would be several percent.
The actual twin-land oil control ring dynamic could be evaluated with the average model
by Tian et al. [38]. One of the future topics of this project is to integrate the deterministic
model developed in this thesis work into Tian's model.
Film Thickness Ratio X = 3
x lO
,xr0
IU
i
i
I
I
i
i
I
r
I
I
1
',
', Hydro
* Deterministic
Pressure
',
Hydro Pressure
-E-Deterministic
Average Model Hydro Pressure
'--e--',
-e--Average Model Hydro Pressure i.........
i........
::
....
•Y i........
--.-.r,------ ,...,.,·,-~---I-----. -------------------/------------------
- - ---- - -------
IIII
. .. - - - -- - - -
- --------J.
- - - -- --------- .-. . . . - . . . .
-.-..
- - -.--
2
-11
0
-8
-6
-4
-2
0
2
4
Tilt Angle (minute)
Fig.3. 26 Tilt Effect (.=3)
6
8
10
=- 4
Film Thickness Ratio
X1o0
-B----
J
Average Model Hydro Pressure
Deterministic Hydro Pressure
Cl;45--------
---
2.5 ........----.......
----........
,~------- ;-........
--3 5
----------
------------.....-----------------------------
o
-10
-8
-6
-4
-2
0
2
4
6
8
10
Tilt Angle (minute)
Fig.3. 27 Tilt Effect (X=4)
3.6.4 Influence of Liner Surface Measurement Resolution
One of the most important factors for the accuracy of the twin-land oil control ring model
is the resolution of the liner finish measurement. The general resolution issue could be
cracked down to three different yet related questions:
(1)
What resolution should be used in the liner finish measurement to present an
accurate enough geometric profile for numerical calculation?
(2)
Based on the existed liner finish measurement, what resolution should be further
used in the numerical calculation to obtain an accurate evaluation of both the
hydrodynamic and asperity contact pressures?
(3)
How large should the total measurement domain be to offer enough geometric
information of the liner finish for numerical calculation?
In this section, I would try to attack these three questions independently. Allow me to
answer the second question first since it's purely a numerical problem thus is the most
clear and straightforward.
Now assume a liner finish measurement in Fig. 3.28 is accurate and could fully represent
the measured liner surface of its surface height profile. The measurement is based on a
mesh of 41im grid size. The question would be what resolution should be used in the
deterministic calculation to restrain the numerical error in the tolerable range. The
coarsest mesh is of course just 4gmx4im. To attain higher numerical accuracy for the
hydrodynamic pressure evaluation, the surface could be interpolated into a finer mesh.
The penalty to do so would be the vast amount of time cost increase in the deterministic
calculation. Considering the current deterministic solver, doubling the spatial resolution,
namely decreasing the grid size by a factor of 2, would more or less increase the
calculation time by a factor of 10. Obviously this would soon become unaffordable.
y01x
n Iv
-6
1
0
,N
"o
-1
rU..
V
-2
-3
al
(-91
4
Di
I
-5()
C.)1
c-
-6
,,m
-7
or
0
100
200
300
400
500
Sliding Direction (Grid Size = 4prn)
Fig.3. 28 Liner Surface Measurement
The numerical scheme used, by definition, is linearly convergent. In order to evaluate the
numerical error of the deterministic solver with an affordable time cost, the surface in Fig.
3.28 is equally divided into four sub-surfaces as in Fig. 3.29. Each of these four subsurfaces are interpolated onto a 2 pm, 1 im and 0.5 m grid size mesh from their original
4pgm mesh, and the hydrodynamic pressures of each of these case are evaluated using the
deterministic hydro solver.
x10
II
II
4)
N
-1
-2
v
8
-3
C13
4,
05
M
I.-
m.•;d
I
4.
C
G)
U
c5
)0
Sliding Direction (Grid Size = 4pm)
Fig.3. 29 Division of Four Sub-surfaces
4 pm mesh
12
0.4
m mesh
0.41
mesh
nm
1
>.80
25
0.35
0.35
03E
0.5 p~ mesh
0.4
25
F·*
0.3
E
2
0.25
0,25
0.25
1.5
0.15
0.15
E
Q
E
0,2
0.1
, ,
0,1
0
0.05
0.05
0.05
0.1
0,05
0.1
0.05
0.1
0.05
0.1
Sliding Direction (mm)
Fig.3. 30 Local Clearances under Different Mesh Sizes
0.5 jm mesh
1 tin mesh
2 im mesh
•r
1000 bar
0.4
0.35
C
0.3
0
0.25
100 bar
ED
Z
E
0.2
oc
4._)
0.15
U
9!
E
:t
5)
10 bar
0.1
0.05
I bar
0.05
0.1
0.05
0.1
0.05
U.1
U.U5
J
U.1
Sliding Direction (mm)
Fig.3. 31 Local Hydro Pressures under Different Mesh Sizes
4 rnm mesh
2 Am mesh
1 4m mesh
r4
1
0.9
0.8
0.7
C
o
0.6
0.5
a
C
0.4
E
0.3
I-
G
0.2
0,1
0.05
0.1
0,05
0,1
005
0.1
0.05
0.1
Sliding Direction (mm)
Fig.3. 32 Local Oil Densities under Different Mesh Sizes
v
The hydro solver is used to test these four sub-surfaces under the standard test
configuration of a 0.1 mm wide ring face sliding over the surface with a film thickness
ratio of 3 and a sliding speed of 3m/s.
Fig. 3.30 to Fig. 3.32 show the comparisons of local surface clearance, hydrodynamic
pressure, and oil density under different mesh sizes. It is obvious that the finer mesh
could yield a more detailed thus more accurate solution of the pressure and density.
A comparison of hydrodynamic pressure values at different locations on sub-surface 1
under these mesh sizes is plotted in Fig. 3.33. The results show that the four mesh sizes
generate the same general trend of hydrodynamic pressure, while there are still
differences exist.
Now assume the 0.5 [tm mesh gives an accurate solution of the hydrodynamic pressure.
The numerical error on the other mesh sizes could be approximated by comparing to the
result on the 0.5 ýtm mesh. Furthermore, we can assume that the numerical error of the
hydrodynamic pressure on each mesh size at different surface locations applies to a
random distribution whose mean value is zero. From the hydrodynamic pressure
evaluations on the four sub-surfaces we can evaluate the standard deviation of the errors
of hydrodynamic pressure on different mesh sizes. Table 3.3 lists the standard deviations.
The numerical error is bounded in a smaller and smaller band if taking three times of its
standard deviation as a bound. Since the average hydrodynamic pressure in this test case
configuration is approximately 5E5 pa, the error band width of the different mesh size
would be:
4 p.m mesh: -44% to +44%
2 p.m mesh: -14% to +14%
1 .mmesh: -5% to +5%
In order to evaluate the local pressure generation accurately enough, a very high
resolution needs to be used. The accordant time cost would be extremely high.
fo
t1)
=3
~I)
C
V
0
I
0
1
0.5
1.5
Sliding Direction (mm)
2
x103
Fig.3. 33 Hydrodynamic Pressures under Different Mesh Sizes
Table 3. 3 Standard Deviations of the Errors of Hydrodynamic Pressures on
Different Mesh Sizes
Standard Deviation
(pa)
Error on 4 .tm mesh
Error on 2 [tm mesh
Error on 1 itm mesh
7.4E4
2.3E4
0.81E4
Now there comes the beauty of the correlation method. We rely on the average
performance of the ring on the full surface measurement in the twin-land oil control ring
model. The deviation of the average value of these hydrodynamic pressures in Fig. 3.33
would be much smaller than the original deviation of these hydrodynamic pressures.
Each of the pressures in Fig. 3.33 covers an area of 0.4mmx0.lmm, while the total area
of the surface in Fig. 3.28 would be 1.6mmx2mm, and therefore 80 times as large.
Assuming the errors of these hydrodynamic pressures apply to a same normal distribution
and are linearly independent under different surface geometries, the deviation of the error
on the average hydrodynamic pressure generation we are interested in will be
1
of the
local errors. Therefore, for each of the mesh sizes, the error band of the average
hydrodynamic pressure on the entire surface in Fig. 3.28 would be:
4 tm mesh: -5% to +5%
2 [im mesh: -1.6% to +1.6%
1 [tm mesh: -0.56% to +0.56%
As a result, we can bound the numerical errors in a 5% band with a coarsest 4 im mesh.
Both the accuracy and the efficiency could be achieved.
At this point we can go ahead and try to attack the third question: "How large should the
total measurement domain be to offer enough geometric information of the liner finish
for numerical calculation?"
Consider the hydrodynamic pressure generation on different locations of the liner surface
as an independent random distribution. The pressures of 4pm mesh basis exhibited in Fig.
3.33 have a standard deviation of 1.8E5 pa. This is rather large comparing to the average
pressure level of 5E5 pa, which means the circumferential width of the sub-surfaces is
too small so that the local pressures are so variant. This appears to be another important
problem for most of the local deterministic methods. While constrained by the
measurement and calculation capability, the calculation domain is so small on each ring
coverage and a huge error band could be applied on the results. Yet with the correlation
method, we only need a hydrodynamic pressure value that represents the specific liner
patch. Again, the average calculation on the hydrodynamic pressure over the entire liner
measurement would constraint the error band by a large amount. We would use the three
times of the standard deviation of the hydrodynamic pressure as a band width. In order to
limit the error of the average pressure in a band of -5% to 5%, a surface size of 18.7 mm2
is needed. This relates to roughly 6 times as large a surface as presented in Fig. 3.28.
The preceding analysis is based on an assumption that the average hydrodynamic
pressure evaluated on the sub-surfaces separately actually represents the value evaluated
on the entire surface in Fig. 3.28. Since the use of periodical boundary condition, the
surface geometry actually changes at the boundary when the surface is divided and
evaluated separately. Therefore this assumption needs to be validated.
cc
ra
v
I0'I)
a.
