Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
1.0 INTRODUCTION
Fuel Injection plays a very important role in a Diesel Engine. The timing and pressure of
injection influences the performance parameters of the engine. Injection pressure was previously
attained by having a pump for every cylinder of the engine. The drawback of this type of
injection mechanism is that the flow of fuel into the cylinder couldn’t be monitored and varied
during the operation of the engine. To overcome this problem and to promote the cleaner burning
of diesel, the concept of Common Rail Direct Injection was developed around 1950. The concept
was way ahead of its time and couldn’t be physically realised.
Due to the advances in the Electronic industry, the complicated concept was realised with the
help of sensors, programmable circuitry etc. Till the recent past Common Rail technology was
highly advantageous in medium sized vehicles, i.e. vehicles with 4 or more cylinders. As the
emission norms are becoming more stringent, the Common Rail technology is being introduced
in small sized vehicles.
Common rail direct fuel injection is a modern variant of direct fuel injection system for petrol
and diesel engines.
On diesel engines, it features a high-pressure (over 1,000 bar/15,000 psi) fuel rail feeding
individual solenoid valves, as opposed to low-pressure fuel pump feeding unit injectors (Pumpe
Düse or pump nozzles). Third-generation common rail diesels now feature piezoelectric injectors
for increased precision, with fuel pressures up to 1,800 bar/26,000 psi.
1.0.1 History:
The common rail system prototype was developed in the late 1960s by Robert Huber of
Switzerland and the technology further developed by Dr. Marco Ganser at the Swiss Federal
Institute of Technology in Zurich later of Ganser-Hydromag AG (est.1995) in Oberägeri.
The first successful usage in production vehicle began in Japan by the mid-1990s. Dr. Shohei
Itoh and Masahiko Miyaki of the Denso Corporation, a Japanese automotive parts manufacturer,
developed the common rail fuel system for heavy duty vehicles and turned it into practical use on
their ECD-U2 common-rail system mounted on the Hino Rising Ranger truck and sold for
general use in 1995. Denso claims the first commercial high pressure common rail system in
1995
Modern common rail systems, whilst working on the same principle, are governed by an engine
control unit (ECU) which opens each injector electronically rather than mechanically. This was
extensively prototyped in the 1990s with collaboration between Magneti Marelli, Centro
Ricerche Fiat and Elasis. After research and development by the Fiat Group, the design was
acquired by the German company Robert Bosch GmbH for completion of development and
refinement for mass-production. In hindsight the sale appeared to be a tactical error for Fiat as
the new technology proved to be highly profitable. The company had little choice but to sell,
however, as it was in a poor financial state at the time and lacked the resources to complete
development on its own. In 1997 they extended its use for passenger cars. The first passenger car
that used the common rail system was the 1997 model Alfa Romeo 156 1.9 JTD, and later on
that same year Mercedes-Benz C 220 CDI.
Common rail engines have been used in marine and locomotive applications for some time. The
Cooper-Bessemer GN-8 (circa 1942) is an example of a hydraulically operated common rail
diesel engine, also known as a modified common rail.
Vickers used common rail systems in submarine engines circa 1916. Doxford Engines Ltd.
(opposed piston heavy marine engines) used a common rail system (from 1921 to 1980) whereby
a multi-cylinder reciprocating fuel pump generated a pressure of approximately 600bar with the
fuel being stored in accumulator bottles. Pressure control was achieved by means of an
adjustable pump discharge stroke and a "spill valve". Camshaft operated mechanical timing
valves were used to supply the spring loaded Brice/CAV/Lucas injectors which injected through
the side of the cylinder into the chamber formed between the pistons. Early engines had a pair of
timing cams, one for ahead running and one for astern. Later engines had two injectors per
cylinder and the final series of constant pressure turbocharged engines were fitted with four
injectors per cylinder. This system was used for the injection of both diesel oil and heavy fuel oil
(600cSt heated to a temperature of approximately 130°C).
1.0.2 Principles:
Solenoid or piezoelectric valves make possible fine electronic control over the fuel injection time
and quantity, and the higher pressure that the common rail technology makes available provides
better fuel atomisation. In order to lower engine noise the engine's electronic control unit can
inject a small amount of diesel just before the main injection event ("pilot" injection), thus
reducing its explosiveness and vibration, as well as optimising injection timing and quantity for
variations in fuel quality, cold starting, and so on. Some advanced common rail fuel systems
perform as many as five injections per stroke.
Common rail engines require no heating up time and produce lower engine noise and emissions
than older systems.
Diesel engines have historically used various forms of fuel injection. Two common types include
the unit injection system and the distributor/inline pump systems. While these older systems
provided accurate fuel quantity and injection timing control they were limited by several factors:
They were cam driven and injection pressure was proportional to engine speed. This
typically meant that the highest injection pressure could only be achieved at the highest
engine speed and the maximum achievable injection pressure decreased as engine speed
decreased. This relationship is true with all pumps, even those used on common rail
systems; with the unit or distributor systems, however, the injection pressure is tied to the
instantaneous pressure of a single pumping event with no accumulator and thus the
relationship is more prominent and troublesome.
They were limited on the number of and timing of injection events that could be
commanded during a single combustion event. While multiple injection events are
possible with these older systems, it is much more difficult and costly to achieve.
For the typical distributor/inline system the start of injection occurred at a pre-determined
pressure (often referred to as: pop pressure) and ended at a pre-determined pressure. This
characteristic results from "dummy" injectors in the cylinder head which opened and
closed at pressures determined by the spring preload applied to the plunger in the
injector. Once the pressure in the injector reached a pre-determined level, the plunger
would lift and injection would start.
