Anda di halaman 1dari 43

CHAPTER 1

INTRODUCTION
1.1 Background
In Tribology, one studies the design, friction, wear and lubrication of interacting
surfaces in relative motion. Perhaps unexpectedly, this is not a small field of research. In fact,
it includes phenomena that occur in everyday life, from trying to walk on a slippery bathroom
floor, but also high-tech applications like the bearings used in spacecraft. The understanding
of the underlying mechanisms requires a combination of knowledge from fields as varied as
physics, chemistry, materials science, mechanical engineering and mathematics and makes
tribology all the more interesting.
In this thesis, we will restrict ourselves to a small, but important, tribological
phenomenon: Elastohydrodynamic lubrication. The term Elastohydrodynamically lubricated
contacts, or EHL contacts, is used to describe the situation in which two solids are pressed
against each other and the lubricant, present in the gap between the solids, prevents the two
surfaces (the asperities) from touching. In addition, the contact pressures are so large that the
elastic deformation of (one of) the solids, is of the order of the thickness of the lubricant film
or larger. EHL contacts can be found in, e.g. gears and rolling element bearings.
Elastohydrodynamic lubrication, (EHL) is a comparatively recent branch of Tribology
which only became properly established in the early nineteen sixties. It is the mechanism that
describes the separation by a lubricant film between two elastic machine elements loaded
against each other, and in relative motion. In this context, machine elements means two
contiguous bodies whose surface geometries would put them in point or line, light touching
contact. Furthermore, they are not necessarily purpose-built to be supported by a lubricant
film, as are journal or thrust bearings. Examples of machine elements are: a pair of spur gear
involutes teeth, ("line" contact) or a ball bearing ball and its raceway groove ("point"
contact). Within the field of elastohydrodynamic lubrication, two main geometrical
arrangements can be distinguished. In the first arrangement, two cylinders may represent
elements with parallel axes, which contact along a common generator. This configuration is
termed as line contact and is typical of cylindrical rolling element bearings, spur gears and
cams. A second type of contact occurs where the lubricated elements touch at a single point.
In this case, cylinders may represent bearing elements with their axes crossed or by the
contact between a sphere and a plane surface. This arrangement is called point contact and
occurs in spherical rolling element bearings. Elastohydrodynamic Lubrication, hereinafter
referred to as EHL, is a topic which is concerned with understanding and modelling
lubrication problems in which solid metal surfaces deform under large loads. The problems
considered occur most commonly in combustion engines, although the ideas apply to many
other regimes. An engine is comprised of large numbers of individual elements, many of
which are in motion relative to each other. Surfaces will therefore be in contact. Elementary
mechanics demands that when such a motion is occurring then there will be a frictional force
opposing the movement. The friction not only reduces the efficiency of the component, since
work must be done to overcome friction, but also increases the wear. In order to reduce the
frictional force, a lubricant (oil) is applied between the surfaces. This separates the two
contacting surfaces only slightly, but this is enough to stop them impacting upon each other.
Friction is reduced to a tenth of the dry contact (unlubricated) case, and thus the wear is also
dramatically reduced. This situation is called hydrodynamic lubrication. One particular
component of interest is the journal bearings of a car. In this situation a very large pressure is
applied over a very small surface area. Once the pressure exceeds about 0.3 GPa (i.e. 3*10
8
Pascal) the contact behaviour moves from being hydrodynamically lubricated to the
elastohydrodynamic regime. Elastohydrodynamic lubrication (EHL) is different from
hydrodynamic problems because here there is actual deformation of the contacting surfaces.
This may sound unlikely but pressures in a journal bearing or gear commonly reach up to 3
GPa. Assuming pressure is force over area, this would correspond to three elephants
balancing on the end of a pen!
With such a wide range of operating pressures in the contact, it is not difficult to
understand that the properties of the lubricant itself will change across the contact. It is,
however, of great importance to lubricant manufacturers that the oils being developed are as
efficient as possible for the operating conditions for which they are intended. It is therefore
necessary for designers of both lubricants and components to obtain performance results for a
variety of lubricants in different operating conditions. The range of scales in EHL problems is
great. Applied load causes pressure distributions across the contact of the order of giga
Pascal, minimum film thickness are in the micrometer range, and lubricant molecules pass
through the contact in a hundredth of a second. This illustrates how difficult it is to conduct
physical experiments into the behaviour of EHL contacts. That consistent results are
achievable at all is a great accomplishment and a testament to the skills of those people
whose experimental work pioneered the techniques now used.
The success of the lubrication system depends upon the existence of a fluid film,
which has a thickness on the order of a few hundred nanometers or even less. Therefore, it is
important to be able to predict the fluid film thickness along with the other associated
parameters such as pressure distribution, coefficient of friction under a given set of operating
conditions at the design stage. In order to meet the growing demand of industry for highly
advanced technologies, the operating loads are increasing, the fluid films are becoming
thinner and special purpose lubricants are being employed. Moreover, due to increasing
competition in the global market, the emphasis is on lower costs and longer lives of the
components. In view of these facts, it is necessary to develop a better understanding of EHL
as well as to update and simplify the existing methods for analyzing the problems of
elastohydrodynamic lubrication.
1.2 Problem Statement
Elastohydrodynamic lubrication can be defined as a form of hydrodynamic lubrication
where the elastic deformation of the contacting bodies and the changes of viscosity with
pressure play fundamental roles. The influence of elasticity is not limited to second-order
changes in load capacity or friction as described for pivoted pad and journal bearings.
Instead, the deformation of the bodies has to be included in the basic model of
elastohydrodynamic lubrication. The same refers to the changes in viscosity due to pressure.
The existence of elastohydrodynamic lubrication was suspected long before it could be
proved or described using specific scientific concepts. The lubrication mechanisms in
conformal contacts such as those encountered in hydrodynamic and hydrostatic bearings were
well described and defined and the reasons for their effectiveness were well understood.
However, the mechanism of lubrication operating in highly loaded non-conformal contacts,
such as those which are found in gears, rolling contact bearings, cams and tappets, although
effective was poorly understood. The wear rates of these devices were very low which
implied the existence of films sufficiently thick to separate the opposing surfaces, yet this
conclusion was in direct contradiction to the calculated values of hydrodynamic film
thicknesses. The predicted values of film thickness were so low that it was inconceivable for
the contacting surfaces to be separated by a viscous liquid film. In fact, the calculated film
thicknesses suggested that the surfaces were lubricated by films only one molecule in
thickness. In experiments specifically designed to permit only lubrication by monomolecular
films, much higher wear rates and friction coefficients were obtained. This apparent
contradiction between the empirical observation of effective lubrication and the limits of
known lubrication mechanisms could not be explained for a considerable period of time. The
entire problem acquired an aura of mystery and many elaborate experiments and theories
were developed as a result. From the view point of an engineer, the answers to the questions
of what controls the lubrication mechanism and how it can be optimized are very important,
since heavily loaded point contacts are often found, and provision for effective lubrication of
these contacts is critical.
1.3 Research Objective
In this thesis main question we would like to answer is: how do a EHL point contact
problem affect by the lubricant properties? What is the location of the areas of very thin film
thickness that can initiate surface distress and lead to contact fatigue, and what are the sizes
of these films. Ideally, a most desirable gear lubrication analysis should include the following
features: Effects due to variable load, contact radius, and rolling and sliding velocities. We
would like to see the effect of pressure distribution on various lubricant parameters and
dependence on load, speed and on rolling and sliding speed.
1.4 Research Approach
The detailed survey of literature,reveals that a lot of work has been carried out over
the last few decades in order to replace the conventional elastohydrodynamic lubrication
theory by more realistic approach involving several physical effects. Despite these efforts, a
major aspect of practical EHL behaviour still remains largely unexplored. Hence, a numerical
investigation is performed in the present work for Double Newtonian Model.
A proper form of the Reynolds equation is lacking for the shear-thinning of polymer
blended oil that displays a second Newtonian plateau. These multi grade lubricants respond
linearly to shear stress at low stress, and then follow a power-law at intermediate stress
followed by linear response again at higher stress. The power law models, with a second
Newtonian viscosity incorporated in the usual way, can be an accurate description of this
behaviour and are convenient forms for integration into the Reynolds equation. Perturbation
scheme to derive a modified Reynolds equation incorporating the effect of non-Newtonian
lubricant behavior. The Reynolds equation, which governs the generation of pressure in the
lubricated contact, is discretized using finite differences and solved along with the load
balance equation using Newton-Raphson technique.
1.5 Thesis layout
Chapter 1 discuss Introduction to the thesis and overview of Elastohydrodynamic
Lubrication. The motivation and requirement in the field of Elastohydrodynamic Lubrication
is discussed.
Chapter 2 consist of literature review. Past work on the same field is discussed.
Hertzian theory for dry contact is discussed. Reynolds equation which governs the
generation of pressure in lubricating films and forms the foundation of hydrodynamic
lubrication analysis is discussed.
Chapter 3 includes problem formulation. In this chapter Perturbation scheme to derive
a modified Reynolds equation incorporating the effect of non-Newtonian lubricant behavior.
Newton raphson technique is discussed to solve Reynolds equation
Chapter 4 provides the overall solution procedure adopted in the solution of modified
Reynolds equation. All steps including in the solution are provided as per their sequence.
Chapter 5 is related with the results obtained by the solutions. A discussion of effects
due to load, speed on the pressure, film thickness is discussed.
Chapter 6 is conclusion of the report. A conclusion of effect of all variables is
provided. As the EHL is beyond the study limits of this report so some of future aspects are
discussed.
Chapter 7 includes list of the references.
CHAPTER 2
LITERATURE REVIEW
2.1 Introduction
The principle of fluid-film lubrication was established by Osborne Reynolds in 1886.
This work was itself an attempt to explain the results of the experimental work of Beauchamp
Tower which was the first to detect high pressures in the lubricant film. Much progress has
been reported in the analysis, design, manufacture and operation of effective and reliable
plain bearing systems since that time. Spectacular progress has also been made in the past
half century in the field of elastohydrodynamic lubrication, which is concerned with the
understanding of the lubrication mechanism in highly stressed machine components such as
gears, rolling element bearings, cams and seals. The history of studies of elastohydrodynamic
lubrication is outlined briefly and the essential feature of elastohydrodynamic conjunctions is
described.
2.2 Historical Background
Early attempts to analyze the lubrication of gears on the basis of the Reynolds
equation adopted the normal assumptions which had proved to be so successful in relation to
plain bearings. However, the predicted film thicknesses were much smaller than the known
surface roughness of the gear teeth, yet it was known that original machining marks persisted
for long periods of time, suggesting that contact between the teeth was minimal. Martin [2]
solved the Reynolds equation for two rigid right circular cylinders lubricated by an iso-
viscous fluid. A geometrically equivalent cylinder, or paraboloid, near a plane was
considered and this enabled the minimum film thickness to be written as
H = 4.9 ........... (2.1)
Where H is the dimensionless film thickness (H=h/R
x
), U is the dimensionless Speed
Parameter, (U= u/R
x
), and W is the dimensionless Load Parameter (W=(w/L)/R).
The film thicknesses predicted by equation (2.1) were only about 0.01m for realistic
operating conditions. This presented a problem, since such film thicknesses were much
smaller than the surface roughness in spur gears of about 0.40.8 m. The analysis is
nevertheless of some considerable historical significance, since it represented an early
attempt to extend hydrodynamic analysis to very highly loaded situations. Furthermore,
although little progress was made for several decades following Martins study, it did show
that additional factors needed to be considered in the analysis.
The extension of the classical hydrodynamic theory to include the effect of elastic
deformation was first carried out by Peppler (1936, 1938) [3] who determined the maximum
oil film pressure which could occur between a pair of gear teeth lubricated by an iso-viscous
lubricant. It was concluded that the maximum oil film pressure could never exceed the
corresponding maximum Hertzian contact pressure.
Meldahl (1941) [4] investigated the effect of high pressure on film shape and pressure
distribution for a constant viscosity lubricant. The expressions for the surface displacements
of a semi-infinite elastic solid subjected to an arbitrary loading were derived and by coupling
these equations with Reynolds equation, a satisfactory solution to the problem was
determined. However, due to poor convergence and cumbersome calculations involved, only
one result could be produced.
The second extension to the lubrication theory was first considered by Gatcombe
(1945) [5] when he investigated theoretically the effect of viscosity-pressure relationship on
film formation. Gatcombe solved the Reynolds equation for a particular viscosity-pressure
relationship assuming arbitrary boundary conditions. Although an increase in the lubricant
film thickness was reported, it was not sufficient enough to allow formation of satisfactory
lubricant film between gear teeth.
In an investigation of film thickness between teeth of rigid gears, started by Hersey
and Lowdenslager (1950) [6], a parabolic viscosity-pressure relationship was assumed.
Broadly, the change in the theoretical load carrying capacity from the isoviscous prediction
was found to be in consonance with the improvement reported by Gatcombe for an
exponential relationship. These developments attracted the attention of several investigators
leading to a more rigorous analysis of the problem.
In the 1940s and 1950s it was recognized that the very high pressures generated in the
conjunctions between gear teeth might well produce significant local elastic deformations
which would alter the effective geometry of the interacting solids. When this significant
complication was introduced into the analysis the predicted film thicknesses did indeed
increase, but only by a relatively small amount. Likewise, when the substantial influence of
pressure upon the viscosity of lubricants was considered, the predicted film thicknesses rose
to a maximum of 2.3 times the Martin values. This was still much too small an effect to
provide any support for the concept of fluid film lubrication in gears.
The major breakthrough came in the middle of the 20th century, when Grubin (1949) [7]
presented an approximate analytical solution which took account of both elastic deformation
and pressure viscosity effects. Although his solutions did not satisfy the elasticity and
lubrication equations throughout the entire contact, the analysis of the inlet zone proved to be
particularly useful. The resulting film thickness formula (2) opened up a new era of
confidence in the ability of gears to enjoy fluid-film lubrication, since the values predicted
were one or two orders of magnitude greater than those predicted by the Martin equation (1).