E
cc
*0
o
"1
0
0.5
1
1.5
2
Sliding Direction (mm)
Fig.3. 34 Comparison of Mean Sub-surface Pressures and the Pressure of the Entire
Surface
Fig. 3.34 exhibits both the average of the hydrodynamic pressures evaluated on subsurfaces 1-4 separately, and the hydrodynamic pressure evaluated on the entire surface
presented in Fig. 3.28. The evaluation is based on a 4[m mesh. The results show good
uniformity. The 3a error band of the difference in the two pressure traces is -6% to 6%,
much smaller than the numerical error of the pressure on the same mesh basis (44%) and
the variation of pressure due to the different surface geometries (108%). Therefore, the
analysis based on the sub-surfaces should be legitimate enough.
Now let's get back to the first question, "What resolution should be used in the liner
finish measurement to present an accurate enough geometric profile for numerical
calculation?"
Due to the aliasing effect, the mesh grid used to sample the surface profile would filter
the high frequency component of the surface profile. For example, when a 4 pm mesh is
used, any information on the surface profile with the wavelength shorter than 8 pm would
be lost. The magnitude of the lost high frequency information is important to decide the
error in the deterministic calculation caused by the aliasing effect.
The four surfaces in Fig. 3.35 will be tested for an example of how the results of
deterministic calculation are influenced by the aliasing effect. Surface 1-1 is a liner
surface measurement based on a 1 p.m mesh. To mimic the effect of aliasing on the
surface, the measurement is first restricted onto a 4 ptm mesh, and then interpolated back
onto the original 1 pm mesh to generate surface 1-2. Similarly surface 2-1 is another
liner surface measurement based on a 4 Rpm mesh. And the measurement is first
restricted onto a 16 Rpm mesh, and then interpolated back onto the original 4 ptm mesh to
generate surface 2-2. It needs to be noticed that since the limitation of the profilometer in
the number of total sample points, surface 1-1 and surface 2-1 have more or less the same
amount of sample points. Therefore surface 2-1 is 16 times as large as surface 1-1 in size,
and this appears to have a large influence in the final test results.
Surface 1.1
Surface 1-2
Surface 2-1
Surface 2-2
Fig.3. 35 Test Surface 1-1 ~ 2-2
The Fourier analysis of both the two original surfaces in Fig. 3.36 show that the
magnitude of the information cut off by sampling surface 1-1 with a 4 gtm mesh is about
0.015 gtm. Meanwhile, the magnitude of the information cut off by sampling surface 2-1
with a 1 gm mesh is over twice as large. Therefore, the error caused by the aliasing for
surface 1-2 should be smaller than the error for surface 2-2.
Surface 1-1
8
933.116
0.12 [ ........
Wavelength (pm)
4
Surface 2-1
2
88
Wavelength (pjm)
16
8
0 ....... ...
.......................
...................
0
.
.
...
.
........
. ......................
. ...i..... i. ....
o.•
i ....... ........ ... . .... ..... .
"'i"'r`"~~'"""~"
0.04 ------
32
........................
.......................
014
0.086
64
------- ----------
... .. .. .. .. .. .. . ...............
0
..........
...
.......... .....................
002
i
50
0
002 -
[
0.02
8
100
150
200
Wave Number (Hz)
v
100 150 200 250 300 350 400 450
Wave Number (Hz)
Fig.3. 36 Fourier Analysis of The Two Original Surfaces
Fig 3.37 shows a comparison between surface 1-1 and 1-2 of the local clearance profile
and hydrodynamic pressure distribution in the ring land coverage. It can be seen that
both the clearance and the corresponding hydrodynamic pressure distribution profiles are
identical for these two surfaces. Obviously the aliasing effect exists. Comparing the
hydrodynamic pressure trace in Fig. 3.38, we can find that the error of hydrodynamic
pressure caused by the aliasing effect of surface measurement is mild. Using the result
for surface 1-1 as a criterion, the average value of the errors is 2E4 pa (defined as the
mean of the absolute value of all the errors). How representative is this value is a bit
questionable, since the surface domain is rather small due to the constraint of the
measurement technique. It can be seen that the pressures generated on surface 1-2 seems
to be monotonously higher than those of surface 2-2. While the opposite trends have
been observed in other surface measurements, there hasn't been enough study to reveal
any reliable conclusion. What we can be certain of is the error of aliasing depends on the
specific surface geometry and needs large enough geometry information to display its
randomness.
This conclusion is backed up by the tests results of surface 2-1 and 2-2. Since the surface
domains of these two surfaces are much larger than those of the previous two surfaces
(16 times as large), the results should be more representative.
Height Pr-ofile (m)
Hydrodynamic Pressure
.6
8,l
""'
1o00
bar
'7
6
5
100 bar
0
4
3
10 bar
2
U,
1
0.05
0.1
0.05
0.1
0.1
0.05
Ring Width Direction (mrm)
0.1
0.05
01
1 bar
Ring Width Direction (ran)
Fig.3. 37 Comparison of Clearance and Hydrodynamic Pressure for Surfaces 1-1
and 1-2
xl0
s
i.
(.
o(n
I-
0
Sliding Direction (rnm)
Fig.3. 38 Hydrodynamic Pressure Generation for Surface 1-1 and 1-2
The comparison between surface 2-1 and 2-2 of the local clearance profile and
hydrodynamic pressure distribution in the ring land coverage can be seen in Fig. 3.39.
It's clear that the aliasing effects on the height profile and the corresponding
hydrodynamic pressure generation are very strong since the surface is measured in a
coarse mesh of 4 itm grid size and re-sampled by an even coarser mesh of 16 gim grid
size. Comparing to the results of the previous two surfaces, the hydrodynamic pressure
profiles in Fig. 3.39 show mostly similar general patterns. The high pressures are
generated at identical locations, while the exact patterns are rather different.
Comparing the hydrodynamic pressure traces in Fig. 3.40, one can find that the two
surfaces have similar capability of generating hydrodynamic pressures, since the general
trend of the trace look alike. Meanwhile the errors are eminent. Again using the pressure
generation of the original surface 2-1 as the true value, the average value of the errors is
3.8E4 pa (defined as the mean of the absolute value of all the errors). This time the
aliasing effect is almost twice as strong as the previous case. This coincides with our
previous analysis.
Furthermore, since the surface domain is much larger, the randomness of the
hydrodynamic pressure errors caused by the aliasing effect of the surface measurement is
clearly displayed. Fig. 3.41 shows the distribution of the aliasing errors.
Since again in the twin-land oil control ring model, we are concerned of the average
hydrodynamic pressure generation capability of the liner surface, the errors of the
measurement aliasing effect can be toned down by the larger surface domain. As shown
in Fig. 3.39 and Fig. 3.40, even a very small sample rate could in principle define the
general magnitude of the hydrodynamic pressure generation capability. Of course this
should have a limit. Crossing the limit would probably cause the surface to lost its very
specific uniqueness and generate intolerable errors in pressure generation. Therefore a
suitable mesh resolution with large enough surface domain covered should be advised.
Height Profile (m)
Hychdrodynamic Pressure
S10
"S
1000 bar
I6
100 bar
5
.
4n
maq
3
10 bar
2
0.05
0.1
0.05
I bar
0.1
0.05
Ring Width Direction (mm)
0.1
Ring Width Direction (mm)
Fig.3. 39 Comparison of Clearance and Hydrodynamic Pressure for Surfaces 2-1
and 2-2
v
1f 5
500
1000
1500
2000
2500
3000
3500
4000
Sliding Direction (prm)
Fig.3. 40 Hydrodynamic Pressure Generation for Surface 2-1 and 2-2
-1.5
-1
-0.5
D
0
-D
2-2
0.5
Ino(
I
1
1.5
2
,s
x 1U
Fig.3. 41 Distribution of Aliasing Errors for Surface 2-2
69
Now let's sum up the answers for these three questions raised at the very beginning of
this section.
The most commonly used method of liner finish measurement has its range of the
resolution selection. However, the total sample numbers in both spatial directions are
usually limited. In order to give a finer depict of the surface profile, the total surface size
are limited. Therefore, there is a trade-off between profiling the surface accurately and
representing the surface features completely.
The test results in this section show that the finer mesh in both surface measurement and
numerical calculation could help to minimize the error of local hydrodynamic pressure
generation. However the increase of the total surface domain could in general help to
reduce the error in evaluating the average hydrodynamic pressure generation capability of
the tested surfaces. For the twin-land oil control ring model, a large enough surface
domain is needed to ensure sufficient information to represent the original liner surface
features.
From the current test results, a 4 jtm grid size mesh for surface measurement as in Fig.
3.28 seems to be able to minimize the total error coming from different sources. While
the surface measurement would better be interpolated onto a finer mesh in the numerical
calculation to achieve a higher numerical accuracy, this is also accompanied by the much
higher time cost. Therefore, considering all the aspects mentioned above, the 4 jIm grid
size mesh is chosen for both measurement and calculation for the current twin-land oil
control ring model.
3.7
Validation of Model Results
Up to this point, not much experimental work has been done to validate the results of the
twin-land oil control ring model for several reasons. First of all, on the model side, the
exact correlation between the oil passage rate and the actual oil consumption is still not
clear. Secondly, on the experimental side, it's hard to separate the friction of oil control
ring with the friction of the other piston rings and the piston. Thirdly, it takes a lot of
time and effort, as well as money to set up the experimental apparatus and do the
measurement. Currently, the needed time, effort and money is still not available.
However, the model has been successfully used to explain some phenomenon observed
through the experiments of other researchers which can not be explained by the
traditional method.
P-2
...................................