In common rail systems a high pressure pump stores a reservoir of fuel at high pressure up to and
above 2,000 bars (29,000 psi). The term "common rail" refers to the fact that all of the fuel
injectors are supplied by a common fuel rail which is nothing more than a pressure accumulator
where the fuel is stored at high pressure. This accumulator supplies multiple fuel injectors with
high pressure fuel. This simplifies the purpose of the high pressure pump in that it only has to
maintain a commanded pressure at a target (either mechanically or electronically controlled).
The fuel injectors are typically ECU-controlled. When the fuel injectors are electrically activated
In medium and large duty engines the fuel is being pumped using Rotary Piston pumps. These
pumps if miniaturized would result in higher costs which would render the implementation of
Common Rail Technology in Small sized vehicles as economically unfeasible. The concept of
single reciprocating pumps was developed keeping the small sized as the target.
The reciprocating pump is driven by a cam which is mounted on the cam shaft of the engine. The
mechanism used is the cam and roller mechanism. The reciprocating plunger is connected to the
roller through a tappet which is enclosed by a guide. The load on the roller is very high which
leads to contact loading in the elements. The rolling of the elements effects a variable load on the
component, which leads to fatigue in the elements.
Delphi Unit Pump Diesel Common Rail (UPCR) System is an innovative engine management
concept that leverages advanced common rail technology a proven "green" strategy for very
small diesel engine programs. The system offers manufacturers a cost-effective, robust solution
to help them achieve optimal fuel efficiency and meet stringent emissions standards, such as
Euro 4.
The Delphi Unit Pump Diesel Common Rail System is specifically designed for 1-, 2-, and 3-
cylinder engine applications. It is ideal for small engine vehicles destined for emerging markets
and for entry-level vehicles that will be marketed in developed regions. It is also well-suited for
other nonautomotive diesel engine products such as small agricultural and industrial equipment.
Key features of the Delphi Diesel UPCR System include fast solenoid diesel injectors and a
common rail, a program-tailored engine control module (ECM), robust unit fuel pump with an
inlet metering valve, as well as an efficient, low cost fuel filter.
1. Pump assembly
2. Tappet assembly
4. Nozzle assembly
5. Rail
Benefits:
Low cost engine management system designed to provide the necessary high value
essential to the market success of low-end products and economy segment vehicle
programs.
Light weight, compact package size enables relative ease of application in very small
engine programs.
High system pressure (up to 1,600 bar) achieves high efficiency and excellent engine
performance.
Proven, compact fast response solenoid diesel fuel injectors are capable of up to five
injections per cycle. The multi-injection capability enables precise tuning of combustion
to help meet emissions, fuel economy and noise targets.
Simple, robust unit pump provides reliable performance and contributes to an overall
economical system cost.
The UPCR System is able to drive devices such as an exhaust gas recirculation (EGR)
valve and intake throttle control. Air management capability helps meet emissions
standards, such as Euro 4.
Long fatigue life is one of the most important concerns while designing cylindrical rolling
elements. The stress induced in the bearing components mainly influences the life of the element.
It therefore becomes necessary to exactly determine stresses developed in rolling elements and
raceways. Roller bearings with rollers having flat contour surfaces suffer from edge
concentration. It was suggested as a remedy to use a contour profile also known as crown.
At Delphi-TVS the development of UPCR was hindered by frequent failure of the cam and roller
follower. Initially when the roller was indigenously developed the interface was failing
aggressively with the cam profile being completely worn out. The roller was outsourced to a
leading bearing manufacturer and prototypes were used in the pump. The life of the rolling
Manipal Institute of Technology 7
EVALUATION OF ROLLER PROFILE FOR SPECIFIC REQUIREMENTS
element has been drastically improved but the elements still fail before 1e7 revolutions. The
standards specify that the elements must maintain its integrity up to 1000 hours of operation.
The recent failures in the rolling elements have been in the form of pitting on the rollers and on
the lobe of the cam. The pitting pattern when observed has been spread over the roller and its
distribution has been uneven around the circumference of the roller. The distribution of pitting
on the roller can be described as progressively increasing and then diminishing back to the initial
point. The failure in the cam is observed over the lobe of the cam, where the load is the highest
and the radius of curvature is the least. Hence the failure can be ascertained to be load induced.
This project aims to evaluate the stresses, for the present dimensions of the cam-roller assembly.
Then various profiles are to be evaluated both theoretically and analytically keeping the
dimensions of the elements constant. This would help in narrowing down to the best profile
under the present scenario.
Redesigning the cam roller assembly, if necessary, to bring the dimensions well within the
strengths of the material. Testing of the prototypes will be undertaken to confirm the compliance
of the design to real time situation
There are 4 basic types of roller profiles that are used in rolling elements. These profiles have
their own benefits and drawbacks. This project would compare all the profiles and their
variations to shortlist the best profile for this application.
The first step would be to calculate the stresses theoretically in the present combination of the
cam-roller assembly and then evaluate the same, using an analysis software. The next step would
involve calculating the stresses for various profiles under the same load conditions.
The material properties such as Tensile strength, Compressive strength, Fatigue Strength are to
be evaluated along with other associated properties. A comparison of stresses in the elements and
strength of the material would help in narrowing down to the cause of failure.
Once the failure causing stress has been identified, the cam-roller assembly will be redesigned to
get the stresses well within the limit. Then the elements will be undertaken for prototyping and
real time experimentation.