11
1
11
8
11
8
95 . 1
L
W
G U
H
(2.2)
It should be noted that this elastohydrodynamic lubrication representation of film thickness in
nominal line contacts involves a new dimensionless group (G =) reflecting the properties
of both the solids () and the viscosity-pressure characteristics of the lubricant (). Grubin
also discussed the distribution of pressure in the contact region qualitatively. It was
concluded that the pressure curve would exhibit a second maximum near the outlet end of the
contact zone. Grubin's work represented the first successful examination of the combined
effects of high pressure on the lubricant and the contacting surfaces. Petrusevich [8] obtained
three numerical solutions to the problem before Dowson and Higginson [9] introduced an
inverse procedure which enabled them to analyze a wider range of conditions. A formula for
minimum film thickness in a line contact lubricated by an incompressible fluid obeying the
Barus pressure viscosity law emerged in the form

13 . 0
7 . 0 6 . 0
6 . 1
L
W
U G
H
(2.3)
The outstanding features of equations (2) and (3) were that they demonstrated:
* The dominant role of the speed parameter.
* The insignificant effect of load upon film thickness.
This equation has been widely used in the analysis and design of systems represented
by line contacts, but as further solutions became available, it was modified to read
13 . 0
7 . 0 54 . 0
65 . 2
W
U G
H (2.4)
The predictions of film thickness arising from equations (3) and (4) were remarkably
consistent with experimental findings from disk machines. This generated confidence, for the
first time, in the analysis of lubrication in many realistic, highly stressed machine elements
which presented essentially line contact geometries. However, it was recognised that many
machine components exhibited point rather than line contact geometries, but the lack of
adequate computing power delayed the generation of such solutions until the 1970s.
Dorr (1954) [10] re-investigated the problem of elastic contacts lubricated by an
isoviscous fluid. He developed an iterative solution procedure in which the pressure
distribution obtained in one stage was used to calculate the surface displacements, which
were added to the initial shape of the contacting solids to produce a slightly different
lubricant film shape. The new film profile was used to re-calculate the pressure distribution
and the process was repeated until the convergence of film shape and pressure profile.
Weber and Saalfeld [11] found out an interesting closed form solution to the
Elastohydrodynamic lubrication problem, which considered both constant and pressure
dependent viscosity. The solution was, however, limited to small deformations and it could
not distinguish between near Hertzian solutions which frequently occur in the practical
situations.
The development of the theory of elastohydrodynamic lubrication required reliable
quantitative experimental data pertaining to lubricating films in heavily loaded elastic
contacts. In this regard, disc machines were built to simulate gear tooth contact and were used
primarily to find the suitability of metals and lubricants for gear operating conditions. A
revolution in disc machine experiments was initiated by Crook (1958) [12] as the direct
measurement of oil film thickness was reported by capacitance and oil flow methods. Crook
found that at low loads the oil film thickness was inversely proportional to the load and
directly proportional to speed as predicted by Martin's theory for rigid cylinders lubricated by
an isoviscous fluid. At higher loads, which are representative of the operating conditions in
gears, the film thickness in pure rolling was found to be almost independent of the load. This
was in consonance with the predictions of Grubin's film thickness formula. Furthermore, the
measured film thickness at high loads was found to be much higher than those predicted by
Martin's theory and the typical value of 1 m quoted by Crook was consistent with the idea
that a continuous oil film may exist between gear teeth.
Crook (1961) [13] further extended the above investigations by considering the
influence of speed and viscosity on film thickness. One of the major conclusions was that the
viscosity of the lubricant at the surface temperature of the discs was of the greatest
importance in the determination of the oil film thickness.
Subsequent investigations by Crook (1963) [14] revealed that rolling friction under
elastohydrodynamic conditions is independent of the load and is proportional to the oil film
thickness. The measurements of effective viscosity led Crook to suggest that the visco-elastic
behaviour of the lubricant may be important.
An alternative experimental approach was described by Sibley and Orcutt (1963) [15]
when presented film thickness measurements based upon X-ray transmission technique. The
film thicknesses so measured, were in agreement with the Crook's measurements and the
current elastohydrodynamic theory. Hence, all these experimental investigations supplied
valuable data useful for the development of elastohydrodynamic lubrication theory.
It is understandable that emphasis was placed upon film thickness prediction in the
first few decades of studies of elastohydrodynamic lubrication. The major concern in the
initial years was to understand how highly stressed machine elements were protected by fluid
films under such severe conditions of operation and to analyse and design systems to take
advantage of the process. This had been achieved in the third quarter of the 20th century and
experimental confirmation of the film thickness predictions was reassuring. It was only in the
last decade or so, when studies of exceptionally thin (nano-metre) elastohydrodynamic films
and sliding, rather than rolling, conditions became the subjects of interest that it became
necessary to re-visit the problem of elastohydrodynamic lubrication, particularly in relation to
point contacts. One of the reasons for this has been the growing emphasis upon efficiency of
lubricated contacts and the need to minimise power losses in machinery.
Dowson and Higginson (1959) [16] presented a new approach to elastohydrodynamic
theory. By introducing a solution of the inverse hydrodynamic lubrication problem, Dowson
and Higginson were able to obtain satisfactory solution of the elastic hydrodynamic equations
after a small number of cycles. The viscosity-pressure relationship for an actual lubricant was
utilized in the analysis and the solution was obtained for a range of loads at a constant speed.
Although most of the characteristics of the solution were consistent with Petrusevish's results,
a noteworthy difference was the absence of the second high pressure peak.
The discrepancy was explained when Dowson and Higginson (1960) [17] reported
further solutions which covered a wide range of loads, speeds and material properties. It was
found that the existence and form of the second peak was a function of all these variables. In
particular, the value of a dimensionless material parameter formed by the exponential
viscosity-pressure coefficient for the lubricant and the effective elastic modulus for the solids
was shown to be influential in the determination of pressure distribution in a heavily loaded
tact.
Dowson and Higginson (1961) [18] analysed their solutions and found that a
convenient dimensionless formula for minimum film thickness could be written in terms of
the load, speed and material parameters. The close agreement between the theoretical
minimum film thickness predictions and the experimental results obtained by Crook and
Sibley demonstrated that the gap between theory and experiments had been largely closed. In
order to show the influence of speed and lubricant compressibility on a theoretical solution to
the elastohydrodynamic lubrication a survey of isothermal solutions was presented by
Dowson et al. (1962)[19]. Speed was shown to be the most important variable in the problem
and the lubricant compressibility was found to have an appreciable influence on pressure
distribution but little effect on minimum film thickness.
Sternlicht et al. (1961) [20] considered the distribution of pressure and temperature in
highly loaded contacts. The introduction of an energy equation considerably increased the
computational work required, but by employing finite difference method and a digital
computer a successful iterative procedure was established.
The improved understanding of highly loaded lubricated contacts called for re-
examination of the mechanism of fatigue failure. The well known type of failure known as
pitting was investigated and the role of lubricant clearly indicated by Dowson (1961,62) [21].
The Hertzian ellipse of dry contact was previously employed as the distribution of surface
stress associated with highly loaded contacts. But the elastohydrodynamic theory showed that
this distribution of pressure may be considerably modified in the presence of a lubricant.
Bell (1961) [22] contributed an interesting analysis of the effects of the non-
Newtonian behaviour by considering the lubrication of rolling surfaces by sinh-law based
rheological model which is popularly known as Ree-Eyring fluid. It was concluded that at
high rates of shear the effect of velocity on film thickness should diminish, whereas that of
load should increase.
Milne (1957) [23], Tanner(1960) [24] and Burton (1960)[25] also studied the effect of
non-Newtonian behaviour of the lubricant in rolling contacts. Once the existence of a
coherent fluid film in some highly loaded contacts had been established, the coefficient of
friction became the subject of several investigations.
Misharin(1958) [26] found that the coefficient of friction between two rollers obeyed
the laws of elastohydrodynamic lubrication theory as developed by Petrusevich. He found
that the friction coefficient increased as the operating temperature was raised and that it
decreased slightly with an increase in either the sliding or rolling velocity.
Greenwood, and Kauzlarich (1973) [27] presented a theoretical analysis for the
temperature rise of oil in the inlet of rollers, and the results were applied to predict the
subsequent film thickness. Comparisons with experimental data support the conclusion that
the thinning of the film thickness below that predicted by isothermal theory is substantially
explained by inlet shear heating of the lubricant. Murch and Wilson (1975) [28] extended this
work and presented a thermal EHL inlet zone analysis. The analysis indicated that viscous
heating in the inlet zone could result in substantial reduction in the lubricant film thickness.
In due course Hamrock and Dowson (1977) [29] generated numerical solutions for
point (elliptical) contacts lubricated by a Newtonian fluid which enabled equations equivalent
to (Eq. 3,4) to be developed. Within the range of conditions considered the following
relationships were found to provide good estimates of central and minimum film thicknesses.
(2.5)
(2.6)
Where
,
_

,
_

,
_

b
a
k
ER
w
W
R
h
H
x x
, ,
2
*
According to this theory an interesting feature of elastohydrodynamic point contacts
was that the regions of minimum film thickness can occur in side lobes rather than at the
outlet on the central axis aligned with the entraining direction. For point contacts,
interferometry, rather than capacitance, was proved to be the most effective experimental
technique for studying the film shape and thickness.
Over some thirty years, equations, like (4), (5) and (6) became established as
reasonable representations for the film thicknesses in elastohydrodynamic conjunctions, even
though it was increasingly recognised that a number of important physical characteristics
such as thermal effects and non-Newtonian behaviour of the lubricant had been omitted from
the analysis. It is quite remarkable that these equations proved to be so useful, in view of the
restrictive assumptions upon which they were based. The reason is, of course, that the flow of
lubricant into the conjunctions, and hence the film thicknesses generated, are determined by
the conditions at entry, where neither non-Newtonian nor thermal actions have an opportunity
to modify the conditions to any significant extent. Hamrock and Dowson extended their
previous work on the theory and a numerical procedure developed by them in an earlier
publication, investigated the influence of lubricant starvation on the minimum film thickness.
This study of lubricant starvation was performed by moving the inlet boundary closer to the
contact center.
Hamrock and Jacobson (1983) [30] developed a procedure for the numerical solution
of the complete isothermal elastohydrodynamic lubrication problem for rectangular contacts.
This procedure calls for the simultaneous solution of the elasticity and Reynolds equations.
The Newtonian fluid model assumes that the shear stress is directly proportional to
the shear strain rate for simplifying the solution of lubrication equation. This results in very
high shear stresses at high shear strain rates, whereas, many of the realistic lubricants exhibit
shear thinning behaviour at high shear strain rates which causes deviation from the
Newtonian values of shear stress. Therefore, the analyses based on Newtonian fluid model, in
general, fail to explain the experimentally measured traction coefficients for various slide to
roll ratios. In order to obtain a fundamental understanding of lubrication performance and
failure in various triboelements, such as rolling element bearings and gears, it is essential to
incorporate the effect of non- Newtonian behaviour of fluid in the numerical scheme. Several
efforts [29-51] have been made to accomplish this task, as discussed below.
In order to understand the effect of the constitutive properties of lubricants, various
rheological models have been investigated under heavily loaded EHL conditions. The sinh-
law based model often attributed to Ree and Eyring (1954) [31] is as follows:

,
_


o
o
dy
du

sinh
(2.7)
Ree-Eyring fluid model has been very popular and used by Hirst and Moore (1974)
[32], Johnson and Tevaarwerk(1977) [33], Berthe et al (1979) [34] and Johnson and
Greenwood (1980) [35]. The full EHL solutions under isothermal conditions with Ree-Eyring
fluid for various slide to roll ratios have been presented by Houpert and Hamrock (1985)
[36], Conry et al (1987) [37] and Chang et al (1988) [38]. Houpert and Hamrock [36]
reported no significant ifference in the film thickness between the Newtonian and non-
Newtonian models for any slide/roll ratio. This is quite unexpected since all previous
calculations for a non-Newtonian lubricant have reported some reduction in the film
thickness. Houpert and Hamrock [36] argued that film thickness is established by
hydrodynamic action in the inlet region alone and since the lubricant in the inlet region
remains Newtonian for the range of speeds and loads considered, the non-Newtonian model
gives identical film thicknesses to those by Newtonian model. Similar results have been
reported by Conry et al [37]. Hence, it is established that for pure rolling condition, the
minimum film thickness should be independent of lubricant rheology.
Bair and Winer (1979) [39] proposed a well known form of non-linear constitutive
equation where the limiting shear strength has been incorporated from the measurements of
EHL experiments as:
( )
1
1


In
t d
d
G dy
du
L
(2.8)
where,
L L
,
=limiting shear strength.
Gecim and Winer (1980) [40] simplified the non-Newtonian fluid model of equation (2.20)
as:
( )

tanh
L
dy
du

(2.9)
Using the above rheological model, Gecim and Winer [40] performed film thickness
calculations and reported up to 40% reduction in film thickness below the corresponding
values computed with Newtonian fluid model in rolling/sliding contacts. The first attempt to
analyse the full solution of a non-Newtonian fluid model with the limiting shear strength in
elastohydrodynamic lubrication was by Jacobson and Hamrock (1984) [41]. Moreover, the
equations (2.8) and (2.9) are difficult to incorporate in the Reynolds equation. Iivonen and
Hamrock (1989) [42] proposed a new and simpler rheological model as:
( ) [ ] 1 1
1