~..........~..,... ,.........
(Plate au h nn
P"2 (Pl~ateau hraning)S25Snihnng
7
P-2
3
01
0
S-3 i
S 1.99Rz, 0.11 Rpk. 0.98Rvk, U.SSK
0.0
.
.0
1,5 2101
U
2
-
3.0
.0
Ift*
0,0
2.57Rz, 0.35Rpk
O,'
10
.Rk,.SR
tO
2.5
3.0
4.0
..
[mimM
-----
1500rpm
zu
15
1800rpm
VU
4)
0
2,n
U. 300
P-2
_
400
600
-360
(Plateu)
S-2.5
-180
TE0G 180
360
Crank angle (deg]
,
(Plateau)
S-2.5
(Sinle)
S-2.5
Cyl.nPressure
. _
Cyl. Pressure
J.......
P-2
P-2
(Plateau)
-360
-18AO
TflC
ISO
-qn
Crank angle [deg]
10
Cyl. Pressure
-360
-ISO
---
TInrV
IAn
So
lan
Crank angle [deg]
Full Load
Fig.3. 42 Effects of Liner Finish Roughness on Piston Ring Pack Friction
Fig. 3.42 shows the experimental measurement of the piston ring pack friction for two
different liners by the research group of Prof. Takiguchi of Musashi Institute of
Technology [30]. It has been noticed that the liner surface with a smaller roughness (P-2)
can generate a higher total friction in the mid stroke. The phenomenon can not be
explained with the traditional method. Any average Reynolds model would predict that
the rougher surface would generate the higher friction for the piston ring pack. Since the
twin-land oil control ring is arguably the part in the piston ring packs that is most
influenced by the liner finish and is responsible for the major part of the overall piston
ring pack friction [1], we may consider the difference of the friction in the experimental
results in Fig. 3.42 comes mainly, if not solely, from the twin-land oil control ring.
It is a shame that the deterministic measurements of the original liner surface used in the
experiment are not available. To verify the possibility of the experimental results on the
numerical side, two liner surface measurements with the similar statistical roughness
parameters have been tested for their performances.
The surface measurement profiles of the two surfaces are shown in Fig. 3.43.
Smooth Surface
700
0
IM
Tf~4o0O
100
0
u 200
5
cyJ
U,
100
M
N
0
0
100
200
300
400
Sliding Direction (pm)
Rough Surface
a
0m
3C)
.4
0
100
200
300
400
500
Sliding Direction (pum)
Fig.3. 43 Surface Profiles of a Smooth and Rough Liner Surfaces
The comparison of the roughness statistical parameters between these two surfaces and
the surfaces used in the experiments are listed in Table 3.4.
Table 3. 4 The Comparison of the Roughness Statistical Parameters [21]
Type of
parameter
Smooth Surfaces
Rough Surfaces
Numerical
Experimental
Numerical
Experimental
Rpk (rtm)
0.11
0.11
0.35
0.35
Rk (tjm)
0.25
0.58
1.05
1.53
Rvk (pm)
0.70
0.98
0.88
1.11
The most important parameter to the twin-land oil control ring is Rpk [21], since it's on
the peak area that all the asperity contact occurs and high hydrodynamic pressure is
developed. Both the smooth surfaces have rather low Rpk, and rough surfaces have quite
high Rpk.
The cycle calculation results of TLOCR frictions for the two tested surfaces can be seen
in Fig. 3.44 to Fig. 3.49. The unit load pressure is fixed at 15 bar and the ring land width
is chosen to be 0.1mm. The asperity contact friction coefficient is set to be 0.14. The
calculation is run under the engine speed of 1500 rpm as in the experiment and an
additional 3000 rpm for high speed situation.
Crank Angle (degree)
Fig.3. 44 Numerical Results onTLOCR Frictions for the Smooth Surface at 1500
rpm
hydro friction
-
---- contact friction
total friction
*
//
-
0
-- I
20
I
40
-- I
60
--I
80
I
I
I
I
100
120
140
160
180
Crank Angle (degree)
Fig.3. 45 Numerical Results on TLOCR Frictions for the Rough Surface at 1500
rpm
i .'
I
I
r
r
I
i
1
r
1.1
_C~-T~-~L-----__-~-_c0.9
E
0.8
Smooth Surface
0.7
-
Rough Surface
0.6
0.5
l
n .2) 0
-
-
0I
20
40
60
80
100
120
140
160
180
crank angle (degree)
Fig.3. 46 Comparison of Average Clearance for the Two Surfaces at 1500 rpm
The results for the engine speed of 1500 rpm could be seen in Fig. 3.44 to Fig. 3.46.
Under this speed, the two surfaces generate similar magnitude of total friction.
The smooth surface generates high enough hydrodynamic pressure to sustain the ring
load, and reduces the asperity contract friction to almost zero in the mid stroke when the
ring sliding speed is high. However, since the surface is too smooth and the clearance is
too small (Fig. 3.46), the hydrodynamic shear force is so large that the overall friction
coefficient is even larger than the asperity contact friction coefficient in the mid stroke.
The rough surface generates low hydrodynamic pressure, therefore the mid stroke
asperity contact friction is still high. However since the clearance is large due to the
higher roughness, the hydrodynamic friction is not as high. The total friction is still
slightly higher than smooth surface.
..-
t-
0
O
S20
0e
a
iL
o:
a
o
W
Crank Angle (degree)
Fig.3. 47 Numerical Results onTLOCR Frictions for the Smooth Surface at 3000
rpm
#:.]
hydro friction
contact friction
total friction
_-
20
15
10
I
SI
20
40
I
I
I
60
80
100 120
Crank Angle (degree)
I
140
160
180
Fig.3. 48 Numerical Results on TLOCR Frictions for the Rough Surface at 3000
rpm
1.6
1.4I
SRough Surface
Smooth Surface
0.8
0.6
I-
0.4
0.2 3
-0
I
20
I
I
I
I
I
I
I
I
I
I
40
60
80
100
120
140
160
I
I
I
I
180
crank angle (degree)
Fig.3. 49 Comparison of Average Clearance for the Two Surfaces at 3000 rpm
The results in Fig. 3.47 to Fig. 3.49 show different trend under high engine speed. Since
the ring sliding speed is high, the hydrodynamic pressure appears to be more dominant
for both surfaces.
The asperity contact friction completely disappears in the mid stroke for the smooth
surface due to the high ring sliding speed. However, the overall friction of the smooth
surface is higher than the rough surface, which is the same with the trend observed in the
experiment.
It's dangerous to make conclusions solely based on the statistical parameters listed in
Table 3.4, since the performance of a liner surface depends on the entire surface topology
(Chapter 5) and the specific TLOCR design (Chapter 4). Meanwhile, the two surface
measurements used in the numerical tests are not the same with the liner surfaces in the
experiments. That's probably the reason why the numerical results do not show the same
trend with the experiment under low engine speed.
However, information conveyed by the test is that smooth surfaces could generate higher
total frictions for twin-land oil control ring than rough surfaces under certain
circumstances. And this trend can not be explained by the traditional average models for
ring dynamics.
4.
Effects of Twin-land Oil Control Ring Design
Twin-land oil control ring (TLOCR) design parameters, such as ring land width, unit load
pressure and ring tension would strongly affect the performance of the ring in all aspects.
In this chapter, some of these design parameters are evaluated using the TLOCR model
and their effects are discussed.
The first two sections cover each of the parameters of ring land width and ring tension,
while the third section discusses the competing effects of several different ring design
parameters and gives some guidelines of the TLOCR design based on the model results.
4.1
Effect of Ring Land Width
The finite ring land width has an important influence on the inter-asperity hydrodynamic
pressure generation. Different ring land widths have been tested based on the liner finish
measurement shown in Fig. 4.1 using Li's deterministic method [31]. All the other
parameters that might affect the inter-asperity hydrodynamic pressure generation are
fixed in the test. As shown in Fig. 4.2, based on the same liner finish measurement,
larger oil control ring land width covers larger area of surface profile. Therefore it has
more chance to build up higher hydrodynamic pressure.
1
- 2
E
0
0
.12
.Z-,
s-
-.3
4L:
I
C
0
(I)
£
E
0
0
I]1
U
UZ
1
1
z
Z.,0
3
35
Sliding Direction (mm)
Fig.4. 1 Liner Finish Measurement
0.05mm 0.1mm
0.2mm
100 bar
I:
1
1.3
4)
10 bar
1.5
4--
U
i
S.020J•4
OAS
9.1
C
Iba
1 bar
055•
0A.
0.15
0.2
Ring Width Direction (nun)
Fig.4. 2 Hydrodynamic Pressure Distribution under Different Ring Widths
Fig. 4.3 presents the time average hydrodynamic pressure distribution along the axial
direction on the OD surface of the TLOCR land under different ring land widths. Due to
the asymmetry boundary condition on the two edges of the oil control ring land (one side
oil flowing in and the other side oil flowing out), the pressure distribution displays an
asymmetric profile. The hydrodynamic pressures at the boundaries are fixed at the
atmospheric pressure. The liner finish micro geometries help to build up the interasperity hydrodynamic pressure in the middle of the ring land coverage over a completely
flat ring face. The larger the ring land width is, the higher the hydrodynamic pressure
could be. The relationship between the average hydrodynamic pressure and the ring land
width is plotted in Fig. 4.4. The average hydrodynamic pressure is doubled when the ring
land width increase from 0.05 mm to 0.3 mm. This trend is similar to the macro shape
dominant pressure generation, where a larger face width would generate a much higher
hydrodynamic pressure. Only in the inter-asperity pressure generation case, the ring land
width is less dominant comparing to its proportional influence for the macro geometry
dominant flow.