The various types of profiles that are used in a roller bearing are:
1. Flat Profile
2. Spherical Profile
3. Logarithmic Profile
Flat Profile:
This is the simplest type of profile. This profile is preferred when the length of the roller and the
corresponding element is the same. The flat profile of the roller results in an edge stress
concentration at roller ends. It is shown in the figure, where the contact stress is very high at
edges. In flat-profiled cylindrical bearings as the load increases the contact stress increases
linearly. For the present case the maximum contact stresses obtained at the corner of the roller at
the inner raceway. The horn shape shows the edge stress concentration. As the load increases the
edge stress also increases. Contact stresses at the middle part of the roller are constant and
relatively low. (P.V.V.N. Prasad, et al 2004)
Spherical Profile:
To prevent high stresses at edges, the roller surface incorporated with a circular crowning. This
eliminates the edge stress concentration at the low and moderate loads. With increase in the
crowning radius, the maximum contact stress decreases up to a certain limit. On further increase
in the crowning radius, the contact stress increases with increase in the crowning. The maximum
contact stress is at the middle of the roller surface. With the increase in the load, the contact
length also increases. At high loads the edge concentration takes place. This profile is good for
low and moderate loads. (P.V.V.N. Prasad, et al 2004)
Logarithmic Profile:
A logarithmic profile is generated using a mathematical logarithmic function. Under all loading
conditions, the logarithmic profile (LP) uses more contact length of the roller than either the fully
crowned or partially crowned roller profiles. Edge loading tends to be avoided in LP especially at
heavy loads. (P.V.V.N. Prasad, et al 2004)
Fig 2.3a: A cylindrical roller with logarithmic profile Fig 2.3b: Logarithmic profile zoomed in h(x) direction
There are various types of logarithmic profiles, which are differentiated by the formula used to
generate the profile. The first formulation for the profile was generated by Lundberg, according
to which the deviation from the straight profile is given as
4 p2 h l
h( x) 2 ln 2
E hl g
1 e (2 x / l )
Where
4p
g 0.5 ln
lE
4 p2 4p
hl 0.5 ln
E
2
lE
h
Where is the multiplier of a Lundberg’s profile deviation, h is the deviation of the roller
hl
generator from the straight profile, hl is the maximum value of the deviation of the profile
according to Lundberg, ε is the corrective exponent. p is the radial load on the roller, Σρ is the
curvature sum, E is the Young’s modulus of elasticity, l is the length of the roller and is the
distance from the centre of the roller.
The roller profile, which is generated for higher load than the applied load, gives the lower
contact stress than the profile generated at the applied (operating) load. After certain limit the
contact stress increases even though the profile generated at higher than the applied load. As
multiplying factor increases the contact length of the roller decreases and the contact stresses
increases with increase in the correction factor. At lower loads than the profile designed load the
contact length is small. As the load increases the contact length increased. For loads above the
designed load, the contact length increases but the contact stress is not uniform and slight edge
stress concentration occurs. The LP of roller results no edge stress concentration at the low,
medium and high loads and contact stresses are distributed uniformly along the length of the
roller.
Lundberg’s Profile is very difficult to manufacture. (Fujiwara, et al, 2007) Hence Johan’s Gohar
profile which is derived from Lundberg’s profile is preferred when it comes to ease of
manufacturing.
2Q 1
z ( x) ln
lE ' 1 (1 .3033 b / a)( 2 x / l ) 2
Where
E
E’- Effective Young’s Modulus, E '
1 2
E- Young’s Modulus
υ- Poisson’s Ratio
Q- Load
However, when this equation is applied, carrying out the calculation using the technique to be
described later results in edge stress occasionally. This trend becomes more remarkable under the
presence of misalignment. Providing a cylindrical roller or conical roller with a straight section
may be desirable for machining or functional reasons; however, using this equation does not
allow a straight section to be set up.
To solve this problem, three design parameters, K1, Km, and zm, are introduced into equation.
2 K1Q 1
z ( x) ln
lE ' zmlE ' 2
2 K Q y a
1 1 e 1 1
K a
K1- Multiple of Q 2
K2- Ratio of crowning length to a
This profile is a partially crowned profile where the crowning is logarithmic in nature. Other type
of partially crowned profile can have spherical or tapered profiles.
In the flat profile of a cylindrical roller as the applied load increases the maximum stress also
increases linearly. Whereas in the circular and logarithmic profiles, the variation of the contact
stresses with the load is not linear. As the radius of the crown of roller surface increases the
maximum stress decreases for the same applied load. In the trend is same in logarithmic profile
also i.e. the maximum stress decreases with the increases design load for the profile up to a
particular limit, afterwards it increases. In the circular crowned profile as the load increases the
uniformity of the stress along the length of the roller vanishes rapidly. In logarithmic profiled
roller as the load increases the length of the contact increases. In the flat profile as the load
increases the edge loading increases.
The logarithmic profile is the remedy for the edge loading at high-applied loads. The circular
crown also eliminated the edge loading but the contact length remains very less. Logarithmic
profile generated at the correction factor of 1 and multiplying factor of 3 gives uniform
distribution of contact stresses along the length of the roller. For increase in the multiplying
factor the contact length increases and with increase in the correction factor the contact stress
increases. Even in logarithmic profile bearings for very heavy loads the small edge stress
induces. It remains to be seen how LP behaves for combined loadings in the bearings i.e. with
non-uniform distribution of the applied load on bearing rings.
The maximum Hertz stress-life exponents were determined for the individual roller profiles and
the resultant individual lives were compared. The following results were obtained (Brian L.
Vlcek, et al, 2000):
1. With the closed form solution and not considering edge or stress concentrations, the flat roller
profile has the longest predicted life followed by the end-tapered profile, the aerospace profile
and the crowned profile, respectively. The full crowned profile produces the lowest lives. While
there are life differences between the end tapered profile and the aerospace profile, these
differences may not be significant. For the FEA solution which considered stress concentrations
the end tapered profile produced the highest lives but not significantly different from that of the
aerospace profile followed by the crowned profile and the flat roller profile, respectively.