L
dy
du
(2.10)
It was further extended to the following general form [42]:
( ) [ ] 1 1
1

n
L
n
dy
du

(2.11)
The above equation (2.11) is close to equation (2.8) when n is large. Therefore, it is
more difficult to incorporate into Reynolds equation for large values of n. In order to
overcome this problem Lee and Hamrock [43] proposed a more appropriate lubricant
rheological circular model:
( )
2 1
2
1

L
dy
du
(2.12)
A more generalized non-Newtonian fluid model [44, 45] in the form of power law has
been used to represent the rheology of practical lubricants. It is given as:
dy
du
dy
du
n 1
1
]
1


(2.13)
where, n is called as the power law index which is less than unity for shear thinning fluids
and is the reference viscosity. This fluid model has been used by several researchers in
their recent works.
Gecim and Winer (1984) [51] derived the two dimensional, transient temperature
distribution in the vicinity of a small, stationary, circular heat source. The source was
assumed to be acting on the surface of a relatively much larger body, which can be treated as
semiinfinite. Gecim and Winer utilized an integral transform technique in order to solve the
heat conduction equation in cylindrical coordinates. The major assumption was that of high-
speed.
Bair.et al. [52] utilized the viscosity-strain rate relationship obtained from this unique
combination of experimental and simulation data along with high-pressure viscometer
measurements to calculate the viscous traction curve in the elastohydrodynamic lubrication
(EHL) regime. he followed two general methodologies: In the first, the traction curve itself
was regarded as a rheological flow curve and the pressure-averaged rheological properties are
extracted directly from the traction curve. The second method requires that properties and
constitutive behaviour was measured out-of-contact. Traction was then calculated by
integration over the contact area with pressure-dependent properties. Bair was one of the first
who had successfully calculated traction force in EHL from the liquid shear response
obtained from molecular dynamics and rheometry. It is important to recognize that without
the availability of both the simulation and experimental data, leading to the parameterization
of the Carreau curve over a wide range of strain rateincluding the non-Newtonian shear-
thinning regimethis would not have been possible.
Kumar et.al. [53] Investigates the traction behaviour in heavily loaded thermo-
elastohydrodynamic lubrication (EHL) line contacts using the Doolittle free-volume equation,
(developed a so-called free-volume viscosity model based on the physical concept that the
resistance to flow in a liquid depends on the relative volume of molecules present per unit
free volume) which closely represents the experimental viscosity-pressure-temperature
relationship and has recently gained attention in the field of EHL, along with Taits equation
of state for compressibility. The well-established Carreau viscosity model has been used to
describe the simple shear-thinning encountered in EHL. The simulation results have been
used to develop an approximate equation for traction coefficient as a function of operating
conditions and material properties. This equation successfully captures the decreasing trend
with increasing slide to roll ratio caused by the thermal effect. The traction-slip
characteristics are expected to be influenced by the limiting shear stress and pressure
dependence of lubricant thermal conductivity, which need to be incorporated in the future.
The Eyring sinh law is presently the most well-accepted model for shear-thinning of
EHD lubricants at high pressure. The Eyring found that a sinh-law could describe, in a
limited way, the stress-increasing portion of thixotropic response of a lubricant at elevated
pressure. Then, it is extremely important that these models receive a critical review using data
obtained by modern methods to evaluate the actual Eyring models and show how they differ
from the model in use today. Bair [54] through his work validated the applicability of the
Hahn-Eyring model for thixotropy and the Ree-Eyring model for shear-thinning. Data from
high-pressure viscometers were used to validate the actual Eyring models. The Ree- Eyring
model for shear-thinning obeys time-temperature-pressure superposition. A film thickness
calculation shows it to suffer from the same anomalous behaviour in sliding contact as does
the sinh law. Ree-Eyring law can also be used to describe the shear thinning behaviour of
fluid by using a multiple flow units, each with a unique Eyring stress. Then the Ree-
Eyring model is essentially the same over the range of available measurements as any of the
conventional models, such as Carreau, that naturally incorporate power-law response. But the
Ree-Eyring model has the disadvantages of requiring many more parameters and of being
useless for extrapolation to higher shear rate.
Kumar and khonsari [55] provided that The Eyring sinh law, used to describe
rheological shear thinning properties of a non-newtonian EHL lubricants, fails to replicate
the experimentally measured flow curves for shear-thinning lubricants. Interestingly, this law
was rejected by Eyring for shear-thinning fluids and, in fact, it was found useful only for
fluids thought to exhibit thixotropy. Kumar et al. suggested the actual ReeEyring model
for shear-thinning involves multiple flow units with appropriate relaxation times. He
provided an extensive set of full EHL line and point contact simulations to investigate the
usefulness of the actual ReeEyring model in EHL applications with shear-thinning
lubricants. Kumar et al. found that actual ReeEyring model predicted the results much closer
to experimental data. Also presented is the application of an appropriate shifting rule
expected to improve the agreement between simulations and experiments.
( )

N
i
i i i
x
1
1
sinh
(2.14)
Bair et.al (1993) [56] studied the lubricant rheological properties at high pressure.
They studied the effect of shear stress for various liquid lubricants by use of various
rheometers. Bair propose an improved relation for the variation of viscosity through the glass
transition, and present refined relations for shear modulus and limiting shear stress. They
reported Newtonian behavior in two ranges of shear stress - at low shear stress a first
Newtonian associated with the viscosity of the blend, and above about 10
4
Pa shear stress, a
transition to a second Newtonian for which viscosity was again rate independent, but with a
value reduced to nearly that of the mineral oil solvent.
Bair et.al.(2004) [57] had provided work on lubricant shear thinning properties. He
provided the conclusive demonstration that the nature of the shear thinning, that affects both
film thickness and traction in EHL contacts, follows the ordinary power-law rule that has
been described by many empirical models of which Carreau is but one example. This was
accomplished by accurate measurements in viscometers of the shear response of a PAO
(poly-alpha-olefin) that possesses a very low critical stress for shear-thinning and accurate
measurements in-contact of film thickness and traction under conditions which accentuate the
shear-thinning effect. The in-contact central film thickness and traction were entirely
predictable from the rheological properties obtained from viscometers using simple
calculations. In addition, a new model has been introduced that may be useful for the
rheological characterization of mixtures.

,
_

1
1
]
1

,
_

+
N
i
n
i
i
i
G
x
1
2
1
1
2
1


(2.15)
Table 2.1
( ) & ( )

Newtonian
1

n

Ostwald-de waele
1 ,
1 ,
1



n
Spriggs
n
+

+
1
2
2
1

Cross
( ) [ ]
2
1
2
2
2
1
n
+

Carreau
( ) [ ]
a
n
a

+

+
1
2
2
1

Carreau-yasuda
( )
1 , 1
sinh
1
1
1
>

N f
f
N
i
i
N
i
i
i
i

Ree-Eyring

Newtonian
n
G
1
1



Ostwald-deWaele
1 ,
1 ,
1
1

G G
G
n

Sprigs

,
_

+
1
1
2
2
1
n
G



Ellis
G

+
1
2
2
Ferry
2
2
2
1
,
_

+
G



Rabinowitsch
a
n
G
a 1
1
1
2
2

1
1
]
1

,
_



Bair
2.2 Governing Equations
2.2.1 Reynolds equation
The equation which governs the generation of pressure in lubricating films is known
as Reynolds equation and it forms the foundation of hydrodynamic lubrication analysis. It
was derived for a Newtonian fluid by neglecting the effects due to curvature of the fluid film.
This assumption is well justified as the effective radius of bearing components is generally
very large compared with the film thickness. This enables the analysis to consider an
equivalent curved surface near a plane.
The derivation of Reynolds equation involves the application of the basic equations of
motion and continuity to the lubricant. The full equations of motion for a Newtonian fluid in
Cartesian coordinates are:

,
_

+
,
_


y
v
x
u
x z
w
x
u
x x
p
X
dt
du

3
2
3
2

,
_

,
_

+
x
w
z
u
z x
v
y
u
y

(2.15 a)

,
_

,
_


z
w
y
v
x x
u
y
v
y y
p
Y
dt
dv

3
2
3
2

,
_

,
_

+
y
u
x
v
x y
w
z
v
z

(2.15 b)

,
_

,
_


y
v
z
w
z x
u
z
w
z z
p
Z
dt
dw

3
2
3
2

,
_

,
_

+
z
v
y
w
y z
u
x
w
x

(2.15 c)
The terms on the left hand side represent inertia effects and on the right hand side are
the body force, pressure and viscous terms in that order. The inertia and body forces are
negligible as compared to the viscous and inertia forces.
The equation of continuity representing conservation of mass is:
0

z
w
y
v
x
u
t

(2.16)
The above Equations 2.l5a, 2.15b, 2.l5c and 2.16, neglecting the relatively small
valued terms like inertia and body force terms, pressure gradient across the film and
introducing the values of the boundary velocities the following form of the Reynolds
equation is obtained.
( ) phu
x z
p h
z x
p h
x