C
0
6
5
4
3
2
so
I
0
•
0.05
0.1
!i|d
0.15
0.2
0.25
Ring Land Width Direction X (mm)
Fig.4. 3 Ring Width Effect on Hydrodynamic Pressure Distribution
I-
C
V
Sr
0.
I-
E
3:
'U
0
0.05
0.1
0.15
0.2
0.25
Ring Land Width (mm)
Fig.4. 4 Ring Width Effect on Average Hydrodynamic Pressure Generation
The effect of ring land width in the development of hydrodynamic pressure is reflected in
the cycle performance of the twin-land oil control ring. Fig. 4.5 shows the results of twinland oil control ring model cycle calculation for different ring land widths. In the
calculation, the ring load and the engine speed are fixed at 50 N and 2000 rpm. Two
performance parameters are evaluated, the oil control ring friction mean effective
pressure (FMEP) and oil passage rate. Here, the oil passage rate is defined as the total
amount of oil left on the liner per unit time by the oil control ring during the piston down
strokes, which is taken as an indicator to the oil consumption. Since the ring tension is
maintained at 50N, the increase in ring land width relates to a decrease in normal load
pressure. Since the larger ring land width also suggests a stronger capability of
hydrodynamic pressure generation, the larger ring land width would relate to a higher
mid-stroke average film thickness, thus a lower asperity contact friction. While the
hydrodynamic friction rises with the increasing ring land width due to the dominant
effect of a larger ring land area, the total friction of the oil control ring shows a complex
trend. On the oil control side, the oil passage rate monotonically increases with the ring
land width since it is mainly controlled by the average film thickness.
X""II
LOUU
2600
O
O
0
2400
x
+
+
+
Ik 2200
Hydro
C-ontact FMEPs
Total
0.05mm
0.10mm Ring
0.15mm Land
0.20mm Width
0.25mm
0.30mm
2000
0.
1800
0
00
0
2000
1 A00
40
00
80
00
4000
6000
8000
10000
Oil Control Ring FMEP (pa)
Fig.4. 5 Influence of ring land width on oil control ring FMEPs
4.2
Effect of Ring Tension
Ring tension has been recognized as one of the most important parameters in controlling
the performance of the twin-land oil control ring (TLOCR), especially on the friction side.
When the ring width land is fixed, the ring tension (tangential force Ft) is proportional to
the unit load pressure:
F, = BwP
in which B denotes the cylinder bore, w represents the ring land width, while Ft and P
denote ring tension and unit load pressure.
This section would discuss the effect of ring tension (the same effect with unit load
pressure when the ring land width is fixed) on the performance of twin-land oil control
ring. How the three vital ring design parameters, including ring land width, ring tension
and unit load pressure could jointly affect the TLCOR performance would be discussed
specifically in the following sections.
The TLOCR model is used to evaluate the performance of the oil control ring under a
typical engine running speed range of 600 rpm to 3000 rpm (from a design of a heavy
duty engine). The liner surface measurement in Fig. 4.1 is used in the evaluation. The
ring land width is fixed at 0.2 mm, and the asperity contact friction coefficient is assigned
as 0.1.
Fig. 4.6 to Fig. 4.9 plot the stribeck curves of the TLOCR cycle average friction
coefficient, average friction (FMEP), velocity weighted asperity contact pressure, and oil
passage rate under the unit load pressure of 10 bar to 40 bar. The X axis of these plots
represents the oil dynamic viscosity [t times engine speed N divided by unit load pressure
Pload, which physically weights the ratio of a characteristic hydrodynamic pressure (p.N)
and the unit load pressure (Pload). Each curve in the plots shows the variation of a specific
physical value with the engine speed changing from 600 rpm to 3000 rpm under a
specific unit load pressure.
The effect of the ring unit load pressure (equivalent to the effect of ring tension here) can
be clearly seen from the four figures.
0.085
0.08
0.075
-
0.07
E 0.065
0
O
0.06
S0.055
LL
S0.05
0.045
0 A4
'0
1
0.5
l
N / Pload
1.5
x 1 'S
Fig.4. 6 Stribeck Curves of Friction Coefficient
The effect of engine speed on the cycle average friction coefficient of the TLOCR has
been discussed in the last chapter. It has been noticed that as the traditional conclusions
drawn from the average models, the friction coefficient is minimized at a certain engine
speed while deviating from this speed either increase the friction of the asperity contact
or increase the friction of hydrodynamic shear force.
The same trend has been noticed on the curves in Fig. 4.6. Furthermore, the unit load
pressure has an eminent influence on this switch point. Increasing the unit load pressure
tend to also increase the critical engine speed where the friction coefficient is the minimal.
And for this specific liner ring land width and engine configuration, using a very small
unit load pressure would even increase the cycle average friction coefficient. This is
because the reduction of hydrodynamic shear force which is related to the mean clearance
between the ring land and the liner surface is not as fast as the reduction of the unit load
pressure.
AFLnhn
IDUUU
,
I
I
- 10 bar, Total
14000
- 10 bar, Contact
12000
- 20 bar, Contact
- 30 bar, Total
- 30 bar, Contact
10000
-40 bar, Total
-40.bar, Contact
-20 bar, Total
CL 8000
LL
6000
4000
2000
0
1
0.5
t N / Pload
1.5
x10-5
Fig.4. 7 Stribeck Curves of FMEP
It has to be mentioned that the friction coefficient is not equivalent to the friction. And
although the friction coefficient gets higher under small unit load pressures, the average
friction (FMEP) still decreases with the reduction of unit load pressure in general. This
of course is due to the dominant change of the unit load pressure. As can be seen in Fig.
4.7, both the total FMEP and the FMEP of asperity contact friction get higher when the
unit load pressure increases, regardless of the variation of engine speed.
Meanwhile under the low engine speed and high unit load pressure, the asperity contact
pressure could be rather high and it would certainly increase the wear of the ring face.
I
x 106
6
E5
S4
U
t-3
1
cc 2
0
1.5
1
0.5
LN /Pload
x 10- 5
Fig.4. 8 Stribeck Curves of Velocity Weighted Asperity Contact Pressure
Another important parameter often used to evaluate the wear rate of the ring face is the
cycle average product of asperity contact pressure and ring sliding speed. It can be seen
from Fig. 4.8 that again, the increase of unit load pressure would monotonously increase
the value of this wear index, and thus would raise the ring face wear rate.
Another interesting phenomenon of the wear rate index is that although higher engine
speed would arbitrarily increase the sliding speed in the product, it would also drastically
enhance the hydrodynamic lubrication. The reduction of the asperity contact pressure
overrates the increase of ring sliding speed in general, and thus the total value of the
product is rather small under very high engine speeds.
It has to be noted that this wear rate index could better represents the condition of the ring
face than that of the liner surface. The wear rate on the liner is obviously different at
different locations of the stroke in an average sense, since the local sliding speed is so
different. The severe wear of the area near the top and bottom center of the oil control
ring on the liner surface has already been noticed and understood and does not need to be
further discussed in this thesis.
A tenn
0O
ca
0)
CU
Co
0,
UI)
u,
5
LtN / Pload
x 10-
Fig.4. 9 Stribeck Curves of Oil Passage Rate
In every sense, the reduction of the TLOCR unit load pressure seems to be beneficial to
the reduction of friction, as has been discussed above. However, since the function of the
oil control ring is still mainly to control the oil consumption, the reduction of unit load
pressure should certainly be constrained.
Fig. 4.9 shows the relationship between the oil passage rate and the stribeck curve X
index. Here the oil passage rate is defined as the total amount of oil left on the liner by
the oil control ring during the piston down strokes per unit time. The reduction of unit
load pressure goes with an increase in the oil passage rate. As has been discussed in the
last chapter, the oil passage rate is considered to be related to the oil consumption, while
the value of oil passage rate is two orders of magnitude larger than the typical diesel
engine oil consumption rate, which indicates only a small fraction of oil left by the oil
control ring is burned in the combustion chamber. The exact relationship between the oil
passage rate and the oil consumption is still not clear. Meanwhile it has been discovered
in previous studies that the unit load pressure is very important to control the oil
consumption. The current model just tells the fact the reduction of the unit load pressure
would accompany an increase of oil consumption as the previous experiences.
4.3
Competing Effects of Different Parameters
The three most important design parameters of the twin-land oil control ring (TLOCR):
ring tension (tangential force Ft), ring land width and unit load pressure are related to
each other by the following relationship:
F, = BwP
in which B denotes the cylinder bore, w represents the ring land width, while Ft and P
denote ring tension and unit load pressure.
As has been discussed in the previous two sections of this chapter, the larger ring land
width would benefit the generation of inter-asperity hydrodynamic pressure generation.
When the ring tension is fixed, it would also reduce the unit load pressure. Both of these
two effects would increase the cycle average clearance between the TLOCR land and the
liner surface. Therefore it would help to reduce the asperity contact friction of the
TLOCR. However, it has been observed that since the area of the ring face has also
increased, the hydrodynamic friction would rise, and so would the total friction force.
Meanwhile the oil passage rate would also increase.
The combined effect of ring land width and ring tension can be seen in Fig. 4.10. The
TLOCR model is used to evaluate the ring land width from 0.05 mm to 0.3 mm, and the
ring tension from 40 N to 80 N. The engine speed is fixed at 2000 rpm in the calculation.
It can be seen that in this specific case while the total friction is mainly controlled by the
ring tension, the ring land width also has certain influence on the friction, especially
when the ring tension is high. Reducing the ring land width appears to be a good
approach to cut the oil passage rate, if not considering its side effects in worsening the
wear by increasing the asperity contact friction.
·MIUI
JUUU ..
-
40N
-50N
Ring
Tension
60N
ON
ON
1*
2500 i.
).05mm
.10mm Ring
.15mm Land
!20mm Width
i25mm
.30mm
0 2000
a.