2. The effect of edge loading on the flat roller profile is to reduce life at the higher load by as
much as 98 and 82 percent at the lower load. The actual percentage calculated depends on the
analysis used.
3. The resultant predicted life at each stress condition not only depends on the life equation used
but also on the Weibull slope assumed. The least variation in predicted life with Weibull slope
comes with the Zaretsky equation. At all conditions calculated for a Weibull slope of 1.11, the
ANSI/ABMA/ISO standard result in the lowest lives. Except for the Weibull slope of 1.11 at
which the Weibull equation predicts the highest lives, the highest lives are predicted by the
Zaretsky equation. For Weibull slopes of 1.5 and 2, both the Lundberg-Palmgren and Ioannides-
Harris (where τu equal 0) equations predict lower lives than the ANSI/ ABMA/ISO standard.
4. Based upon the Hertz stresses for line contact, the load-life exponent p of 10/3, results in a
maximum Hertz stress-life exponent n equal to 6.6. This value is inconsistent that experienced in
the field. Lundberg and Palmgren’s justification for a p of 10/3 was that a roller bearing can
experience “mixed contact,” that is, one raceway can experience “line contact” and the other
raceway “point contact.” This is certainly not consistent with the vast majority of cylindrical
roller and tapered roller bearings designed and used today.
Coating on surfaces:
There is a wide range of coating techniques and careful selection of the appropriate coating
material and method is a pre-requisite for an effective coating. Prior to selecting the coating
material and method the first question to be asked is whether wear or friction is of greater
concern. If the prime objective is to reduce friction then a solid lubricant coating should be
selected and the coating method will, in most cases, be either sputtering or a combination of
painting and baking (Gwidon W. Stachowiak).
To suppress wear by the application of coatings, it is first necessary to determine the mechanism
of wear occurring, e.g. whether abrasive wear or some other form of wear is present. Although
most coatings can suppress several forms of wear, each type of coating is most effective at
preventing a few specific wear mechanisms. Therefore during the selection process of the most
effective coating to suppress wear in a particular situation, i.e. coating optimization, the
prevailing wear mechanism must first be recognized and assessed. The basic characteristics of
the coatings which can be achieved by the methods described in the previous section in terms of
wear control are summarized in the figure.
It can be seen from figure that while the optimization of a coating to resist abrasive wear is
relatively simple, i.e. it is sufficient to produce a thick hard surface layer with toughness high
enough to prevent coating fracture, other wear mechanisms require much greater care in coating
optimization.
Studies of wear resistant coatings reveal that hard coatings are most effective in suppressing
abrasive wear. An example of this finding is illustrated in figure which shows the wear rate of a
pump rotor as a function of the hardness of the coating applied to the surface. It can be seen from
figure that the abrasive wear rate declines to a negligible value once a PVD coating of titanium
nitride, which is characterized by extremely high hardness, is employed.
In this example abrasive wear was caused by very fine contaminants present in the pumped fluid
and the size of the abrasives was sufficiently small for a thin PVD coating to be effective. In
other applications where the abrasive particles are much larger, thicker coatings are more
appropriate.
It was also found that thin films of ceramics such as titanium nitride are quite effective in
suppressing adhesive wear in poorly lubricated and high stress contacts. For example, when a
cutting tool is coated with titanium nitride, adhesion and seizure between the tool and metal chip
does not occur even when cutting is performed in a vacuum. Titanium nitride coatings were also
applied to gears and the scuffing tests on coated and uncoated gears revealed that the critical load
and scuffing resistance for coated gears is much higher.
This coating also reduces the coefficient of friction in unlubricated sliding as well as wear rates,
e.g. coefficients of friction close to 0.1 between titanium nitride and zirconium nitride coatings
on hardened bearing steel have been observed. Unfortunately titanium nitride coatings do not
provide corrosion resistance. Since zirconium and hafnium belong to the same IVB group of the
periodic table of chemical elements as titanium, some similarity in wear properties of their
compounds can be expected. In fact, hafnium nitride was found to give the best wear resistance
performance in tests on cutting tools. Zirconium nitride is also extremely useful as a coating. It
should also be mentioned that for hard coatings to be effective, adequate substrate hardness is
essential. Therefore hardened steels and materials such as stellite are generally used as a
substrate for this type of coating.
Fretting wear can be mitigated by the use of hard coatings, e.g. carbides, especially at small
amplitudes of fretting movement. However, at higher fretting amplitudes, spalling of the carbide
coatings renders them ineffective.
Coatings produced by ion implantation, in certain applications, can also provide large reductions
in wear. Since the coatings produced by this technique are very thin they are only effective in
reducing wear at low load levels as illustrated.
Fig 2.7: Effect of nitrogen ion implantation on wear rates of stainless steel in unlubricated sliding
The performance of non-metallic coatings such as tungsten carbide used for rolling elements is
related to the operating conditions. For example, it was found that 100-200 [μm] thick plasma-
sprayed coatings on steel and ceramic balls fail by surface wear when lubrication is poor or by
sub-surface delamination when lubrication is effective.
3.0 Methodology
The load on the roller is the force required to pressurize the fuel to 1600 bar. During pumping, in
a reciprocating pump, the forces that need to be overcome are the frictional force, spring force,
and the fluid force.
F=PxA
bd 2
A=
4
= 4.419x10-5
F =160x106 x 4.419x10-5
=7068.59 N
7068.59
F=
.85
= 8316 N
=6+5= 11mm
Spring Force,
Fs = K x d
= 11 x 29.85
= 328.35 N
Acceleration in m/s2,
36 N 2
acc = .01164 x
1000
= 6704.64 m/s2
Inertia Forces,
FI = m x acc
= 6704.64 x 0.1635
= 1096 N
L = F + Fs + FI
= 9.74 kN
The Load on the roller is transferred indirectly and is in the form of a distributed load on the pin
over the length of contact with the tappet. The loading is offset and cannot be used directly in the
equations stated in the references. The first step would be to convert the offset load into a point
load on to the centre of the roller. A 2-dimensional image is shown in figure3.1, and the steps
involved in the conversion are shown.