,
_

,
_

12 12
3 3
(2.17)
where
( )
2
2 1
u u
u
+

.
The second term in the above Equation (2.17) corresponds to side-leakage which
refers to flow along z-direction. On neglecting the side-leakage term:
( ) phu
x x
p h
x

,
_

12
3
(2.18)
This equation is integrated to yield the familiar integrated form of Reynolds equation:
( ) ( )
3
12
e
dp u
h h
dx h

1
]
(2.19)
Where the subscript e refers to the conditions at the point where the pressure gradient dp/dx
vanishes.
2.2.2 Elasticity equations
The lubricated contacting surfaces undergo elastic deformation due to which the
pressure distribution and film shape are differing from the prediction of hydrodynamic
lubrication theory. It is, therefore, important to evaluate the elastic displacements of the
contacting surfaces. In the case of a line contact the pressure zone is very small compared
with the radius of the solids. So the displacements are uniform along the length and the solids
are in a condition of plane strain. The contact problem is usually simplified by calculating the
displacements for a semi-infinite flat solid and adding to the curved surface of the roller.
The calculation begins from the Boussinesq function for the stresses due to a normal
line load on the surface of a semi-infinite solid. The stresses and displacements under this line
load can be integrated over the entire contact zone to give the corresponding quantities for a
distributed pressure. The stress components in Cartesian co-ordinates are given as follows:
( )
2
2 2
2
2
y x
y Px
x
+

( )
2
2 2
3
2
y x
Py
y
+

( )
2
2 2
2
2
y x
Pxy
xy
+

(2.20)
where P is the normal load per unit length
Using the relation between strain and stress components given by Hooke's law, the
displacements u and v along x and y directions respectively, are obtained as follows:
( ) ( )
( ) y f dx
E
v v
dx
E
v
u
y x
+
+



1 1
2
(2.21a)
( ) ( )
( ) x g dy
E
v v
dy
E
v
v
x y
+
+



1 1
2
(2.21b)
where v is the Poisson's ratio
Using the above equations, the surface normal displacement, v , due to a pressure
distribution p(s) is given by
( )
( )
( ) ( )

2
1
constant
1
2
2
s
s
ds s x In s p
E
v
x v

(2.22)
where s
1
and s
2
mark the boundaries of the pressure zone
CHAPTER 3
ANALYSIS
3.1 Introduction
The contacts in mechanical components, such as gears, roller bearings and cams are
simulated by EHL line contacts. The extensive review of the relevant literature, presented in
Chapter 2, reveals numerous efforts to include realistic physical effects in EHL
investigations. Despite these attempts, many important issues related to EHL line contacts
need the attention of researchers. Therefore, the present study attempts to investigate the
effects of starvation in EHL contacts. This chapter describes the mathematical model
employed in the present study. Therefore the present study attempts to investigate full EHL
Simulations using double Newtonian model. This chapter describe the mathematical model
employed in present study.
3.2 Problem Formulation
The analysis of elastohydrodynamic lubrication of line contacts, as shown in Fig 3.1
involves the simultaneous solution of Reynolds equation, elasticity equation and load
equilibrium equation subject to appropriate boundary conditions with due consideration to the
variation of the operating conditions i.e. load , speed and slide to roll ratio. Therefore, the
effect on dimensionless minimum film thickness, dimensionless central film thickness and
coefficient of friction is studied by the help of these equations.
3.2.1 Rheological Model of Lubricant
An early validation of a classical Newtonian EHD film thickness formula found it to
be accurate for mineral oils but not to be accurate for mineral oil/polymer blends and some
synthetic oils. It was assumed that shear-thinning was responsible for films being thinner than
predicted by Newtonian theory. A recent calculation based upon the measured shear response
of lubricants concludes that film thinning is due to a power law type shear thinning behaviour
of lubricant.
A second Newtonian regime with a viscosity equal to
2
often appears in
measurements. The parameter, n, is the power-law exponent and is the slope on a loglog plot
of stress versus rate of the shear-thinning regime when the low shear viscosity, >>
2
. The
parameter, a, controls the breadth of the transition between Newtonian and power-law
regimes.
(3.1)
Where is strain rate,
2
is second Newtonian viscosity and

is the low shear viscosity.


Fig 3.1 Contact geometry and co-ordinate axes
3.2.2 Velocity and Velocity Gradient
The fluid velocity distribution along the rolling direction is given below in
dimensional form
( ) ( )
x
p hy y
y
h
u u
u u
a b
a

+
2
2
(3.2)
The velocity gradient along the film is given by:
( ) ( )
x
p h y
h
u u
y
u
a b

2
2
(3.3)
where

is the effective lubricant viscosity derived using perturbation method described


subsequently.
3.2.3 Reynolds equation
The fluid pressure at the inlet as well as the outlet of an EHL conjunction is equal to
the ambient pressure [57, 58]. In addition, the decline of pressure to the ambient value at the
outlet is gradual, which implies that the pressure gradient at the outlet approaches zero.
Therefore, the Reynolds equation is solved for the pressure distribution subject to the
following boundary conditions: Based upon the assumptions used by Dien and Elrod [61],
several researchers use a perturbation scheme to derive a modified Reynolds equation
incorporating the effect of non-Newtonian lubricant behavior. The basic methodology used in
a perturbation scheme is shown below for the case of a line contact. The scheme begins with
the introduction of equivalent viscosity,
e

, which is given by
I
e
/
(3.4)
where
I u y /
. Now, velocity u is expanded in terms of

, which is a small
non-dimensional amplitude parameter:
u u u +
0 1

(3.5)
Then
` 1 0
I I I +
(3.6)
where
I
u
y
I
u
y
0
0
1
1

,
(3.7)
Expanding the equivalent viscosity
e

in the region near


I
0
into a Taylor series:
1 0
+
e

(3.8)
where
( )
0 0
I
e

and
0
1 1
I
e
I
I

,
_


(3.9)
The momentum equation is:

y
p
x

(3.10)
Using Equations (3.1), (3.3), (3.5) and neglecting
2
:
( ) ( ) ( )
` 1 0 0 1 0 0 ` 1 0 1 0
I I I I I + + + +

(3.11)
Expanding
p
,
p p + 0
(3.12)
Substituting Equations (3.11) and (3.12) in Equation (3.10):
( )
x
p
y
I I
y
I

+
+



` 1 0 0 1 0
0
(3.13)
0
2
0
2
0

y
u
(3.14)
and
( )
x
p
y
I I

+
` 1 0 0 1


(3.15)
Integrating Equation (3.11) under the boundary conditions,
a
u u
0
at
0 y
and
b
u u
0
at
h y
:
( )
y
h
u u
u u
a b
a

+
0
(3.16)
where
a
u
,
b
u
are the velocities of the lower and upper surfaces respectively and h is the
film thickness. Substituting Equation (3.9) into Equation (3.15) gives
x
p
y
u

2
1
2

(3.17)
where,

,
_

,
_


0 0
0

I
e
I
I

(3.18)
Integrating Equation (3.14) under the boundary conditions,
0
1
u
at
0 y
and
b
u 0
1

at
h y
:
( )
x
p hy y
u

2
2
1
(3.19)
From Equations (3.5), (3.12), (3.16) and (3.18)
( ) ( )
x
p hy y
y
h
u u
u u
a b
a

+
2
2
(3.20)
( ) ( )
x
p h y
h
u u
y
u
a b

2
2
(3.21)
Applying the above perturbation scheme to double Newtonian shear-thinning model:
( )
( )
2
1
2
1
2 1 2
1 /

1
1
]
1

,
_

+ +
n
cr
e
G



(3.22)
( ) ( )
( )
2
3
2
1
2 1 2
2
1
1 1

1
1
]
1

,
_

,
_


n
cr cr
e
G
I
G
I
n
I


(3.23)
( ) ( )
( )
( )
( )
2
1
2
0 1
2 1 2
2
3
2
0 1
2 1
2
0 1
0 0
1 1 1
0

1
1
]
1

,
_

+ + +
1
1
]
1

,
_

,
_

,
_

,
_


n
cr
n
cr cr
I
e
G
I
G
I
G
I
n
I
I


(3.24)

( )
( )
1
1
]
1

,
_

+
1
1
]
1

,
_

+ +

2
0 1
2
3
2
0 1
2 1 2
1 1
cr
n
cr
G
I
n
G
I

Therefore, modified Reynolds equation is:
( ) h
x
u
x p h
x
o

,
_

1
3
12
/

(3.25)
( )
( )
1
1
]
1

,
_

+
1
1
]
1

,
_

+ +

2
0 1
2
3
2
0 1
1 2 1 2
1 1 / 1 /
cr
n
cr
G
I
n
G
I

(3.26)
In dimensionless form:
( )
( )
1
1
]
1

,
_


+
1
1
]
1

,
_


+ +

2
1
2
3
2
1
1 2 1 2
8
1
8
1 / 1 /
cr
n
cr
WHG
E SU
n
WHG
E SU

(3.27)
( )
( )
( ) ( )
1
]
1

1
1
1
]
1

1
1
]
1

,
_

+ +

x
p h y
h
u u
G
a b
n
cr



2
2
1
2
1
2
1
2 1 2

(3.28)
( ) ( )
( )
( )
( )
1
]
1

1
1
1
]
1

1
1
]
1

,
_

'