0
500 60
4mn
A
---
5000
II
70
00
7000
8000
I g"
6000
ii
mI
00
I
100
10
m
II
9000
10000
Ii
11000
O11 Control Ring FMEP (pa)
Fig.4. 10 The Competing Effect of Ring Land Width and Ring Tension 1
40N
----,•lN
6-0N
I-M
Ring
Tension
S70N
-0. 80N
0
0
2500
O.OSmm
ing
and
idth
O
Irm
Qdw
o:
02000
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
TLOCR Contact FMEP (pa)
Fig.4. 11 The Competing Effect of Ring Land Width and Ring Tension 2
The effect of ring land width and ring tension on the asperity contact friction can be seen
in Fig. 4.11. It's clear that increasing the ring tension while maintaining the ring land
width would increase the average asperity contact FMEP. Furthermore, although
reducing the ring land width while maintaining the ring tension could help to control the
oil passage rate and would not necessarily increase the total friction of the oil control ring,
it would certainly increase the average asperity contact friction by a large amount.
Therefore, the ring face would be worn down much faster.
3000
2500
---
10 bar
_-
30 bar
------
50 bar
70 bar
-
90 bar
~
~
110 bar
o 0.05 mm
2000
E
d
o
0.1 mm
x
0.15 timm
*
+
+
0.2 mm
0.25 mm
0.3 mm
~--"*
0/0
15;nn
0J
I
0.5
~
1
I
~
_-
-A..
1.5
2
I
..
2.5
TLOCR FMEP (pa)
3
3.5
4
x 104
Fig.4. 12 Competing Effect of Ring Land Width and Unit Load Pressure 1
Fig. 4.12 shows the competing effect of ring land width and unit load pressure. Again the
X and Y axes represent the two most important indices to evaluate the twin-land oil
control ring performance, the cycle average friction (FMEP) and the oil passage rate for
oil consumption.
It can be seen that the common notion that the oil consumption is mainly controlled by
the unit load pressure is more or less correct under relatively high unit pressures. In this
case, when the unit load pressure is over 50 bar, the oil passage rate is not changing so
much for different ring land widths. However, when the unit load pressure is particularly
small, i.e. 10 bar, the ring land width would make a tremendous difference in oil passage
rate. Meanwhile increasing unit load pressure would certainly reduce the oil passage rate
and raise the friction.
.JUU
2800
0
E 2600
O
2000
1ANN
0
0.5
1
1.5
2
2.5
Cycle Average Asperity Contact Pressure (pa) x 10s
Fig.4. 13 Competing Effect of Ring Land Width and Unit Load Pressure 2
A widely accepted design criterion is to reduce the ring land width for a lower friction,
and maintain the unit load pressure to keep the oil control. From Fig. 4.12 we can find
that in general, reducing the ring land width and maintaining the unit load pressure would
even reduce the oil passage rate and thus help to control the oil consumption better.
However, since the reduction of oil passage rate would always go with an increase in
cycle average asperity contact pressure (Fig. 4.13). And this would certainly increase the
rate of ring face wear. Fig. 4.10 shows that when ring land width increases due to the
wear down and the ring tension is more or less fixed, the ring would soon lose its oil
control, and the rate of oil consumption would deteriorate very fast.
4.4
Effect of Liner Finish Evolution
It needs to be noticed that all the analysis in the previous sections are based on one liner
finish measurement. In reality, except for the ring face, the liner surface is also being
worn down constantly. And differed from the ring face wear, the liner surface wear is not
evenly distributed on the whole stroke. Close to the top and bottom center of the oil
control ring, the wear would be more severe due to the low sliding speed and therefore
low hydrodynamic lift on the ring face. The mid stroke is found to be usually worn down
much more slowly. In general, the liner surface wear down would also depend on the
engine running conditions. And since the liner wear down is related to all the three rings
and the piston lubrications, the process is very complex and hard to predict.
Since the liner surface profile is actually evolving, and the twin-land oil control ring
(TLOCR) performance is highly based on the liner finish micro geometries, the evolution
of the TLOCR performance in both friction and oil control should be considered carefully.
The current TLOCR model could be easily extended by inputting the liner finish
measurement at different locations of the stroke and different stage of the liner wear
down. The detailed effect of different liner finish measurement on the TLOCR
performance would be discussed and analyzed in the next chapter. Meanwhile, the model
also has a potential in being used to predict the process of liner wear down since all the
asperity contact and oil flow could be well predicted.
5.
Effects of Liner Surface Micro Geometry
In this chapter, the importance of liner surface micro geometry to the twin-land oil
control ring (TLOCR) and liner surface interaction is discussed and certain features of the
liner surface micro topology are explored.
The first section continues the discussion of effect of different liner surfaces on the
TLOCR friction. The following section explores the effects of four detailed liner surface
micro geometry, including the plateau wavelength, the valley distance, the valley depth,
and the cross-hatch angle.
The tests in section one is conducted with the cycle calculation model of TLOCR
introduced in Chapter 3, while the tests in section two is conducted with the deterministic
hydrodynamic method by Li [31] introduced in Chapter2.
5.1
Importance of Liner Finish
The importance of the liner finish to the twin-land oil control ring (TLOCR) and liner
surface interaction has already been shown in the sections of model results (3.5) and
validation of model results (3.7) in Chapter 3. In this section, the discussion in section
3.7 (validation of model results) would be continued to further demonstrate the function
of liner finish in ring pack friction minimization.
Section 3.7 evaluates the total friction of TLOCR with two liner surface measurements,
one very rough and the other very smooth, and compared the numerical results with some
experimental results with the liner surfaces of the similar roughness. The results show
that both the smooth and rough liner surfaces would generate large overall friction under
medium engine speed (1500 rpm), and the smooth surface could generation higher
friction than the rough one under high engine speed (3000 rpm).
However, the overall friction coefficients for both the two surfaces are close to (1500 rpm)
or higher (3000 rpm) than the asperity contact friction coefficient. Obviously, this
indicates a low efficiency for both of these two liner surfaces. In this section, a third liner
surface with the medium roughness is evaluated and its performance in TLOCR friction
is compared to the previous two surfaces. The potential of friction minimization is then
discussed.
2
E
,E
C
0
0
C
2
I%
_4.
r,m
1.5
0
0I
0
U)
=1.
E 0.5
CO
U
G
0.5
1
1.5
2
3
2.5
3.5
Sliding Direction (mm)
Fig.5. 1 Surface Profile of the Medium Rough Liner
The surface with the medium roughness is shown in Fig. 5.1. The comparison of the
roughness statistical parameters between the medium rough surface and the previous two
surfaces are listed in Table 5.1.
Table 5. 1 The Comparison of the Roughness Statistical Parameters [21]
Type of parameter
Smooth Surfaces
Medium Surfaces
Rough Surfaces
Rpk (gm)
0.11
0.17
0.35
Rk ([im)
0.25
0.50
1.53
Rvk (pm)
0.70
1.48
1.11
As in the section 3.7, the unit load pressure is fixed at 15 bar and the ring land width is
chosen to be 0.1mm. The asperity contact friction coefficient is set to be 0.14. The
calculation is run under the engine speed of 1500 rpm as in the experiment and an
additional 3000 rpm for high speed situation.
0.9
Smooth Surface
. - Medium Surface
·-- Rough Surface
0.8
0.7
0.6
...
02. 0'
0
..
.
.
.
.
.
I
I
I
I
I
I
I
I
20
40
60
80
100
120
140
160
180
crank angle (degree)
Fig.5. 2 Comparison of Average Clearance for the Three Surfaces at 1500 rpm
0
Crank Angle (degree)
Fig.5. 3 Comparison of Total Friction for the Three Surfaces at 1500 rpm
1.2
-
1.1
,z
----
---
-
1
--
Smooth Surface
.............
Medium Surface
..........
Rough Surface
0.9
0.8
0.7
S,
0.6
----...
.,.
,.
-
0.5
•
•..
0.4
0.3
0.2
CI
-
20
40
60
80
100 120
crank angle (degree)
140
160
180
Fig.5. 4 Comparison of Average Clearance for the Three Surfaces at 3000 rpm
0
20
40
60
80
100
120
140
160
180
Crank Angle (degree)
Fig.5. 5 Comparison of Total Friction for the Three Surfaces at 3000 rpm
The medium rough surface generates the least total friction for TLOCR under both
medium and high engine speeds as shown in Fig. 5.3 and Fig. 5.5. The TLOCR
clearance for the medium rough surface is not as small as the smooth surface, so that the
hydrodynamic friction is lower. Meanwhile the surface could generate enough amount of
hydro lift to reduce the asperity contact friction in the mid stroke. Consequently, the
medium surface generates the least total friction.
On the oil control side, the medium rough surface leak much lower amount of oil into the
third land than the rough surface, which suggests a better oil control. Comparing to the
smooth surface, it's slightly worse.
Again, it's somewhat arbitrary to judge the surface only by the roughness parameters
listed in Table 5.1, since the surface topology beyond the surface height distribution
would strongly affect both the hydrodynamic pressure generation and the asperity contact
pressure generation as will be discussed in the following section of this chapter.
However, the surface height distribution is still the most dominant factor to the
performance of the liner surface in both friction and oil control. It has been observed that
in general, the rough surface is less capable of generating hydrodynamic pressure to
sustain the ring load, and the asperity contact friction would be consequently high in the
mid stroke. The smooth surface, on the contrary, usually has very low asperity contact
friction but high hydrodynamic shear force. Since the total friction of the TLOCR is the
sum of both asperity contact friction and hydrodynamic shear force, there is obviously
certain potential to choose the right roughness height distribution of the liner surface to
minimize the total friction of the TLOCR.