Roller
Load Load
PIN
BUSH
Fig 3.1: Schematic diagram of load transfer
The load distribution over the pin is shown in figure3.2. The load from the tappet is in the form
of a UDL of 1132.56 N/mm and the length of contact between the pin and the tappet is 4.3mm.
The pin is supported by the bush whose length is 13.35mm. But during loading, pin acts as an
overhanging beam and due to the deformation of the pin the support from the bush is converted
into a 2 point support at the ends of the bush.
The UDL is converted into a point load, at the centre of distribution. The Reaction R2 is the
reaction force on from the bush and is the main point where the load is transferred to the bush
and then to the roller. R1 is the Reaction that is generated, when the pin is constrained at its
midpoint. The reaction R1 is the load that would act at the centre of the roller under the present
loading conditions.
Fig 3.3: Free body diagram over half the length of the pin
On solving the force distribution diagram
4870 R1 R2
Hence,
R1 2316.44N
L 2 R1
L 4632.88N
F
0.
564
con
1/
R
l
1
/R
1 2
Where
1 1
1 2
2 2
E1 E2
1 2
.29 1 2
.3
202e9 205 e
9
167 .822 10 6
=
l
σx = σz = - σcon
σy = - 2υ x σcon
100 .693 10 6
=
l
The length of the roller is calculated for the Maximum normal stress and is
2
167.822 10 6
l
2000 10
6
l = 7.04 mm
The present roller at the length of 14.35 mm suffices the requirement of minimum roller length
of 7.04mm. This is just under the consideration of normal stresses and not any other form of
stresses.
σx = σz = - 1395.925 MPa
σy = 837.55 MPa
= 424.36 Mpa
The Principal Stresses in the material for pure rolling is the normal X,Y,Z axis stresses, due to
the absence of applied shear.
The roller is then evaluated under the Von Mises Stress Criteria
σmax =
2
x
2
y
2
z xy yz zx
= 558.375 MPa.
1 1 1
B 171.48
2 R1 R2
m1 m2
2 m1 m2 F
a
B l
2.616 10 5
a
l
a 0.0001037m
2 F
con
al
Since the elements undergo complete contact with each other, the co-efficient of friction is
μ=0.33.
fmaxcon
= 465.261 MPa
x
0.3
a
2
x
1
con
n
x
2
a
= -1331.63 MPa
x
x 2fmax
t
a
= -279.156 MPa
2
z
n
1
con
x
2
a
= -1331.63MPa
z 0
t
xz 0
n
x2
xz fmax 1
t
a2
= -443.83 Mpa
x x x 1775.46MPa
n t
z z z 1331.63MPa
n t
xz 443.83MPa
y ( x z ) 932.127MPa
In a flat profile roller, end stresses are generated as the cam width is larger than the roller width.
The end stresses are quite larger than the maximum contact stresses. End stresses can be reduced
by changing the profile of the roller surface. A crowned surface reduces the end stress effect to
an extent. The limiting factor the profile is the load. As the load on the roller, contact length
increases. When the contact length equals the length of the roller, end stress is caused in the
roller.
In spherical crowning the profile is crowned at a particular radius, to negate the effect of edge
stresses. This type of evaluation is an iterative process as there is no particular formulation to
provide the best possible crowning radius for the element dimensions. For a crowned profile,
initially the dimensions are generated for a flat profile and then keeping the dimensions constant
the profile is varied to investigate on the effect of variation in profile parameters.
The contact region with a spherical profile is in the form of an ellipse, unlike a rectangle in a flat
profile. The contact dimensions are vastly influenced by the radius of curvature of the elements.
The geometric co-efficient such as A and B are used to find an parameter ϕ which will help in
generating the contact patch dimensions.
3 F
con
2 ab
1 1 1
A '` '
1 1
2 R1 R1 R2 R2
2 2
1 1 1 1 1 1 1 1 1
B ' ' 2 ' ' cos 2
2 R1 R1 R2 R2 R1 R1 R2 R2
B
cos 1
A
3 F ( m1 m2 )
a ka 3
4A
3F ( m1 m2 )
b kb 3
4A
x 2 (1 2 )
b
con
a b
a
y 2 (1 2 ) con
a b
z con
b
k3
a
1 2 2
k4 a b
a
These applied stresses are also the Principal Stresses that occur at the point of contact.
At the end of the major axis of the contact ellipse the shear stress at the surface is
(
1
2
xz
k
)
1
tanh
k1
3
2
1
4 con
kk 4 4
At the ends of the minor axis of the contact ellipse the shear stress at the surface is
k3 k3 1 k 4
xz (1 2 ) 1 tan con
k 42 k4 k3
These stresses are static stresses and are principal stresses. During the calculation of the stresses
the effect of friction has not been taken into consideration. During analysis of dynamic stresses
in the crowned profile the stress calculation tends to be more complex and is best done using any
analytic software. Hence the theoretical stress evaluation is limited to static stresses.
Φ 0 10 20 30 35 40 45 50 55 60 65 70 75 80 85 90
Ka ∞ 6.612 3.778 2.731 2.397 2.136 1.926 1.757 1.611 1.486 1.378 1.284 1.202 1.128 1.061 1
Kb 0 0.319 0.408 0.493 0.53 0.567 0.604 0.641 0.678 0.717 0.759 0.802 0.846 0.893 0.944 1
As the radius of crowning is increased, the contact width and the contact length for a
particular load increases. The length of contact is limited to the length of the roller and the
width to the width of contact for a flat profile. The roller length is limited to the present
length l = .01435m, as the any change in length would require changes in the manufacturing
process parameters.