+ +

x
p h
h
u u
x
p h
h
u u
G
h y
a b
n
a b
cr


2 2
1
2
1
2
1
2 1 2
(3.29)
In dimensionless form:
( ) ( )
( )
1
]
1

+
1
1
1
]
1

1
1
]
1

,
_

+ +

X
P WH
WH
SU
X
P
G
E WH
WHG
E SU
h y
n
cr cr





8 8
1 / 1 /
1
2
1
2
1
1 2 1 2
(3.30)
Inlet boundary condition
P=0 at X=X
in
(3.31)
Outlet boundary condition
0
0 X X at
X
P
P

(3.32)
3.2.4 Finite difference formulation
The Reynolds equation is discretized by using a finite differencing scheme to obtain
the equations
0
i
f
(2 to N) as follows:
( ) ( ) [ ]
X
H H
K
X
p P
X
p P
f
i i i i
i
i i
i i

+
+
1
2
1
2 / 1
2
1
2 / 1


(3.33)
where
i
i
H

,
_

3

3.2.5 Boundary conditions
Since the first node lies at X = X
in
, P
1
is kept fixed at 0 in order to satisfy the inlet
boundary condition imposed by equation (3.29). The outlet boundary coordinate X
o
is
determined by following a procedure in which Xo is initially set equal to the maximum value
of X -coordinate, X
max
, used in the numerical simulation. The new values of pressure are
calculated. X
N
is then located such that P < 0 for X > X
N
, where X
N
represents the X-coordinate
of the node nearest to X
o
and X
N
< X
o
. P
N-1
, where P
N
and P
N+1
are used to define a second
degree polynomial P(X) as follows:
( ) C BX AX X P + +
2
(3.34)
where A, B, C are unknown constants to be determined by solving the following equations:

1 1
2
1
+ +
N N N
P C BX AX

N N N
P C BX AX + +
2
1 1
2
1 + + +
+ +
N N N
P C BX AX (3.35)
The solution of the above equations gives

( )
( )
2
1 1
2
2
X
P P P
A
N N N



( )
( )
( )
( )
2
1 1 1 1
2
2
X
P P P
X
X
P P
B
N N N
N
N N

+ +
( )
( )
( )
( )
N
N N
N
N N N
N
P
X
P P
X
X
P P P
X C +

+ +
2
2
2
1 1
2
1 1 2
(3.36) X
0
is
calculated such that
( ) 0
0
X P
,
0
0
2
0
+ + C BX AX (3.37)
The new value of X
N
is given by the X-coordinate of the node nearest to the outlet
boundary defined by the above equation (3.35) such that X
N
and X
0.
After a few iterations X
0

and X
N
attain steady values.
3.2.6 Film thickness equation
The film thickness, h, at any point in a rough EHL conjunction is:
v
R
x
h h + +
2
2
0
(3.38)
Where h
0
is the offset film thickness, v contributes to the surface normal displacement and
the method adopted for its calculation is discussed subsequently.
The film thickness in non-dimensional form is given by:
( ) v
X
H X H + +
2
2
0
(3.39)
where,
2
b vR v is the non-dimensional surface displacement.
3.2.7 Elastic deformation
The elasticity integral is expressed in non-dimensional form [50] as follows:
( )


0
2
2
1
X
X
in
dS S X In P v

(3.40)
The above equation is represented in discrete form as follows:

N
j
j ij i
P D v
1

(3.41)
where, ij
D
is the influence coefficient at node i due to a unit pressure at node j.
Venner et al [57] used a simple expression for influence coefficients for evaluation of
elastic deformation in EHL line contacts. This expression, as given below, is applicable when
a uniform mesh is used for discretization:

'


,
_

,
_

+ 1
2
1
2
1 1
X j i In X j i D
ij


1
]
1

'


,
_

,
_

1
2
1
2
1
X j i In X j i
(3.42)
where X is the grid size of the uniform mesh used.
The non-dimensional elastic deformation computed from above equation (3.40) is
used for the calculation of fluid film thickness, which is required for the calculation of
pressures.
3.2.8 Density-Pressure Relationship
The present analysis uses Dowson and Higginson [3] density-pressure relationship for
the lubricants in the dimensionless form.

,
_

h
h
p P
p P
. 10 7 . 1 1
. 10 6 . 0
1
9
9
(3.43)
3.2.9 Viscosity-Pressure Relationship
The present analysis uses the following relation for viscosity in the dimensionless for
( ) ( )
{ }
9
0
exp 9.67 1 1 5.1 10 .
z
h
In P p
1
+ + +
1
]
(3.44)
3.2.10 Load Equilibrium Equation
The pressure developed in the lubricant supports the applied load. Therefore, the
pressure distribution obtained from the Reynolds equation should satisfy the following
condition
w pdx
a
i
x
x

(3.45)
where w is the applied load per unit width. The above equation is expressed in non-
dimensional form as follows
2

a
i
x
x
pdX
(3.46)
The integral in equation (3.44) is calculated using Simpson's rule and it can be written
in the following form:
0
2
2

N
j
j j
P C W
(3.47)
where,

'

... 7 , 5 , 3 3 2
... 6 , 4 , 2 3 4
1 3
j X
j X
j X
C
j
(3.48)
3.2.11 Newton-Raphson Formulation
The simultaneous system of N equations by discretized Reynolds equation (3.30) and
discretized load equilibrium equation (3.47) are solved using the Newton-Raphson technique.
The N system unknowns are
N
P P P P ...., , , ,
4 3 2
and H
0
. The matrix equation of this
system is
1
1
1
1
1
1
1
]
1


1
1
1
1
1
1
1
]
1

1
1
1
1
1
1
1
1
1
1
]
1

0
3
2
0
3
2
3 2
0 3 2
0
3 3
3
3
2
3
0
2 2
3
2
2
2
0
W
f
f
f
H
P
P
P
C C C
H
f
P
f
P
f
P
f
H
f
P
f
P
f
P
f
H
f
P
f
P
f
P
f
N N
N
N
N
N N N
N
N

(3.49)
The elements of the Jacobian matrix
[ ] J
are calculated as described in Appendix 1.
3.2.12 Coefficient of friction
The coefficient of friction is taken as the ratio of total shear force of the lubricant at
the surface boundaries and the applied normal load as expressed by the following equation
( ) 0
o
i
x
x
y dx
COF
w

(3.50)
CHAPTER 4
SOLUTION PROCEDURE
4.1 Introduction
The preceding chapter describes the governing equations for the analysis of the
double Newtonian model for shear thinning lubricants in elastohydrodynamic lubrication. A
computer program is developed and the solution procedure is adopted for the numerical
implementation of the governing equations, are described in this chapter.
The primary aim of the computer program is to determine the pressure distribution
and fluid profile within the EHL contact. Therefore, pressure, minimum film thickness,
central film thickness and coefficient of friction are the primary variables which are
calculated by using the secondary variables namely fluid viscosity, density and elastic
deformation. It is apparent from the nature of the governing equations that the above
mentioned primary and secondary variables are interdependent. For example, pressure is
calculated from Reynolds equation by using film thickness, viscosity and density; the film
thickness depends upon elastic deformation which is calculated by using the pressure
distribution.
4.2 Overall Solution Procedure
The steps involved in the overall solution scheme are given below
1. The pressure distribution [P], offset film thickness H
o
and outlet boundary co-ordinate X
o
are initialized to some reference values.
2. The current distribution is used to calculate surface displacements [ v ] using equation
(3.41).
3. The surface displacement [ v ] are used along with the offset film thickness H
o
to evaluate
the fluid film thickness, at every node by using film thickness equation (3.38)
4. The fluid density ( ) and viscosity ( are calculated using equation (3.43) and (3.44)
respectively.
5. This step executes the calculation of viscosity modification factor

given by equation
(3.33).
6. The residual vector [f ] is calculated from equation (3.33).