5.2
Effects of Detailed Liner Surface Micro Geometry
As indicated by the previous section, the liner surface micro topology beyond the surface
height distribution would strongly affect both the hydrodynamic pressure generation and
the asperity contact pressure generation, and thus affect the performance of twin-land oil
control ring liner interaction.
In the section, the effects of four detailed liner surface micro geometry, including the
plateau wavelength, the valley distance, the valley depth, and the cross-hatch angle are
evaluated and discussed using the deterministic hydrodynamic method by Li et al [31].
In order to study the influences of certain liner honing surface geometrical features on the
interaction between the TLOCR land and the cylinder liner, it is necessary to generate
artificial liner honing surfaces numerically. To single out the effect of each certain
feature, the artificial surfaces are designed in this thesis work such that the plateau part of
the surface is composed of two combined sinusoidal wave patterns with a certain crosshatch angle, while the valley area is represented by wedge shape deep grooves equally
spaced on the surface (see Fig. 5.6). The plateau wavelength, and the distance between
the deep grooves are modified to study their effects on the inter-asperity hydrodynamic
pressure generation using Li's model [31]. However, cautions have to be made as the
regularly shaped asperities lack many features existing in the real surfaces.
............
%
..........
00.
600
00
10)
I
09
CO
Sliding Direction (pr)
Fig.5. 6 Artificial Liner Surface
5.2.1 Effect of Plateau Wavelength
As shown in Fig. 5.7, three artificial surfaces with different sinusoidal wavelengths are
tested for their capability of hydrodynamic pressure generation. The surfaces share the
same plateau amplitude and valley area. Therefore they have identical height distribution
functions.
Surface 2
Surface 1
24
-40
C
ru
x (prm)
200
200 U Y (pm)
x (pm)
Surface 3
10
C
:
c
2
- - -surfacel
8
surface2
surface3
6
0
-2
200
00
100
X(pm) 2X0 0
(pjm) =
02
.2
-1.5
-1
0.5
0
0.5
H (pm)
Fig.5. 7 Artificial Surfaces of Different Plateau Wavelengths
However, although the surfaces have the same height distribution, the difference in
plateau wavelength induces a large difference in hydrodynamic pressure generation. Fig.
5.8 shows the hydrodynamic pressure distribution for each of the three surfaces with the
same ring width of 0.3mm and fixed average film thickness, oil viscosity as well as ring
sliding speed. As shown in Fig. 5.8, the plateau wavelength has a significant influence
on the hydrodynamic pressure generation. A larger wavelength, related to a larger size
for each asperity requires larger amount of oil to be pushed around at each blockage, and
this enables a higher average hydrodynamic pressure to be built up. This trend should
also be true for the real liner surfaces with all the randomness involved.
Surface I
Surface 2
---
bar
400
1000
1000
400
400
200
100
0
100
200
i
300
0
100
I-I
200
300
X (pm)
x (pm)
Surface 3
1 25
600
10
400
20
1.
01
200
1
S,
0
0
100
200
300
1
1.5
2
0.5
Normalized Wavelength
4
x (pm)
Fig.5. 8 Hydro Pressure Generation for Artificial Surfaces of Different Plateau
Wavelengths
The size of asperities also influences the asperity contact pressure generation. The
increase of the plateau wavelength would enlarge the asperity radius of curvature.
Consequently it would increase the local asperity contact force for each contact zone
according to the Hertz contact model [41]. However, since the increase of wavelength
also relates to a lower number of asperities per unit area, in average, the nominal asperity
contact pressure drops for a larger plateau wavelength.
Considering the actual oil control ring and liner interaction, the hydrodynamic pressure
and asperity contact pressure are jointly responsible for sustaining the normal load from
the oil control ring tension. The current results imply that increase of plateau asperity
size would enhance hydrodynamic pressure and decrease asperity contact pressure. As a
100
result, larger asperity size would bring less contact and thus lower rate of wear, if not
necessarily the reduction of friction.
5.2.2 Effect of Valley Distance
In order to examine the effect of deep valley distribution on the liner surface, two
important parameters of deep valley are evaluated, the valley distance and the valley
depth. In this sub-section, the effect of the valley distance is discussed. The evaluation of
valley depth would be left for the next sub-section.
Fig. 5.9 shows the patterns of four artificial surfaces with identical plateau profiles, and
different groove densities on the surface. Here groove densities are defined as the ratio of
groove area to the entire surface area, which indicates the availability of the deep grooves
on the liner surface.
Surface I
200'
200
150
150
TvTT
Surface 2
-• T
yyTrIF
A'TLYAY1t¶
*A,'*
2
100
50
"'"::'
:~::"::~:::~~::::i~;::i":-:"'
::~:
~~:'g:I:Z'~"b-~-~~-~~a
""':":
:~::::I::~:::I:
____
____
_____
50
100
150
ItpA
200
50
X ()Sur
Surface 3
100
150
200
02
OA
Surface 4
200
8.6
150
150
0.8
500
100
50
50
I
0
0
50
100
150
200
50
x (pm)
100
150
200
x (Pm)
Fig.5. 9 Artificial Surfaces of Different Groove Densities
101
."
=
x (raM)
'I:
=,
The higher density of deep grooves ensures a more sufficient oil supply to the plateau
area, thus would help the plateau to build up a higher hydrodynamic pressure (see Fig.
5.10). However, the increase of groove density also suggests a reduction of the plateau
area. Since hydrodynamic pressure is usually low in the groove area due to the larger
local clearance, the average hydrodynamic pressure on the entire surface doesn't
necessarily go up with the increase of groove density (also see Fig. 5.10). In the sense of
friction control, the higher groove density implies a lower proportion of plateau area with
high shear stress. Therefore as long as increasing the groove density is not accompanied
with a decrease in hydrodynamic pressure generation, it would certainly be helpful for
hydrodynamic friction reduction.
14
12
-e
S10
CL
8
0
L-
•
Average Hydro Pressure
-- Average Hydro Pressure of Plateau Area
4
o
n
0
II
5
|
I
i
10
15
20
25
Percentage of Groove Area (%)
30
Fig.5. 10 Influences of groove densities on the hydrodynamic pressure generation
5.2.3 Effect of Valley Depth
Another important parameter of the valley is the depth of the grooves. To generalize the
attained conclusions, surfaces measurements of the real liner surfaces are evaluated with
the surface height of their valley part modified.
Fig. 5.11 demonstrates the surface height distribution of a liner surface measurement.
The surface is divided numerically into two parts: the plateau part formed by the final
fine honing process, and the valley part coming from the early honing processes. The
102
plateau part is cut so that the mean value of the heights is equivalent to the median of
them.
Zmean = Zmedian
The rest part is identified as the valley part. In order to study the effect of valley depth,
the valley depth factor A is defined so that the liner surface is modified according to it.
valley = min ( Z ptateau ) +
Zvaiey -min( Z plteau
When the valley depth factor A is equal to 1, the surface is the same with the original
liner measurement, and when A is larger or smaller than 1, the depth of the valley part of
the surface is increased or reduced proportionally (Fig. 5.12).
Valley
· Be--~-------·--
£
·--·
·---
·--
Plateau
ArlCA
Z0UU
ii
2000
1500
1000
500
-
A ~~~_,_~.,~~,._...,...~~.,...
......... ~.~
W&NOMWOMMM"
-LU
-15
-1U
-s
0
5
Zm11ean=Zmedian
Fig.5. 11 Hist Graph of the Surface Height
103
Plateau
I
Valley
,
|
-X
'
-- · h =1.5
=1
I
Fig.5. 12 Definition of Valley Depth Factor A
Five surfaces of different valley depth factors are evaluated with the deterministic
hydrodynamic method by Li et al. [31 ]. The comparisons of the profiles of local
clearance, oil density and hydrodynamic pressure generation are shown in the following
three figures (Fig. 5.13 to Fig. 5.15).
It can be seen that both the profiles of oil density and hydrodynamic pressure
development are maintained similar with five very different valley depth factors. And the
average hydrodynamic pressure development in Fig. 5.16 show very slight differences
despite the eminent difference in the depth of deep valleys shown in Fig. 5.13. The same
phenomenon has been observed in the tests of two other very different liner surface
measurements with their valley part modified accordingly.
On the asperity contact side, since the valley part barely participates in the asperity
contact and the plateau surface area is not affected by the change of valley depth factor,
the asperity contact pressure is irrelevant of the value of the valley depth factor.
The analysis shows that unlike the valley distance, the parameter of valley depth barely
affects either the development of hydrodynamic pressure or asperity contact pressure.
Therefore changing the depth of the valley part would not affect the friction performance
of the twin-land oil control ring, and might increase the oil passage rate since the increase
of valley volume.
104
However, let's not forget another important function of these deep valleys, containing
and transporting the debris of the wear between the piston ring pack and the liner surface.
Reducing the depth of valley might diminish this functionality and thus cause the liner
surface to be more sensitive to wear and scuffing.
%=
0.1
x=2
1=1
X=0.6
X=3
•c10
9
7
0
00
6
5
r-
4
3500
E
U 200
200
2
C
0
C0
0 50 100 0 50 100 0 50 100
0 50 100
0 50 100
Sliding Direction (im)
Fig.5. 13 Clearance Profile of Different Valley Depth Factor A
A=0.1
A=0.6
;L=3
X= 1
1
0.9
'E
0.8
0
0.7
lzý60C
C50
0.6
G0400
0.5
a
3C
0,4
E
1200
2M
0.3
0,2
o I
0.1
00
_n
0
mi
SO
5~
n
•rm'
100
a
ou
su0
Sliding
0
~
50
nr
100
-
rrru
0
50
~ rr
'
100
0
50
·
n3
n
100
Direciton
(pm)
Fig.5. 14 Oil Density Profile of Different Valley Depth Factor A
105
700
700
60O
600
0· i5ax0
500
A=3
X-=2
X=1
=0.6
0.1
1000 bar
700
C
300
400
-
400
400
300
100
300
200
200
100
_ 0
100
0
I
0 50 100
t00 bar
U 50 100 O
50 100
U
wu 1UJ
10 bar
I bar
0 50 100
Sliding Direciton (pm)
Fig.5. 15 Hydrodynamic Pressure Profile of Different Valley Depth Factor A
Sx 10
r
x=0.1. P
--
=4.62E5 pa
X--0.6, PaV=4.55E5 pa
8
A,=1, P
47
Q.