Hence for a given maximum length, the crowning radius at which minimum contact stresses
occurs is to be evaluated. It has been observed that the contact pressure reduces with the
increase in crowning radius up to a limiting crown radius. After the limiting crown radius, the
contact stress increases with the increase in crown radius.
Theoretical evaluation of spherical profile under static load has generated stresses to be above
the design limit of 2000 MPa. This calls in for evaluation of other possible profile such as the
logarithmic profile. The stress calculation for a logarithmic profile cannot be done
theoretically; hence analytic software is used to evaluate the stresses generated during
loading.
For the purpose of this project, two types of profiles have been evaluated and their stress
results have been compared with the profile used in the failed roller. The profiles have been
generated for a design load of 9740N. The two profiles are John’s-Gohar profile and
modified John’s-Gohar profile. These profiles have been selected for their ease of
manufacturing.
2Q 1
z ( x) ln
lE ' 1 (1 .3033 b / a)( 2 x / l ) 2
The parameters a, b, l are the results of flat profile calculation.
The parameters taken for design in the modified John’s-Gohar profile is zm=0.025, K1=1,
K2=0.696.
Calculation of the deviation is for one side from the midpoint of the axis and the other is the
mirror image.
For modified John’s-Gohar profile, deviation of the profile from the flat profile starts 5mm
away from the midpoint of the axis, on either side.
Modelling: All the four components i.e. pin, bush, roller and cam were modelled as separate
volumes. Pin, bush and roller was partially modelled (2 quadrants) where as the cam was
modelled as a cylinder whose outer radius is the lobe radius and only one quadrant is
modelled.
Meshing: The model was meshed with two different materials. The material properties for
the pin, bush and roller was the same, where as the second material was applied to the cam.
The element type was 10-noded tetrahedron with three degrees of freedom. Due to the
complexity and time consumption, element size was restricted to .7 mm.
Boundary Conditions: Symmetric boundary conditions were placed on the surface where
ever required. The cam was constrained at the inner diameter in all directions.
Contact Elements: The various volumes were stand alone and there was no surface or node
relation between them. Contact elements were introduced at the following interfaces:
1. Pin-Bush interface
2. Bush-Roller interface
3. Roller-Cam interface
The first volume corresponds to the contact element whereas the second volume corresponds
to the target element. The target and contact elements were surface to surface.
Since there are three interfaces of contact elements, use of friction is compulsory, which
makes evaluation of static loading difficult. Contact algorithm was set to penalty method due
to its simplicity. Due to the restriction in mesh size, the elements were placed apart by a
particular distance. Hence changes were made with initial adjustment. Changes can be
established with initial penetration or initial contact closure, though the latter is more
appropriate. Sometimes, even though initial penetration has been specified, the contact used
to report open status. Here closing the contact would generate the result. If either of the two is
not employed or the status of contact was open then the software would generate an error of
rigid body motion. Element time increment was maintained to be reasonable.
Loading:
Initially, the load was applied as surface pressure over the nodes. Pressure was applied on
either sides of the pin over the length of contact with the tappet. On analysis it was found that
the stresses generated were much less than the theoretically calculated stresses.
The load application was changed from uniform pressure to uniformly distributed loading.
The load distribution was calculated as follows:
Q
UDL
Nn
Q- total load
On application of UDL on the nodes, the stresses generated were comparable to the
theoretical stresses. The only drawback in this type of load application is there will be stress
singularity around the point of load, due to the intensity of loading in that particular region.
Analysis Controls:
Analysis is of static type without any variation of loading properties. Since the analysis is
complex and is iterative in nature loading steps need to be specified. The controls varied in
the analysis are listed below
Number of substeps: 10
1
ELEMENTS
JUN 19 2010
U
16:20:41
F
1000
1002
1001
Z X
1003
Fig 3.4: Isometric view of the meshed component with Boundary Conditions
1
ELEMENTS
JUN 19 2010
U
16:21:37
F
1002
1001
1000
Y
X Z
1003
Fig 3.5: Side View of the meshed component with Boundary Conditions
3.1.2 Results:
1
NODAL SOLUTION
DEC 3 2009
STEP=1 09:52:17
SUB =7
TIME=50
SY (AVG)
RSYS=0
DMX =.025445
SMN =-1692
SMX =168.58
MN
Fig 3.6: Stress distribution Flat profile with the presence of edge loading
1
NODAL SOLUTION
DEC 3 2009
STEP=1 10:12:12
SUB =7
TIME=50
SY (AVG)
RSYS=0
DMX =.030104
SMN =-1751
SMX =189.364
Y
MN Z
MX X
1
NODAL SOLUTION
DEC 3 2009
STEP=1 10:48:07
SUB =7 1000
1002
TIME=50
SY (AVG)
RSYS=0
DMX =.028684
SMN =-1959
SMX =128.77
Y
MN MX Z X
1
NODAL SOLUTION
DEC 3 2009
STEP=1 10:26:53
SUB =7
TIME=50
SY (AVG)
RSYS=0
DMX =.026458
SMN =-1888
SMX =204.852
Y
MX MN Z X
ANSYS results proved the previously discussed problem of Edge Loading in Flat profile.
Comparing the stresses generated, Flat Profile with edge loading and Spherical profile
developed the same amount of maximum stress. Compressive Stresses in John’s-Gohar
Profile is higher than the all the profiles. Compressive Stresses in all the profiles is under the
design limit of 2000MPa.