7. The residual vector W is calculated from the load equilibrium equation (3.47).
8. The residual vectors calculated in the steps 6 and 7 are assembled in a single vector [F]
to facilitate execution of Newton-Raphson scheme.
9. This is followed by computation of Jacobian coefficients, as described in Appendix-1, to
get the Jacobian matrix [J].
10. The corrections to the system variables, specified by vectors on the left hand side of
equation (3.49) are computed by inverting the Jacobin matrix using Gauss elimination.
11. The corrections calculated in step 10, are added to the corresponding system variables to
get the new values of pressure distribution and offset film thickness.
12. The outlet boundary co-ordinate X
o
is corrected by using equations (3.31) and (3.32).
13. The X-coordinate of the node nearest to the outlet boundary is set equal to X
N
such that
X
N
< X
o
. This decides the number of nodes N in the solution domain.
14. The pressure gradients along the fluid film are calculated. The termination of the iterative
loops in the EHL analysis of line contacts requires the fulfilment of the predefined
convergence criteria to arrive at an accurate solution. In order to check the convergence
of the pressure distribution, the sum of the nodal pressures corresponding to the current
iteration (say nth) is calculated. If the fractional difference between this value and that
corresponding to the previous iteration is less than the prescribed tolerance TOL, the
pressure distribution is assumed to have converged. Thus,
TOL
P
P P
n
N
i
i
n
N
i
i
n
N
i
i

1
]
1

1
]
1

1
]
1


1
1
1
1 1
(4.1)
The offset film thickness is assumed to converge if the fractional change in its value
becomes less than the prescribed tolerance in successive iterations
[ ] [ ]
TOL
H
H H
n
o
n n o