=4.51 Es pa
1=2, P =4.48E5 pa
-=3,
PaV=4.40E5 pa
(0
0o6
C,-
-AA
tý
ILI
"o
100
·
200
·
300
·
400
·
500
Sliding Direction (rm)
Fig.5. 16 Average Hydrodynamic Pressure Trace of Different Valley Depth Factor
A
106
5.2.4 Effect of Cross-hatch Angle
This sub-section includes the study of another important liner finish parameter, the crosshatch angle, also named honing angle.
In the study of cross-hatch angle effect, a 3D liner surface measurement is used. The
surface is stretched and shrunk in both the axial and circumferential directions to simulate
the change of cross-hatch angle.
It is assumed in this paper that the change of cross-hatch angle on a surface equalizes the
change of wavelengths in both the axial (X) and circumferential (Y) directions. Fig. 5.17
shows the influence of the cross-hatch angle on a local surface asperity. It is assumed
that the length scale of the asperity normal to the honing direction is determined only by
the shape of the honing stones, thus is independent of the cross-hatch angle. With two
different cross-hatch angles a and a', the length scales of the related asperity satisfy the
following rules:
d=d',
so that
AX / AX' = cos(a'/2) / cos(a/2),
(1)
AY / AY' = sin(a'/2) / sin(a/2).
(2)
Based on the relations (1) and (2), the cross-hatch angle effect could be correlated from
the X and Y direction wavelength effects. Li's model [31] is used to examine the X and
Y direction wavelength effects separately with a 3D liner surface measurement. Grid
sizes of the surface in each of the X and Y directions are modified to simulate the change
of wavelengths. Fig. 5.18 gives an illustration how changing the grid size on the sample
of a curve could shrink the curve while maintaining the shape of it.
107
*~/
'I
I.,
A
/'
Fig.5. 17 Change of Cross-Hatch Angle
II
I
0.8
0.8
r---- r----r ---r--r-rI III
-- - -
0.2 . .... ... .-
-04-02
I --r----IL-------L..........----*----- --.0.4
-,
-
0.6 - AifIL
0.4
-,
0,4
0.2
rI
I t
LJ
_ IL
11
-0.2
-064
-0.6
-0.48
0
-086
4
8
12
16
20
24
28
X (Wn)
32
.
0 2 4 6 810121416
X(IPm)
Fig.5. 18 Shrink the Curve by Changing the Grid Size of the Measured Sample
The liner surface measurement used in the study of honing angle's effect is shown in Fig.
5.1. Its cross-hatch angle is 40 degrees, and the measurement grid size of both the X and
Y directions is 41•m. According to (1) and (2), the evolutions of grid sizes AX and AY
with the change of cross-hatch angle are plotted in Fig. 5.19. Due to the small value of
108
the original honing angle, the change of cross-hatch angle induces a mild change of grid
size in the sliding direction AX, and a severe change of AY.
E
4)
N
0'
20
30
40
50
60
70
80
Honing Angle (degree)
Fig.5. 19 Grid sizes related to different honing angles
The influences of changing grid sizes on the average hydrodynamic pressure generation
are tested. In the test, the parameters including oil viscosity, ring sliding speed, ring land
axial width and average oil film thickness are fixed. The results are plotted in Fig. 5.20.
The hydrodynamic pressure development shows a similar response to the stretch and
shrink of the surface in X and Y directions. The increase of the grid size in both
directions relates to the increase of the surface wavelength. As discussed in the subsection 5.2.1, the increase of the surface wavelength would enhance the hydrodynamic
pressure development.
109
0_ý
0L.
CA
a0)
V
a)
I-
e4)
o
oi
-2
3
4
5
6
7
8
Grid Size (pIm)
Fig.5. 20 Influences of the wavelengths in X and Y directions on the hydrodynamic
pressure generation
It is assumed that the influences of changing AX and AY are independent of each other.
Combining the results in Fig. 5.19 and Fig. 5.20, the influence of cross-hatch angle on the
hydrodynamic pressure generation is shown in Fig. 5.21. Obviously, the reduction of
honing angle enhances the hydrodynamic pressure generated, thus would help to reduce
the oil control ring friction. Based on this particular liner, a reduction of cross-hatch
angle from 40 degrees to 20 degrees obtains a roughly 10 percent increase in the
hydrodynamic pressure generation. However, since the deep grooves are responsible of
transporting the particles on the liner surface, a reduction in its cross-hatch angle might
undermine this functionality. Therefore a liner with different honing angles in the plateau
and valley area might be beneficial.
110
I-C
41.
u,S4.E
U
E
" 4.;
0o
>I
L_
-31
S3.114
.. _t
20
30
40
50
60
70
80
Honing Angle (degree)
Fig.5. 21 Influence of different honing angles on the hydrodynamic pressure
generation
It should be reminded that here as mentioned previously, the result is obtained based on
certain assumptions on the mechanism of a changing cross-hatch angle. In reality, the
change of cross-hatch angle might induce some more complex fundamental changes in
the surface geometry. These changes together with the effect of cross-hatch angle in
surface wavelength would jointly affect the performance of the liner surface.
111
6.
Conclusions and Future Research
6.1
Twin-land Oil Control Ring Model Development
A lubrication model based on the deterministic method has been developed to examine
the full cycle performance of the twin land oil control ring (TLOCR) liner interaction,
where the liner micro geometries are solely responsible for the hydrodynamic pressure
generation. By decoupling the hydrodynamic and asperity contact parts, and using a
scaling strategy in the hydrodynamic pressure evaluation, the model achieves both the
deterministic accuracy and calculation time efficiency. The model allows one to examine
both detailed local inter-asperity phenomena and whole cycle performance. With the
current approach, one can realistically examine the influence of the liner finish micro
geometries and specific ring design parameters on the oil control ring performance.
The application of the model is fairly straightforward. First, one needs to obtain the
relationship between the hydrodynamic pressure and average clearance by using the
deterministic method by Li, et al. [31] with a representative measured liner surface,
specified TLOCR land width, and the product of oil viscosity and sliding speed as input.
At the same time, one needs to obtain asperity contact pressure and average clearance
relationship. Then, using the proposed method, one can conduct the calculation for the
entire engine cycle with given ring tension, liner temperature distribution, oil
viscosity/temperature relationship at any engine speed. Moreover, the model is flexible
and can be easily extended to consider different liner finish patches at different liner axial
locations.
It is assumed in the model that the ring could well conform to the liner and the ring face
is parallel to the liner nominal surface. Furthermore, the normal load force to be
balanced by the hydrodynamic and asperity contact pressures is assumed to be evenly
distributed along the circumference as well as between the two lands. In reality, these are
not completely true. Piston dynamic tilt can introduce uneven force distribution between
the two lands [38], while the cylinder bore distortion could cause an uneven distribution
of the load force along the circumference. The effects of these certain assumptions on the
model accuracy are discussed. Nonetheless, using a flat surface on the ring side is one
113
plausible way to isolate the effects of geometrical features on the liner. In a well behaved
engine, considering the dynamics of TLOCR and wear evolution of the two lands, the
configuration of a flat surface and rough liner most likely is representative of essential
physics of the interaction between a TLOCR and a rough liner.
Furthermore, some other effects that will affect the correctness and accuracy of the model
including the negligence of the squeeze oil effect and the surface measurement resolution
are discussed and their influences are evaluated.
In the end, the model is used to evaluate and explain some experimental results in piston
ring pack friction, which could not be explained using the traditional method of average
Reynolds' equation.
6.2
Effects of Twin-land Oil Control Ring Design
Several key design parameters of the TLOCR, including ring tension, unit load pressure
and land axial width are studied using the TLOCR lubrication model. The results show
that land axial width, additional to ring-tension and unit-pressure, is important in
controlling the oil flow passing the oil control ring and severity of the asperity contact.
Most importantly, under the same unit pressure reducing land width results in less oil
film thickness and more asperity contact. In real applications optimization of TLOCR
needs to consider the competing effects of ring tension and land axial width in
compromising friction, wear, and oil control. The specifics of liner finish shifts the
balance of these competing effects. The developed tools have potential to help engineers
to reduce the amount of tests and development lead-time in power cylinder system
optimization for friction, wear, and oil consumption.
6.3
Effects of Liner Surface Micro Geometry
Studies have been done to examine the influence of several important liner surface
features on the hydrodynamic pressure generation between the TLOCR land and the liner
surface using Li's deterministic hydrodynamic model [31]. The studied liner features
include the plateau wavelength, the valley distance, the valley depth and the cross-hatch
angle. The results show significant effect of the plateau wavelength as well as the crosshatch angle on hydrodynamic pressure generation. Additionally, the frequency of the
114
deep valleys on the liner finish is found to have an important influence on the
hydrodynamic pressure generation in the plateau area through the oil supply, while the
depth of these deep valleys have little, if any influence on the TLOCR liner interaction.
6.4
Discussion of Probable Future Work
While the models developed could be very useful for the designers of piston ring packs
and liner to help them optimizing the performance of the TLOCR liner interaction, and it
has been used in this thesis work to study some key parameters of liner finish micro
geometry, many important topics in this field of research are still left open.