The Shear stress in John’s-Gohar profile is the least and is comparatively lesser than flat and
spherical profile. Both Flat and Spherical profiles have almost the same shear stress generated
through them. The stress limit for shear stress is not defined. Hence, material tests are
required to be carried out, to confirm the values of shear stress limit.
All the details on material properties like Young’s Modulus, Tensile strength, Ultimate
strength, Compressive strength, Endurance limit is either unknown or undocumented. Hence
to provide a comprehensive study on the best possible profile and also suitable dimensional
changes to the Cam-Roller interface it is important to conduct the required material tests.
Four specimens were submitted to the materials lab at Delphi-TVS for conducting material
test. Test report for two specimens was provided by them which are provided in fig 4.2a and
4.2b. Tensile test is just a basis for starting fatigue testing of the material. Using the tensile
strength report, the range for loading the fatigue specimen can be narrowed down to.
Compressive tests were not carried out for the material, as there was a documented report
available with the Design Dept. The document was from a leading bearing manufacturer
which stated that the compressive stresses in the roller should not exceed 2000 MPa.
The most important cause of failure in contact fatigue is the shear stress which causes the
material to shear off from the surface of the material. Shear endurance limit cannot be
directly generated using any existing tests. Hence, tensile endurance limit is evaluated from
R.R. Moore’s rotating bending test. From the Tensile Endurance strength, Shear endurance
strength can be derived using suitable formula. Total 16 specimens were prepared. For the
load of 9 kg 4 specimens were tested according to the standards.
5.0 REDESIGN
The endurance tests revealed that the shear stress generated in the contact elements was
greater than the endurance shear strength. Considering all the profiles analysed, even John’s-
Gohar profile generated shear stress larger than the strength. This can be one of the main
causes for failure. Since, varying the type of profile didn’t solve the cause of reducing the
stresses; hence a change in dimension is required to get the stresses.
Since the compressive stresses are within the limit, any increase in dimension is intended to
reduce the shear stresses. The roller reciprocates at a high speed; hence varying the
dimensions would cause higher vibrations to be generated due to the larger reciprocating
mass. Hence it is important to know that the limiting factor for increasing dimension will be
vibration.
Volume is directly proportional to the length and to the square of radius. Hence increasing the
radius would increase the volume rapidly than what is caused by increasing the length of the
roller. When comparing the variation in contact stresses, stresses reduce at a faster rate with
the increase in length against the same increase in radius.
The other choice for dimension change will be the increase in the minimum radius of the
cam. The radius of the cam is the least at the lobe of the cam. Hence to attain a higher lobe
radius the mass of the cam has to be increased. In other words the minimum cam radius needs
to be increased.
To start off with the redesign, the three variables are roller length, roller radius and the cam
lobe radius. Here it is preferred to keep the roller radius constant due to the problems
associated with reciprocating vibrations.
Having just one equation and two unknowns it is quite confusing to play with dimensions.
Hence an excel sheet was developed with all the influencing parameters, where in any change
in a dimension would generate the subsequent results. This table is made flexible to
incorporate material changes. Due to the usage of this table, the stress reduction due to
variation in any particular dimension or properties, hence giving a clear interpretation to the
solution. As discussed in the previous chapters we initially start the design for a flat profile,
the dimensions are further incorporated to other profiles. The various parts of the table are
listed below.
Material Properties
Poissons ratio for Roller 0.30
Poissons ratio for Cam 0.29
Youngs Modulus of Roller 2.05E+05
Youngs Modulus of Cam 2.03E+05
Material Constant for Roller 4.44E-06
Material Constant for Cam 4.51E-06
Load Conditions
Load contact length 4.60
half length 8.50
distance of load
mid- point from 3.30
bush end
R1 6570.25
R2 1837.44
Stresses
axes X Z Y Shear
normal -1013.78 -1013.78 0.00
transverse -210.42 0.00 -334.55
total -1224.20 -1013.78 -671.39 -334.55
Dimensions in mm
Stress and Pressure in M Pa
Mass in kg
Calculated quantities
Variables
Final Results
Previously the crowning radius having the least contact pressure was wrongly predicted due
to the usage of linear interpolation. During analysis, the best crowning radius was narrowed
down and results were tabulated.
From the previous analysis, the best crowning radius was evaluated to be 1900 mm.
1 1 1 1 1
A ' '
2 R1 R1 R2 R2
0.1630
2 2
1 1 1 1 1 1 1 1 1
B ' ' 2 ' '
2 R1 R1 R2 R2 R1 R1 R2 R2
.1619
B
Cos
A
0.9935
Using the above mentioned value of Cosϕ we generate an equation with three unknowns i.e.
R1 , R1' and R 2 . The equation generated will be having an unknown R1' . The other two values
were determined previously while designing a flat profile.
where
C (Cos 2 1)
1 1
R
R1 R2
Solving the above equation for the values R1 10 , R2 18 , the value of R1' is obtained as
1984mm. The roller profile was crowned spherically for the radius specified.
In the previous analysis it was noted that John’s Gohar profile, generated the least amount of
shear stress within the roller, even though the contact pressure was higher than the flat profile
and spherical profile. Hence it was decided to analyse the contact stresses for the present
roller dimensions for the logarithmic profile.
X 4.5kN Y(x) 5kN Y(x) 7.9kN Y(x) 12kN Y(X) 15kN Y(x)
0.00 0 0 0 0 0
0.90 6.87E-06 7.63E-06 1.21E-05 1.83E-05 2.29E-05
2.25 4.41E-05 4.90E-05 7.74E-05 1.18E-04 1.47E-04
3.60 1.19E-04 1.32E-04 2.09E-04 3.17E-04 3.97E-04
5.40 3.04E-04 3.38E-04 5.34E-04 8.11E-04 1.01E-03
6.75 5.62E-04 6.25E-04 9.87E-04 1.50E-03 1.87E-03
9.00 3.24E-03 3.60E-03 5.69E-03 8.64E-03 1.08E-02
The table listed above displays the deviation from a flat profile for various loads. This study
is an extra addition to the process, to know the behaviour of the profiles under the same
loading.