1
1 0
(4.2)
The value of TOL adopted in the analysis is
4
10 1

as it has been found that a lower
value does not contribute to improve the accuracy of the solution. The iterative loop
terminates and the current values are considered as the final solution only if all the
relevant convergence criteria are satisfied simultaneously.
15. If any one or more of the relevant criteria are not satisfied, the next iteration begins and
the control is shifted back to the step 2.
Chapter 5
RESULTS AND DISCUSSION
5.1 Introduction
The present chapter evaluates the results and simulation of EHL with a double
Newtonian model to describe the shear thinning effect on lubricant. The solution for the EHL
characteristics is being done by the steps suggested in the previous chapter.
S
Sr.
No.
Input Parameters Output Characteristics
1 Dimensionless load (W=2x10
-5
to 5x10
-
4
)
Dimensionless central Film Thickness
2 Dimensionless speed (U=5x10
-12
to
5x10
-10
)
Film Thickness Distribution
3 Slide to roll ratio(0 to1) Pressure Distribution
4 Double Newtonian model parameter :-
power law index(n), G
cr
,
2
/
1
Coefficient of friction (COF)
Table 5.1 Output characteristics to be studied
5.1 Effect of viscosity ratio:-
Figure (5.1) compares the variations of generalized Newtonian viscosity () with
shear-rate obtained using actual double-Newtonian model. Graphs are plotted with four
different values of second Newtonian viscosity. From the figure it can be shown that for
viscosity ratio (
2
/
1
) equals to 0.8 shows least shear thinning property and curve 2
nd
with
viscosity ratio(
2
/
1
) equals to 0.2 shows maximum shear thinning effect.
Figure 5.2 shows the variation of pressure with respect to X. Pressure spikes shows
approximately similar type of behaviour. Graph is shown with four different values of
combinations of viscosity ratio (
2
/
1
) and velocity. As we approach towards positive values
2
nd
plot (
2
/
1
=0.4) shows largest pressure but abrupt spike is nearly absent in this. 4
th
curve
(
2
/
1
=0.8) shows highest pressure spike relative to other three.
Figure 5.3 predicts the distribution of dimensionless film thickness (H) with respect to
X. The graphs are plotted for different combinations of dimensionless speed parameter (u)
and viscosity ratio (
2
/
1
). The value of S is taken as unity, power law index (n) =0.5,
dimensionless load parameter is constant and taken as W=1.310
-4
. Figure shows that value
of H decreases sharply at inlet of contact zone and also slight variation is shown at outlet. The
value of H slightly moves up at outlet of contact zone. Comparatively for larger value of
dimensionless speed parameter the value of dimensionless film thickness (H) is more.
Between the two factors the velocity is more pronounced factor to effect shear thinning
property. For relatively at higher value of viscosity ratio the curve shows similar nature but
higher value of H. 2
nd
curve with low velocity depicts the character of relatively minimum
film thickness. We can show that at lower value of velocity parameter lubricant shows more
shear thinning property than lower value of viscosity ratio at higher speed. All rest curves are
plotted with same velocity and different values of viscosity ratio (
2
/
1
). Graph depicts that
for larger viscosity a relatively higher value of film thickness is obtained. Similar to figure
(5.1) at same value of velocity parameter the shear thinning property is more pronounced in
the lower viscosity ratio. At the exit of contact region film thickness variation moves up
slightly.
Figure 5.4 compares the central film thickness (h
c
/R) with respect to dimensionless
speed parameter (U). The value of slide to roll ratio is taken as unity. The value of power law
index n=0.5, value of critical shear stress G
cr
=310
4
Pa, value of dimensionless load
parameter is taken as W=1.310
-4
. Logarithmic graph shows that h
c
/R increase linearly to
speed parameter U. h
c
/R increase with increase in value of U. The graphs are plotted for
different values of viscosity ratio (
2
/
1
). When value of viscosity ratio is more than zero than
the slope of variation is comparatively more. For value
2
/
1
=0.2,0.4,0.8 the slope is nearly
equivalent but initial value of h
c
/R is more at higher value of
2
/
1
. These graphs in
consonance with the figure (5.1) shows that 3
rd
curve (
2
/
1
=0.8) shows least shear thinning
property and the 4
th
curve shows maximum shear thinning property.
Figure 5.5 shows the variation between slide to roll ratio(S) and central film thickness
(h
c
/R) with respect to different values of viscosity ratio (
2
/
1
). At same value of
2
/
1,
Figure
shows that at relatively lower value of speed parameter the initial value of central film
thickness (h
c
/R) is comparatively lower. Central film thickness decreases with increase in
slide to roll ratio. This is due to the fact that shear rate increases with increasing slide to roll
ratio which, in turn, leads to more pronounced decrease in lubricant viscosity on account of
shear-thinning behaviour. At the same value of dimensionless speed parameter, for lower
value of
2
/
1
the slope of decrement in value of central film thickness (h
c
/R) is sharp with S
but after a value curve shows a very small change. At same velocity parameter, lower
viscosity ratio shows higher shear thinning behaviour. Figure depicts that at higher speed
parameter, lubricant shows least shear thinning effect, central film thickness (h
c
/R) decrease
with S but at very slow rate or approximately variation plot is horizontal. At very low value
of dimensionless speed parameter (10
-12
), the change in h
c
/R with
2
/
1
is very negligible but
curve 2
nd
(
2
/
1
=0.4) slopes downwards slightly as proceeds with S compared to curve 6
th
(
2
/
1
=0.8).
Figure 5.6 shows the variation between slide to roll ratio(S) and traction coefficient
(COF).graphs are plotted for different combination of viscosity ratio (
2
/
1
) and
dimensionless speed parameter (u). At same speed parameter, Figure shows that at lower
value of viscosity ratio slope of increment in traction coefficient is comparatively lower.
Figure depicts traction coefficient increase with increase in value of S. At the same value of
viscosity ratio, for lower value of speed parameter traction coefficient is lower.
Figure 5.7 shows the variation of pressure with respect to X. The graphs are plotted
for different combinations of dimensionless load parameter (w) and viscosity ratio (
2
/
1
).
Pressure spikes shows approximately similar type of behaviour except for maximum value of
load (w=5.2*10
-4
). Graph is shown with four different values of combinations of viscosity
ratio (
2
/
1
) and dimensionless load parameter. As we approach towards positive values 3
rd
plot (
2
/
1
=0.4, w=5.2*10
-4
) shows largest pressure but abrupt spike is nearly absent in this.
4
th
curve (
2
/
1
=0.8, w=2*10
-5
) shows highest pressure spike relative to other three.
Figure 5.9 compares the central film thickness (h
c
/R) with respect to dimensionless
load parameter (w). The value of slide to roll ratio is taken as unity. The value of power law
index n=0.5, value of critical shear stress G
cr
=310
4
Pa, value of dimensionless load
parameter is taken as U=110
-11
. Logarithmic graph shows that h
c
/R decrease linearly to load
parameter w. The graphs are plotted for different values of viscosity ratio (
2
/
1
). h
c
/R
decrease with increase in value of W. When value of viscosity ratio is more than zero than the
slope of variation is comparatively more. For value
2
/
1
=0.2,0.4,0.8 the slope is nearly
equivalent but initial value of h
c
/R is more at higher value of
2
/
1
. The graphs produce the
similar result to figure (5.1), the lower viscosity ratio shows higher shear thinning effect.
Figure 5.10 shows the variation between slide to roll ratio(S) and central film
thickness (h
c
/R) with respect to variation of viscosity ratio (
2
/
1
) and dimension less load
parameter (w). At same value of viscosity ratio (
2
/
1
), Figure shows that at low value of load
parameter the initial value of central film thickness (h
c
/R) is comparatively more Central film
thickness decreases with increase in slide to roll ratio. This is due to the fact that shear rate
increases with increasing slide to roll ratio which, in turn, leads to more pronounced decrease
in lubricant viscosity on account of shear-thinning behaviour. At the same value of
dimensionless load parameter, for low value of
2
/
1
the decrement in value of central film
thickness (h
c
/R) is very sharp with S asymptotically but after a value curve shows a very
small change. Figure depicts that at higher load parameter central film thickness (h
c
/R)
decrease with S but at very slow rate or approximately variation plot is horizontal. Figure
shows that highest load parameter with lowest value of viscosity ratio gives highest shear
thinning behaviour. There is also an interesting fact that at medium load and highest viscosity
ratio (w=1.3*10
-4
,
2
/
1
=0.8) shear thinning effect reduces at higher value of S as compared
with low load and low viscosity ratio (w=2*10
-4
,
2
/
1
=0.2).
Figure 5.11 shows the variation between slide to roll ratio(S) and traction coefficient
(COF). Graphs are plotted for different combination of viscosity ratio (
2
/
1
) and
dimensionless load parameter (w). At same load parameter, Figure shows that at lower value
of viscosity ratio rate of increment in traction coefficient is comparatively lower. At the same
value of viscosity ratio, for lower value of speed parameter traction coefficient is lower.
Effect of G
cr
:-
Figure (5.12) compares the variations of generalized Newtonian viscosity () with
shear-rate obtained using actual double-Newtonian model. Graphs are plotted with four
different values of G
cr
. From the figure it can be shown that for critical stress (Gcr) equals to