First of all, in this thesis work only the most common performance parameters such as
friction, average oil passage, and contact are discussed. Many other secondary
performance parameters such as oil film variation along the circumference can be
extracted from the deterministic method and they can be critical to oil transport and
overall oil consumption, which will be left for further research. Furthermore, liner finish
features especially the deep valleys may play a critical role to hide and transport worn
metal debris. In this thesis, only the oil flows are modeled and however worn particles
are transported is a subject that bears other practical significance.
Secondly, in the TLOCR lubrication model, the normal load force in the radial direction
is assumed to be evenly distributed along the circumference as well as between the two
lands. In reality, piston dynamic tilt can introduce uneven force distribution between the
two lands [38], while the cylinder bore distortion could cause an uneven distribution of
the load force along the circumference. The influences of these assumptions have been
evaluated in the thesis, while the details are not considered in the current model, and
would be future topics for model improvement.
Thirdly, due to the wear between the piston ring packs and the liner surface, the
roughness profile of the liner finish is actually changing during the engine operation.
Meanwhile, the wear rate of the liner surface at different axial locations is different since
the change of piston speed. The current TLOCR model has the freedom in inputting the
liner measurement patches after different engine operation time and at different liner
axial locations, and gives a more continuous evaluation of the TLOCR performance.
115
Together with Li's deterministic method [31], the model also has the potential to study
and predict the evolution of the liner surface profile during the engine operation. And
this may help to cast a light in solving the mystery of liner finish break in and wear
processes.
At last, by continuing the study of liner finish surface textures, the researchers might
eventually be able to set up certain guidelines for the liner manufacturing to optimize the
performance of the TLOCR and liner interaction in all the aspects including friction,
duration and oil control.
116
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
Richardson, D. E., 2000 "Review of Power Cylinder Friction for Diesel Engines",
ASME Journal of Engineering for Gas Turbines and Power, Vol. 122, pp 506-519.
Hill, S., H., Sytsma, S., J., 1991 "A Systems Approach to Oil Consumption", SAE
paper 910743.
Yilmaz, E., 2003 "Sources and Characteristics of Oil Consumption in a SparkIgnition Engine", Ph.D. thesis, Massachusetts Institute of technology.
Heywood, J. B., 1998 Internal Combustion Engine Fundamentals,McGraw-Hill.
Mayer, W. J., Lechman, D. C., and Hilden, D. L., 1980 "The Contribution of
Engine Oil to Diesel Exhaust Particulate Emissions", SAE paper 800256.
Drury, C., and Withehouse, S., 1994 "The effect of Lubricant Phosphorus Level
on Exhaust Emissions in a Field Trial of Gasoline Engine Vehicles", SAE paper
940233.
Ueda, F., Sugiyama, S., Arimura, K., Hamaguchi, S., and Akiyama, K., 1994
"Engine Additive Effects on Deactivation of Monolithic Three-Way Catalysts and
Oxygen Sensors", SAE paper 940746.
Wentworth, J., 1982 "Effect of Top Compression Ring Profile on Oil
Consumption and Blowby with the Sealed Ring-Orifice Design", SAE paper
820089.
Fesser, R., 1988 "Small Engine Piston Ring Design for Reduced Oil
Consumption", SAE paper 881327.
Jakobs, R. J., and Westbrooke, K., 1990 "Aspects of Influencing Oil Consumption
in Diesel Engines for Low Emissions", SAE paper 900587.
Yoshida, H., Kusama, K., and Kobayashi, H., 1991 "Diesel Engine Oil
Consumption Depending on Piston Ring Design", SAE paper 911699.
Mihara, K, and Inoue, H., 1995 "Effect of Piston Top Ring Design on Oil
Consumption", SAE paper 950937.
Hill, S. H., Troy, C. K., Brown, J. R., and Hamelink, J. C., 1995 "An
Experimental Study of the Effect of Cylinder Bore Finish on Engine Oil
Consumption", SAE paper 950938.
Hegemeier, T., and Stewart, M., 1993 "Some Effects of Liner Finish on Diesel
Engine Operating Characteristics", SAE paper 930716.
Hill, S. H., 2001 "Cylinder Bore Finish and their Effect on Oil Consumption",
SAE paper 2001-01-3550.
Schneider, E.W., Blossfeld, D.H., Lechman, D.C., Hill, R., Reising, R., and
Brevick, J. E., 1993 "Effect of Cylinder Bore Out-of Roundness on Piston Ring
Rotation and Engine Oil Consumption", SAE paper 930796.
Hitosugi, H., Nagoshi, K., Komada, M., and Furuhama, S., 1996 "Study on
Mechanism of Lubricating Oil Consumption Caused by Cylinder bore
Deformation", SAE paper 960305.
Furuhama, S., Hiruma, M., and Yoshida, H., 1981 "An Increase of Engine Oil
Consumption at High Temperature of Piston and Cylinder", SAE paper 810976.
Orrin, D. S., and Coles, B. W., 1971 "Effects of Engine Oil Composition on Oil
Consumption", SAE paper 710141.
117
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
Stewart, R. M., and Saely, T. W., 1977 "The Relationship between Oil Viscosity
Engine Performance - A Literature Search", SAE paper 770372.
ISO 13565-3:1996/1998, "Geometrical Product Specifications (GPS) -- Surface
texture: Profile method; Surfaces having stratified functional properties -- Part 3:
Height characterization using the material probability curve".
Jocsak, J., Li, Y., Tian, T., and Wong, V., 2006 "Modeling and Optimizing
Honing Texture for Reduced Friction in Internal Combustion Engines, SAE paper
2006-01-0647, Transactions, V. 115, Fuels & Lubricants, SAE World Congress &
Exhibition, Detroit, MI,April 3-6.
Takata, R., Li, Y., Wong, V.W., 2006 "Effect of Liner Surface Texturing on
Ring/Liner Friction in Large-Bore IC Engines", Paper ICEF2006-1525,
Proceedings of ICEF06 ASME Internal Combustion Engine Division 2006 Fall
Technical Conference, Sacramento, California, November 5-8.
Y. Ren Jeng, 1996 "Impact of plateaued surfaces on tribological performance".
Tribol. Perform. 39 (1996), pp. 354-361.
P. Pawlus, 1997 "Change of cylinder surface topography in the initial stage of
engine life". Wear 209 (1997), pp. 69-83.
B.-G. Rosen, R. Ohlsson and T.R. Thomas, 1996 "Wear of cylinder bore micro
topography". Wear 198 (1996), pp.271-279.
Bolander, N. W., Sadeghi, F., and Gerber, G. R., 2005 "Piston Ring Friction
Reduction Through Surface Modification", ASME Internal Combustion Engine
Division 2005 Fall Technical Conference.
Hill, S. H., Brown, J. R., Hamelink, J. C., 1995, "An Experimental Study of the
Effect of Cylinder Bore Finish on Engine Oil Consumption", SAE paper 950938.
Hill, S. H., 2001 "Cylinder Bore Finishes and their Effect on Oil Consumption",
SAE paper 2001-01-3550.
Sato, O., Takiguchi, M., Aihara, T., Seki, Y., Fujimura, K., Tateishi, Y., 2004
"Improvement of Piston Lubrication in a Diesel Engine by Means of Cylinder
Surface Roughness", SAE Paper 2004-01-0604.
Li, Y., Chen, H. and Tian, T., 2008 "A deterministic model for lubricant transport
within complex geometry under sliding contact and its application in the
interaction between the oil control ring and rough liner in internal combustion
engines", SAE International Powertrains, Fuels and Lubricants Congress,
SAE2008-01-1615.
Chen, H., Li, Y. and Tian, Ti., 2008 "A novel approach to model the lubrication
and friction between the twin-land oil control ring and liner with consideration of
micro structure of the liner surface finish in internal combustion engines", SAE
International Powertrains, Fuels and Lubricants Congress, SAE2008-01-1613.
Chen, H. and Tian, T., 2008 "The influences of cylinder liner honing patterns and
oil control ring design parameters on the interaction between the twin-land oil
control ring and the cylinder liner in internal combustion engines", SAE
International Powertrains, Fuels and Lubricants Congress, SAE2008-01-1614.
Bolander, N. W., Sadeghi, F., and Gerber, G. R., 2005 "Piston Ring Friction
Reduction Through Surface Modification", ASME Internal Combustion Engine
Division 2005 Fall Technical Conference.
118
[35]
[36]
[37]
[38]
[39]
[40]
[41]
Elrod, H.G., 1981 "A Cavitation Algorithm", Journal of Lubrication Technology,
Transactions of the ASME, 103 (3), 350-354.
Vijayaraghavan, D., Keith, T.G. Jr., 1989 "Development and Evaluation of a
Cavitation Algorithm", STLE Tribology Transactions, Vol. 32 No.2, pp. 2 2 5 - 3 3 .
Payvar, P., and Salant, R. F., 1992, "A Computational Method for Cavitation in a
Wavy Mechanical Seal," ASME J. Tribol., 114, pp. 199-204.
Tian T. and Wong, V.W., 1998, "Modeling the Lubrication, Dynamics, and
Effects of Piston Dynamic Tilt of Twin-Land Oil Control Rings in IC Engines,"
1998 Fall ASME ICE Technical Conference, September 1998, Clymer, NY.
ASME Transaction, Journal of Engineering for Gas Turbines and Power, Vol. 122,
pp 119-129, January 2000.
Greenwood, J. A. and Tripp, J., 1971 "The Contact of Two Nominally Flat
Surfaces", Proc. Inst. Mech. Engrs., 185, 625-633.
Priest, M., Dowson, D. and Taylor, C.M., 1999 "Predictive wear modeling of
lubricated piston rings in a diesel engine", Wear 231, pp. 89-101.
Modern Tribology Handbook, Vols 1 (2001) (Bhushan B), 621.89 BHU.
119