1
NODAL SOLUTION
MAY 21 2010
STEP=1
15:59:47
SUB =7
TIME=50
SYZ (AVG)
RSYS=0
DMX =.022115
SMN =-264.235
SMX =112.208
MX
Y
X Z MN
Fig. 5.1: Shear stress distribution for roller with spherical profile with radius 4950mm
1
NODAL SOLUTION
MAY 24 2010
STEP=1
10:15:10
SUB =7
TIME=50
SYZ (AVG)
RSYS=0 1002
1001
DMX =.020625
SMN =-333.132
SMX =116.346
MX
Y
X Z MN
Fig. 5.2: Shear stress distribution over the length of the roller for logarithmic profile designed for 4.5kN
1
NODAL SOLUTION
MAY 24 2010
STEP=1
10:14:06
SUB =7
TIME=50
SYZ (AVG)
RSYS=0 1002
1001
DMX =.020657
SMN =-328.339
SMX =116.787
MX
Y
X Z MN
Fig. 5.3: Shear stress distribution over the length of the roller for logarithmic profile designed for 5kN
1
NODAL SOLUTION
MAY 24 2010
STEP=1
10:11:36
SUB =7
TIME=50
SYZ (AVG)
RSYS=0 1002
1001
DMX =.020876
SMN =-300.751
SMX =118.802
MX
Y
X Z
MN
Fig. 5.4: Shear stress distribution over the length of the roller for logarithmic profile designed for 7.9kN
1
NODAL SOLUTION
MAY 31 2010
STEP=1
06:16:43
SUB =7
TIME=50
SYZ (AVG)
RSYS=0 1002
1001
DMX =.021192
SMN =-277.864
SMX =111.986
MX
Y
X Z MN
Fig. 5.5: Shear stress distribution over the length of the roller for logarithmic profile designed for 12kN
1
NODAL SOLUTION
JUN 29 2010
STEP=1
16:53:49
SUB =7
TIME=50
SYZ (AVG)
RSYS=0 1002
1001
DMX =.020603
SMN =-268.617
SMX =114.672
MX
Y
X Z MN
Fig. 5.6: Shear stress distribution over the length of the roller for logarithmic profile designed for 15kN
On initial observation it is found that the shear stress in the roller with spherical profile is the
least. But it is higher than the shear endurance limit. It is important to remember that the
tested material is EN31 where as the material used in the roller is 100Cr6. Hence a
competitors profile was tested, and comparison was made. It was noted that the stresses
generated in the spherical profile is much lesser than the competitor’s profile.
The material used in the roller is 100Cr6 where as the material used for testing is EN31. This
is so because the 100Cr6 is not readily available in India. To validate the endurance limit and
make the design more realistic, the first step would be to test cleaner material 100Cr6.
For any fatigue test the surface finish plays a very important role. Due to the unavailability of
centreless grinding facility with the service providers around the establishment, the best
surface finish that was attained was using finishing emery. It is recommended to finish the
test specimen after heat treatment process using a centreless grinding machine.
The rollers nowadays are coated with a harder material to delay the advent of surface cracks.
The fatigue specimen is to be tested with and without surface coating to compare the benefits
of coating in the roller, when it comes to failures with surface cracks.
Whenever it comes to reciprocating components, the mass of the components is kept bare
minimum. As this design advocates increase in dimensions, it throws the components into a
region of higher vibration. It is therefore recommended to conduct vibration analysis and also
to optimize the mass of the elements. Mass away from the surface can be reduced, as the high
intensity stresses occur at the surface and not within the material.
This project fell short in validation of the obtained results due to the paucity of time and
unavailability of the test rigs. Thus, the most important recommendation would be to validate
the redesigned components to provide a better understanding of the results.
The behaviour of the profiles under loading is just as studied from various references. With
the flat profile generating edge stresses and hence going above the limits, it is not preferred
for the present applications. Spherical profile if properly designed, will give the desired
results with stresses lesser than the flat profile.
It has been noted in circular and logarithmic profiles, the variation of the contact stresses with
the load is not linear. As the radius of the crown of roller surface increases the maximum
stress decreases for the same applied load. In the trend is same in logarithmic profile also i.e.
the maximum stress decreases with the increases design load for the profile up to a particular
limit, afterwards it increases. In the circular crowned profile as the load increases the
uniformity of the stress along the length of the roller vanishes rapidly. In logarithmic profiled
roller as the load increases the length of the contact increases. If we keep the logarithmic
design load at 3 times the load acting at the centre of the roller, then logarithmic profile will
suffice any requirement.
On comparing spherical and logarithmic profile, it can be stated that the stress increase, after
the optimal profile, begins at an earlier stage in a spherical profile than in logarithmic profile.
There are no existing formulae to confirm the optimal crowning radius for a spherical profile.
Hence logarithmic profile can be preferred over spherical and flat profile.
Even though at the dimensions recommended shear stress is not under the tested endurance
limit, the design is still safe. It is not wrong to advocate the design as the stresses are much
lesser than the stresses in the competitor’s product which has sustained the required number
of testing hours, with the roller having a flat profile. This is due to the difference in the
material used for testing and the material used in the component. The roller material is a
cleaner material with lesser non metallic impurities.
Hence at the end, it can be recommended to design rollers with logarithmic profile with a
load factor of 2.75 to 3, which is the commonly used profile by the leading bearing
manufacturer.
REFERENCES
14. www.wikipedia.org.