3*10
6
shows least shear thinning property (curve 4
th
) and curve 1st with Gcr equals to 3*10
3
shows maximum shear thinning effect.
Figure 5.13 compares the central film thickness (h
c
/R) with respect to dimensionless
speed parameter (U). The value of slide to roll ratio is taken as unity. The value of power law
index n=0.5, value of critical shear stress
2
/
1
=0.5, value of dimensionless load parameter is
taken as W=1.310
-4
. Logarithmic graph shows that h
c
/R increase linearly to speed parameter
U. h
c
/R increase with increase in value of U. The graphs are plotted for different values of
viscosity ratio (G
cR
). Figure provides the idea that for lower value of G
cr
the central film
thickness is lower. The figure shows at lowest value of G
cr
the graph is most sear thinning
and at highest value of G
cr
is least shear thinning.
Figure 5.14 shows the variation between slide to roll ratio (S) and central film
thickness (h
c
/R) with different combination of critical stress limit (G
cr
) and dimensionless
speed parameter (u). Figure shows that at low value of speed parameter the initial value of
central film thickness (h
c
/R) is comparatively less. Figure depicts variation of film thickness
asymptotically with S. At the same value of dimensionless speed parameter, for low value of
G
cr
the decrement in value of central film thickness (h
c
/R) is very sharp with S but after a
value curve shows a very small change. Figure depicts that at higher speed parameter central
film thickness (h
c
/R) decrease with S but at very slow rate or approximately variation plot is
horizontal. Figure provides the conclusion that velocity parameter dominates in inducing
shear thinning effect than G
cr
. This is shown from the fact that at lowest value of speed
parameter with higher G
cr
(u=10
-12
, G
cr
=3*10
4
) is more shear thinning than (u=10
-10
,
G
cr
=3*10
3
). Figure is also consistent with previous result that at same velocity parameter,
lower the value of G
cr
higher be the shear thinning behaviour in the lubricant.
Figure 5.16 compares the central film thickness (h
c
/R) with respect to dimensionless
load parameter (w). The value of slide to roll ratio is taken as unity. The value of power law
index n=0.5, value of critical shear stress
2
/
1
=0.5, value of dimensionless load parameter is
taken as U=110
-11
. Logarithmic graph shows that h
c
/R decrease linearly to load parameter w.
The graphs are plotted for different values of critical shear stress (G
cr
). h
c
/R decrease with
increase in value of W linearly. For a larger value of G
cr
the initial value of h
c
/R is
comparatively larger and slope of all curves are approximately same. Graph is consistent with
figure (12 &13) and depicts that at lower value of G
cr
, lubricant shows higher shear thinning
behaviour.
Figure 5.17 shows the variation between slide to roll ratio (S) and central film
thickness (h
c
/R) with different combination of critical stress limit (G
cr
) and dimensionless
load parameter (w). Central film thickness decreases with increase in slide to roll ratio. This
is due to the fact that shear rate increases with increasing slide to roll ratio which, in turn,
leads to more pronounced decrease in lubricant viscosity on account of shear-thinning
behaviour. Figure shows that at larger value of load parameter the initial value of central film
thickness (h
c
/R) is comparatively less, means higher shear thinning behaviour. At the same
value of dimensionless load parameter, for low value of G
cr
the decrement in value of central
film thickness (h
c
/R) is very sharp with S but after a value curve shows a very small change.
Figure shows that the results are in much consonance with the previous result and state that
lubricant show higher shear thinning behaviour at condition of higher load parameter and
lower value of G
cr
. One more interesting point is that for combination of higher load
parameter and higher G
cr
(curve 6
th
), the shear thinning effect drops at higher value of S
compared to lower load parameter and lower value of G
cr
(curve 2
nd
). So at higher value of S
the effect of G
cr
dominates than the load parameter.
Figure 5.18 shows the variation between slide to roll ratio(S) and traction coefficient
(COF). Graphs are plotted for different combination of critical shear stress (G
cr
) and
dimensionless load parameter (w). At same load parameter, Figure shows that at lower value
of critical shear stress slope of increase in traction coefficient is comparatively more.
Although the value is approximately matches at lower value of S (for s=0.1). At the same
value of critical shear stress (G
cr
), for lower value of load parameter slope of increment in
traction coefficient is lower.
Effect of Power law index ( n )
Figure 5.19 compares the variation of viscosity with respect to strain rate. Graph
shows that value of viscosity is very same at low value of . Graph is plotted for different
values of power law index (n). At higher value of power law index the rate of increase in
viscosity with respect to strain and at lower power law index the viscosity decrease
significantly with respect to strain rate. Figure shows that at n=0.3, the lubricant shows
highest shear thinning behaviour.
Figure 5.20 compares the central film thickness (h
c
/R) with respect to dimensionless
speed parameter (U). The value of slide to roll ratio is taken as unity. Value of critical shear
stress G
cr
=310
4
Pa, value of dimensionless load parameter is taken as W=1.310
-4
, values of
viscosity ratio (
2
/
1
) is taken as 0.5. Logarithmic graph are plotted for different values of
power law index (n). Logarithmic graph shows that h
c
/R varies linearly to speed parameter U.
h
c
/R increase with increase in value of U. Figure is consistent with figure (19) and provides
that at low power law index (n) the central film thickness is minimum. So the lubricant shows
higher shear thinning behaviour with a double Newtonian model of lower value of power law
index.
Figure 5.21 shows the variation between slide to roll ratio(S) and central film
thickness (h
c
/R). Graphs are plotted for different combination of power law index (n) and
dimensionless speed parameter (u). Value of viscosity ratio (
2
/
1
) is take as 0.5, critical
shear stress G
cr
is taken as 3*10
4
, value of dimensionless load parameter is taken as 1.3*10
4
.
At same value of power law index (n), Figure shows that at low value of speed parameter the
initial value of central film thickness (h
c
/R) is comparatively more. Figure depicts variation of
film thickness asymptotically with S. At the same value of dimensionless speed parameter,
for lower value of power law index (n) the slope of decrement in value of central film
thickness (h
c
/R) is lower with S but after a value curve shows a very small change. At much
lower speed parameter (10
-12
) variation in central film thickness (h
c
/R) with different values of
power law index (n) is nearly absent and curve is almost horizontal. Figure shows that
between combination of power law index and velocity, velocity is more predominant factor.
Curve 1
st
shows very low value of central film thickness. At same value of velocity, the figure
is consonance with figure (19) that at low value of power law index shear thinning effect is
more.
Figure 5.23 compares the central film thickness (h
c
/R) with respect to dimensionless
load parameter (w). The value of slide to roll ratio is taken as unity. The value of viscosity
ratio is taken
2
/
1
=0.5, value of critical shear stress G
cr
=310
4
Pa, value of dimensionless
load parameter is taken as U=110
-11
. Logarithmic graph shows that h
c
/R decrease linearly to
dimension less load parameter w. The graphs are plotted for different values of power law
index (n). h
c
/R decrease with increase in value of W. When value of power law index (n) is
more than initial value of h
c
/R is comparatively more. Graph is consistent with previous result
that at low value of power law index model shows high shear thinning effect.
Figure 5.24 shows the variation between slide to roll ratio(S) and central film
thickness (h
c
/R). Graph is plotted for different combination of power law index (n) and
dimensionless load parameter (w). Central film thickness decreases with increase in slide to
roll ratio. This is due to the fact that shear rate increases with increasing slide to roll ratio
which, in turn, leads to more pronounced decrease in lubricant viscosity on account of shear-
thinning behaviour. At the same value of dimensionless load parameter, for lower value of n
the slope of decrement in value of central film thickness (h
c
/R) is sharp with S at starting
values and after that slope start to decrease. At same dimensionless load parameter, at lower
value of power law index model shows high shear thinning behaviour. Figure shows that
effect of load parameter dominates rather than power law index. For example at high load
parameter and comparatively high power law index (curve 4
th
) shows high shear thinning
characteristics than at low load and lower power law index (curve 1
st
). Figure shows an
interesting fact that at higher value of slide to roll ratio effect of power law index enhance
than load parameter. For example at high load and high power law index (curve 6
th
) has less
shear thinning effect at high slide to roll ratio than at low load and low power law index
(curve 3
rd
).
Figure 5.25 shows the variation between slide to roll ratio(S) and traction coefficient
(COF). Graphs are plotted for different combination of power law index (n) and
dimensionless load parameter (w). At same load parameter, Figure shows that at lower value
of power law index slope of increment in traction coefficient is comparatively more.
Although the value approximately matches at lower value of S (for s=0.1).

Anda mungkin juga menyukai